diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4492 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.99968,
+  "eval_steps": 300,
+  "global_step": 1562,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1847133757961784e-09,
+      "logits/generated": 0.17403794825077057,
+      "logits/real": -0.19873479008674622,
+      "logps/generated": -514.6786499023438,
+      "logps/real": -454.4227294921875,
+      "loss": 0.15,
+      "rewards/accuracies": 0.0,
+      "rewards/generated": 0.0,
+      "rewards/margins": 0.0,
+      "rewards/real": 0.0,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.592356687898089e-08,
+      "logits/generated": 0.1870264708995819,
+      "logits/real": -0.25900763273239136,
+      "logps/generated": -526.2941284179688,
+      "logps/real": -490.1958923339844,
+      "loss": 0.1507,
+      "rewards/accuracies": 0.3984375,
+      "rewards/generated": -0.0001907353289425373,
+      "rewards/margins": 0.0010350469965487719,
+      "rewards/real": 0.0008443119004368782,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.184713375796178e-08,
+      "logits/generated": 0.1984216719865799,
+      "logits/real": -0.26348429918289185,
+      "logps/generated": -505.8414611816406,
+      "logps/real": -491.9839782714844,
+      "loss": 0.151,
+      "rewards/accuracies": 0.48750001192092896,
+      "rewards/generated": 0.006021722219884396,
+      "rewards/margins": -0.0053602345287799835,
+      "rewards/real": 0.000661488447804004,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.777070063694268e-08,
+      "logits/generated": 0.17304742336273193,
+      "logits/real": -0.21200346946716309,
+      "logps/generated": -538.6023559570312,
+      "logps/real": -442.2162170410156,
+      "loss": 0.1509,
+      "rewards/accuracies": 0.49687498807907104,
+      "rewards/generated": 0.0030143880285322666,
+      "rewards/margins": -0.0018971532117575407,
+      "rewards/real": 0.0011172344675287604,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.369426751592356e-08,
+      "logits/generated": 0.1914023607969284,
+      "logits/real": -0.22652260959148407,
+      "logps/generated": -523.87451171875,
+      "logps/real": -461.0634765625,
+      "loss": 0.1504,
+      "rewards/accuracies": 0.5375000238418579,
+      "rewards/generated": 0.0033149768132716417,
+      "rewards/margins": 0.004283360205590725,
+      "rewards/real": 0.007598336786031723,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.961783439490445e-08,
+      "logits/generated": 0.18474087119102478,
+      "logits/real": -0.2507132589817047,
+      "logps/generated": -501.36993408203125,
+      "logps/real": -479.93634033203125,
+      "loss": 0.151,
+      "rewards/accuracies": 0.4906249940395355,
+      "rewards/generated": 0.010729875415563583,
+      "rewards/margins": -0.005274372640997171,
+      "rewards/real": 0.005455502774566412,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.554140127388536e-08,
+      "logits/generated": 0.20316779613494873,
+      "logits/real": -0.21709361672401428,
+      "logps/generated": -534.5445556640625,
+      "logps/real": -470.92254638671875,
+      "loss": 0.1506,
+      "rewards/accuracies": 0.484375,
+      "rewards/generated": 0.005789478309452534,
+      "rewards/margins": -0.00014353431470226496,
+      "rewards/real": 0.0056459433399140835,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1146496815286624e-07,
+      "logits/generated": 0.1715896874666214,
+      "logits/real": -0.24265193939208984,
+      "logps/generated": -517.3651123046875,
+      "logps/real": -451.755615234375,
+      "loss": 0.1503,
+      "rewards/accuracies": 0.515625,
+      "rewards/generated": -0.0016872085398063064,
+      "rewards/margins": 0.008004234172403812,
+      "rewards/real": 0.006317025516182184,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2738853503184713e-07,
+      "logits/generated": 0.15530803799629211,
+      "logits/real": -0.23413880169391632,
+      "logps/generated": -523.96826171875,
+      "logps/real": -444.64239501953125,
+      "loss": 0.1502,
+      "rewards/accuracies": 0.559374988079071,
+      "rewards/generated": 0.0005639827577397227,
+      "rewards/margins": 0.007568719331175089,
+      "rewards/real": 0.008132701739668846,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.43312101910828e-07,
+      "logits/generated": 0.1671060025691986,
+      "logits/real": -0.22391800582408905,
+      "logps/generated": -550.3038330078125,
+      "logps/real": -439.77947998046875,
+      "loss": 0.1505,
+      "rewards/accuracies": 0.53125,
+      "rewards/generated": 0.004643476102501154,
+      "rewards/margins": 0.001223881496116519,
+      "rewards/real": 0.005867358297109604,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.592356687898089e-07,
+      "logits/generated": 0.17757461965084076,
+      "logits/real": -0.24453218281269073,
+      "logps/generated": -519.4007568359375,
+      "logps/real": -458.79669189453125,
+      "loss": 0.1506,
+      "rewards/accuracies": 0.49687498807907104,
+      "rewards/generated": -0.00017246263450942934,
+      "rewards/margins": 0.0037554509472101927,
+      "rewards/real": 0.0035829886328428984,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7515923566878978e-07,
+      "logits/generated": 0.18612250685691833,
+      "logits/real": -0.21440061926841736,
+      "logps/generated": -539.6683959960938,
+      "logps/real": -446.758056640625,
+      "loss": 0.1503,
+      "rewards/accuracies": 0.512499988079071,
+      "rewards/generated": 0.002787713659927249,
+      "rewards/margins": 0.008473692461848259,
+      "rewards/real": 0.011261406354606152,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9108280254777072e-07,
+      "logits/generated": 0.169862300157547,
+      "logits/real": -0.22989757359027863,
+      "logps/generated": -498.18560791015625,
+      "logps/real": -442.7648010253906,
+      "loss": 0.1501,
+      "rewards/accuracies": 0.565625011920929,
+      "rewards/generated": -0.012974532321095467,
+      "rewards/margins": 0.017906008288264275,
+      "rewards/real": 0.004931476432830095,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.070063694267516e-07,
+      "logits/generated": 0.19755002856254578,
+      "logits/real": -0.23432913422584534,
+      "logps/generated": -492.85595703125,
+      "logps/real": -494.6519470214844,
+      "loss": 0.1498,
+      "rewards/accuracies": 0.5718749761581421,
+      "rewards/generated": -0.0050980569794774055,
+      "rewards/margins": 0.018167994916439056,
+      "rewards/real": 0.01306993793696165,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.2292993630573247e-07,
+      "logits/generated": 0.16374804079532623,
+      "logits/real": -0.25221407413482666,
+      "logps/generated": -521.1048583984375,
+      "logps/real": -445.358642578125,
+      "loss": 0.1503,
+      "rewards/accuracies": 0.5562499761581421,
+      "rewards/generated": -0.0037274628411978483,
+      "rewards/margins": 0.011061501689255238,
+      "rewards/real": 0.007334038615226746,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.388535031847134e-07,
+      "logits/generated": 0.18414834141731262,
+      "logits/real": -0.23798434436321259,
+      "logps/generated": -539.92138671875,
+      "logps/real": -464.05462646484375,
+      "loss": 0.1502,
+      "rewards/accuracies": 0.5406249761581421,
+      "rewards/generated": -0.008644538931548595,
+      "rewards/margins": 0.01328897662460804,
+      "rewards/real": 0.00464443676173687,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.5477707006369425e-07,
+      "logits/generated": 0.18885524570941925,
+      "logits/real": -0.2001829594373703,
+      "logps/generated": -549.0367431640625,
+      "logps/real": -452.0292053222656,
+      "loss": 0.1493,
+      "rewards/accuracies": 0.609375,
+      "rewards/generated": -0.005654294043779373,
+      "rewards/margins": 0.023647544905543327,
+      "rewards/real": 0.017993250861763954,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.7070063694267513e-07,
+      "logits/generated": 0.16570258140563965,
+      "logits/real": -0.22671516239643097,
+      "logps/generated": -527.3927001953125,
+      "logps/real": -434.76971435546875,
+      "loss": 0.1495,
+      "rewards/accuracies": 0.6156250238418579,
+      "rewards/generated": -0.009726697579026222,
+      "rewards/margins": 0.026334956288337708,
+      "rewards/real": 0.016608258709311485,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.86624203821656e-07,
+      "logits/generated": 0.16409741342067719,
+      "logits/real": -0.2301245927810669,
+      "logps/generated": -514.0899658203125,
+      "logps/real": -469.2464904785156,
+      "loss": 0.1489,
+      "rewards/accuracies": 0.6343749761581421,
+      "rewards/generated": -0.008600350469350815,
+      "rewards/margins": 0.033345777541399,
+      "rewards/real": 0.024745427072048187,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.0254777070063694e-07,
+      "logits/generated": 0.18098126351833344,
+      "logits/real": -0.19588425755500793,
+      "logps/generated": -511.7598571777344,
+      "logps/real": -406.40911865234375,
+      "loss": 0.1488,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/generated": -0.009649967774748802,
+      "rewards/margins": 0.03349680081009865,
+      "rewards/real": 0.023846831172704697,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.184713375796178e-07,
+      "logits/generated": 0.17149221897125244,
+      "logits/real": -0.24450810253620148,
+      "logps/generated": -530.5223388671875,
+      "logps/real": -459.1062927246094,
+      "loss": 0.1487,
+      "rewards/accuracies": 0.659375011920929,
+      "rewards/generated": -0.014194811694324017,
+      "rewards/margins": 0.03900580853223801,
+      "rewards/real": 0.024810992181301117,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.343949044585987e-07,
+      "logits/generated": 0.19177064299583435,
+      "logits/real": -0.1881466656923294,
+      "logps/generated": -520.2600708007812,
+      "logps/real": -438.0816345214844,
+      "loss": 0.1482,
+      "rewards/accuracies": 0.671875,
+      "rewards/generated": -0.01024587918072939,
+      "rewards/margins": 0.04250962659716606,
+      "rewards/real": 0.032263752073049545,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5031847133757957e-07,
+      "logits/generated": 0.16849389672279358,
+      "logits/real": -0.2514097988605499,
+      "logps/generated": -541.0388793945312,
+      "logps/real": -480.9339904785156,
+      "loss": 0.1482,
+      "rewards/accuracies": 0.668749988079071,
+      "rewards/generated": -0.008990215137600899,
+      "rewards/margins": 0.041446976363658905,
+      "rewards/real": 0.03245675936341286,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.6624203821656045e-07,
+      "logits/generated": 0.19658248126506805,
+      "logits/real": -0.2052055150270462,
+      "logps/generated": -525.2179565429688,
+      "logps/real": -448.5086364746094,
+      "loss": 0.1479,
+      "rewards/accuracies": 0.6812499761581421,
+      "rewards/generated": -0.014174744486808777,
+      "rewards/margins": 0.050183337181806564,
+      "rewards/real": 0.03600858896970749,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.8216560509554143e-07,
+      "logits/generated": 0.1979866474866867,
+      "logits/real": -0.24727025628089905,
+      "logps/generated": -540.0928955078125,
+      "logps/real": -480.67047119140625,
+      "loss": 0.1477,
+      "rewards/accuracies": 0.6968749761581421,
+      "rewards/generated": -0.016895266249775887,
+      "rewards/margins": 0.0549207404255867,
+      "rewards/real": 0.03802547603845596,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.980891719745223e-07,
+      "logits/generated": 0.21044392883777618,
+      "logits/real": -0.2199120819568634,
+      "logps/generated": -508.4944763183594,
+      "logps/real": -454.1514587402344,
+      "loss": 0.1476,
+      "rewards/accuracies": 0.715624988079071,
+      "rewards/generated": -0.013310949318110943,
+      "rewards/margins": 0.055444687604904175,
+      "rewards/real": 0.042133744806051254,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.140127388535032e-07,
+      "logits/generated": 0.19091640412807465,
+      "logits/real": -0.23178955912590027,
+      "logps/generated": -527.3304443359375,
+      "logps/real": -436.0814514160156,
+      "loss": 0.1475,
+      "rewards/accuracies": 0.659375011920929,
+      "rewards/generated": -0.013830204494297504,
+      "rewards/margins": 0.05727874115109444,
+      "rewards/real": 0.04344853386282921,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.2993630573248406e-07,
+      "logits/generated": 0.17117801308631897,
+      "logits/real": -0.2260015904903412,
+      "logps/generated": -530.6170654296875,
+      "logps/real": -450.876220703125,
+      "loss": 0.1466,
+      "rewards/accuracies": 0.75,
+      "rewards/generated": -0.02331192046403885,
+      "rewards/margins": 0.07541900128126144,
+      "rewards/real": 0.052107084542512894,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.4585987261146494e-07,
+      "logits/generated": 0.19855618476867676,
+      "logits/real": -0.23370346426963806,
+      "logps/generated": -541.4176025390625,
+      "logps/real": -485.81463623046875,
+      "loss": 0.146,
+      "rewards/accuracies": 0.7406250238418579,
+      "rewards/generated": -0.01844293251633644,
+      "rewards/margins": 0.08036132156848907,
+      "rewards/real": 0.061918385326862335,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6178343949044587e-07,
+      "logits/generated": 0.16755443811416626,
+      "logits/real": -0.24783170223236084,
+      "logps/generated": -502.64849853515625,
+      "logps/real": -458.10296630859375,
+      "loss": 0.1461,
+      "rewards/accuracies": 0.7593749761581421,
+      "rewards/generated": -0.004401802085340023,
+      "rewards/margins": 0.06796350330114365,
+      "rewards/real": 0.06356170028448105,
+      "step": 145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.777070063694267e-07,
+      "logits/generated": 0.17052355408668518,
+      "logits/real": -0.26604199409484863,
+      "logps/generated": -535.1319580078125,
+      "logps/real": -480.93682861328125,
+      "loss": 0.1451,
+      "rewards/accuracies": 0.7749999761581421,
+      "rewards/generated": -0.021760020405054092,
+      "rewards/margins": 0.09478209912776947,
+      "rewards/real": 0.07302206754684448,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.936305732484076e-07,
+      "logits/generated": 0.18140792846679688,
+      "logits/real": -0.24586549401283264,
+      "logps/generated": -495.56951904296875,
+      "logps/real": -445.80426025390625,
+      "loss": 0.1446,
+      "rewards/accuracies": 0.815625011920929,
+      "rewards/generated": -0.010536590591073036,
+      "rewards/margins": 0.09246022999286652,
+      "rewards/real": 0.08192362636327744,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.98932384341637e-07,
+      "logits/generated": 0.18911007046699524,
+      "logits/real": -0.2553741931915283,
+      "logps/generated": -516.3770751953125,
+      "logps/real": -471.3077087402344,
+      "loss": 0.1446,
+      "rewards/accuracies": 0.7875000238418579,
+      "rewards/generated": -0.018946032971143723,
+      "rewards/margins": 0.10126407444477081,
+      "rewards/real": 0.0823180302977562,
+      "step": 160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.97153024911032e-07,
+      "logits/generated": 0.19667308032512665,
+      "logits/real": -0.15881827473640442,
+      "logps/generated": -534.4208374023438,
+      "logps/real": -442.3114318847656,
+      "loss": 0.1441,
+      "rewards/accuracies": 0.8218749761581421,
+      "rewards/generated": -0.01733066886663437,
+      "rewards/margins": 0.10644586384296417,
+      "rewards/real": 0.0891151949763298,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.953736654804271e-07,
+      "logits/generated": 0.17530401051044464,
+      "logits/real": -0.19269545376300812,
+      "logps/generated": -517.9406127929688,
+      "logps/real": -430.65789794921875,
+      "loss": 0.1435,
+      "rewards/accuracies": 0.8125,
+      "rewards/generated": -0.014586791396141052,
+      "rewards/margins": 0.11395077407360077,
+      "rewards/real": 0.09936399012804031,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.935943060498221e-07,
+      "logits/generated": 0.20905160903930664,
+      "logits/real": -0.19150669872760773,
+      "logps/generated": -523.1032104492188,
+      "logps/real": -445.8037109375,
+      "loss": 0.1424,
+      "rewards/accuracies": 0.8656250238418579,
+      "rewards/generated": -0.021403489634394646,
+      "rewards/margins": 0.13283729553222656,
+      "rewards/real": 0.11143380403518677,
+      "step": 175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.918149466192171e-07,
+      "logits/generated": 0.20821765065193176,
+      "logits/real": -0.18395385146141052,
+      "logps/generated": -523.7911987304688,
+      "logps/real": -444.244384765625,
+      "loss": 0.1427,
+      "rewards/accuracies": 0.8187500238418579,
+      "rewards/generated": -0.011090211570262909,
+      "rewards/margins": 0.12333385646343231,
+      "rewards/real": 0.1122436374425888,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900355871886121e-07,
+      "logits/generated": 0.2085740864276886,
+      "logits/real": -0.18884018063545227,
+      "logps/generated": -514.5075073242188,
+      "logps/real": -455.51153564453125,
+      "loss": 0.1418,
+      "rewards/accuracies": 0.8218749761581421,
+      "rewards/generated": -0.008907720446586609,
+      "rewards/margins": 0.13150587677955627,
+      "rewards/real": 0.12259815633296967,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.882562277580071e-07,
+      "logits/generated": 0.19080451130867004,
+      "logits/real": -0.18134233355522156,
+      "logps/generated": -535.5404052734375,
+      "logps/real": -420.12054443359375,
+      "loss": 0.142,
+      "rewards/accuracies": 0.809374988079071,
+      "rewards/generated": -0.01622314192354679,
+      "rewards/margins": 0.13654197752475739,
+      "rewards/real": 0.12031883001327515,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.864768683274021e-07,
+      "logits/generated": 0.19419755041599274,
+      "logits/real": -0.2099919319152832,
+      "logps/generated": -506.61297607421875,
+      "logps/real": -456.75201416015625,
+      "loss": 0.142,
+      "rewards/accuracies": 0.84375,
+      "rewards/generated": -0.005563541315495968,
+      "rewards/margins": 0.13059118390083313,
+      "rewards/real": 0.1250276416540146,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.846975088967971e-07,
+      "logits/generated": 0.2110450267791748,
+      "logits/real": -0.17780761420726776,
+      "logps/generated": -504.53253173828125,
+      "logps/real": -446.33013916015625,
+      "loss": 0.1415,
+      "rewards/accuracies": 0.84375,
+      "rewards/generated": -0.020040716975927353,
+      "rewards/margins": 0.14582987129688263,
+      "rewards/real": 0.12578915059566498,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.829181494661922e-07,
+      "logits/generated": 0.19535906612873077,
+      "logits/real": -0.23168042302131653,
+      "logps/generated": -529.9983520507812,
+      "logps/real": -478.4393615722656,
+      "loss": 0.1408,
+      "rewards/accuracies": 0.840624988079071,
+      "rewards/generated": -0.01576526090502739,
+      "rewards/margins": 0.15367695689201355,
+      "rewards/real": 0.13791170716285706,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.811387900355872e-07,
+      "logits/generated": 0.22139854729175568,
+      "logits/real": -0.18635587394237518,
+      "logps/generated": -528.14404296875,
+      "logps/real": -444.3465270996094,
+      "loss": 0.1397,
+      "rewards/accuracies": 0.878125011920929,
+      "rewards/generated": -0.016462773084640503,
+      "rewards/margins": 0.1693098396062851,
+      "rewards/real": 0.1528470814228058,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.793594306049822e-07,
+      "logits/generated": 0.16860654950141907,
+      "logits/real": -0.26878511905670166,
+      "logps/generated": -514.0562744140625,
+      "logps/real": -470.86279296875,
+      "loss": 0.14,
+      "rewards/accuracies": 0.862500011920929,
+      "rewards/generated": -0.03441483899950981,
+      "rewards/margins": 0.1793655902147293,
+      "rewards/real": 0.1449507623910904,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775800711743772e-07,
+      "logits/generated": 0.2018643319606781,
+      "logits/real": -0.22842028737068176,
+      "logps/generated": -509.4456481933594,
+      "logps/real": -471.5657653808594,
+      "loss": 0.1407,
+      "rewards/accuracies": 0.871874988079071,
+      "rewards/generated": -0.02802978828549385,
+      "rewards/margins": 0.16557763516902924,
+      "rewards/real": 0.1375478357076645,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.758007117437722e-07,
+      "logits/generated": 0.19222237169742584,
+      "logits/real": -0.17731007933616638,
+      "logps/generated": -525.8961181640625,
+      "logps/real": -429.556396484375,
+      "loss": 0.1391,
+      "rewards/accuracies": 0.8843749761581421,
+      "rewards/generated": -0.026082757860422134,
+      "rewards/margins": 0.18103452026844025,
+      "rewards/real": 0.15495173633098602,
+      "step": 225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7402135231316723e-07,
+      "logits/generated": 0.20608314871788025,
+      "logits/real": -0.196997731924057,
+      "logps/generated": -517.6672973632812,
+      "logps/real": -437.429931640625,
+      "loss": 0.1391,
+      "rewards/accuracies": 0.878125011920929,
+      "rewards/generated": -0.026546889916062355,
+      "rewards/margins": 0.18050511181354523,
+      "rewards/real": 0.15395823121070862,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.722419928825623e-07,
+      "logits/generated": 0.18294331431388855,
+      "logits/real": -0.19551487267017365,
+      "logps/generated": -539.04052734375,
+      "logps/real": -430.70318603515625,
+      "loss": 0.1372,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/generated": -0.03303760290145874,
+      "rewards/margins": 0.21606886386871338,
+      "rewards/real": 0.18303126096725464,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7046263345195726e-07,
+      "logits/generated": 0.18518558144569397,
+      "logits/real": -0.2292046993970871,
+      "logps/generated": -509.7601623535156,
+      "logps/real": -452.63836669921875,
+      "loss": 0.139,
+      "rewards/accuracies": 0.8687499761581421,
+      "rewards/generated": -0.01840498112142086,
+      "rewards/margins": 0.18604739010334015,
+      "rewards/real": 0.16764239966869354,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.686832740213523e-07,
+      "logits/generated": 0.19558075070381165,
+      "logits/real": -0.22530385851860046,
+      "logps/generated": -520.8900756835938,
+      "logps/real": -470.57257080078125,
+      "loss": 0.1386,
+      "rewards/accuracies": 0.8843749761581421,
+      "rewards/generated": -0.027829840779304504,
+      "rewards/margins": 0.19147595763206482,
+      "rewards/real": 0.1636461317539215,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.669039145907473e-07,
+      "logits/generated": 0.19283290207386017,
+      "logits/real": -0.19199484586715698,
+      "logps/generated": -538.2701416015625,
+      "logps/real": -440.3575134277344,
+      "loss": 0.1376,
+      "rewards/accuracies": 0.8968750238418579,
+      "rewards/generated": -0.03322785347700119,
+      "rewards/margins": 0.21201416850090027,
+      "rewards/real": 0.17878632247447968,
+      "step": 250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.651245551601423e-07,
+      "logits/generated": 0.21194609999656677,
+      "logits/real": -0.15735319256782532,
+      "logps/generated": -533.7866821289062,
+      "logps/real": -436.27215576171875,
+      "loss": 0.1373,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/generated": -0.030345138162374496,
+      "rewards/margins": 0.21164409816265106,
+      "rewards/real": 0.18129894137382507,
+      "step": 255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.633451957295373e-07,
+      "logits/generated": 0.1993105113506317,
+      "logits/real": -0.20154225826263428,
+      "logps/generated": -525.4606323242188,
+      "logps/real": -439.7225646972656,
+      "loss": 0.1375,
+      "rewards/accuracies": 0.862500011920929,
+      "rewards/generated": -0.019895639270544052,
+      "rewards/margins": 0.20292837917804718,
+      "rewards/real": 0.18303272128105164,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.615658362989324e-07,
+      "logits/generated": 0.17055116593837738,
+      "logits/real": -0.22003407776355743,
+      "logps/generated": -537.1515502929688,
+      "logps/real": -442.8724670410156,
+      "loss": 0.1371,
+      "rewards/accuracies": 0.90625,
+      "rewards/generated": -0.027774950489401817,
+      "rewards/margins": 0.21557831764221191,
+      "rewards/real": 0.18780337274074554,
+      "step": 265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.597864768683274e-07,
+      "logits/generated": 0.1934186965227127,
+      "logits/real": -0.16990774869918823,
+      "logps/generated": -537.012451171875,
+      "logps/real": -409.61419677734375,
+      "loss": 0.1366,
+      "rewards/accuracies": 0.903124988079071,
+      "rewards/generated": -0.03637081757187843,
+      "rewards/margins": 0.23026354610919952,
+      "rewards/real": 0.19389274716377258,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.580071174377224e-07,
+      "logits/generated": 0.1830703169107437,
+      "logits/real": -0.22130803763866425,
+      "logps/generated": -523.718994140625,
+      "logps/real": -452.82818603515625,
+      "loss": 0.1363,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.036309439688920975,
+      "rewards/margins": 0.230547234416008,
+      "rewards/real": 0.19423779845237732,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5622775800711743e-07,
+      "logits/generated": 0.20362897217273712,
+      "logits/real": -0.2156350165605545,
+      "logps/generated": -535.4857788085938,
+      "logps/real": -458.9664001464844,
+      "loss": 0.1364,
+      "rewards/accuracies": 0.909375011920929,
+      "rewards/generated": -0.032396264374256134,
+      "rewards/margins": 0.22477588057518005,
+      "rewards/real": 0.1923796534538269,
+      "step": 280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5444839857651244e-07,
+      "logits/generated": 0.18986502289772034,
+      "logits/real": -0.2057947814464569,
+      "logps/generated": -524.5361328125,
+      "logps/real": -449.27734375,
+      "loss": 0.135,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.039955057203769684,
+      "rewards/margins": 0.25269564986228943,
+      "rewards/real": 0.21274061501026154,
+      "step": 285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.526690391459074e-07,
+      "logits/generated": 0.21750202775001526,
+      "logits/real": -0.18711628019809723,
+      "logps/generated": -533.2612915039062,
+      "logps/real": -452.9281311035156,
+      "loss": 0.1353,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.02985314093530178,
+      "rewards/margins": 0.2417806088924408,
+      "rewards/real": 0.21192745864391327,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5088967971530247e-07,
+      "logits/generated": 0.1709345579147339,
+      "logits/real": -0.2246251404285431,
+      "logps/generated": -536.2772216796875,
+      "logps/real": -442.66571044921875,
+      "loss": 0.1353,
+      "rewards/accuracies": 0.9156249761581421,
+      "rewards/generated": -0.03336368873715401,
+      "rewards/margins": 0.24980218708515167,
+      "rewards/real": 0.21643848717212677,
+      "step": 295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.491103202846975e-07,
+      "logits/generated": 0.1961456835269928,
+      "logits/real": -0.21015064418315887,
+      "logps/generated": -516.0350341796875,
+      "logps/real": -440.63934326171875,
+      "loss": 0.1357,
+      "rewards/accuracies": 0.903124988079071,
+      "rewards/generated": -0.027470093220472336,
+      "rewards/margins": 0.237370565533638,
+      "rewards/real": 0.20990045368671417,
+      "step": 300
+    },
+    {
+      "epoch": 0.19,
+      "eval_logits/generated": 0.19990801811218262,
+      "eval_logits/real": -0.20589132606983185,
+      "eval_logps/generated": -521.6344604492188,
+      "eval_logps/real": -455.5482482910156,
+      "eval_loss": 0.13489334285259247,
+      "eval_rewards/accuracies": 0.9054999947547913,
+      "eval_rewards/generated": -0.03345545381307602,
+      "eval_rewards/margins": 0.25203606486320496,
+      "eval_rewards/real": 0.21858063340187073,
+      "eval_runtime": 445.7957,
+      "eval_samples_per_second": 4.486,
+      "eval_steps_per_second": 1.122,
+      "step": 300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.473309608540925e-07,
+      "logits/generated": 0.17669954895973206,
+      "logits/real": -0.22454991936683655,
+      "logps/generated": -525.3490600585938,
+      "logps/real": -443.7996520996094,
+      "loss": 0.1344,
+      "rewards/accuracies": 0.893750011920929,
+      "rewards/generated": -0.03830927610397339,
+      "rewards/margins": 0.2601231336593628,
+      "rewards/real": 0.2218138426542282,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.455516014234875e-07,
+      "logits/generated": 0.21098542213439941,
+      "logits/real": -0.16925989091396332,
+      "logps/generated": -529.2811889648438,
+      "logps/real": -446.4693908691406,
+      "loss": 0.135,
+      "rewards/accuracies": 0.909375011920929,
+      "rewards/generated": -0.05097696930170059,
+      "rewards/margins": 0.2641257643699646,
+      "rewards/real": 0.213148832321167,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.4377224199288253e-07,
+      "logits/generated": 0.19224326312541962,
+      "logits/real": -0.19469952583312988,
+      "logps/generated": -522.6043090820312,
+      "logps/real": -442.81683349609375,
+      "loss": 0.1343,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.03429437428712845,
+      "rewards/margins": 0.26410582661628723,
+      "rewards/real": 0.22981147468090057,
+      "step": 315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.4199288256227754e-07,
+      "logits/generated": 0.1928466409444809,
+      "logits/real": -0.20203690230846405,
+      "logps/generated": -560.1980590820312,
+      "logps/real": -428.7918395996094,
+      "loss": 0.1333,
+      "rewards/accuracies": 0.9156249761581421,
+      "rewards/generated": -0.035711534321308136,
+      "rewards/margins": 0.2725156843662262,
+      "rewards/real": 0.23680415749549866,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.402135231316726e-07,
+      "logits/generated": 0.20243430137634277,
+      "logits/real": -0.16769111156463623,
+      "logps/generated": -504.7337951660156,
+      "logps/real": -429.45672607421875,
+      "loss": 0.1329,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.030326668173074722,
+      "rewards/margins": 0.27559566497802734,
+      "rewards/real": 0.2452690154314041,
+      "step": 325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.3843416370106763e-07,
+      "logits/generated": 0.2106173038482666,
+      "logits/real": -0.22538356482982635,
+      "logps/generated": -517.978759765625,
+      "logps/real": -474.59735107421875,
+      "loss": 0.1332,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.033002205193042755,
+      "rewards/margins": 0.27451324462890625,
+      "rewards/real": 0.2415110170841217,
+      "step": 330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.3665480427046264e-07,
+      "logits/generated": 0.21016332507133484,
+      "logits/real": -0.2604514956474304,
+      "logps/generated": -520.3605346679688,
+      "logps/real": -490.40631103515625,
+      "loss": 0.1336,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/generated": -0.044487275183200836,
+      "rewards/margins": 0.2778502106666565,
+      "rewards/real": 0.23336295783519745,
+      "step": 335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.348754448398576e-07,
+      "logits/generated": 0.22852489352226257,
+      "logits/real": -0.18696899712085724,
+      "logps/generated": -523.6672973632812,
+      "logps/real": -495.40057373046875,
+      "loss": 0.133,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.03802080079913139,
+      "rewards/margins": 0.2772233188152313,
+      "rewards/real": 0.23920252919197083,
+      "step": 340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.330960854092526e-07,
+      "logits/generated": 0.17415957152843475,
+      "logits/real": -0.19011852145195007,
+      "logps/generated": -513.3272705078125,
+      "logps/real": -403.3451232910156,
+      "loss": 0.1332,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.045419953763484955,
+      "rewards/margins": 0.2882402539253235,
+      "rewards/real": 0.24282029271125793,
+      "step": 345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.3131672597864763e-07,
+      "logits/generated": 0.2256070077419281,
+      "logits/real": -0.17997749149799347,
+      "logps/generated": -534.0467529296875,
+      "logps/real": -483.41937255859375,
+      "loss": 0.1314,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.049169979989528656,
+      "rewards/margins": 0.3096661865711212,
+      "rewards/real": 0.26049619913101196,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.295373665480427e-07,
+      "logits/generated": 0.2115151435136795,
+      "logits/real": -0.1779278963804245,
+      "logps/generated": -524.0692749023438,
+      "logps/real": -455.04046630859375,
+      "loss": 0.1326,
+      "rewards/accuracies": 0.890625,
+      "rewards/generated": -0.034146808087825775,
+      "rewards/margins": 0.2888898253440857,
+      "rewards/real": 0.2547430098056793,
+      "step": 355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.277580071174377e-07,
+      "logits/generated": 0.19143882393836975,
+      "logits/real": -0.1881864219903946,
+      "logps/generated": -514.4266357421875,
+      "logps/real": -449.707763671875,
+      "loss": 0.1332,
+      "rewards/accuracies": 0.909375011920929,
+      "rewards/generated": -0.028750786557793617,
+      "rewards/margins": 0.27422600984573364,
+      "rewards/real": 0.24547524750232697,
+      "step": 360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.2597864768683273e-07,
+      "logits/generated": 0.14517921209335327,
+      "logits/real": -0.23158793151378632,
+      "logps/generated": -524.9425048828125,
+      "logps/real": -423.65985107421875,
+      "loss": 0.131,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.02722376585006714,
+      "rewards/margins": 0.3096860945224762,
+      "rewards/real": 0.28246229887008667,
+      "step": 365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.2419928825622774e-07,
+      "logits/generated": 0.19077686965465546,
+      "logits/real": -0.206539124250412,
+      "logps/generated": -510.04681396484375,
+      "logps/real": -430.26513671875,
+      "loss": 0.1325,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/generated": -0.028535595163702965,
+      "rewards/margins": 0.28712087869644165,
+      "rewards/real": 0.25858527421951294,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.2241992882562276e-07,
+      "logits/generated": 0.20149891078472137,
+      "logits/real": -0.20975959300994873,
+      "logps/generated": -516.36083984375,
+      "logps/real": -456.06915283203125,
+      "loss": 0.1315,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.03869970887899399,
+      "rewards/margins": 0.3044109344482422,
+      "rewards/real": 0.26571124792099,
+      "step": 375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.2064056939501777e-07,
+      "logits/generated": 0.20944401621818542,
+      "logits/real": -0.20790605247020721,
+      "logps/generated": -513.2515869140625,
+      "logps/real": -470.3443298339844,
+      "loss": 0.1333,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/generated": -0.038152649998664856,
+      "rewards/margins": 0.2787010371685028,
+      "rewards/real": 0.24054837226867676,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.188612099644128e-07,
+      "logits/generated": 0.19982407987117767,
+      "logits/real": -0.22643396258354187,
+      "logps/generated": -531.8196411132812,
+      "logps/real": -455.9195861816406,
+      "loss": 0.1316,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.047723181545734406,
+      "rewards/margins": 0.31384533643722534,
+      "rewards/real": 0.26612216234207153,
+      "step": 385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.170818505338078e-07,
+      "logits/generated": 0.18060657382011414,
+      "logits/real": -0.2101529836654663,
+      "logps/generated": -522.3262939453125,
+      "logps/real": -430.184326171875,
+      "loss": 0.1328,
+      "rewards/accuracies": 0.909375011920929,
+      "rewards/generated": -0.05798162892460823,
+      "rewards/margins": 0.306007444858551,
+      "rewards/real": 0.2480258047580719,
+      "step": 390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.153024911032028e-07,
+      "logits/generated": 0.21487073600292206,
+      "logits/real": -0.17722031474113464,
+      "logps/generated": -508.89166259765625,
+      "logps/real": -457.6886291503906,
+      "loss": 0.132,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.033996812999248505,
+      "rewards/margins": 0.30001968145370483,
+      "rewards/real": 0.26602286100387573,
+      "step": 395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1352313167259783e-07,
+      "logits/generated": 0.21322698891162872,
+      "logits/real": -0.18131455779075623,
+      "logps/generated": -532.228271484375,
+      "logps/real": -413.98114013671875,
+      "loss": 0.1313,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.039401743561029434,
+      "rewards/margins": 0.30484768748283386,
+      "rewards/real": 0.2654459774494171,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1174377224199284e-07,
+      "logits/generated": 0.22022435069084167,
+      "logits/real": -0.16830001771450043,
+      "logps/generated": -524.0745849609375,
+      "logps/real": -444.0736389160156,
+      "loss": 0.1323,
+      "rewards/accuracies": 0.9156249761581421,
+      "rewards/generated": -0.05002317577600479,
+      "rewards/margins": 0.3070148527622223,
+      "rewards/real": 0.2569916844367981,
+      "step": 405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.0996441281138786e-07,
+      "logits/generated": 0.2300226241350174,
+      "logits/real": -0.19325582683086395,
+      "logps/generated": -515.1468505859375,
+      "logps/real": -462.9017639160156,
+      "loss": 0.1321,
+      "rewards/accuracies": 0.878125011920929,
+      "rewards/generated": -0.0361974723637104,
+      "rewards/margins": 0.2977278232574463,
+      "rewards/real": 0.261530339717865,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.081850533807829e-07,
+      "logits/generated": 0.21782374382019043,
+      "logits/real": -0.16392473876476288,
+      "logps/generated": -521.0572509765625,
+      "logps/real": -434.34857177734375,
+      "loss": 0.1308,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.05005559325218201,
+      "rewards/margins": 0.32770293951034546,
+      "rewards/real": 0.27764734625816345,
+      "step": 415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0640569395017794e-07,
+      "logits/generated": 0.20849747955799103,
+      "logits/real": -0.19438520073890686,
+      "logps/generated": -506.94970703125,
+      "logps/real": -476.08905029296875,
+      "loss": 0.1318,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.04421483352780342,
+      "rewards/margins": 0.30931296944618225,
+      "rewards/real": 0.2650981545448303,
+      "step": 420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.0462633451957295e-07,
+      "logits/generated": 0.2435588538646698,
+      "logits/real": -0.20184215903282166,
+      "logps/generated": -525.3304443359375,
+      "logps/real": -493.5686950683594,
+      "loss": 0.129,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.040996938943862915,
+      "rewards/margins": 0.3434499204158783,
+      "rewards/real": 0.3024529814720154,
+      "step": 425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0284697508896797e-07,
+      "logits/generated": 0.21247819066047668,
+      "logits/real": -0.2371760904788971,
+      "logps/generated": -513.602783203125,
+      "logps/real": -488.39349365234375,
+      "loss": 0.1314,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.032181791961193085,
+      "rewards/margins": 0.3022555410861969,
+      "rewards/real": 0.27007371187210083,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0106761565836293e-07,
+      "logits/generated": 0.21260352432727814,
+      "logits/real": -0.18493767082691193,
+      "logps/generated": -530.4390869140625,
+      "logps/real": -451.08502197265625,
+      "loss": 0.1303,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.04360523074865341,
+      "rewards/margins": 0.3299635052680969,
+      "rewards/real": 0.2863582968711853,
+      "step": 435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.9928825622775795e-07,
+      "logits/generated": 0.17910648882389069,
+      "logits/real": -0.21472099423408508,
+      "logps/generated": -506.29864501953125,
+      "logps/real": -440.6806640625,
+      "loss": 0.1298,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.049231477081775665,
+      "rewards/margins": 0.3400951623916626,
+      "rewards/real": 0.29086369276046753,
+      "step": 440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.97508896797153e-07,
+      "logits/generated": 0.2306959182024002,
+      "logits/real": -0.22861719131469727,
+      "logps/generated": -518.2430419921875,
+      "logps/real": -484.080322265625,
+      "loss": 0.1309,
+      "rewards/accuracies": 0.9156249761581421,
+      "rewards/generated": -0.04008597880601883,
+      "rewards/margins": 0.31413963437080383,
+      "rewards/real": 0.2740536630153656,
+      "step": 445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9572953736654803e-07,
+      "logits/generated": 0.19337113201618195,
+      "logits/real": -0.1461922973394394,
+      "logps/generated": -520.7548828125,
+      "logps/real": -412.48834228515625,
+      "loss": 0.1303,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.04457572102546692,
+      "rewards/margins": 0.33213964104652405,
+      "rewards/real": 0.28756392002105713,
+      "step": 450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9395017793594304e-07,
+      "logits/generated": 0.19213825464248657,
+      "logits/real": -0.17207488417625427,
+      "logps/generated": -508.79742431640625,
+      "logps/real": -431.6582946777344,
+      "loss": 0.1309,
+      "rewards/accuracies": 0.8812500238418579,
+      "rewards/generated": -0.03421945124864578,
+      "rewards/margins": 0.315326452255249,
+      "rewards/real": 0.2811070382595062,
+      "step": 455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.9217081850533806e-07,
+      "logits/generated": 0.20467457175254822,
+      "logits/real": -0.18655052781105042,
+      "logps/generated": -534.2474365234375,
+      "logps/real": -451.9898986816406,
+      "loss": 0.1296,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/generated": -0.045324601233005524,
+      "rewards/margins": 0.33579733967781067,
+      "rewards/real": 0.29047271609306335,
+      "step": 460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9039145907473307e-07,
+      "logits/generated": 0.1963758021593094,
+      "logits/real": -0.24852268397808075,
+      "logps/generated": -519.084716796875,
+      "logps/real": -475.10467529296875,
+      "loss": 0.1284,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.04257890582084656,
+      "rewards/margins": 0.3550090789794922,
+      "rewards/real": 0.31243017315864563,
+      "step": 465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.886120996441281e-07,
+      "logits/generated": 0.18512029945850372,
+      "logits/real": -0.20866036415100098,
+      "logps/generated": -530.0177612304688,
+      "logps/real": -454.27496337890625,
+      "loss": 0.1285,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.03352942317724228,
+      "rewards/margins": 0.34436360001564026,
+      "rewards/real": 0.310834139585495,
+      "step": 470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.8683274021352315e-07,
+      "logits/generated": 0.20876750349998474,
+      "logits/real": -0.23259572684764862,
+      "logps/generated": -542.6410522460938,
+      "logps/real": -462.28558349609375,
+      "loss": 0.1296,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.028329987078905106,
+      "rewards/margins": 0.32887953519821167,
+      "rewards/real": 0.30054956674575806,
+      "step": 475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8505338078291817e-07,
+      "logits/generated": 0.20735135674476624,
+      "logits/real": -0.18274542689323425,
+      "logps/generated": -531.2545776367188,
+      "logps/real": -445.79254150390625,
+      "loss": 0.1297,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.049115560948848724,
+      "rewards/margins": 0.34622785449028015,
+      "rewards/real": 0.29711228609085083,
+      "step": 480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8327402135231313e-07,
+      "logits/generated": 0.22061915695667267,
+      "logits/real": -0.1666678935289383,
+      "logps/generated": -512.0516967773438,
+      "logps/real": -454.59765625,
+      "loss": 0.1287,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.031470268964767456,
+      "rewards/margins": 0.3416782021522522,
+      "rewards/real": 0.31020793318748474,
+      "step": 485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8149466192170814e-07,
+      "logits/generated": 0.20676842331886292,
+      "logits/real": -0.22948101162910461,
+      "logps/generated": -517.1962890625,
+      "logps/real": -462.365478515625,
+      "loss": 0.1289,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/generated": -0.05391014367341995,
+      "rewards/margins": 0.35990110039711,
+      "rewards/real": 0.30599093437194824,
+      "step": 490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.7971530249110316e-07,
+      "logits/generated": 0.21802881360054016,
+      "logits/real": -0.20144328474998474,
+      "logps/generated": -512.7889404296875,
+      "logps/real": -446.5421447753906,
+      "loss": 0.1284,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.04316414147615433,
+      "rewards/margins": 0.35774049162864685,
+      "rewards/real": 0.3145763874053955,
+      "step": 495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.7793594306049817e-07,
+      "logits/generated": 0.1871260106563568,
+      "logits/real": -0.18488526344299316,
+      "logps/generated": -512.2315063476562,
+      "logps/real": -420.43865966796875,
+      "loss": 0.1275,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.0390264093875885,
+      "rewards/margins": 0.3577350676059723,
+      "rewards/real": 0.3187086582183838,
+      "step": 500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.7615658362989324e-07,
+      "logits/generated": 0.236494779586792,
+      "logits/real": -0.15096931159496307,
+      "logps/generated": -507.55548095703125,
+      "logps/real": -445.41046142578125,
+      "loss": 0.1275,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.048680149018764496,
+      "rewards/margins": 0.3694431185722351,
+      "rewards/real": 0.3207629919052124,
+      "step": 505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7437722419928825e-07,
+      "logits/generated": 0.22562535107135773,
+      "logits/real": -0.19512876868247986,
+      "logps/generated": -531.6770629882812,
+      "logps/real": -446.9190979003906,
+      "loss": 0.1277,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.04859489947557449,
+      "rewards/margins": 0.3696749806404114,
+      "rewards/real": 0.3210800588130951,
+      "step": 510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7259786476868327e-07,
+      "logits/generated": 0.2176874577999115,
+      "logits/real": -0.1850893795490265,
+      "logps/generated": -514.0935668945312,
+      "logps/real": -441.65728759765625,
+      "loss": 0.1298,
+      "rewards/accuracies": 0.8968750238418579,
+      "rewards/generated": -0.03823425993323326,
+      "rewards/margins": 0.3405149579048157,
+      "rewards/real": 0.3022806644439697,
+      "step": 515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.708185053380783e-07,
+      "logits/generated": 0.24159689247608185,
+      "logits/real": -0.2040737122297287,
+      "logps/generated": -501.77117919921875,
+      "logps/real": -471.8893127441406,
+      "loss": 0.1273,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.048200882971286774,
+      "rewards/margins": 0.37571340799331665,
+      "rewards/real": 0.3275125324726105,
+      "step": 520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.690391459074733e-07,
+      "logits/generated": 0.2472068816423416,
+      "logits/real": -0.1870822012424469,
+      "logps/generated": -508.8600158691406,
+      "logps/real": -473.0227966308594,
+      "loss": 0.1285,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.04576011374592781,
+      "rewards/margins": 0.36241215467453003,
+      "rewards/real": 0.3166520297527313,
+      "step": 525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6725978647686836e-07,
+      "logits/generated": 0.2242005616426468,
+      "logits/real": -0.22695569694042206,
+      "logps/generated": -510.89080810546875,
+      "logps/real": -479.42108154296875,
+      "loss": 0.1283,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -0.049571048468351364,
+      "rewards/margins": 0.36083894968032837,
+      "rewards/real": 0.3112679123878479,
+      "step": 530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.654804270462633e-07,
+      "logits/generated": 0.21036097407341003,
+      "logits/real": -0.1436823010444641,
+      "logps/generated": -523.9207153320312,
+      "logps/real": -429.45458984375,
+      "loss": 0.1265,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.06013381481170654,
+      "rewards/margins": 0.3962857127189636,
+      "rewards/real": 0.3361518979072571,
+      "step": 535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6370106761565834e-07,
+      "logits/generated": 0.24738578498363495,
+      "logits/real": -0.13808973133563995,
+      "logps/generated": -509.6529235839844,
+      "logps/real": -439.38531494140625,
+      "loss": 0.128,
+      "rewards/accuracies": 0.903124988079071,
+      "rewards/generated": -0.0440841019153595,
+      "rewards/margins": 0.36679965257644653,
+      "rewards/real": 0.32271555066108704,
+      "step": 540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6192170818505336e-07,
+      "logits/generated": 0.22692394256591797,
+      "logits/real": -0.17329858243465424,
+      "logps/generated": -526.3966064453125,
+      "logps/real": -431.3544921875,
+      "loss": 0.125,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -0.05675189942121506,
+      "rewards/margins": 0.4089035987854004,
+      "rewards/real": 0.35215169191360474,
+      "step": 545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6014234875444837e-07,
+      "logits/generated": 0.24121804535388947,
+      "logits/real": -0.20156726241111755,
+      "logps/generated": -524.36572265625,
+      "logps/real": -451.2276306152344,
+      "loss": 0.1267,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.06758119910955429,
+      "rewards/margins": 0.39545828104019165,
+      "rewards/real": 0.32787710428237915,
+      "step": 550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.583629893238434e-07,
+      "logits/generated": 0.22711153328418732,
+      "logits/real": -0.16746626794338226,
+      "logps/generated": -536.3843994140625,
+      "logps/real": -448.345703125,
+      "loss": 0.1257,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.04590682312846184,
+      "rewards/margins": 0.38996079564094543,
+      "rewards/real": 0.3440540134906769,
+      "step": 555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.565836298932384e-07,
+      "logits/generated": 0.19922706484794617,
+      "logits/real": -0.21461109817028046,
+      "logps/generated": -515.5289306640625,
+      "logps/real": -459.3103942871094,
+      "loss": 0.1278,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.05231200531125069,
+      "rewards/margins": 0.3805178105831146,
+      "rewards/real": 0.32820582389831543,
+      "step": 560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5480427046263347e-07,
+      "logits/generated": 0.22783759236335754,
+      "logits/real": -0.20611587166786194,
+      "logps/generated": -521.995849609375,
+      "logps/real": -476.81585693359375,
+      "loss": 0.1268,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.04818415641784668,
+      "rewards/margins": 0.3822416663169861,
+      "rewards/real": 0.3340575098991394,
+      "step": 565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.530249110320285e-07,
+      "logits/generated": 0.2178882360458374,
+      "logits/real": -0.25764960050582886,
+      "logps/generated": -535.8800659179688,
+      "logps/real": -476.8208923339844,
+      "loss": 0.1274,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.0434628501534462,
+      "rewards/margins": 0.370759516954422,
+      "rewards/real": 0.3272966742515564,
+      "step": 570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.512455516014235e-07,
+      "logits/generated": 0.18462668359279633,
+      "logits/real": -0.2151033580303192,
+      "logps/generated": -523.6564331054688,
+      "logps/real": -465.83929443359375,
+      "loss": 0.1281,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.0575849823653698,
+      "rewards/margins": 0.3719588816165924,
+      "rewards/real": 0.3143738806247711,
+      "step": 575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4946619217081846e-07,
+      "logits/generated": 0.21648356318473816,
+      "logits/real": -0.22641988098621368,
+      "logps/generated": -503.7261657714844,
+      "logps/real": -470.4150390625,
+      "loss": 0.1279,
+      "rewards/accuracies": 0.903124988079071,
+      "rewards/generated": -0.035186249762773514,
+      "rewards/margins": 0.36030828952789307,
+      "rewards/real": 0.32512202858924866,
+      "step": 580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4768683274021347e-07,
+      "logits/generated": 0.22231841087341309,
+      "logits/real": -0.2200026959180832,
+      "logps/generated": -551.32373046875,
+      "logps/real": -477.9534606933594,
+      "loss": 0.1272,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -0.07955564558506012,
+      "rewards/margins": 0.3987639546394348,
+      "rewards/real": 0.3192083239555359,
+      "step": 585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.459074733096085e-07,
+      "logits/generated": 0.23221631348133087,
+      "logits/real": -0.17455917596817017,
+      "logps/generated": -522.5137329101562,
+      "logps/real": -466.8882751464844,
+      "loss": 0.1259,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.07725702226161957,
+      "rewards/margins": 0.4179654121398926,
+      "rewards/real": 0.3407083749771118,
+      "step": 590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4412811387900355e-07,
+      "logits/generated": 0.24756267666816711,
+      "logits/real": -0.20915552973747253,
+      "logps/generated": -514.363037109375,
+      "logps/real": -500.83721923828125,
+      "loss": 0.1266,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.05856259539723396,
+      "rewards/margins": 0.3967771828174591,
+      "rewards/real": 0.33821457624435425,
+      "step": 595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4234875444839857e-07,
+      "logits/generated": 0.21352288126945496,
+      "logits/real": -0.20619618892669678,
+      "logps/generated": -510.033203125,
+      "logps/real": -440.34423828125,
+      "loss": 0.1261,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.03017326630651951,
+      "rewards/margins": 0.3813417851924896,
+      "rewards/real": 0.35116851329803467,
+      "step": 600
+    },
+    {
+      "epoch": 0.38,
+      "eval_logits/generated": 0.22333942353725433,
+      "eval_logits/real": -0.18551431596279144,
+      "eval_logps/generated": -521.822509765625,
+      "eval_logps/real": -454.307861328125,
+      "eval_loss": 0.12625987827777863,
+      "eval_rewards/accuracies": 0.9365000128746033,
+      "eval_rewards/generated": -0.0522722452878952,
+      "eval_rewards/margins": 0.3948952555656433,
+      "eval_rewards/real": 0.3426230549812317,
+      "eval_runtime": 467.1537,
+      "eval_samples_per_second": 4.281,
+      "eval_steps_per_second": 1.07,
+      "step": 600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.405693950177936e-07,
+      "logits/generated": 0.20984432101249695,
+      "logits/real": -0.19126197695732117,
+      "logps/generated": -518.5199584960938,
+      "logps/real": -439.51324462890625,
+      "loss": 0.1265,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.05940445512533188,
+      "rewards/margins": 0.4006901681423187,
+      "rewards/real": 0.34128570556640625,
+      "step": 605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.387900355871886e-07,
+      "logits/generated": 0.23297259211540222,
+      "logits/real": -0.1324882060289383,
+      "logps/generated": -533.5361938476562,
+      "logps/real": -436.12335205078125,
+      "loss": 0.1264,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.044547099620103836,
+      "rewards/margins": 0.39039477705955505,
+      "rewards/real": 0.3458476662635803,
+      "step": 610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.370106761565836e-07,
+      "logits/generated": 0.21869917213916779,
+      "logits/real": -0.16005317866802216,
+      "logps/generated": -505.5008239746094,
+      "logps/real": -440.06341552734375,
+      "loss": 0.1251,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.03350327908992767,
+      "rewards/margins": 0.3943425714969635,
+      "rewards/real": 0.360839307308197,
+      "step": 615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.352313167259787e-07,
+      "logits/generated": 0.20472922921180725,
+      "logits/real": -0.17530474066734314,
+      "logps/generated": -510.0504455566406,
+      "logps/real": -431.83587646484375,
+      "loss": 0.1268,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.05266762897372246,
+      "rewards/margins": 0.3853222131729126,
+      "rewards/real": 0.33265453577041626,
+      "step": 620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.334519572953737e-07,
+      "logits/generated": 0.2204224169254303,
+      "logits/real": -0.1700909435749054,
+      "logps/generated": -510.51141357421875,
+      "logps/real": -437.6064453125,
+      "loss": 0.1247,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.03798319026827812,
+      "rewards/margins": 0.40347009897232056,
+      "rewards/real": 0.36548691987991333,
+      "step": 625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3167259786476865e-07,
+      "logits/generated": 0.21306562423706055,
+      "logits/real": -0.19046835601329803,
+      "logps/generated": -513.0300903320312,
+      "logps/real": -442.71124267578125,
+      "loss": 0.1259,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.03914360702037811,
+      "rewards/margins": 0.38992029428482056,
+      "rewards/real": 0.35077667236328125,
+      "step": 630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2989323843416367e-07,
+      "logits/generated": 0.21032293140888214,
+      "logits/real": -0.21575979888439178,
+      "logps/generated": -509.75128173828125,
+      "logps/real": -469.04351806640625,
+      "loss": 0.1265,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.06580056995153427,
+      "rewards/margins": 0.40863609313964844,
+      "rewards/real": 0.3428354859352112,
+      "step": 635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.281138790035587e-07,
+      "logits/generated": 0.24854688346385956,
+      "logits/real": -0.14409920573234558,
+      "logps/generated": -514.2542724609375,
+      "logps/real": -451.86846923828125,
+      "loss": 0.1268,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.02604399248957634,
+      "rewards/margins": 0.3732038140296936,
+      "rewards/real": 0.34715980291366577,
+      "step": 640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.263345195729537e-07,
+      "logits/generated": 0.1697693169116974,
+      "logits/real": -0.22252459824085236,
+      "logps/generated": -515.5403442382812,
+      "logps/real": -433.65496826171875,
+      "loss": 0.1261,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.06866597384214401,
+      "rewards/margins": 0.40779757499694824,
+      "rewards/real": 0.3391316533088684,
+      "step": 645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.245551601423487e-07,
+      "logits/generated": 0.24601443111896515,
+      "logits/real": -0.2113780677318573,
+      "logps/generated": -532.38232421875,
+      "logps/real": -487.1578063964844,
+      "loss": 0.1259,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.06285814940929413,
+      "rewards/margins": 0.40850821137428284,
+      "rewards/real": 0.3456500470638275,
+      "step": 650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.227758007117438e-07,
+      "logits/generated": 0.21627071499824524,
+      "logits/real": -0.16779956221580505,
+      "logps/generated": -538.037109375,
+      "logps/real": -454.26580810546875,
+      "loss": 0.1257,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.07498253881931305,
+      "rewards/margins": 0.4246435761451721,
+      "rewards/real": 0.34966105222702026,
+      "step": 655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.209964412811388e-07,
+      "logits/generated": 0.22846777737140656,
+      "logits/real": -0.23057064414024353,
+      "logps/generated": -534.3760986328125,
+      "logps/real": -467.74365234375,
+      "loss": 0.1251,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.05258546024560928,
+      "rewards/margins": 0.4155358672142029,
+      "rewards/real": 0.3629503846168518,
+      "step": 660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.192170818505338e-07,
+      "logits/generated": 0.22176751494407654,
+      "logits/real": -0.2189159095287323,
+      "logps/generated": -512.0242309570312,
+      "logps/real": -471.23724365234375,
+      "loss": 0.126,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -0.02907324768602848,
+      "rewards/margins": 0.3788103461265564,
+      "rewards/real": 0.34973710775375366,
+      "step": 665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.174377224199288e-07,
+      "logits/generated": 0.22437652945518494,
+      "logits/real": -0.14511239528656006,
+      "logps/generated": -529.4261474609375,
+      "logps/real": -408.4437561035156,
+      "loss": 0.125,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.09781909734010696,
+      "rewards/margins": 0.45350271463394165,
+      "rewards/real": 0.3556835353374481,
+      "step": 670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1565836298932384e-07,
+      "logits/generated": 0.2222731113433838,
+      "logits/real": -0.20278367400169373,
+      "logps/generated": -516.1197509765625,
+      "logps/real": -449.39227294921875,
+      "loss": 0.1256,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.05577481910586357,
+      "rewards/margins": 0.41228675842285156,
+      "rewards/real": 0.3565119206905365,
+      "step": 675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.138790035587188e-07,
+      "logits/generated": 0.19292616844177246,
+      "logits/real": -0.1925380676984787,
+      "logps/generated": -521.6808471679688,
+      "logps/real": -438.78778076171875,
+      "loss": 0.1251,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.060262132436037064,
+      "rewards/margins": 0.4185408055782318,
+      "rewards/real": 0.35827872157096863,
+      "step": 680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1209964412811387e-07,
+      "logits/generated": 0.23866550624370575,
+      "logits/real": -0.1537286341190338,
+      "logps/generated": -510.27069091796875,
+      "logps/real": -429.9480895996094,
+      "loss": 0.1261,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.056734926998615265,
+      "rewards/margins": 0.4073144495487213,
+      "rewards/real": 0.35057950019836426,
+      "step": 685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.103202846975089e-07,
+      "logits/generated": 0.22526803612709045,
+      "logits/real": -0.2068435698747635,
+      "logps/generated": -523.3880615234375,
+      "logps/real": -460.9934997558594,
+      "loss": 0.1252,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.05335558205842972,
+      "rewards/margins": 0.41251659393310547,
+      "rewards/real": 0.35916104912757874,
+      "step": 690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.085409252669039e-07,
+      "logits/generated": 0.2166483849287033,
+      "logits/real": -0.16389943659305573,
+      "logps/generated": -533.65478515625,
+      "logps/real": -447.15252685546875,
+      "loss": 0.1273,
+      "rewards/accuracies": 0.909375011920929,
+      "rewards/generated": -0.04129910469055176,
+      "rewards/margins": 0.37630194425582886,
+      "rewards/real": 0.3350028395652771,
+      "step": 695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.067615658362989e-07,
+      "logits/generated": 0.22569017112255096,
+      "logits/real": -0.1831541508436203,
+      "logps/generated": -510.294921875,
+      "logps/real": -446.5750427246094,
+      "loss": 0.1271,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/generated": -0.0428488552570343,
+      "rewards/margins": 0.3756479024887085,
+      "rewards/real": 0.3327990472316742,
+      "step": 700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.049822064056939e-07,
+      "logits/generated": 0.23136644065380096,
+      "logits/real": -0.23328185081481934,
+      "logps/generated": -552.6895751953125,
+      "logps/real": -474.44781494140625,
+      "loss": 0.1242,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.07789872586727142,
+      "rewards/margins": 0.44054466485977173,
+      "rewards/real": 0.3626458942890167,
+      "step": 705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.03202846975089e-07,
+      "logits/generated": 0.2204451858997345,
+      "logits/real": -0.19231361150741577,
+      "logps/generated": -524.7732543945312,
+      "logps/real": -432.150390625,
+      "loss": 0.1256,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.05935334041714668,
+      "rewards/margins": 0.4120023250579834,
+      "rewards/real": 0.3526490032672882,
+      "step": 710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.01423487544484e-07,
+      "logits/generated": 0.2073640078306198,
+      "logits/real": -0.21677975356578827,
+      "logps/generated": -534.759765625,
+      "logps/real": -462.5953674316406,
+      "loss": 0.1256,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.0705971047282219,
+      "rewards/margins": 0.4193173944950104,
+      "rewards/real": 0.3487202823162079,
+      "step": 715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.99644128113879e-07,
+      "logits/generated": 0.23698802292346954,
+      "logits/real": -0.12353632599115372,
+      "logps/generated": -526.7183227539062,
+      "logps/real": -405.3436584472656,
+      "loss": 0.1245,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.03835119307041168,
+      "rewards/margins": 0.4056209623813629,
+      "rewards/real": 0.3672698140144348,
+      "step": 720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.97864768683274e-07,
+      "logits/generated": 0.21073408424854279,
+      "logits/real": -0.13484127819538116,
+      "logps/generated": -521.523193359375,
+      "logps/real": -420.841796875,
+      "loss": 0.1234,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.0601874478161335,
+      "rewards/margins": 0.44554439187049866,
+      "rewards/real": 0.38535696268081665,
+      "step": 725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.96085409252669e-07,
+      "logits/generated": 0.2359226495027542,
+      "logits/real": -0.139388769865036,
+      "logps/generated": -498.6105041503906,
+      "logps/real": -436.9283142089844,
+      "loss": 0.1267,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.06179014965891838,
+      "rewards/margins": 0.4049047827720642,
+      "rewards/real": 0.3431146442890167,
+      "step": 730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.94306049822064e-07,
+      "logits/generated": 0.22477808594703674,
+      "logits/real": -0.18658998608589172,
+      "logps/generated": -539.577392578125,
+      "logps/real": -436.072998046875,
+      "loss": 0.1252,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.06574585288763046,
+      "rewards/margins": 0.4261474609375,
+      "rewards/real": 0.36040163040161133,
+      "step": 735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.92526690391459e-07,
+      "logits/generated": 0.24180345237255096,
+      "logits/real": -0.10471512377262115,
+      "logps/generated": -533.6973266601562,
+      "logps/real": -423.3241271972656,
+      "loss": 0.1236,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.08005066215991974,
+      "rewards/margins": 0.45733848214149475,
+      "rewards/real": 0.3772878050804138,
+      "step": 740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.907473309608541e-07,
+      "logits/generated": 0.21116897463798523,
+      "logits/real": -0.19643299281597137,
+      "logps/generated": -534.4854125976562,
+      "logps/real": -429.40313720703125,
+      "loss": 0.1253,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.03764469176530838,
+      "rewards/margins": 0.40410351753234863,
+      "rewards/real": 0.36645883321762085,
+      "step": 745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.889679715302491e-07,
+      "logits/generated": 0.2296733409166336,
+      "logits/real": -0.18100515007972717,
+      "logps/generated": -523.2075805664062,
+      "logps/real": -454.98931884765625,
+      "loss": 0.1249,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.06059467792510986,
+      "rewards/margins": 0.42317479848861694,
+      "rewards/real": 0.3625801205635071,
+      "step": 750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.871886120996441e-07,
+      "logits/generated": 0.269220769405365,
+      "logits/real": -0.13889244198799133,
+      "logps/generated": -497.45379638671875,
+      "logps/real": -437.2528381347656,
+      "loss": 0.1245,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.04436558485031128,
+      "rewards/margins": 0.41556286811828613,
+      "rewards/real": 0.37119728326797485,
+      "step": 755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8540925266903914e-07,
+      "logits/generated": 0.21270576119422913,
+      "logits/real": -0.18475386500358582,
+      "logps/generated": -528.8832397460938,
+      "logps/real": -440.03057861328125,
+      "loss": 0.1238,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.0362856462597847,
+      "rewards/margins": 0.42006000876426697,
+      "rewards/real": 0.38377436995506287,
+      "step": 760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8362989323843415e-07,
+      "logits/generated": 0.23695051670074463,
+      "logits/real": -0.18097969889640808,
+      "logps/generated": -525.389404296875,
+      "logps/real": -460.3702697753906,
+      "loss": 0.1235,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.06530066579580307,
+      "rewards/margins": 0.4440096318721771,
+      "rewards/real": 0.37870892882347107,
+      "step": 765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.818505338078292e-07,
+      "logits/generated": 0.24917741119861603,
+      "logits/real": -0.16412577033042908,
+      "logps/generated": -509.0914001464844,
+      "logps/real": -463.69830322265625,
+      "loss": 0.1257,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.0415273942053318,
+      "rewards/margins": 0.3965379595756531,
+      "rewards/real": 0.3550105392932892,
+      "step": 770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.800711743772242e-07,
+      "logits/generated": 0.25012919306755066,
+      "logits/real": -0.1794908493757248,
+      "logps/generated": -524.6221923828125,
+      "logps/real": -469.295166015625,
+      "loss": 0.1252,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.06365501135587692,
+      "rewards/margins": 0.42465314269065857,
+      "rewards/real": 0.36099809408187866,
+      "step": 775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.782918149466192e-07,
+      "logits/generated": 0.2226579487323761,
+      "logits/real": -0.19419345259666443,
+      "logps/generated": -530.0787963867188,
+      "logps/real": -469.27606201171875,
+      "loss": 0.1259,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.051626600325107574,
+      "rewards/margins": 0.4078029990196228,
+      "rewards/real": 0.3561764359474182,
+      "step": 780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.765124555160142e-07,
+      "logits/generated": 0.1991543471813202,
+      "logits/real": -0.21607446670532227,
+      "logps/generated": -518.56005859375,
+      "logps/real": -447.11962890625,
+      "loss": 0.1247,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.05867942050099373,
+      "rewards/margins": 0.42213720083236694,
+      "rewards/real": 0.3634577691555023,
+      "step": 785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.747330960854092e-07,
+      "logits/generated": 0.21239633858203888,
+      "logits/real": -0.15555164217948914,
+      "logps/generated": -504.6756286621094,
+      "logps/real": -419.6802673339844,
+      "loss": 0.1243,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.03540899232029915,
+      "rewards/margins": 0.4143758714199066,
+      "rewards/real": 0.3789668679237366,
+      "step": 790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7295373665480424e-07,
+      "logits/generated": 0.23810231685638428,
+      "logits/real": -0.17116566002368927,
+      "logps/generated": -507.6982421875,
+      "logps/real": -465.0140686035156,
+      "loss": 0.1252,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.035895101726055145,
+      "rewards/margins": 0.40210455656051636,
+      "rewards/real": 0.3662094473838806,
+      "step": 795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.711743772241993e-07,
+      "logits/generated": 0.23017612099647522,
+      "logits/real": -0.19639410078525543,
+      "logps/generated": -513.0866088867188,
+      "logps/real": -443.845947265625,
+      "loss": 0.1243,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.06835327297449112,
+      "rewards/margins": 0.44039884209632874,
+      "rewards/real": 0.3720455765724182,
+      "step": 800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.693950177935943e-07,
+      "logits/generated": 0.2244982272386551,
+      "logits/real": -0.19841459393501282,
+      "logps/generated": -542.2571411132812,
+      "logps/real": -450.25,
+      "loss": 0.1255,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.07537367194890976,
+      "rewards/margins": 0.42732158303260803,
+      "rewards/real": 0.35194793343544006,
+      "step": 805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6761565836298933e-07,
+      "logits/generated": 0.2113502472639084,
+      "logits/real": -0.18353787064552307,
+      "logps/generated": -529.5034790039062,
+      "logps/real": -447.72357177734375,
+      "loss": 0.1234,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.06783580780029297,
+      "rewards/margins": 0.4514337480068207,
+      "rewards/real": 0.3835979402065277,
+      "step": 810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6583629893238435e-07,
+      "logits/generated": 0.23711976408958435,
+      "logits/real": -0.16799500584602356,
+      "logps/generated": -506.9942932128906,
+      "logps/real": -472.8714294433594,
+      "loss": 0.124,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.043571460992097855,
+      "rewards/margins": 0.4209679961204529,
+      "rewards/real": 0.3773965537548065,
+      "step": 815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.6405693950177936e-07,
+      "logits/generated": 0.21126489341259003,
+      "logits/real": -0.19328458607196808,
+      "logps/generated": -506.70159912109375,
+      "logps/real": -458.549560546875,
+      "loss": 0.1239,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.05301676318049431,
+      "rewards/margins": 0.42876943945884705,
+      "rewards/real": 0.375752717256546,
+      "step": 820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.622775800711743e-07,
+      "logits/generated": 0.23474998772144318,
+      "logits/real": -0.1575848013162613,
+      "logps/generated": -540.1917114257812,
+      "logps/real": -445.42156982421875,
+      "loss": 0.123,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.07524801790714264,
+      "rewards/margins": 0.4695201516151428,
+      "rewards/real": 0.3942721486091614,
+      "step": 825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6049822064056934e-07,
+      "logits/generated": 0.20971938967704773,
+      "logits/real": -0.18155238032341003,
+      "logps/generated": -521.19775390625,
+      "logps/real": -440.2428283691406,
+      "loss": 0.1251,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.05046185106039047,
+      "rewards/margins": 0.4103214144706726,
+      "rewards/real": 0.3598595857620239,
+      "step": 830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.587188612099644e-07,
+      "logits/generated": 0.2089037150144577,
+      "logits/real": -0.15210075676441193,
+      "logps/generated": -532.7290649414062,
+      "logps/real": -409.13787841796875,
+      "loss": 0.1242,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.06394266337156296,
+      "rewards/margins": 0.4383109509944916,
+      "rewards/real": 0.37436825037002563,
+      "step": 835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.569395017793594e-07,
+      "logits/generated": 0.2460833042860031,
+      "logits/real": -0.13932695984840393,
+      "logps/generated": -526.330322265625,
+      "logps/real": -448.05377197265625,
+      "loss": 0.1235,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.07719500362873077,
+      "rewards/margins": 0.4495556950569153,
+      "rewards/real": 0.3723606467247009,
+      "step": 840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5516014234875444e-07,
+      "logits/generated": 0.20458194613456726,
+      "logits/real": -0.17320473492145538,
+      "logps/generated": -524.3390502929688,
+      "logps/real": -426.5902404785156,
+      "loss": 0.1223,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.0741974413394928,
+      "rewards/margins": 0.46985602378845215,
+      "rewards/real": 0.39565858244895935,
+      "step": 845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5338078291814945e-07,
+      "logits/generated": 0.26021358370780945,
+      "logits/real": -0.15163478255271912,
+      "logps/generated": -508.49951171875,
+      "logps/real": -436.19061279296875,
+      "loss": 0.1249,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.051833074539899826,
+      "rewards/margins": 0.4227888584136963,
+      "rewards/real": 0.37095582485198975,
+      "step": 850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5160142348754447e-07,
+      "logits/generated": 0.22410514950752258,
+      "logits/real": -0.19840450584888458,
+      "logps/generated": -527.9317016601562,
+      "logps/real": -448.7506408691406,
+      "loss": 0.1231,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.05459089204668999,
+      "rewards/margins": 0.44405126571655273,
+      "rewards/real": 0.38946038484573364,
+      "step": 855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.498220640569395e-07,
+      "logits/generated": 0.21920056641101837,
+      "logits/real": -0.1736256182193756,
+      "logps/generated": -534.7623291015625,
+      "logps/real": -443.5054626464844,
+      "loss": 0.1238,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.06553854048252106,
+      "rewards/margins": 0.4489002823829651,
+      "rewards/real": 0.3833617568016052,
+      "step": 860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.480427046263345e-07,
+      "logits/generated": 0.21249528229236603,
+      "logits/real": -0.14947716891765594,
+      "logps/generated": -537.5214233398438,
+      "logps/real": -431.114013671875,
+      "loss": 0.1227,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.06390459835529327,
+      "rewards/margins": 0.4571428894996643,
+      "rewards/real": 0.3932383060455322,
+      "step": 865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.462633451957295e-07,
+      "logits/generated": 0.22841641306877136,
+      "logits/real": -0.1430717259645462,
+      "logps/generated": -520.1146240234375,
+      "logps/real": -417.46124267578125,
+      "loss": 0.1243,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.051337797194719315,
+      "rewards/margins": 0.4231833815574646,
+      "rewards/real": 0.37184563279151917,
+      "step": 870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.444839857651246e-07,
+      "logits/generated": 0.2303360402584076,
+      "logits/real": -0.21451357007026672,
+      "logps/generated": -516.4577026367188,
+      "logps/real": -471.78485107421875,
+      "loss": 0.1254,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.04709147289395332,
+      "rewards/margins": 0.4153119921684265,
+      "rewards/real": 0.3682205379009247,
+      "step": 875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.427046263345196e-07,
+      "logits/generated": 0.24512319266796112,
+      "logits/real": -0.1643613874912262,
+      "logps/generated": -521.9609985351562,
+      "logps/real": -444.45794677734375,
+      "loss": 0.1233,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.07158510386943817,
+      "rewards/margins": 0.45351576805114746,
+      "rewards/real": 0.3819306790828705,
+      "step": 880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.4092526690391455e-07,
+      "logits/generated": 0.24102430045604706,
+      "logits/real": -0.13085922598838806,
+      "logps/generated": -524.731689453125,
+      "logps/real": -430.0142517089844,
+      "loss": 0.1247,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.0607861764729023,
+      "rewards/margins": 0.42205628752708435,
+      "rewards/real": 0.36127012968063354,
+      "step": 885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.391459074733096e-07,
+      "logits/generated": 0.23209047317504883,
+      "logits/real": -0.15809141099452972,
+      "logps/generated": -519.98095703125,
+      "logps/real": -441.5606994628906,
+      "loss": 0.1234,
+      "rewards/accuracies": 0.953125,
+      "rewards/generated": -0.06819422543048859,
+      "rewards/margins": 0.45197123289108276,
+      "rewards/real": 0.383776992559433,
+      "step": 890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.373665480427046e-07,
+      "logits/generated": 0.25702834129333496,
+      "logits/real": -0.1923447549343109,
+      "logps/generated": -512.0235595703125,
+      "logps/real": -448.59381103515625,
+      "loss": 0.1233,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -0.0629502683877945,
+      "rewards/margins": 0.449089914560318,
+      "rewards/real": 0.3861396312713623,
+      "step": 895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.3558718861209965e-07,
+      "logits/generated": 0.20865292847156525,
+      "logits/real": -0.18462467193603516,
+      "logps/generated": -508.04833984375,
+      "logps/real": -440.5543518066406,
+      "loss": 0.1243,
+      "rewards/accuracies": 0.903124988079071,
+      "rewards/generated": -0.025387108325958252,
+      "rewards/margins": 0.4082257151603699,
+      "rewards/real": 0.3828386664390564,
+      "step": 900
+    },
+    {
+      "epoch": 0.58,
+      "eval_logits/generated": 0.22340521216392517,
+      "eval_logits/real": -0.18402215838432312,
+      "eval_logps/generated": -521.901123046875,
+      "eval_logps/real": -453.87786865234375,
+      "eval_loss": 0.12333940714597702,
+      "eval_rewards/accuracies": 0.9375,
+      "eval_rewards/generated": -0.06012633815407753,
+      "eval_rewards/margins": 0.4457506537437439,
+      "eval_rewards/real": 0.38562434911727905,
+      "eval_runtime": 467.7656,
+      "eval_samples_per_second": 4.276,
+      "eval_steps_per_second": 1.069,
+      "step": 900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.3380782918149466e-07,
+      "logits/generated": 0.23081421852111816,
+      "logits/real": -0.17550137639045715,
+      "logps/generated": -508.828369140625,
+      "logps/real": -436.1839904785156,
+      "loss": 0.1238,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.056590236723423004,
+      "rewards/margins": 0.44170355796813965,
+      "rewards/real": 0.38511329889297485,
+      "step": 905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.3202846975088965e-07,
+      "logits/generated": 0.2366132289171219,
+      "logits/real": -0.2016211748123169,
+      "logps/generated": -530.8549194335938,
+      "logps/real": -485.4046325683594,
+      "loss": 0.1242,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.06151513382792473,
+      "rewards/margins": 0.43632203340530396,
+      "rewards/real": 0.3748069107532501,
+      "step": 910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.302491103202847e-07,
+      "logits/generated": 0.2326919585466385,
+      "logits/real": -0.1740848273038864,
+      "logps/generated": -510.0120544433594,
+      "logps/real": -443.95330810546875,
+      "loss": 0.1235,
+      "rewards/accuracies": 0.953125,
+      "rewards/generated": -0.05962473154067993,
+      "rewards/margins": 0.4408292770385742,
+      "rewards/real": 0.38120460510253906,
+      "step": 915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.284697508896797e-07,
+      "logits/generated": 0.21932096779346466,
+      "logits/real": -0.19047066569328308,
+      "logps/generated": -500.6233825683594,
+      "logps/real": -449.2893981933594,
+      "loss": 0.1234,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.03953690081834793,
+      "rewards/margins": 0.43253955245018005,
+      "rewards/real": 0.3930026888847351,
+      "step": 920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.2669039145907472e-07,
+      "logits/generated": 0.24997837841510773,
+      "logits/real": -0.1812470406293869,
+      "logps/generated": -535.9525146484375,
+      "logps/real": -481.9747009277344,
+      "loss": 0.1234,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.05714315176010132,
+      "rewards/margins": 0.4409334659576416,
+      "rewards/real": 0.3837903141975403,
+      "step": 925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.2491103202846974e-07,
+      "logits/generated": 0.20015868544578552,
+      "logits/real": -0.21435590088367462,
+      "logps/generated": -501.24798583984375,
+      "logps/real": -457.22003173828125,
+      "loss": 0.1258,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.03714684024453163,
+      "rewards/margins": 0.3973380923271179,
+      "rewards/real": 0.3601912558078766,
+      "step": 930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.2313167259786475e-07,
+      "logits/generated": 0.21653036773204803,
+      "logits/real": -0.20086181163787842,
+      "logps/generated": -521.1724853515625,
+      "logps/real": -462.53448486328125,
+      "loss": 0.1239,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.04712541028857231,
+      "rewards/margins": 0.43625393509864807,
+      "rewards/real": 0.3891284763813019,
+      "step": 935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.2135231316725976e-07,
+      "logits/generated": 0.2084265649318695,
+      "logits/real": -0.1898992955684662,
+      "logps/generated": -528.6702880859375,
+      "logps/real": -452.29473876953125,
+      "loss": 0.123,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.05745628476142883,
+      "rewards/margins": 0.4446054995059967,
+      "rewards/real": 0.38714921474456787,
+      "step": 940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.195729537366548e-07,
+      "logits/generated": 0.22444197535514832,
+      "logits/real": -0.16234365105628967,
+      "logps/generated": -524.1942749023438,
+      "logps/real": -446.8998107910156,
+      "loss": 0.1226,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.04743706434965134,
+      "rewards/margins": 0.4473690390586853,
+      "rewards/real": 0.39993205666542053,
+      "step": 945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1779359430604982e-07,
+      "logits/generated": 0.25939130783081055,
+      "logits/real": -0.14666108787059784,
+      "logps/generated": -547.7173461914062,
+      "logps/real": -455.5703125,
+      "loss": 0.1219,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -0.07788354158401489,
+      "rewards/margins": 0.47866588830947876,
+      "rewards/real": 0.4007822871208191,
+      "step": 950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.160142348754448e-07,
+      "logits/generated": 0.2458912432193756,
+      "logits/real": -0.17810478806495667,
+      "logps/generated": -541.607666015625,
+      "logps/real": -476.3172912597656,
+      "loss": 0.1238,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.060118354856967926,
+      "rewards/margins": 0.4450649619102478,
+      "rewards/real": 0.38494664430618286,
+      "step": 955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1423487544483985e-07,
+      "logits/generated": 0.22042623162269592,
+      "logits/real": -0.17981268465518951,
+      "logps/generated": -528.3734130859375,
+      "logps/real": -457.7159729003906,
+      "loss": 0.124,
+      "rewards/accuracies": 0.9156249761581421,
+      "rewards/generated": -0.05706105753779411,
+      "rewards/margins": 0.4376358091831207,
+      "rewards/real": 0.3805747330188751,
+      "step": 960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.1245551601423486e-07,
+      "logits/generated": 0.23842720687389374,
+      "logits/real": -0.18174445629119873,
+      "logps/generated": -519.1848754882812,
+      "logps/real": -455.0909729003906,
+      "loss": 0.1218,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.059157829731702805,
+      "rewards/margins": 0.4687219560146332,
+      "rewards/real": 0.4095641076564789,
+      "step": 965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.1067615658362988e-07,
+      "logits/generated": 0.23483705520629883,
+      "logits/real": -0.1849731206893921,
+      "logps/generated": -523.4705200195312,
+      "logps/real": -474.61761474609375,
+      "loss": 0.1238,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.047459594905376434,
+      "rewards/margins": 0.43443241715431213,
+      "rewards/real": 0.3869728147983551,
+      "step": 970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0889679715302492e-07,
+      "logits/generated": 0.21335165202617645,
+      "logits/real": -0.1933227926492691,
+      "logps/generated": -509.9913635253906,
+      "logps/real": -453.110107421875,
+      "loss": 0.1227,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.04094861447811127,
+      "rewards/margins": 0.4360763430595398,
+      "rewards/real": 0.3951277732849121,
+      "step": 975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.071174377224199e-07,
+      "logits/generated": 0.22970063984394073,
+      "logits/real": -0.16639523208141327,
+      "logps/generated": -530.9884033203125,
+      "logps/real": -476.9608459472656,
+      "loss": 0.1234,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.055165134370326996,
+      "rewards/margins": 0.4414592683315277,
+      "rewards/real": 0.3862941563129425,
+      "step": 980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0533807829181492e-07,
+      "logits/generated": 0.23509879410266876,
+      "logits/real": -0.17442622780799866,
+      "logps/generated": -527.2897338867188,
+      "logps/real": -449.77783203125,
+      "loss": 0.123,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.0715959221124649,
+      "rewards/margins": 0.45488494634628296,
+      "rewards/real": 0.38328900933265686,
+      "step": 985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0355871886120996e-07,
+      "logits/generated": 0.22457948327064514,
+      "logits/real": -0.16395524144172668,
+      "logps/generated": -541.068359375,
+      "logps/real": -445.13232421875,
+      "loss": 0.1224,
+      "rewards/accuracies": 0.953125,
+      "rewards/generated": -0.07380439341068268,
+      "rewards/margins": 0.46186861395835876,
+      "rewards/real": 0.3880642354488373,
+      "step": 990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0177935943060498e-07,
+      "logits/generated": 0.233793705701828,
+      "logits/real": -0.17521928250789642,
+      "logps/generated": -499.760009765625,
+      "logps/real": -418.1783142089844,
+      "loss": 0.1227,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.03161894157528877,
+      "rewards/margins": 0.43319034576416016,
+      "rewards/real": 0.4015713632106781,
+      "step": 995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2e-07,
+      "logits/generated": 0.24164748191833496,
+      "logits/real": -0.17329975962638855,
+      "logps/generated": -509.35968017578125,
+      "logps/real": -438.0674743652344,
+      "loss": 0.1232,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.04283856227993965,
+      "rewards/margins": 0.43587201833724976,
+      "rewards/real": 0.3930334448814392,
+      "step": 1000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.98220640569395e-07,
+      "logits/generated": 0.24707035720348358,
+      "logits/real": -0.19441106915473938,
+      "logps/generated": -523.8375244140625,
+      "logps/real": -466.98876953125,
+      "loss": 0.1239,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.059231944382190704,
+      "rewards/margins": 0.4405438005924225,
+      "rewards/real": 0.38131183385849,
+      "step": 1005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9644128113879002e-07,
+      "logits/generated": 0.21787595748901367,
+      "logits/real": -0.1625303477048874,
+      "logps/generated": -511.28656005859375,
+      "logps/real": -433.72943115234375,
+      "loss": 0.1228,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.04570590704679489,
+      "rewards/margins": 0.4410758912563324,
+      "rewards/real": 0.3953699767589569,
+      "step": 1010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9466192170818503e-07,
+      "logits/generated": 0.2413444072008133,
+      "logits/real": -0.13660605251789093,
+      "logps/generated": -518.0386962890625,
+      "logps/real": -424.8408203125,
+      "loss": 0.1239,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.06294144690036774,
+      "rewards/margins": 0.4464820325374603,
+      "rewards/real": 0.3835405707359314,
+      "step": 1015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9288256227758008e-07,
+      "logits/generated": 0.2317664921283722,
+      "logits/real": -0.17541226744651794,
+      "logps/generated": -501.98565673828125,
+      "logps/real": -448.27301025390625,
+      "loss": 0.1226,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.03287245333194733,
+      "rewards/margins": 0.43534618616104126,
+      "rewards/real": 0.4024737477302551,
+      "step": 1020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.911032028469751e-07,
+      "logits/generated": 0.2325422316789627,
+      "logits/real": -0.15674057602882385,
+      "logps/generated": -521.0519409179688,
+      "logps/real": -445.3016052246094,
+      "loss": 0.1231,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.054769836366176605,
+      "rewards/margins": 0.4464344382286072,
+      "rewards/real": 0.3916645646095276,
+      "step": 1025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8932384341637008e-07,
+      "logits/generated": 0.21268360316753387,
+      "logits/real": -0.2170007973909378,
+      "logps/generated": -517.19140625,
+      "logps/real": -450.4443359375,
+      "loss": 0.1234,
+      "rewards/accuracies": 0.96875,
+      "rewards/generated": -0.05523901432752609,
+      "rewards/margins": 0.442660391330719,
+      "rewards/real": 0.3874213695526123,
+      "step": 1030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8754448398576512e-07,
+      "logits/generated": 0.20065391063690186,
+      "logits/real": -0.18983721733093262,
+      "logps/generated": -516.1802978515625,
+      "logps/real": -441.12176513671875,
+      "loss": 0.1224,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -0.04751238971948624,
+      "rewards/margins": 0.44967880845069885,
+      "rewards/real": 0.4021664261817932,
+      "step": 1035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8576512455516013e-07,
+      "logits/generated": 0.23106245696544647,
+      "logits/real": -0.12498452514410019,
+      "logps/generated": -529.2099609375,
+      "logps/real": -408.1026916503906,
+      "loss": 0.1238,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.05271226167678833,
+      "rewards/margins": 0.4329907298088074,
+      "rewards/real": 0.38027846813201904,
+      "step": 1040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8398576512455515e-07,
+      "logits/generated": 0.2200096845626831,
+      "logits/real": -0.16534483432769775,
+      "logps/generated": -525.7679443359375,
+      "logps/real": -423.7900390625,
+      "loss": 0.1222,
+      "rewards/accuracies": 0.965624988079071,
+      "rewards/generated": -0.07986018806695938,
+      "rewards/margins": 0.4750327467918396,
+      "rewards/real": 0.3951725959777832,
+      "step": 1045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.822064056939502e-07,
+      "logits/generated": 0.21786005795001984,
+      "logits/real": -0.12979988753795624,
+      "logps/generated": -517.2242431640625,
+      "logps/real": -415.1058654785156,
+      "loss": 0.1227,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -0.044034309685230255,
+      "rewards/margins": 0.4487292170524597,
+      "rewards/real": 0.40469488501548767,
+      "step": 1050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8042704626334518e-07,
+      "logits/generated": 0.19789546728134155,
+      "logits/real": -0.18105196952819824,
+      "logps/generated": -516.8419799804688,
+      "logps/real": -411.0154724121094,
+      "loss": 0.1233,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.07462586462497711,
+      "rewards/margins": 0.4560207426548004,
+      "rewards/real": 0.3813949227333069,
+      "step": 1055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.786476868327402e-07,
+      "logits/generated": 0.23341619968414307,
+      "logits/real": -0.16988156735897064,
+      "logps/generated": -524.7384643554688,
+      "logps/real": -435.165771484375,
+      "loss": 0.1241,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.04136037454009056,
+      "rewards/margins": 0.41689807176589966,
+      "rewards/real": 0.3755376935005188,
+      "step": 1060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7686832740213523e-07,
+      "logits/generated": 0.21485109627246857,
+      "logits/real": -0.22060473263263702,
+      "logps/generated": -524.8414916992188,
+      "logps/real": -476.2647399902344,
+      "loss": 0.1232,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -0.04514962434768677,
+      "rewards/margins": 0.43863219022750854,
+      "rewards/real": 0.39348262548446655,
+      "step": 1065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7508896797153025e-07,
+      "logits/generated": 0.2269206941127777,
+      "logits/real": -0.1672908365726471,
+      "logps/generated": -529.927978515625,
+      "logps/real": -418.10711669921875,
+      "loss": 0.1237,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.061279088258743286,
+      "rewards/margins": 0.4455617070198059,
+      "rewards/real": 0.384282648563385,
+      "step": 1070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7330960854092526e-07,
+      "logits/generated": 0.23159179091453552,
+      "logits/real": -0.21682953834533691,
+      "logps/generated": -519.7667236328125,
+      "logps/real": -459.9512634277344,
+      "loss": 0.1242,
+      "rewards/accuracies": 0.903124988079071,
+      "rewards/generated": -0.05122289061546326,
+      "rewards/margins": 0.4344315528869629,
+      "rewards/real": 0.383208692073822,
+      "step": 1075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7153024911032028e-07,
+      "logits/generated": 0.22269535064697266,
+      "logits/real": -0.19031396508216858,
+      "logps/generated": -514.4034423828125,
+      "logps/real": -453.30755615234375,
+      "loss": 0.1233,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.036404937505722046,
+      "rewards/margins": 0.4338456988334656,
+      "rewards/real": 0.3974407911300659,
+      "step": 1080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.697508896797153e-07,
+      "logits/generated": 0.23956215381622314,
+      "logits/real": -0.15907862782478333,
+      "logps/generated": -536.2980346679688,
+      "logps/real": -470.60064697265625,
+      "loss": 0.1221,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.07160317897796631,
+      "rewards/margins": 0.4722219407558441,
+      "rewards/real": 0.4006187319755554,
+      "step": 1085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.679715302491103e-07,
+      "logits/generated": 0.2078259289264679,
+      "logits/real": -0.2215595692396164,
+      "logps/generated": -505.8096618652344,
+      "logps/real": -450.19012451171875,
+      "loss": 0.1237,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.060281794518232346,
+      "rewards/margins": 0.43966370820999146,
+      "rewards/real": 0.3793818950653076,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6619217081850535e-07,
+      "logits/generated": 0.220962256193161,
+      "logits/real": -0.17713703215122223,
+      "logps/generated": -537.6134643554688,
+      "logps/real": -442.96917724609375,
+      "loss": 0.1226,
+      "rewards/accuracies": 0.953125,
+      "rewards/generated": -0.07121272385120392,
+      "rewards/margins": 0.46468058228492737,
+      "rewards/real": 0.39346787333488464,
+      "step": 1095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6441281138790033e-07,
+      "logits/generated": 0.2232249677181244,
+      "logits/real": -0.23343662917613983,
+      "logps/generated": -530.177978515625,
+      "logps/real": -474.0650939941406,
+      "loss": 0.1232,
+      "rewards/accuracies": 0.953125,
+      "rewards/generated": -0.07620830088853836,
+      "rewards/margins": 0.4571600556373596,
+      "rewards/real": 0.38095176219940186,
+      "step": 1100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6263345195729535e-07,
+      "logits/generated": 0.22311291098594666,
+      "logits/real": -0.1836051195859909,
+      "logps/generated": -513.8787841796875,
+      "logps/real": -432.8919982910156,
+      "loss": 0.1223,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.05465202406048775,
+      "rewards/margins": 0.45078739523887634,
+      "rewards/real": 0.3961353898048401,
+      "step": 1105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.608540925266904e-07,
+      "logits/generated": 0.20984773337841034,
+      "logits/real": -0.2067348212003708,
+      "logps/generated": -502.0240783691406,
+      "logps/real": -450.7012634277344,
+      "loss": 0.1232,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.053631316870450974,
+      "rewards/margins": 0.4417099952697754,
+      "rewards/real": 0.3880787193775177,
+      "step": 1110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.590747330960854e-07,
+      "logits/generated": 0.2537307143211365,
+      "logits/real": -0.15498527884483337,
+      "logps/generated": -517.6680908203125,
+      "logps/real": -448.23541259765625,
+      "loss": 0.1233,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.033959295600652695,
+      "rewards/margins": 0.42226237058639526,
+      "rewards/real": 0.38830310106277466,
+      "step": 1115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5729537366548042e-07,
+      "logits/generated": 0.21758441627025604,
+      "logits/real": -0.1972777396440506,
+      "logps/generated": -523.7211303710938,
+      "logps/real": -442.3697814941406,
+      "loss": 0.1221,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -0.05707954615354538,
+      "rewards/margins": 0.46116429567337036,
+      "rewards/real": 0.4040847420692444,
+      "step": 1120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5551601423487543e-07,
+      "logits/generated": 0.2385714054107666,
+      "logits/real": -0.15557484328746796,
+      "logps/generated": -534.7314453125,
+      "logps/real": -459.919189453125,
+      "loss": 0.122,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.061738379299640656,
+      "rewards/margins": 0.46245869994163513,
+      "rewards/real": 0.40072035789489746,
+      "step": 1125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5373665480427045e-07,
+      "logits/generated": 0.22526773810386658,
+      "logits/real": -0.23086409270763397,
+      "logps/generated": -515.3248291015625,
+      "logps/real": -480.4952697753906,
+      "loss": 0.1237,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/generated": -0.047265470027923584,
+      "rewards/margins": 0.4348309636116028,
+      "rewards/real": 0.3875654339790344,
+      "step": 1130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5195729537366546e-07,
+      "logits/generated": 0.2394159585237503,
+      "logits/real": -0.1480354219675064,
+      "logps/generated": -539.8045654296875,
+      "logps/real": -449.03851318359375,
+      "loss": 0.1219,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -0.06060241535305977,
+      "rewards/margins": 0.46144571900367737,
+      "rewards/real": 0.40084323287010193,
+      "step": 1135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.501779359430605e-07,
+      "logits/generated": 0.20694789290428162,
+      "logits/real": -0.1889125555753708,
+      "logps/generated": -497.44586181640625,
+      "logps/real": -446.2994079589844,
+      "loss": 0.1233,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.05379456281661987,
+      "rewards/margins": 0.4522787928581238,
+      "rewards/real": 0.3984842300415039,
+      "step": 1140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4839857651245552e-07,
+      "logits/generated": 0.25509268045425415,
+      "logits/real": -0.1534079909324646,
+      "logps/generated": -549.9916381835938,
+      "logps/real": -462.14581298828125,
+      "loss": 0.1219,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -0.07747825235128403,
+      "rewards/margins": 0.47670111060142517,
+      "rewards/real": 0.39922288060188293,
+      "step": 1145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.466192170818505e-07,
+      "logits/generated": 0.22593429684638977,
+      "logits/real": -0.13438035547733307,
+      "logps/generated": -525.7147216796875,
+      "logps/real": -415.84014892578125,
+      "loss": 0.1229,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.07066598534584045,
+      "rewards/margins": 0.45950788259506226,
+      "rewards/real": 0.3888419568538666,
+      "step": 1150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4483985765124555e-07,
+      "logits/generated": 0.24762868881225586,
+      "logits/real": -0.15819518268108368,
+      "logps/generated": -552.5275268554688,
+      "logps/real": -452.3013610839844,
+      "loss": 0.1225,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.05504069849848747,
+      "rewards/margins": 0.45216450095176697,
+      "rewards/real": 0.3971238136291504,
+      "step": 1155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4306049822064056e-07,
+      "logits/generated": 0.2369476854801178,
+      "logits/real": -0.14361664652824402,
+      "logps/generated": -522.8367919921875,
+      "logps/real": -437.43280029296875,
+      "loss": 0.1234,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.04768730327486992,
+      "rewards/margins": 0.44329744577407837,
+      "rewards/real": 0.39561015367507935,
+      "step": 1160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4128113879003557e-07,
+      "logits/generated": 0.21606259047985077,
+      "logits/real": -0.17811116576194763,
+      "logps/generated": -521.3794555664062,
+      "logps/real": -449.33306884765625,
+      "loss": 0.1223,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.06894604861736298,
+      "rewards/margins": 0.4709966778755188,
+      "rewards/real": 0.402050644159317,
+      "step": 1165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3950177935943062e-07,
+      "logits/generated": 0.2552662491798401,
+      "logits/real": -0.16689833998680115,
+      "logps/generated": -517.5098876953125,
+      "logps/real": -453.36199951171875,
+      "loss": 0.1235,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.050110749900341034,
+      "rewards/margins": 0.43731746077537537,
+      "rewards/real": 0.38720670342445374,
+      "step": 1170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.377224199288256e-07,
+      "logits/generated": 0.19484691321849823,
+      "logits/real": -0.15271511673927307,
+      "logps/generated": -537.5327758789062,
+      "logps/real": -430.572509765625,
+      "loss": 0.1225,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.07170198112726212,
+      "rewards/margins": 0.478828489780426,
+      "rewards/real": 0.4071265161037445,
+      "step": 1175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3594306049822062e-07,
+      "logits/generated": 0.21546511352062225,
+      "logits/real": -0.1496323198080063,
+      "logps/generated": -523.5759887695312,
+      "logps/real": -446.5785217285156,
+      "loss": 0.122,
+      "rewards/accuracies": 0.9593750238418579,
+      "rewards/generated": -0.07428407669067383,
+      "rewards/margins": 0.4780218005180359,
+      "rewards/real": 0.40373772382736206,
+      "step": 1180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3416370106761566e-07,
+      "logits/generated": 0.25350579619407654,
+      "logits/real": -0.1811196357011795,
+      "logps/generated": -509.49334716796875,
+      "logps/real": -469.2283630371094,
+      "loss": 0.1231,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.04726102203130722,
+      "rewards/margins": 0.44062432646751404,
+      "rewards/real": 0.3933633267879486,
+      "step": 1185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3238434163701067e-07,
+      "logits/generated": 0.19710242748260498,
+      "logits/real": -0.195444256067276,
+      "logps/generated": -510.51953125,
+      "logps/real": -427.2699279785156,
+      "loss": 0.1235,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.052245985716581345,
+      "rewards/margins": 0.44260525703430176,
+      "rewards/real": 0.3903593122959137,
+      "step": 1190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.306049822064057e-07,
+      "logits/generated": 0.23108847439289093,
+      "logits/real": -0.19040106236934662,
+      "logps/generated": -526.1116943359375,
+      "logps/real": -457.45245361328125,
+      "loss": 0.1205,
+      "rewards/accuracies": 0.96875,
+      "rewards/generated": -0.07810448855161667,
+      "rewards/margins": 0.499803364276886,
+      "rewards/real": 0.4216988682746887,
+      "step": 1195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.288256227758007e-07,
+      "logits/generated": 0.23808474838733673,
+      "logits/real": -0.20193028450012207,
+      "logps/generated": -532.6949462890625,
+      "logps/real": -462.358642578125,
+      "loss": 0.1232,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.05180268734693527,
+      "rewards/margins": 0.4387210011482239,
+      "rewards/real": 0.3869183659553528,
+      "step": 1200
+    },
+    {
+      "epoch": 0.77,
+      "eval_logits/generated": 0.2313157618045807,
+      "eval_logits/real": -0.1777036488056183,
+      "eval_logps/generated": -521.87158203125,
+      "eval_logps/real": -453.7703857421875,
+      "eval_loss": 0.12271205335855484,
+      "eval_rewards/accuracies": 0.9415000081062317,
+      "eval_rewards/generated": -0.05717482790350914,
+      "eval_rewards/margins": 0.45354583859443665,
+      "eval_rewards/real": 0.3963710367679596,
+      "eval_runtime": 467.21,
+      "eval_samples_per_second": 4.281,
+      "eval_steps_per_second": 1.07,
+      "step": 1200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2704626334519572e-07,
+      "logits/generated": 0.22476062178611755,
+      "logits/real": -0.17389290034770966,
+      "logps/generated": -521.6498413085938,
+      "logps/real": -443.27880859375,
+      "loss": 0.1235,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.0522431917488575,
+      "rewards/margins": 0.4405241906642914,
+      "rewards/real": 0.3882810175418854,
+      "step": 1205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2526690391459073e-07,
+      "logits/generated": 0.23659996688365936,
+      "logits/real": -0.19962027668952942,
+      "logps/generated": -530.6138305664062,
+      "logps/real": -465.6412658691406,
+      "loss": 0.1223,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.08992975950241089,
+      "rewards/margins": 0.48291119933128357,
+      "rewards/real": 0.3929814398288727,
+      "step": 1210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2348754448398575e-07,
+      "logits/generated": 0.24056227505207062,
+      "logits/real": -0.17281261086463928,
+      "logps/generated": -519.2171630859375,
+      "logps/real": -464.92913818359375,
+      "loss": 0.1233,
+      "rewards/accuracies": 0.903124988079071,
+      "rewards/generated": -0.04756608605384827,
+      "rewards/margins": 0.43203601241111755,
+      "rewards/real": 0.3844699263572693,
+      "step": 1215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.217081850533808e-07,
+      "logits/generated": 0.23706042766571045,
+      "logits/real": -0.1741897165775299,
+      "logps/generated": -483.63140869140625,
+      "logps/real": -462.0602111816406,
+      "loss": 0.1228,
+      "rewards/accuracies": 0.953125,
+      "rewards/generated": -0.035449109971523285,
+      "rewards/margins": 0.4352191388607025,
+      "rewards/real": 0.39976999163627625,
+      "step": 1220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.199288256227758e-07,
+      "logits/generated": 0.23114576935768127,
+      "logits/real": -0.14803336560726166,
+      "logps/generated": -507.52008056640625,
+      "logps/real": -441.3143615722656,
+      "loss": 0.1225,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.04358341172337532,
+      "rewards/margins": 0.4518836438655853,
+      "rewards/real": 0.4083002209663391,
+      "step": 1225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1814946619217082e-07,
+      "logits/generated": 0.23625341057777405,
+      "logits/real": -0.22369003295898438,
+      "logps/generated": -506.97796630859375,
+      "logps/real": -472.6698303222656,
+      "loss": 0.1229,
+      "rewards/accuracies": 0.953125,
+      "rewards/generated": -0.05624117702245712,
+      "rewards/margins": 0.44402360916137695,
+      "rewards/real": 0.3877824544906616,
+      "step": 1230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1637010676156583e-07,
+      "logits/generated": 0.22912919521331787,
+      "logits/real": -0.18156306445598602,
+      "logps/generated": -509.318359375,
+      "logps/real": -449.29425048828125,
+      "loss": 0.1234,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.03210196644067764,
+      "rewards/margins": 0.4248197078704834,
+      "rewards/real": 0.39271771907806396,
+      "step": 1235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1459074733096084e-07,
+      "logits/generated": 0.19215625524520874,
+      "logits/real": -0.18478159606456757,
+      "logps/generated": -532.9630737304688,
+      "logps/real": -453.92376708984375,
+      "loss": 0.1221,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.048404209315776825,
+      "rewards/margins": 0.45223093032836914,
+      "rewards/real": 0.4038267135620117,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1281138790035587e-07,
+      "logits/generated": 0.2604306638240814,
+      "logits/real": -0.16049329936504364,
+      "logps/generated": -514.5932006835938,
+      "logps/real": -452.393310546875,
+      "loss": 0.1231,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.03789561614394188,
+      "rewards/margins": 0.4359763562679291,
+      "rewards/real": 0.3980807363986969,
+      "step": 1245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1103202846975089e-07,
+      "logits/generated": 0.2561563551425934,
+      "logits/real": -0.18810555338859558,
+      "logps/generated": -523.1640625,
+      "logps/real": -469.3626403808594,
+      "loss": 0.1232,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.047877997159957886,
+      "rewards/margins": 0.4385640025138855,
+      "rewards/real": 0.3906860947608948,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.092526690391459e-07,
+      "logits/generated": 0.22691071033477783,
+      "logits/real": -0.16963443160057068,
+      "logps/generated": -533.8391723632812,
+      "logps/real": -441.6620178222656,
+      "loss": 0.1223,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.06845617294311523,
+      "rewards/margins": 0.4655072093009949,
+      "rewards/real": 0.39705103635787964,
+      "step": 1255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0747330960854093e-07,
+      "logits/generated": 0.23466971516609192,
+      "logits/real": -0.1737900674343109,
+      "logps/generated": -523.0357666015625,
+      "logps/real": -466.3274841308594,
+      "loss": 0.1227,
+      "rewards/accuracies": 0.9156249761581421,
+      "rewards/generated": -0.03422842174768448,
+      "rewards/margins": 0.43477416038513184,
+      "rewards/real": 0.40054574608802795,
+      "step": 1260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0569395017793593e-07,
+      "logits/generated": 0.23214109241962433,
+      "logits/real": -0.2207547128200531,
+      "logps/generated": -523.21630859375,
+      "logps/real": -460.16033935546875,
+      "loss": 0.1221,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -0.05244157463312149,
+      "rewards/margins": 0.4536016881465912,
+      "rewards/real": 0.4011601507663727,
+      "step": 1265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0391459074733096e-07,
+      "logits/generated": 0.22818417847156525,
+      "logits/real": -0.18806496262550354,
+      "logps/generated": -534.0014038085938,
+      "logps/real": -461.1249084472656,
+      "loss": 0.1223,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.08126085996627808,
+      "rewards/margins": 0.4781245291233063,
+      "rewards/real": 0.3968636393547058,
+      "step": 1270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0213523131672597e-07,
+      "logits/generated": 0.22448763251304626,
+      "logits/real": -0.19211895763874054,
+      "logps/generated": -502.9695739746094,
+      "logps/real": -461.84539794921875,
+      "loss": 0.1221,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.04918776452541351,
+      "rewards/margins": 0.4540463387966156,
+      "rewards/real": 0.4048585891723633,
+      "step": 1275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0035587188612099e-07,
+      "logits/generated": 0.2139461785554886,
+      "logits/real": -0.1837858110666275,
+      "logps/generated": -507.08953857421875,
+      "logps/real": -462.7340393066406,
+      "loss": 0.1241,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.04183631390333176,
+      "rewards/margins": 0.42796725034713745,
+      "rewards/real": 0.3861309885978699,
+      "step": 1280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.857651245551602e-08,
+      "logits/generated": 0.2099412977695465,
+      "logits/real": -0.2005094587802887,
+      "logps/generated": -509.3089294433594,
+      "logps/real": -458.82098388671875,
+      "loss": 0.1235,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.0423799529671669,
+      "rewards/margins": 0.4318648874759674,
+      "rewards/real": 0.3894849121570587,
+      "step": 1285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.679715302491103e-08,
+      "logits/generated": 0.24018998444080353,
+      "logits/real": -0.16420888900756836,
+      "logps/generated": -515.2698974609375,
+      "logps/real": -465.8897399902344,
+      "loss": 0.1232,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.04603344202041626,
+      "rewards/margins": 0.4456867277622223,
+      "rewards/real": 0.3996533453464508,
+      "step": 1290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.501779359430604e-08,
+      "logits/generated": 0.2142796516418457,
+      "logits/real": -0.23481447994709015,
+      "logps/generated": -517.4632568359375,
+      "logps/real": -457.7532653808594,
+      "loss": 0.1224,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.06851876527070999,
+      "rewards/margins": 0.46548646688461304,
+      "rewards/real": 0.39696773886680603,
+      "step": 1295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.323843416370106e-08,
+      "logits/generated": 0.21380500495433807,
+      "logits/real": -0.1961486041545868,
+      "logps/generated": -520.6165771484375,
+      "logps/real": -449.50152587890625,
+      "loss": 0.1235,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.057625479996204376,
+      "rewards/margins": 0.4460286498069763,
+      "rewards/real": 0.3884032368659973,
+      "step": 1300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.145907473309609e-08,
+      "logits/generated": 0.23146884143352509,
+      "logits/real": -0.1590948849916458,
+      "logps/generated": -525.8096923828125,
+      "logps/real": -447.4246520996094,
+      "loss": 0.1241,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.06719191372394562,
+      "rewards/margins": 0.43803900480270386,
+      "rewards/real": 0.3708471357822418,
+      "step": 1305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.96797153024911e-08,
+      "logits/generated": 0.24106153845787048,
+      "logits/real": -0.17956070601940155,
+      "logps/generated": -509.88909912109375,
+      "logps/real": -451.45361328125,
+      "loss": 0.1215,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/generated": -0.053158439695835114,
+      "rewards/margins": 0.4641904830932617,
+      "rewards/real": 0.411032110452652,
+      "step": 1310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.790035587188612e-08,
+      "logits/generated": 0.23294071853160858,
+      "logits/real": -0.17599083483219147,
+      "logps/generated": -541.8015747070312,
+      "logps/real": -449.78802490234375,
+      "loss": 0.121,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.06215289235115051,
+      "rewards/margins": 0.4786357283592224,
+      "rewards/real": 0.4164828360080719,
+      "step": 1315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.612099644128114e-08,
+      "logits/generated": 0.22666697204113007,
+      "logits/real": -0.17323556542396545,
+      "logps/generated": -543.7659912109375,
+      "logps/real": -444.4297790527344,
+      "loss": 0.1237,
+      "rewards/accuracies": 0.934374988079071,
+      "rewards/generated": -0.07400919497013092,
+      "rewards/margins": 0.4550935626029968,
+      "rewards/real": 0.3810843825340271,
+      "step": 1320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.434163701067614e-08,
+      "logits/generated": 0.21421749889850616,
+      "logits/real": -0.2157014161348343,
+      "logps/generated": -510.8021545410156,
+      "logps/real": -463.74285888671875,
+      "loss": 0.1226,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.055482130497694016,
+      "rewards/margins": 0.45598897337913513,
+      "rewards/real": 0.4005068242549896,
+      "step": 1325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.256227758007117e-08,
+      "logits/generated": 0.2465168535709381,
+      "logits/real": -0.19514866173267365,
+      "logps/generated": -503.04339599609375,
+      "logps/real": -482.16339111328125,
+      "loss": 0.1231,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.05232943966984749,
+      "rewards/margins": 0.4523092210292816,
+      "rewards/real": 0.3999797999858856,
+      "step": 1330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.07829181494662e-08,
+      "logits/generated": 0.2046203315258026,
+      "logits/real": -0.22382350265979767,
+      "logps/generated": -516.7283935546875,
+      "logps/real": -462.3814392089844,
+      "loss": 0.1235,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.061030227690935135,
+      "rewards/margins": 0.4473925530910492,
+      "rewards/real": 0.3863622844219208,
+      "step": 1335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.90035587188612e-08,
+      "logits/generated": 0.23818406462669373,
+      "logits/real": -0.13241663575172424,
+      "logps/generated": -552.3175659179688,
+      "logps/real": -429.982666015625,
+      "loss": 0.1232,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.0787079781293869,
+      "rewards/margins": 0.46521496772766113,
+      "rewards/real": 0.38650697469711304,
+      "step": 1340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.722419928825623e-08,
+      "logits/generated": 0.22872892022132874,
+      "logits/real": -0.1680324524641037,
+      "logps/generated": -528.6109619140625,
+      "logps/real": -445.991455078125,
+      "loss": 0.1222,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.0603509321808815,
+      "rewards/margins": 0.462471067905426,
+      "rewards/real": 0.40212011337280273,
+      "step": 1345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.544483985765124e-08,
+      "logits/generated": 0.24162690341472626,
+      "logits/real": -0.16354595124721527,
+      "logps/generated": -515.4553833007812,
+      "logps/real": -447.14892578125,
+      "loss": 0.1227,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.04346945881843567,
+      "rewards/margins": 0.4477364122867584,
+      "rewards/real": 0.40426692366600037,
+      "step": 1350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.366548042704626e-08,
+      "logits/generated": 0.21535074710845947,
+      "logits/real": -0.19052883982658386,
+      "logps/generated": -536.5697021484375,
+      "logps/real": -434.23016357421875,
+      "loss": 0.1224,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.05363154411315918,
+      "rewards/margins": 0.4495977461338043,
+      "rewards/real": 0.39596620202064514,
+      "step": 1355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.188612099644127e-08,
+      "logits/generated": 0.23940858244895935,
+      "logits/real": -0.153304785490036,
+      "logps/generated": -519.3319702148438,
+      "logps/real": -430.232666015625,
+      "loss": 0.1217,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.07525327801704407,
+      "rewards/margins": 0.48004379868507385,
+      "rewards/real": 0.4047905504703522,
+      "step": 1360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.01067615658363e-08,
+      "logits/generated": 0.26696276664733887,
+      "logits/real": -0.1220448836684227,
+      "logps/generated": -520.37060546875,
+      "logps/real": -437.6148376464844,
+      "loss": 0.1223,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.03939700126647949,
+      "rewards/margins": 0.44409412145614624,
+      "rewards/real": 0.4046971797943115,
+      "step": 1365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.832740213523131e-08,
+      "logits/generated": 0.26054686307907104,
+      "logits/real": -0.16391593217849731,
+      "logps/generated": -517.5562744140625,
+      "logps/real": -444.6243591308594,
+      "loss": 0.1236,
+      "rewards/accuracies": 0.909375011920929,
+      "rewards/generated": -0.03623567894101143,
+      "rewards/margins": 0.4285515248775482,
+      "rewards/real": 0.3923158347606659,
+      "step": 1370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.654804270462633e-08,
+      "logits/generated": 0.19792599976062775,
+      "logits/real": -0.18831519782543182,
+      "logps/generated": -526.7872924804688,
+      "logps/real": -425.1585388183594,
+      "loss": 0.1235,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -0.060060203075408936,
+      "rewards/margins": 0.45477867126464844,
+      "rewards/real": 0.3947184681892395,
+      "step": 1375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.476868327402136e-08,
+      "logits/generated": 0.25231215357780457,
+      "logits/real": -0.12078098207712173,
+      "logps/generated": -523.71435546875,
+      "logps/real": -404.197509765625,
+      "loss": 0.1216,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.055614911019802094,
+      "rewards/margins": 0.4722030758857727,
+      "rewards/real": 0.4165882170200348,
+      "step": 1380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.298932384341636e-08,
+      "logits/generated": 0.22361817955970764,
+      "logits/real": -0.1951177716255188,
+      "logps/generated": -527.4632568359375,
+      "logps/real": -449.739990234375,
+      "loss": 0.123,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -0.04769791662693024,
+      "rewards/margins": 0.44549164175987244,
+      "rewards/real": 0.397793710231781,
+      "step": 1385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.120996441281139e-08,
+      "logits/generated": 0.235061377286911,
+      "logits/real": -0.19142548739910126,
+      "logps/generated": -500.3617248535156,
+      "logps/real": -432.40045166015625,
+      "loss": 0.123,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.03681398928165436,
+      "rewards/margins": 0.4336830675601959,
+      "rewards/real": 0.396869033575058,
+      "step": 1390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.94306049822064e-08,
+      "logits/generated": 0.23789739608764648,
+      "logits/real": -0.20226213335990906,
+      "logps/generated": -524.2302856445312,
+      "logps/real": -473.21124267578125,
+      "loss": 0.1247,
+      "rewards/accuracies": 0.9156249761581421,
+      "rewards/generated": -0.04299360513687134,
+      "rewards/margins": 0.41463667154312134,
+      "rewards/real": 0.3716430068016052,
+      "step": 1395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.765124555160142e-08,
+      "logits/generated": 0.209858700633049,
+      "logits/real": -0.17679789662361145,
+      "logps/generated": -527.78076171875,
+      "logps/real": -436.3880310058594,
+      "loss": 0.1209,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.07421739399433136,
+      "rewards/margins": 0.49320459365844727,
+      "rewards/real": 0.41898712515830994,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.587188612099644e-08,
+      "logits/generated": 0.22804374992847443,
+      "logits/real": -0.17837099730968475,
+      "logps/generated": -531.051513671875,
+      "logps/real": -454.724609375,
+      "loss": 0.1225,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/generated": -0.0480850450694561,
+      "rewards/margins": 0.4538131356239319,
+      "rewards/real": 0.4057280421257019,
+      "step": 1405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4092526690391456e-08,
+      "logits/generated": 0.23690180480480194,
+      "logits/real": -0.19386790692806244,
+      "logps/generated": -530.9876098632812,
+      "logps/real": -470.317138671875,
+      "loss": 0.1228,
+      "rewards/accuracies": 0.953125,
+      "rewards/generated": -0.055680133402347565,
+      "rewards/margins": 0.45267653465270996,
+      "rewards/real": 0.3969964385032654,
+      "step": 1410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.231316725978648e-08,
+      "logits/generated": 0.2339600771665573,
+      "logits/real": -0.10314682871103287,
+      "logps/generated": -542.47265625,
+      "logps/real": -420.5494079589844,
+      "loss": 0.1208,
+      "rewards/accuracies": 0.96875,
+      "rewards/generated": -0.0906255766749382,
+      "rewards/margins": 0.5058237910270691,
+      "rewards/real": 0.4151982367038727,
+      "step": 1415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.053380782918149e-08,
+      "logits/generated": 0.24456700682640076,
+      "logits/real": -0.1966424584388733,
+      "logps/generated": -538.0858154296875,
+      "logps/real": -469.58837890625,
+      "loss": 0.1224,
+      "rewards/accuracies": 0.9312499761581421,
+      "rewards/generated": -0.06548374891281128,
+      "rewards/margins": 0.4630776345729828,
+      "rewards/real": 0.3975939154624939,
+      "step": 1420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.8754448398576507e-08,
+      "logits/generated": 0.23372788727283478,
+      "logits/real": -0.18431547284126282,
+      "logps/generated": -533.2066650390625,
+      "logps/real": -444.8937072753906,
+      "loss": 0.1234,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.05459724739193916,
+      "rewards/margins": 0.4527161717414856,
+      "rewards/real": 0.39811891317367554,
+      "step": 1425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.697508896797153e-08,
+      "logits/generated": 0.23714499175548553,
+      "logits/real": -0.1850479245185852,
+      "logps/generated": -504.242431640625,
+      "logps/real": -466.57415771484375,
+      "loss": 0.1241,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -0.038482390344142914,
+      "rewards/margins": 0.418579638004303,
+      "rewards/real": 0.38009724020957947,
+      "step": 1430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.519572953736655e-08,
+      "logits/generated": 0.21894851326942444,
+      "logits/real": -0.28000861406326294,
+      "logps/generated": -500.18115234375,
+      "logps/real": -503.328369140625,
+      "loss": 0.1218,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.05481521040201187,
+      "rewards/margins": 0.4618280529975891,
+      "rewards/real": 0.40701285004615784,
+      "step": 1435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.341637010676156e-08,
+      "logits/generated": 0.23817463219165802,
+      "logits/real": -0.12684933841228485,
+      "logps/generated": -529.4266357421875,
+      "logps/real": -412.86907958984375,
+      "loss": 0.1207,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -0.05582699924707413,
+      "rewards/margins": 0.4807185232639313,
+      "rewards/real": 0.42489147186279297,
+      "step": 1440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.1637010676156584e-08,
+      "logits/generated": 0.244916170835495,
+      "logits/real": -0.18898700177669525,
+      "logps/generated": -527.6209716796875,
+      "logps/real": -472.571533203125,
+      "loss": 0.1209,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.04574694111943245,
+      "rewards/margins": 0.4711214601993561,
+      "rewards/real": 0.42537444829940796,
+      "step": 1445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.98576512455516e-08,
+      "logits/generated": 0.209895521402359,
+      "logits/real": -0.13670508563518524,
+      "logps/generated": -502.51922607421875,
+      "logps/real": -424.85418701171875,
+      "loss": 0.1238,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.02930234745144844,
+      "rewards/margins": 0.4223732054233551,
+      "rewards/real": 0.39307087659835815,
+      "step": 1450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.807829181494661e-08,
+      "logits/generated": 0.21665044128894806,
+      "logits/real": -0.1499323546886444,
+      "logps/generated": -528.1993408203125,
+      "logps/real": -411.34130859375,
+      "loss": 0.1224,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.053298868238925934,
+      "rewards/margins": 0.46004271507263184,
+      "rewards/real": 0.4067438244819641,
+      "step": 1455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6298932384341634e-08,
+      "logits/generated": 0.22750067710876465,
+      "logits/real": -0.18200306594371796,
+      "logps/generated": -515.8609008789062,
+      "logps/real": -437.69927978515625,
+      "loss": 0.1222,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.037331800907850266,
+      "rewards/margins": 0.4469398558139801,
+      "rewards/real": 0.40960803627967834,
+      "step": 1460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4519572953736656e-08,
+      "logits/generated": 0.22528275847434998,
+      "logits/real": -0.201947420835495,
+      "logps/generated": -522.4163208007812,
+      "logps/real": -452.88427734375,
+      "loss": 0.1233,
+      "rewards/accuracies": 0.9468749761581421,
+      "rewards/generated": -0.05317317321896553,
+      "rewards/margins": 0.4387715458869934,
+      "rewards/real": 0.385598361492157,
+      "step": 1465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.274021352313167e-08,
+      "logits/generated": 0.24460799992084503,
+      "logits/real": -0.15885956585407257,
+      "logps/generated": -528.5340576171875,
+      "logps/real": -438.3102111816406,
+      "loss": 0.1231,
+      "rewards/accuracies": 0.9156249761581421,
+      "rewards/generated": -0.056962721049785614,
+      "rewards/margins": 0.44773292541503906,
+      "rewards/real": 0.39077019691467285,
+      "step": 1470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.096085409252669e-08,
+      "logits/generated": 0.23852023482322693,
+      "logits/real": -0.17653930187225342,
+      "logps/generated": -513.607666015625,
+      "logps/real": -450.1045837402344,
+      "loss": 0.122,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.04695012420415878,
+      "rewards/margins": 0.45696502923965454,
+      "rewards/real": 0.41001492738723755,
+      "step": 1475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9181494661921706e-08,
+      "logits/generated": 0.2057987004518509,
+      "logits/real": -0.19016572833061218,
+      "logps/generated": -544.1546630859375,
+      "logps/real": -443.2815856933594,
+      "loss": 0.1224,
+      "rewards/accuracies": 0.9281250238418579,
+      "rewards/generated": -0.04910854250192642,
+      "rewards/margins": 0.4497135579586029,
+      "rewards/real": 0.4006050229072571,
+      "step": 1480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7402135231316727e-08,
+      "logits/generated": 0.2639480233192444,
+      "logits/real": -0.13512758910655975,
+      "logps/generated": -537.3682861328125,
+      "logps/real": -450.20721435546875,
+      "loss": 0.1217,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -0.07757305353879929,
+      "rewards/margins": 0.4825238287448883,
+      "rewards/real": 0.4049507677555084,
+      "step": 1485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.562277580071174e-08,
+      "logits/generated": 0.22935017943382263,
+      "logits/real": -0.15518589317798615,
+      "logps/generated": -541.0477294921875,
+      "logps/real": -439.45721435546875,
+      "loss": 0.1228,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.06261979043483734,
+      "rewards/margins": 0.4490610957145691,
+      "rewards/real": 0.38644129037857056,
+      "step": 1490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.384341637010676e-08,
+      "logits/generated": 0.2474227398633957,
+      "logits/real": -0.23404696583747864,
+      "logps/generated": -522.2889404296875,
+      "logps/real": -471.97772216796875,
+      "loss": 0.1235,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.05089433863759041,
+      "rewards/margins": 0.43490394949913025,
+      "rewards/real": 0.38400956988334656,
+      "step": 1495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.206405693950178e-08,
+      "logits/generated": 0.22496287524700165,
+      "logits/real": -0.1576504409313202,
+      "logps/generated": -545.5980834960938,
+      "logps/real": -434.0711975097656,
+      "loss": 0.1226,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.08410739153623581,
+      "rewards/margins": 0.47313404083251953,
+      "rewards/real": 0.3890266716480255,
+      "step": 1500
+    },
+    {
+      "epoch": 0.96,
+      "eval_logits/generated": 0.227307990193367,
+      "eval_logits/real": -0.18104109168052673,
+      "eval_logps/generated": -521.93017578125,
+      "eval_logps/real": -453.7297668457031,
+      "eval_loss": 0.12231329083442688,
+      "eval_rewards/accuracies": 0.9465000033378601,
+      "eval_rewards/generated": -0.0630280077457428,
+      "eval_rewards/margins": 0.4634554088115692,
+      "eval_rewards/real": 0.4004274308681488,
+      "eval_runtime": 463.6445,
+      "eval_samples_per_second": 4.314,
+      "eval_steps_per_second": 1.078,
+      "step": 1500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0284697508896795e-08,
+      "logits/generated": 0.2287018746137619,
+      "logits/real": -0.1820969432592392,
+      "logps/generated": -505.2705078125,
+      "logps/real": -460.46954345703125,
+      "loss": 0.1222,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.05245542526245117,
+      "rewards/margins": 0.46576395630836487,
+      "rewards/real": 0.4133085310459137,
+      "step": 1505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8505338078291812e-08,
+      "logits/generated": 0.23314492404460907,
+      "logits/real": -0.18071797490119934,
+      "logps/generated": -516.2825317382812,
+      "logps/real": -446.4068908691406,
+      "loss": 0.1226,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.06283563375473022,
+      "rewards/margins": 0.463799387216568,
+      "rewards/real": 0.40096378326416016,
+      "step": 1510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6725978647686833e-08,
+      "logits/generated": 0.23628249764442444,
+      "logits/real": -0.19212110340595245,
+      "logps/generated": -526.8966064453125,
+      "logps/real": -465.02606201171875,
+      "loss": 0.1221,
+      "rewards/accuracies": 0.953125,
+      "rewards/generated": -0.06296736001968384,
+      "rewards/margins": 0.4659939706325531,
+      "rewards/real": 0.4030265808105469,
+      "step": 1515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.494661921708185e-08,
+      "logits/generated": 0.22333106398582458,
+      "logits/real": -0.1808868944644928,
+      "logps/generated": -537.9901123046875,
+      "logps/real": -449.5810546875,
+      "loss": 0.1213,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.03545105457305908,
+      "rewards/margins": 0.45481738448143005,
+      "rewards/real": 0.41936635971069336,
+      "step": 1520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3167259786476867e-08,
+      "logits/generated": 0.203715518116951,
+      "logits/real": -0.20505018532276154,
+      "logps/generated": -514.4688720703125,
+      "logps/real": -435.12103271484375,
+      "loss": 0.124,
+      "rewards/accuracies": 0.921875,
+      "rewards/generated": -0.062085770070552826,
+      "rewards/margins": 0.444580078125,
+      "rewards/real": 0.3824942708015442,
+      "step": 1525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1387900355871887e-08,
+      "logits/generated": 0.24170592427253723,
+      "logits/real": -0.15465418994426727,
+      "logps/generated": -538.7681884765625,
+      "logps/real": -460.85333251953125,
+      "loss": 0.1218,
+      "rewards/accuracies": 0.9593750238418579,
+      "rewards/generated": -0.0660967007279396,
+      "rewards/margins": 0.47000962495803833,
+      "rewards/real": 0.4039129316806793,
+      "step": 1530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.608540925266903e-09,
+      "logits/generated": 0.2076333463191986,
+      "logits/real": -0.25217288732528687,
+      "logps/generated": -501.05328369140625,
+      "logps/real": -470.36285400390625,
+      "loss": 0.1225,
+      "rewards/accuracies": 0.956250011920929,
+      "rewards/generated": -0.05864990875124931,
+      "rewards/margins": 0.461745023727417,
+      "rewards/real": 0.4030950963497162,
+      "step": 1535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.82918149466192e-09,
+      "logits/generated": 0.21803216636180878,
+      "logits/real": -0.22686973214149475,
+      "logps/generated": -513.02392578125,
+      "logps/real": -461.0308532714844,
+      "loss": 0.1217,
+      "rewards/accuracies": 0.940625011920929,
+      "rewards/generated": -0.046993426978588104,
+      "rewards/margins": 0.4582246243953705,
+      "rewards/real": 0.41123121976852417,
+      "step": 1540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.049822064056939e-09,
+      "logits/generated": 0.23574039340019226,
+      "logits/real": -0.15900997817516327,
+      "logps/generated": -513.01025390625,
+      "logps/real": -451.59881591796875,
+      "loss": 0.121,
+      "rewards/accuracies": 0.9437500238418579,
+      "rewards/generated": -0.06016629934310913,
+      "rewards/margins": 0.48446398973464966,
+      "rewards/real": 0.42429766058921814,
+      "step": 1545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.270462633451957e-09,
+      "logits/generated": 0.21644452214241028,
+      "logits/real": -0.22703060507774353,
+      "logps/generated": -519.6123046875,
+      "logps/real": -449.2998046875,
+      "loss": 0.1224,
+      "rewards/accuracies": 0.9593750238418579,
+      "rewards/generated": -0.05588964372873306,
+      "rewards/margins": 0.45251893997192383,
+      "rewards/real": 0.396629273891449,
+      "step": 1550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.491103202846975e-09,
+      "logits/generated": 0.22004759311676025,
+      "logits/real": -0.19879832863807678,
+      "logps/generated": -508.9859313964844,
+      "logps/real": -448.991943359375,
+      "loss": 0.1229,
+      "rewards/accuracies": 0.90625,
+      "rewards/generated": -0.04875675216317177,
+      "rewards/margins": 0.44634467363357544,
+      "rewards/real": 0.3975878953933716,
+      "step": 1555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.117437722419929e-10,
+      "logits/generated": 0.2379693239927292,
+      "logits/real": -0.15108105540275574,
+      "logps/generated": -523.8177490234375,
+      "logps/real": -426.63323974609375,
+      "loss": 0.1224,
+      "rewards/accuracies": 0.918749988079071,
+      "rewards/generated": -0.03124368190765381,
+      "rewards/margins": 0.4431038498878479,
+      "rewards/real": 0.41186007857322693,
+      "step": 1560
+    },
+    {
+      "epoch": 1.0,
+      "step": 1562,
+      "total_flos": 0.0,
+      "train_loss": 0.12871409855334615,
+      "train_runtime": 38810.2478,
+      "train_samples_per_second": 2.577,
+      "train_steps_per_second": 0.04
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 1562,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 0.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}