diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -3,1145 +3,2269 @@
   "best_model_checkpoint": null,
   "epoch": 1.0,
   "eval_steps": 500,
-  "global_step": 782,
+  "global_step": 1563,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 6.329113924050633e-09,
+      "learning_rate": 3.1847133757961784e-09,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -403.3199157714844,
-      "logps/real": -443.6107177734375,
-      "loss": 4.7453,
-      "rewards/accuracies": 0.5,
-      "rewards/generated": -12.943833351135254,
-      "rewards/margins": -2.53641414642334,
-      "rewards/real": -15.48024845123291,
+      "logps/generated": -539.3351440429688,
+      "logps/real": -367.9522399902344,
+      "loss": 1.6159,
+      "rewards/accuracies": 0.625,
+      "rewards/generated": -21.398611068725586,
+      "rewards/margins": 5.284192085266113,
+      "rewards/real": -16.114418029785156,
       "step": 1
     },
     {
       "epoch": 0.01,
-      "learning_rate": 6.329113924050633e-08,
+      "learning_rate": 3.184713375796178e-08,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -487.2853088378906,
-      "logps/real": -384.8033142089844,
-      "loss": 3.4516,
-      "rewards/accuracies": 0.6111111044883728,
-      "rewards/generated": -19.51491928100586,
-      "rewards/margins": 6.083561420440674,
-      "rewards/real": -13.431358337402344,
+      "logps/generated": -466.6612548828125,
+      "logps/real": -352.3954162597656,
+      "loss": 2.9155,
+      "rewards/accuracies": 0.6944444179534912,
+      "rewards/generated": -16.62398910522461,
+      "rewards/margins": 6.3880696296691895,
+      "rewards/real": -10.235919952392578,
       "step": 10
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.2658227848101266e-07,
+      "epoch": 0.01,
+      "learning_rate": 6.369426751592356e-08,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -473.446044921875,
-      "logps/real": -413.171630859375,
-      "loss": 3.9705,
-      "rewards/accuracies": 0.637499988079071,
-      "rewards/generated": -18.367900848388672,
-      "rewards/margins": 3.9024269580841064,
-      "rewards/real": -14.465472221374512,
+      "logps/generated": -494.484375,
+      "logps/real": -385.1441955566406,
+      "loss": 3.3003,
+      "rewards/accuracies": 0.699999988079071,
+      "rewards/generated": -20.39480972290039,
+      "rewards/margins": 6.397805213928223,
+      "rewards/real": -13.997003555297852,
       "step": 20
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.89873417721519e-07,
+      "epoch": 0.02,
+      "learning_rate": 9.554140127388536e-08,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -510.4585876464844,
-      "logps/real": -408.74212646484375,
-      "loss": 3.6955,
-      "rewards/accuracies": 0.6875,
-      "rewards/generated": -21.059778213500977,
-      "rewards/margins": 7.762242317199707,
-      "rewards/real": -13.297533988952637,
+      "logps/generated": -502.4696350097656,
+      "logps/real": -407.5299987792969,
+      "loss": 4.0551,
+      "rewards/accuracies": 0.5874999761581421,
+      "rewards/generated": -21.359249114990234,
+      "rewards/margins": 6.890555381774902,
+      "rewards/real": -14.4686918258667,
       "step": 30
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.5316455696202533e-07,
+      "epoch": 0.03,
+      "learning_rate": 1.2738853503184713e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -451.734130859375,
-      "logps/real": -368.16998291015625,
-      "loss": 3.9056,
-      "rewards/accuracies": 0.625,
-      "rewards/generated": -16.805126190185547,
-      "rewards/margins": 4.4605584144592285,
-      "rewards/real": -12.344568252563477,
+      "logps/generated": -453.08099365234375,
+      "logps/real": -369.57183837890625,
+      "loss": 3.4836,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/generated": -16.532052993774414,
+      "rewards/margins": 5.033951759338379,
+      "rewards/real": -11.498100280761719,
       "step": 40
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.1645569620253163e-07,
+      "epoch": 0.03,
+      "learning_rate": 1.592356687898089e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -552.9141845703125,
-      "logps/real": -397.95318603515625,
-      "loss": 2.9742,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/generated": -25.695674896240234,
-      "rewards/margins": 10.653547286987305,
-      "rewards/real": -15.042126655578613,
+      "logps/generated": -496.9840393066406,
+      "logps/real": -375.58489990234375,
+      "loss": 3.6846,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/generated": -21.554412841796875,
+      "rewards/margins": 8.116512298583984,
+      "rewards/real": -13.437899589538574,
       "step": 50
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 3.79746835443038e-07,
+      "epoch": 0.04,
+      "learning_rate": 1.9108280254777072e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -509.5240783691406,
-      "logps/real": -390.6533203125,
-      "loss": 2.5682,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/generated": -22.568603515625,
-      "rewards/margins": 7.904494285583496,
-      "rewards/real": -14.664111137390137,
+      "logps/generated": -510.9554138183594,
+      "logps/real": -394.2112731933594,
+      "loss": 2.9311,
+      "rewards/accuracies": 0.7250000238418579,
+      "rewards/generated": -22.012584686279297,
+      "rewards/margins": 7.591399192810059,
+      "rewards/real": -14.421185493469238,
       "step": 60
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.4303797468354424e-07,
+      "epoch": 0.04,
+      "learning_rate": 2.2292993630573247e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -554.5638427734375,
-      "logps/real": -382.84832763671875,
-      "loss": 1.7793,
-      "rewards/accuracies": 0.8374999761581421,
-      "rewards/generated": -28.215290069580078,
-      "rewards/margins": 14.945714950561523,
-      "rewards/real": -13.269571304321289,
+      "logps/generated": -500.39190673828125,
+      "logps/real": -409.89202880859375,
+      "loss": 3.2596,
+      "rewards/accuracies": 0.6499999761581421,
+      "rewards/generated": -21.793880462646484,
+      "rewards/margins": 6.847817897796631,
+      "rewards/real": -14.946063041687012,
       "step": 70
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.992887624466572e-07,
+      "epoch": 0.05,
+      "learning_rate": 2.5477707006369425e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -574.03759765625,
-      "logps/real": -402.69232177734375,
-      "loss": 1.4194,
-      "rewards/accuracies": 0.800000011920929,
-      "rewards/generated": -28.118602752685547,
-      "rewards/margins": 12.523630142211914,
-      "rewards/real": -15.594972610473633,
+      "logps/generated": -502.09320068359375,
+      "logps/real": -387.62890625,
+      "loss": 2.3412,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/generated": -21.89040184020996,
+      "rewards/margins": 7.6944780349731445,
+      "rewards/real": -14.19592571258545,
       "step": 80
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.92176386913229e-07,
+      "epoch": 0.06,
+      "learning_rate": 2.86624203821656e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -577.7903442382812,
-      "logps/real": -401.81561279296875,
-      "loss": 1.2635,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/generated": -30.364093780517578,
-      "rewards/margins": 14.755389213562012,
-      "rewards/real": -15.608701705932617,
+      "logps/generated": -557.3800048828125,
+      "logps/real": -414.1551818847656,
+      "loss": 1.9592,
+      "rewards/accuracies": 0.7875000238418579,
+      "rewards/generated": -26.10390281677246,
+      "rewards/margins": 10.303709983825684,
+      "rewards/real": -15.800193786621094,
       "step": 90
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.850640113798008e-07,
+      "epoch": 0.06,
+      "learning_rate": 3.184713375796178e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -654.4276733398438,
-      "logps/real": -442.893798828125,
-      "loss": 0.9213,
-      "rewards/accuracies": 0.8999999761581421,
-      "rewards/generated": -35.57740020751953,
-      "rewards/margins": 18.7423038482666,
-      "rewards/real": -16.835100173950195,
+      "logps/generated": -543.3717651367188,
+      "logps/real": -368.0298156738281,
+      "loss": 1.7186,
+      "rewards/accuracies": 0.8374999761581421,
+      "rewards/generated": -25.229318618774414,
+      "rewards/margins": 12.685417175292969,
+      "rewards/real": -12.543901443481445,
       "step": 100
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.779516358463727e-07,
+      "epoch": 0.07,
+      "learning_rate": 3.5031847133757957e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -637.9244384765625,
-      "logps/real": -426.7471618652344,
-      "loss": 1.0798,
-      "rewards/accuracies": 0.875,
-      "rewards/generated": -35.024356842041016,
-      "rewards/margins": 18.848251342773438,
-      "rewards/real": -16.17610740661621,
+      "logps/generated": -596.9669189453125,
+      "logps/real": -413.40350341796875,
+      "loss": 1.4592,
+      "rewards/accuracies": 0.8374999761581421,
+      "rewards/generated": -30.341022491455078,
+      "rewards/margins": 15.404217720031738,
+      "rewards/real": -14.936800956726074,
       "step": 110
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.7083926031294454e-07,
+      "epoch": 0.08,
+      "learning_rate": 3.8216560509554143e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -647.3193969726562,
-      "logps/real": -419.98095703125,
-      "loss": 0.7428,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -35.827938079833984,
-      "rewards/margins": 21.43227767944336,
-      "rewards/real": -14.395665168762207,
+      "logps/generated": -569.759033203125,
+      "logps/real": -374.6027526855469,
+      "loss": 1.1146,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/generated": -26.69822120666504,
+      "rewards/margins": 14.171666145324707,
+      "rewards/real": -12.526556015014648,
       "step": 120
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.6372688477951633e-07,
+      "epoch": 0.08,
+      "learning_rate": 4.140127388535032e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -647.6121826171875,
-      "logps/real": -413.42401123046875,
-      "loss": 0.6566,
-      "rewards/accuracies": 0.875,
-      "rewards/generated": -35.408538818359375,
-      "rewards/margins": 20.003904342651367,
-      "rewards/real": -15.404635429382324,
+      "logps/generated": -593.4500122070312,
+      "logps/real": -421.9783630371094,
+      "loss": 1.2263,
+      "rewards/accuracies": 0.862500011920929,
+      "rewards/generated": -31.279077529907227,
+      "rewards/margins": 15.886209487915039,
+      "rewards/real": -15.39286994934082,
       "step": 130
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.5661450924608817e-07,
+      "epoch": 0.09,
+      "learning_rate": 4.4585987261146494e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -697.9991455078125,
-      "logps/real": -401.5552673339844,
-      "loss": 0.9421,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -38.38452911376953,
-      "rewards/margins": 23.595943450927734,
-      "rewards/real": -14.788581848144531,
+      "logps/generated": -648.861328125,
+      "logps/real": -401.9364013671875,
+      "loss": 0.8198,
+      "rewards/accuracies": 0.8500000238418579,
+      "rewards/generated": -34.970481872558594,
+      "rewards/margins": 18.90291976928711,
+      "rewards/real": -16.067562103271484,
       "step": 140
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.4950213371266e-07,
+      "epoch": 0.1,
+      "learning_rate": 4.777070063694267e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -672.8363037109375,
-      "logps/real": -401.61590576171875,
-      "loss": 0.6587,
-      "rewards/accuracies": 0.8999999761581421,
-      "rewards/generated": -38.686241149902344,
-      "rewards/margins": 23.072139739990234,
-      "rewards/real": -15.614102363586426,
+      "logps/generated": -638.9237670898438,
+      "logps/real": -419.0517578125,
+      "loss": 1.0322,
+      "rewards/accuracies": 0.862500011920929,
+      "rewards/generated": -34.63279342651367,
+      "rewards/margins": 18.07627296447754,
+      "rewards/real": -16.556522369384766,
       "step": 150
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.4238975817923186e-07,
+      "epoch": 0.1,
+      "learning_rate": 4.989331436699858e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -641.0380859375,
-      "logps/real": -377.9221496582031,
-      "loss": 0.6261,
-      "rewards/accuracies": 0.8999999761581421,
-      "rewards/generated": -35.72499084472656,
-      "rewards/margins": 21.804719924926758,
-      "rewards/real": -13.920272827148438,
+      "logps/generated": -667.6073608398438,
+      "logps/real": -418.95562744140625,
+      "loss": 1.075,
+      "rewards/accuracies": 0.875,
+      "rewards/generated": -37.12474822998047,
+      "rewards/margins": 20.29035186767578,
+      "rewards/real": -16.834400177001953,
       "step": 160
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 4.3527738264580364e-07,
+      "epoch": 0.11,
+      "learning_rate": 4.953769559032717e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -676.1116943359375,
-      "logps/real": -410.857421875,
-      "loss": 0.6189,
-      "rewards/accuracies": 0.9125000238418579,
-      "rewards/generated": -37.884979248046875,
-      "rewards/margins": 23.008445739746094,
-      "rewards/real": -14.876535415649414,
+      "logps/generated": -700.0158081054688,
+      "logps/real": -375.8499755859375,
+      "loss": 1.0279,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -40.3820915222168,
+      "rewards/margins": 25.6898136138916,
+      "rewards/real": -14.692278861999512,
       "step": 170
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.2816500711237554e-07,
+      "epoch": 0.12,
+      "learning_rate": 4.918207681365576e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -654.6505737304688,
-      "logps/real": -370.24224853515625,
-      "loss": 0.3247,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -36.68430709838867,
-      "rewards/margins": 23.027376174926758,
-      "rewards/real": -13.656933784484863,
+      "logps/generated": -680.1141357421875,
+      "logps/real": -433.60052490234375,
+      "loss": 0.7383,
+      "rewards/accuracies": 0.8999999761581421,
+      "rewards/generated": -36.7303352355957,
+      "rewards/margins": 21.14933967590332,
+      "rewards/real": -15.5809965133667,
       "step": 180
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.2105263157894733e-07,
+      "epoch": 0.12,
+      "learning_rate": 4.882645803698435e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -665.4993286132812,
-      "logps/real": -364.58880615234375,
-      "loss": 0.5709,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -37.38275146484375,
-      "rewards/margins": 25.380136489868164,
-      "rewards/real": -12.002609252929688,
+      "logps/generated": -654.9556274414062,
+      "logps/real": -447.13519287109375,
+      "loss": 0.8003,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/generated": -37.65789031982422,
+      "rewards/margins": 19.25699806213379,
+      "rewards/real": -18.40089225769043,
       "step": 190
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.1394025604551917e-07,
+      "epoch": 0.13,
+      "learning_rate": 4.847083926031294e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -684.380615234375,
-      "logps/real": -392.02288818359375,
-      "loss": 0.6823,
-      "rewards/accuracies": 0.8999999761581421,
-      "rewards/generated": -38.82280349731445,
-      "rewards/margins": 24.800228118896484,
-      "rewards/real": -14.02257251739502,
+      "logps/generated": -667.2574462890625,
+      "logps/real": -390.1756591796875,
+      "loss": 0.4891,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -37.77182388305664,
+      "rewards/margins": 22.858360290527344,
+      "rewards/real": -14.913459777832031,
       "step": 200
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.06827880512091e-07,
+      "epoch": 0.13,
+      "learning_rate": 4.811522048364154e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -741.0132446289062,
-      "logps/real": -383.6390075683594,
-      "loss": 0.5369,
-      "rewards/accuracies": 0.925000011920929,
-      "rewards/generated": -43.17304229736328,
-      "rewards/margins": 28.80326271057129,
-      "rewards/real": -14.369776725769043,
+      "logps/generated": -710.2974853515625,
+      "logps/real": -364.0440673828125,
+      "loss": 0.4879,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -42.257808685302734,
+      "rewards/margins": 28.898040771484375,
+      "rewards/real": -13.359766960144043,
       "step": 210
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 3.9971550497866285e-07,
+      "epoch": 0.14,
+      "learning_rate": 4.775960170697012e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -665.7184448242188,
-      "logps/real": -401.6400146484375,
-      "loss": 0.3696,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -38.721866607666016,
-      "rewards/margins": 24.154865264892578,
-      "rewards/real": -14.567001342773438,
+      "logps/generated": -648.4423828125,
+      "logps/real": -361.45953369140625,
+      "loss": 0.543,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -36.720672607421875,
+      "rewards/margins": 22.87405014038086,
+      "rewards/real": -13.846624374389648,
       "step": 220
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 3.926031294452347e-07,
+      "epoch": 0.15,
+      "learning_rate": 4.7403982930298717e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -681.5037231445312,
-      "logps/real": -394.03509521484375,
-      "loss": 0.2532,
-      "rewards/accuracies": 0.949999988079071,
-      "rewards/generated": -38.438758850097656,
-      "rewards/margins": 23.877817153930664,
-      "rewards/real": -14.560938835144043,
+      "logps/generated": -702.8461303710938,
+      "logps/real": -383.67425537109375,
+      "loss": 0.4983,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -40.814414978027344,
+      "rewards/margins": 26.6862850189209,
+      "rewards/real": -14.128130912780762,
       "step": 230
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 3.8549075391180653e-07,
+      "epoch": 0.15,
+      "learning_rate": 4.7048364153627306e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -705.4946899414062,
-      "logps/real": -414.25146484375,
-      "loss": 0.4719,
-      "rewards/accuracies": 0.925000011920929,
-      "rewards/generated": -40.05225372314453,
-      "rewards/margins": 24.665658950805664,
-      "rewards/real": -15.386594772338867,
+      "logps/generated": -666.6107788085938,
+      "logps/real": -442.3255310058594,
+      "loss": 0.5706,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -37.09008026123047,
+      "rewards/margins": 21.38768196105957,
+      "rewards/real": -15.702404975891113,
       "step": 240
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 3.783783783783784e-07,
+      "epoch": 0.16,
+      "learning_rate": 4.66927453769559e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -718.469970703125,
-      "logps/real": -393.9696350097656,
-      "loss": 0.326,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -43.078765869140625,
-      "rewards/margins": 28.980609893798828,
-      "rewards/real": -14.09815788269043,
+      "logps/generated": -685.1495971679688,
+      "logps/real": -359.0677795410156,
+      "loss": 0.5271,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -38.3968505859375,
+      "rewards/margins": 26.170297622680664,
+      "rewards/real": -12.226548194885254,
       "step": 250
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 3.7126600284495016e-07,
+      "epoch": 0.17,
+      "learning_rate": 4.633712660028449e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -721.9268798828125,
-      "logps/real": -403.2998046875,
-      "loss": 0.4804,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -41.29940414428711,
-      "rewards/margins": 26.352609634399414,
-      "rewards/real": -14.946797370910645,
+      "logps/generated": -679.1738891601562,
+      "logps/real": -367.4692077636719,
+      "loss": 0.4291,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -38.32642364501953,
+      "rewards/margins": 25.107736587524414,
+      "rewards/real": -13.218683242797852,
       "step": 260
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 3.6415362731152206e-07,
+      "epoch": 0.17,
+      "learning_rate": 4.5981507823613085e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -690.9553833007812,
-      "logps/real": -374.5037536621094,
-      "loss": 0.3827,
-      "rewards/accuracies": 0.949999988079071,
-      "rewards/generated": -40.69682693481445,
-      "rewards/margins": 26.715112686157227,
-      "rewards/real": -13.981710433959961,
+      "logps/generated": -681.1062622070312,
+      "logps/real": -395.8605651855469,
+      "loss": 0.6404,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -39.41220474243164,
+      "rewards/margins": 23.57551383972168,
+      "rewards/real": -15.836690902709961,
       "step": 270
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 3.5704125177809385e-07,
+      "epoch": 0.18,
+      "learning_rate": 4.562588904694168e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -725.0104370117188,
-      "logps/real": -406.488037109375,
-      "loss": 0.4101,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -42.66962432861328,
-      "rewards/margins": 27.242467880249023,
-      "rewards/real": -15.427154541015625,
+      "logps/generated": -720.9105224609375,
+      "logps/real": -410.877197265625,
+      "loss": 0.3044,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -43.70952606201172,
+      "rewards/margins": 28.780017852783203,
+      "rewards/real": -14.929506301879883,
       "step": 280
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.4992887624466574e-07,
+      "epoch": 0.19,
+      "learning_rate": 4.5270270270270264e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -697.3206787109375,
-      "logps/real": -400.65753173828125,
-      "loss": 0.3629,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -39.595970153808594,
-      "rewards/margins": 24.689983367919922,
-      "rewards/real": -14.90599250793457,
+      "logps/generated": -714.8857421875,
+      "logps/real": -394.2933654785156,
+      "loss": 0.2007,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -41.24427032470703,
+      "rewards/margins": 28.501968383789062,
+      "rewards/real": -12.742300987243652,
       "step": 290
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 3.4281650071123753e-07,
+      "epoch": 0.19,
+      "learning_rate": 4.491465149359886e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -711.7612915039062,
-      "logps/real": -388.0057678222656,
-      "loss": 0.2122,
-      "rewards/accuracies": 0.949999988079071,
-      "rewards/generated": -42.39947509765625,
-      "rewards/margins": 27.109222412109375,
-      "rewards/real": -15.290254592895508,
+      "logps/generated": -682.9447021484375,
+      "logps/real": -367.23590087890625,
+      "loss": 0.4911,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -39.695213317871094,
+      "rewards/margins": 26.347553253173828,
+      "rewards/real": -13.34765911102295,
       "step": 300
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 3.3570412517780937e-07,
+      "epoch": 0.2,
+      "learning_rate": 4.4559032716927454e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -713.0751953125,
-      "logps/real": -423.7640686035156,
-      "loss": 0.3657,
-      "rewards/accuracies": 0.9125000238418579,
-      "rewards/generated": -41.861183166503906,
-      "rewards/margins": 26.213199615478516,
-      "rewards/real": -15.647982597351074,
+      "logps/generated": -709.5479736328125,
+      "logps/real": -430.88763427734375,
+      "loss": 0.1828,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -42.66682052612305,
+      "rewards/margins": 27.468265533447266,
+      "rewards/real": -15.19856071472168,
       "step": 310
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.285917496443812e-07,
+      "epoch": 0.2,
+      "learning_rate": 4.420341394025605e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -695.4622192382812,
-      "logps/real": -386.4324645996094,
-      "loss": 0.3005,
-      "rewards/accuracies": 0.925000011920929,
-      "rewards/generated": -39.45234298706055,
-      "rewards/margins": 25.696590423583984,
-      "rewards/real": -13.755752563476562,
+      "logps/generated": -755.2057495117188,
+      "logps/real": -418.7355041503906,
+      "loss": 0.3104,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -46.417259216308594,
+      "rewards/margins": 30.384328842163086,
+      "rewards/real": -16.032928466796875,
       "step": 320
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 3.2147937411095305e-07,
+      "epoch": 0.21,
+      "learning_rate": 4.384779516358463e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -774.9369506835938,
-      "logps/real": -426.2724609375,
-      "loss": 0.2325,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -48.17813491821289,
-      "rewards/margins": 31.29937171936035,
-      "rewards/real": -16.87876319885254,
+      "logps/generated": -744.1261596679688,
+      "logps/real": -432.3663024902344,
+      "loss": 0.4376,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/generated": -45.77213668823242,
+      "rewards/margins": 28.370372772216797,
+      "rewards/real": -17.401762008666992,
       "step": 330
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.1436699857752484e-07,
+      "epoch": 0.22,
+      "learning_rate": 4.3492176386913227e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -714.9234008789062,
-      "logps/real": -395.0069885253906,
-      "loss": 0.2981,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -41.547645568847656,
-      "rewards/margins": 27.00480079650879,
-      "rewards/real": -14.54284381866455,
+      "logps/generated": -778.1282348632812,
+      "logps/real": -449.26776123046875,
+      "loss": 0.2628,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -48.66665267944336,
+      "rewards/margins": 31.182621002197266,
+      "rewards/real": -17.484031677246094,
       "step": 340
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 3.0725462304409674e-07,
+      "epoch": 0.22,
+      "learning_rate": 4.313655761024182e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -721.8656005859375,
-      "logps/real": -400.1606750488281,
-      "loss": 0.2019,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -44.031803131103516,
-      "rewards/margins": 28.76715087890625,
-      "rewards/real": -15.264646530151367,
+      "logps/generated": -756.3446655273438,
+      "logps/real": -430.87493896484375,
+      "loss": 0.3925,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -46.581642150878906,
+      "rewards/margins": 29.290760040283203,
+      "rewards/real": -17.290876388549805,
       "step": 350
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 3.001422475106685e-07,
+      "epoch": 0.23,
+      "learning_rate": 4.278093883357041e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -728.89599609375,
-      "logps/real": -470.1971130371094,
-      "loss": 0.5931,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -43.573814392089844,
-      "rewards/margins": 24.9804630279541,
-      "rewards/real": -18.593351364135742,
+      "logps/generated": -777.9512939453125,
+      "logps/real": -414.18450927734375,
+      "loss": 0.4194,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -48.169349670410156,
+      "rewards/margins": 32.56087875366211,
+      "rewards/real": -15.608467102050781,
       "step": 360
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 2.9302987197724037e-07,
+      "epoch": 0.24,
+      "learning_rate": 4.2425320056899e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -667.4353637695312,
-      "logps/real": -369.68328857421875,
-      "loss": 0.4031,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -38.71464920043945,
-      "rewards/margins": 24.013246536254883,
-      "rewards/real": -14.701400756835938,
+      "logps/generated": -776.2547607421875,
+      "logps/real": -459.26971435546875,
+      "loss": 0.1642,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -48.74531173706055,
+      "rewards/margins": 30.371551513671875,
+      "rewards/real": -18.373756408691406,
       "step": 370
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 2.8591749644381226e-07,
+      "epoch": 0.24,
+      "learning_rate": 4.2069701280227595e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -708.6209716796875,
-      "logps/real": -395.7604675292969,
-      "loss": 0.3691,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -42.240318298339844,
-      "rewards/margins": 26.32097816467285,
-      "rewards/real": -15.919347763061523,
+      "logps/generated": -787.847900390625,
+      "logps/real": -431.939697265625,
+      "loss": 0.1822,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -48.518409729003906,
+      "rewards/margins": 31.980022430419922,
+      "rewards/real": -16.538381576538086,
       "step": 380
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.7880512091038405e-07,
+      "epoch": 0.25,
+      "learning_rate": 4.1714082503556185e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -766.4454345703125,
-      "logps/real": -423.30755615234375,
-      "loss": 0.2758,
+      "logps/generated": -782.6195678710938,
+      "logps/real": -409.2000732421875,
+      "loss": 0.1507,
       "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -47.435791015625,
-      "rewards/margins": 30.477214813232422,
-      "rewards/real": -16.958572387695312,
+      "rewards/generated": -49.06304168701172,
+      "rewards/margins": 32.46137237548828,
+      "rewards/real": -16.601669311523438,
       "step": 390
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 2.716927453769559e-07,
+      "epoch": 0.26,
+      "learning_rate": 4.135846372688478e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -768.6455078125,
-      "logps/real": -449.0997009277344,
-      "loss": 0.3455,
-      "rewards/accuracies": 0.949999988079071,
-      "rewards/generated": -47.01398849487305,
-      "rewards/margins": 28.302906036376953,
-      "rewards/real": -18.711084365844727,
+      "logps/generated": -743.0278930664062,
+      "logps/real": -435.90399169921875,
+      "loss": 0.3316,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -46.87841796875,
+      "rewards/margins": 29.552413940429688,
+      "rewards/real": -17.32600975036621,
       "step": 400
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 2.6458036984352773e-07,
+      "epoch": 0.26,
+      "learning_rate": 4.100284495021337e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -734.2765502929688,
-      "logps/real": -447.5357971191406,
-      "loss": 0.375,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -46.2476921081543,
-      "rewards/margins": 27.413455963134766,
-      "rewards/real": -18.834239959716797,
+      "logps/generated": -779.1712646484375,
+      "logps/real": -406.473876953125,
+      "loss": 0.333,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -48.82392883300781,
+      "rewards/margins": 33.1125602722168,
+      "rewards/real": -15.711362838745117,
       "step": 410
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 2.574679943100996e-07,
+      "epoch": 0.27,
+      "learning_rate": 4.064722617354196e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -730.5106201171875,
-      "logps/real": -453.9371643066406,
-      "loss": 0.0816,
-      "rewards/accuracies": 0.925000011920929,
-      "rewards/generated": -45.583648681640625,
-      "rewards/margins": 27.334802627563477,
-      "rewards/real": -18.248844146728516,
+      "logps/generated": -774.0535888671875,
+      "logps/real": -417.33184814453125,
+      "loss": 0.2446,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -49.45975875854492,
+      "rewards/margins": 33.8248405456543,
+      "rewards/real": -15.634923934936523,
       "step": 420
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 2.5035561877667136e-07,
+      "epoch": 0.28,
+      "learning_rate": 4.0291607396870553e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -766.5603637695312,
-      "logps/real": -438.89788818359375,
-      "loss": 0.1735,
-      "rewards/accuracies": 0.9750000238418579,
-      "rewards/generated": -46.9077262878418,
-      "rewards/margins": 29.52130126953125,
-      "rewards/real": -17.386432647705078,
+      "logps/generated": -767.9479370117188,
+      "logps/real": -403.72357177734375,
+      "loss": 0.2645,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -48.79080581665039,
+      "rewards/margins": 33.18938064575195,
+      "rewards/real": -15.601428031921387,
       "step": 430
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 2.4324324324324326e-07,
+      "epoch": 0.28,
+      "learning_rate": 3.993598862019915e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -781.421630859375,
-      "logps/real": -405.1501770019531,
-      "loss": 0.3965,
+      "logps/generated": -754.3994140625,
+      "logps/real": -410.26470947265625,
+      "loss": 0.2166,
       "rewards/accuracies": 0.987500011920929,
-      "rewards/generated": -49.6665153503418,
-      "rewards/margins": 33.62762451171875,
-      "rewards/real": -16.03889274597168,
+      "rewards/generated": -46.57341766357422,
+      "rewards/margins": 30.815637588500977,
+      "rewards/real": -15.757779121398926,
       "step": 440
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 2.3613086770981507e-07,
+      "epoch": 0.29,
+      "learning_rate": 3.9580369843527737e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -760.2469482421875,
-      "logps/real": -404.4407653808594,
-      "loss": 0.4128,
-      "rewards/accuracies": 0.949999988079071,
-      "rewards/generated": -47.4620361328125,
-      "rewards/margins": 30.4351863861084,
-      "rewards/real": -17.026851654052734,
+      "logps/generated": -870.4566650390625,
+      "logps/real": -408.28997802734375,
+      "loss": 0.2063,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -57.19744110107422,
+      "rewards/margins": 40.77977752685547,
+      "rewards/real": -16.41766357421875,
       "step": 450
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 2.290184921763869e-07,
+      "epoch": 0.29,
+      "learning_rate": 3.9224751066856327e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -820.1843872070312,
-      "logps/real": -420.02587890625,
-      "loss": 0.1306,
-      "rewards/accuracies": 1.0,
-      "rewards/generated": -51.192138671875,
-      "rewards/margins": 35.38946533203125,
-      "rewards/real": -15.8026704788208,
+      "logps/generated": -876.7347412109375,
+      "logps/real": -472.7223205566406,
+      "loss": 0.3418,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -57.74506378173828,
+      "rewards/margins": 37.913551330566406,
+      "rewards/real": -19.831506729125977,
       "step": 460
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 2.2190611664295875e-07,
+      "epoch": 0.3,
+      "learning_rate": 3.886913229018492e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -759.2930908203125,
-      "logps/real": -438.888916015625,
-      "loss": 0.2487,
-      "rewards/accuracies": 0.925000011920929,
-      "rewards/generated": -45.420188903808594,
-      "rewards/margins": 27.34377670288086,
-      "rewards/real": -18.0764102935791,
+      "logps/generated": -798.3450317382812,
+      "logps/real": -431.00628662109375,
+      "loss": 0.5482,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -50.90120315551758,
+      "rewards/margins": 32.8853645324707,
+      "rewards/real": -18.01584243774414,
       "step": 470
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 2.1479374110953057e-07,
+      "epoch": 0.31,
+      "learning_rate": 3.851351351351351e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -777.7584838867188,
-      "logps/real": -388.6899108886719,
-      "loss": 0.3922,
-      "rewards/accuracies": 0.949999988079071,
-      "rewards/generated": -48.69025421142578,
-      "rewards/margins": 32.126991271972656,
-      "rewards/real": -16.563264846801758,
+      "logps/generated": -804.5404052734375,
+      "logps/real": -420.87066650390625,
+      "loss": 0.1632,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -51.20261764526367,
+      "rewards/margins": 34.63499069213867,
+      "rewards/real": -16.567626953125,
       "step": 480
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 2.076813655761024e-07,
+      "epoch": 0.31,
+      "learning_rate": 3.8157894736842105e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -804.2822875976562,
-      "logps/real": -438.558837890625,
-      "loss": 0.2558,
-      "rewards/accuracies": 0.987500011920929,
-      "rewards/generated": -50.539390563964844,
-      "rewards/margins": 31.528533935546875,
-      "rewards/real": -19.0108585357666,
+      "logps/generated": -808.2142333984375,
+      "logps/real": -412.75909423828125,
+      "loss": 0.067,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -51.9697265625,
+      "rewards/margins": 33.47379684448242,
+      "rewards/real": -18.49593162536621,
       "step": 490
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 2.0056899004267425e-07,
+      "epoch": 0.32,
+      "learning_rate": 3.7802275960170695e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -799.6202392578125,
-      "logps/real": -483.55096435546875,
-      "loss": 0.2945,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -50.81739807128906,
-      "rewards/margins": 30.668746948242188,
-      "rewards/real": -20.14865493774414,
+      "logps/generated": -833.8155517578125,
+      "logps/real": -421.85693359375,
+      "loss": 0.3303,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -53.8129768371582,
+      "rewards/margins": 36.602169036865234,
+      "rewards/real": -17.210813522338867,
       "step": 500
     },
     {
-      "epoch": 0.64,
+      "epoch": 0.32,
       "eval_logits/generated": -Infinity,
       "eval_logits/real": -Infinity,
-      "eval_logps/generated": -509.32861328125,
-      "eval_logps/real": -313.0279541015625,
-      "eval_loss": 0.1748354285955429,
-      "eval_rewards/accuracies": 0.9442675113677979,
-      "eval_rewards/generated": -21.789350509643555,
-      "eval_rewards/margins": 15.725247383117676,
-      "eval_rewards/real": -6.0641021728515625,
-      "eval_runtime": 590.026,
-      "eval_samples_per_second": 8.474,
-      "eval_steps_per_second": 0.266,
+      "eval_logps/generated": -529.861328125,
+      "eval_logps/real": -300.84613037109375,
+      "eval_loss": 0.20025908946990967,
+      "eval_rewards/accuracies": 0.9371019005775452,
+      "eval_rewards/generated": -23.842620849609375,
+      "eval_rewards/margins": 18.996700286865234,
+      "eval_rewards/real": -4.845921039581299,
+      "eval_runtime": 816.3836,
+      "eval_samples_per_second": 6.125,
+      "eval_steps_per_second": 0.192,
       "step": 500
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.9345661450924607e-07,
+      "epoch": 0.33,
+      "learning_rate": 3.7446657183499284e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -757.3658447265625,
-      "logps/real": -412.8812561035156,
-      "loss": 0.2523,
-      "rewards/accuracies": 0.9750000238418579,
-      "rewards/generated": -45.53099822998047,
-      "rewards/margins": 29.076038360595703,
-      "rewards/real": -16.454959869384766,
+      "logps/generated": -865.2786254882812,
+      "logps/real": -424.654052734375,
+      "loss": 0.113,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -56.876319885253906,
+      "rewards/margins": 39.13337326049805,
+      "rewards/real": -17.74294662475586,
       "step": 510
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.863442389758179e-07,
+      "epoch": 0.33,
+      "learning_rate": 3.709103840682788e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -780.4212036132812,
-      "logps/real": -437.5557556152344,
-      "loss": 0.2147,
-      "rewards/accuracies": 0.987500011920929,
-      "rewards/generated": -49.51309585571289,
-      "rewards/margins": 32.93183517456055,
-      "rewards/real": -16.581256866455078,
+      "logps/generated": -779.5700073242188,
+      "logps/real": -393.60321044921875,
+      "loss": 0.1035,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -49.612606048583984,
+      "rewards/margins": 33.626441955566406,
+      "rewards/real": -15.986169815063477,
       "step": 520
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.7923186344238975e-07,
+      "epoch": 0.34,
+      "learning_rate": 3.6735419630156474e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -735.9564208984375,
-      "logps/real": -407.6651306152344,
-      "loss": 0.2975,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -44.98466491699219,
-      "rewards/margins": 29.423254013061523,
-      "rewards/real": -15.561413764953613,
+      "logps/generated": -776.9521484375,
+      "logps/real": -400.86016845703125,
+      "loss": 0.138,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -49.583370208740234,
+      "rewards/margins": 32.01411819458008,
+      "rewards/real": -17.569255828857422,
       "step": 530
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.721194879089616e-07,
+      "epoch": 0.35,
+      "learning_rate": 3.637980085348506e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -807.4033203125,
-      "logps/real": -429.4000549316406,
-      "loss": 0.2254,
-      "rewards/accuracies": 0.949999988079071,
-      "rewards/generated": -51.46739959716797,
-      "rewards/margins": 32.75578689575195,
-      "rewards/real": -18.711612701416016,
+      "logps/generated": -768.1486206054688,
+      "logps/real": -421.4039611816406,
+      "loss": 0.3281,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -48.634918212890625,
+      "rewards/margins": 31.545923233032227,
+      "rewards/real": -17.088993072509766,
       "step": 540
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.650071123755334e-07,
+      "epoch": 0.35,
+      "learning_rate": 3.602418207681365e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -811.77099609375,
-      "logps/real": -425.84649658203125,
-      "loss": 0.3105,
+      "logps/generated": -798.0943603515625,
+      "logps/real": -438.1573791503906,
+      "loss": 0.1217,
       "rewards/accuracies": 0.949999988079071,
-      "rewards/generated": -52.33929443359375,
-      "rewards/margins": 34.459083557128906,
-      "rewards/real": -17.88020896911621,
+      "rewards/generated": -49.93975067138672,
+      "rewards/margins": 32.817298889160156,
+      "rewards/real": -17.12244987487793,
       "step": 550
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.5789473684210525e-07,
+      "epoch": 0.36,
+      "learning_rate": 3.5668563300142247e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -788.242919921875,
-      "logps/real": -440.695556640625,
-      "loss": 0.1702,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -49.456298828125,
-      "rewards/margins": 31.41439437866211,
-      "rewards/real": -18.041906356811523,
+      "logps/generated": -784.5931396484375,
+      "logps/real": -432.72039794921875,
+      "loss": 0.1697,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -49.35939407348633,
+      "rewards/margins": 30.70552635192871,
+      "rewards/real": -18.65386962890625,
       "step": 560
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.507823613086771e-07,
+      "epoch": 0.36,
+      "learning_rate": 3.5312944523470837e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -752.26611328125,
-      "logps/real": -427.10113525390625,
-      "loss": 0.3281,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -47.79071807861328,
-      "rewards/margins": 30.950246810913086,
-      "rewards/real": -16.84047508239746,
+      "logps/generated": -830.1300659179688,
+      "logps/real": -439.60400390625,
+      "loss": 0.1078,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -53.280181884765625,
+      "rewards/margins": 33.769203186035156,
+      "rewards/real": -19.510984420776367,
       "step": 570
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.436699857752489e-07,
+      "epoch": 0.37,
+      "learning_rate": 3.495732574679943e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -826.0186767578125,
-      "logps/real": -424.66680908203125,
-      "loss": 0.3042,
-      "rewards/accuracies": 0.9750000238418579,
-      "rewards/generated": -51.909454345703125,
-      "rewards/margins": 34.6719856262207,
-      "rewards/real": -17.237468719482422,
+      "logps/generated": -795.7073364257812,
+      "logps/real": -419.2820739746094,
+      "loss": 0.3486,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -51.35248947143555,
+      "rewards/margins": 34.13421630859375,
+      "rewards/real": -17.2182674407959,
       "step": 580
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.3655761024182077e-07,
+      "epoch": 0.38,
+      "learning_rate": 3.460170697012802e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -816.61962890625,
-      "logps/real": -420.100830078125,
-      "loss": 0.2635,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -52.91407012939453,
-      "rewards/margins": 36.203983306884766,
-      "rewards/real": -16.710086822509766,
+      "logps/generated": -867.7032470703125,
+      "logps/real": -411.7303771972656,
+      "loss": 0.1504,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -57.48298263549805,
+      "rewards/margins": 41.10187530517578,
+      "rewards/real": -16.3811092376709,
       "step": 590
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.2944523470839261e-07,
+      "epoch": 0.38,
+      "learning_rate": 3.424608819345661e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -820.4493408203125,
-      "logps/real": -428.4378967285156,
-      "loss": 0.347,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -51.51692581176758,
-      "rewards/margins": 34.30039596557617,
-      "rewards/real": -17.216527938842773,
+      "logps/generated": -883.2786865234375,
+      "logps/real": -425.11041259765625,
+      "loss": 0.26,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -58.78193283081055,
+      "rewards/margins": 40.10378646850586,
+      "rewards/real": -18.67814064025879,
       "step": 600
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.2233285917496443e-07,
+      "epoch": 0.39,
+      "learning_rate": 3.3890469416785205e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -768.8092041015625,
-      "logps/real": -420.62200927734375,
-      "loss": 0.0588,
-      "rewards/accuracies": 0.987500011920929,
-      "rewards/generated": -47.871437072753906,
-      "rewards/margins": 32.33845901489258,
-      "rewards/real": -15.532976150512695,
+      "logps/generated": -849.3904418945312,
+      "logps/real": -451.054931640625,
+      "loss": 0.1136,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -56.99665451049805,
+      "rewards/margins": 37.19718933105469,
+      "rewards/real": -19.79946517944336,
       "step": 610
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.1522048364153626e-07,
+      "epoch": 0.4,
+      "learning_rate": 3.35348506401138e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -786.0013427734375,
-      "logps/real": -442.2757873535156,
-      "loss": 0.2483,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -50.278663635253906,
-      "rewards/margins": 33.101016998291016,
-      "rewards/real": -17.17764663696289,
+      "logps/generated": -894.3114013671875,
+      "logps/real": -414.6637268066406,
+      "loss": 0.2349,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -59.90185546875,
+      "rewards/margins": 41.685302734375,
+      "rewards/real": -18.216556549072266,
       "step": 620
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.0810810810810811e-07,
+      "epoch": 0.4,
+      "learning_rate": 3.3179231863442384e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -787.63818359375,
-      "logps/real": -404.19122314453125,
-      "loss": 0.1336,
+      "logps/generated": -826.98681640625,
+      "logps/real": -439.54559326171875,
+      "loss": 0.1956,
       "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -50.15456008911133,
-      "rewards/margins": 33.99864959716797,
-      "rewards/real": -16.15591049194336,
+      "rewards/generated": -53.87934494018555,
+      "rewards/margins": 35.65498733520508,
+      "rewards/real": -18.2243595123291,
       "step": 630
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.0099573257467994e-07,
+      "epoch": 0.41,
+      "learning_rate": 3.282361308677098e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -830.99462890625,
-      "logps/real": -457.3229064941406,
-      "loss": 0.2925,
-      "rewards/accuracies": 0.949999988079071,
-      "rewards/generated": -53.06385040283203,
-      "rewards/margins": 34.269439697265625,
-      "rewards/real": -18.794404983520508,
+      "logps/generated": -865.1568603515625,
+      "logps/real": -431.68914794921875,
+      "loss": 0.132,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -56.41038131713867,
+      "rewards/margins": 37.95232391357422,
+      "rewards/real": -18.458057403564453,
       "step": 640
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 9.388335704125178e-08,
+      "epoch": 0.42,
+      "learning_rate": 3.2467994310099573e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -777.55517578125,
-      "logps/real": -405.3326110839844,
-      "loss": 0.137,
+      "logps/generated": -910.2918090820312,
+      "logps/real": -463.79034423828125,
+      "loss": 0.0608,
       "rewards/accuracies": 1.0,
-      "rewards/generated": -49.94211196899414,
-      "rewards/margins": 32.77922821044922,
-      "rewards/real": -17.16288948059082,
+      "rewards/generated": -61.92814254760742,
+      "rewards/margins": 42.76028823852539,
+      "rewards/real": -19.167850494384766,
       "step": 650
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 8.677098150782361e-08,
+      "epoch": 0.42,
+      "learning_rate": 3.211237553342817e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -775.7921142578125,
-      "logps/real": -408.35302734375,
-      "loss": 0.0992,
+      "logps/generated": -792.5632934570312,
+      "logps/real": -417.36187744140625,
+      "loss": 0.0828,
       "rewards/accuracies": 0.987500011920929,
-      "rewards/generated": -48.498905181884766,
-      "rewards/margins": 32.695289611816406,
-      "rewards/real": -15.803617477416992,
+      "rewards/generated": -51.489219665527344,
+      "rewards/margins": 34.43842697143555,
+      "rewards/real": -17.050796508789062,
       "step": 660
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 7.965860597439544e-08,
+      "epoch": 0.43,
+      "learning_rate": 3.175675675675675e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -764.4813232421875,
-      "logps/real": -427.140625,
-      "loss": 0.2396,
-      "rewards/accuracies": 0.925000011920929,
-      "rewards/generated": -48.443748474121094,
-      "rewards/margins": 31.320148468017578,
-      "rewards/real": -17.123600006103516,
+      "logps/generated": -870.7483520507812,
+      "logps/real": -379.50860595703125,
+      "loss": 0.2451,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -59.225135803222656,
+      "rewards/margins": 44.224327087402344,
+      "rewards/real": -15.000802993774414,
       "step": 670
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 7.254623044096728e-08,
+      "epoch": 0.44,
+      "learning_rate": 3.1401137980085347e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -803.4662475585938,
-      "logps/real": -435.402099609375,
-      "loss": 0.2362,
-      "rewards/accuracies": 0.987500011920929,
-      "rewards/generated": -51.644203186035156,
-      "rewards/margins": 33.23621368408203,
-      "rewards/real": -18.40799331665039,
+      "logps/generated": -930.4283447265625,
+      "logps/real": -410.5220642089844,
+      "loss": 0.1281,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -63.19865036010742,
+      "rewards/margins": 46.53211212158203,
+      "rewards/real": -16.666542053222656,
       "step": 680
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 6.543385490753911e-08,
+      "epoch": 0.44,
+      "learning_rate": 3.104551920341394e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -863.4392700195312,
-      "logps/real": -439.1971130371094,
-      "loss": 0.1949,
-      "rewards/accuracies": 0.9750000238418579,
-      "rewards/generated": -55.93394088745117,
-      "rewards/margins": 36.83943176269531,
-      "rewards/real": -19.094507217407227,
+      "logps/generated": -847.4844970703125,
+      "logps/real": -407.2476806640625,
+      "loss": 0.2284,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -55.74333572387695,
+      "rewards/margins": 40.407684326171875,
+      "rewards/real": -15.335647583007812,
       "step": 690
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 5.832147937411095e-08,
+      "epoch": 0.45,
+      "learning_rate": 3.068990042674253e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -779.9309692382812,
-      "logps/real": -411.91455078125,
-      "loss": 0.2666,
-      "rewards/accuracies": 0.9375,
-      "rewards/generated": -47.416011810302734,
-      "rewards/margins": 29.500951766967773,
-      "rewards/real": -17.915063858032227,
+      "logps/generated": -811.4222412109375,
+      "logps/real": -391.034912109375,
+      "loss": 0.142,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -53.94512939453125,
+      "rewards/margins": 37.505123138427734,
+      "rewards/real": -16.440000534057617,
       "step": 700
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 5.120910384068278e-08,
+      "epoch": 0.45,
+      "learning_rate": 3.033428165007112e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -826.6921997070312,
-      "logps/real": -447.87762451171875,
-      "loss": 0.258,
-      "rewards/accuracies": 0.949999988079071,
-      "rewards/generated": -53.74137496948242,
-      "rewards/margins": 35.463531494140625,
-      "rewards/real": -18.277841567993164,
+      "logps/generated": -877.0598754882812,
+      "logps/real": -428.3775329589844,
+      "loss": 0.0922,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -59.027740478515625,
+      "rewards/margins": 40.57575988769531,
+      "rewards/real": -18.45196533203125,
       "step": 710
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 4.4096728307254624e-08,
+      "epoch": 0.46,
+      "learning_rate": 2.9978662873399715e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -834.2398681640625,
-      "logps/real": -427.90966796875,
-      "loss": 0.1272,
-      "rewards/accuracies": 0.987500011920929,
-      "rewards/generated": -53.82038497924805,
-      "rewards/margins": 36.249298095703125,
-      "rewards/real": -17.571086883544922,
+      "logps/generated": -855.3162841796875,
+      "logps/real": -435.65301513671875,
+      "loss": 0.2617,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -56.983558654785156,
+      "rewards/margins": 38.14686965942383,
+      "rewards/real": -18.836681365966797,
       "step": 720
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 3.698435277382646e-08,
+      "epoch": 0.47,
+      "learning_rate": 2.9623044096728305e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -838.7642822265625,
-      "logps/real": -436.14642333984375,
-      "loss": 0.2078,
-      "rewards/accuracies": 0.9750000238418579,
-      "rewards/generated": -55.14348602294922,
-      "rewards/margins": 37.05797576904297,
-      "rewards/real": -18.08551597595215,
+      "logps/generated": -818.3048095703125,
+      "logps/real": -438.0257263183594,
+      "loss": 0.281,
+      "rewards/accuracies": 0.887499988079071,
+      "rewards/generated": -54.11011505126953,
+      "rewards/margins": 37.23180389404297,
+      "rewards/real": -16.878313064575195,
       "step": 730
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 2.9871977240398294e-08,
+      "epoch": 0.47,
+      "learning_rate": 2.92674253200569e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -798.5349731445312,
-      "logps/real": -427.6378479003906,
-      "loss": 0.2621,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -52.086219787597656,
-      "rewards/margins": 34.135223388671875,
-      "rewards/real": -17.95099449157715,
+      "logps/generated": -871.099609375,
+      "logps/real": -420.02880859375,
+      "loss": 0.1622,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -58.3928337097168,
+      "rewards/margins": 41.17832565307617,
+      "rewards/real": -17.21451187133789,
       "step": 740
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 2.275960170697013e-08,
+      "epoch": 0.48,
+      "learning_rate": 2.8911806543385494e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -835.6724853515625,
-      "logps/real": -414.00152587890625,
-      "loss": 0.1564,
+      "logps/generated": -907.7469482421875,
+      "logps/real": -472.86614990234375,
+      "loss": 0.2422,
       "rewards/accuracies": 0.9750000238418579,
-      "rewards/generated": -52.597930908203125,
-      "rewards/margins": 36.33824920654297,
-      "rewards/real": -16.259681701660156,
+      "rewards/generated": -61.027503967285156,
+      "rewards/margins": 42.720279693603516,
+      "rewards/real": -18.30722427368164,
       "step": 750
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.564722617354196e-08,
+      "epoch": 0.49,
+      "learning_rate": 2.855618776671408e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -850.904296875,
-      "logps/real": -392.68310546875,
-      "loss": 0.1515,
+      "logps/generated": -920.8482666015625,
+      "logps/real": -464.93218994140625,
+      "loss": 0.1303,
       "rewards/accuracies": 0.9750000238418579,
-      "rewards/generated": -55.202125549316406,
-      "rewards/margins": 38.95097351074219,
-      "rewards/real": -16.25115394592285,
+      "rewards/generated": -63.184059143066406,
+      "rewards/margins": 44.418663024902344,
+      "rewards/real": -18.765398025512695,
       "step": 760
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 8.534850640113798e-09,
+      "epoch": 0.49,
+      "learning_rate": 2.8200568990042673e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -814.2432250976562,
-      "logps/real": -415.373779296875,
-      "loss": 0.1867,
-      "rewards/accuracies": 0.9624999761581421,
-      "rewards/generated": -53.38776397705078,
-      "rewards/margins": 37.246055603027344,
-      "rewards/real": -16.141704559326172,
+      "logps/generated": -957.7092895507812,
+      "logps/real": -442.88897705078125,
+      "loss": 0.3922,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -65.42414093017578,
+      "rewards/margins": 46.468353271484375,
+      "rewards/real": -18.955785751342773,
       "step": 770
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.422475106685633e-09,
+      "epoch": 0.5,
+      "learning_rate": 2.784495021337127e-07,
       "logits/generated": -Infinity,
       "logits/real": -Infinity,
-      "logps/generated": -811.1026611328125,
-      "logps/real": -388.99835205078125,
-      "loss": 0.2403,
-      "rewards/accuracies": 0.9750000238418579,
-      "rewards/generated": -51.92308807373047,
-      "rewards/margins": 35.41986846923828,
-      "rewards/real": -16.503215789794922,
+      "logps/generated": -908.4993286132812,
+      "logps/real": -413.3855895996094,
+      "loss": 0.1754,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -63.04656982421875,
+      "rewards/margins": 44.62030792236328,
+      "rewards/real": -18.426259994506836,
       "step": 780
     },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7489331436699857e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -863.7523193359375,
+      "logps/real": -465.85009765625,
+      "loss": 0.1958,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -56.57744216918945,
+      "rewards/margins": 37.282875061035156,
+      "rewards/real": -19.2945613861084,
+      "step": 790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7133712660028446e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -967.6433715820312,
+      "logps/real": -416.7491149902344,
+      "loss": 0.163,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -66.82258605957031,
+      "rewards/margins": 49.467437744140625,
+      "rewards/real": -17.355154037475586,
+      "step": 800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.677809388335704e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -902.4373168945312,
+      "logps/real": -438.209716796875,
+      "loss": 0.1638,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -61.510841369628906,
+      "rewards/margins": 41.37336349487305,
+      "rewards/real": -20.13747787475586,
+      "step": 810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.642247510668563e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -977.4361572265625,
+      "logps/real": -426.0132751464844,
+      "loss": 0.1951,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -67.98822784423828,
+      "rewards/margins": 50.39772415161133,
+      "rewards/real": -17.59050941467285,
+      "step": 820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.6066856330014225e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -933.1921997070312,
+      "logps/real": -432.3758239746094,
+      "loss": 0.1135,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -64.10530090332031,
+      "rewards/margins": 46.14706802368164,
+      "rewards/real": -17.958240509033203,
+      "step": 830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5711237553342815e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -933.9869995117188,
+      "logps/real": -469.08074951171875,
+      "loss": 0.1554,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -62.14231491088867,
+      "rewards/margins": 42.210784912109375,
+      "rewards/real": -19.931529998779297,
+      "step": 840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.5355618776671404e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -905.9029541015625,
+      "logps/real": -442.23382568359375,
+      "loss": 0.0891,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -61.816429138183594,
+      "rewards/margins": 43.42655944824219,
+      "rewards/real": -18.38986587524414,
+      "step": 850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.5e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -903.47314453125,
+      "logps/real": -426.867431640625,
+      "loss": 0.2286,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -61.971595764160156,
+      "rewards/margins": 43.64350891113281,
+      "rewards/real": -18.328088760375977,
+      "step": 860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4644381223328594e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -904.4534301757812,
+      "logps/real": -448.7220153808594,
+      "loss": 0.0816,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -62.04343795776367,
+      "rewards/margins": 42.920806884765625,
+      "rewards/real": -19.122634887695312,
+      "step": 870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4288762446657183e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -967.4172973632812,
+      "logps/real": -439.7066955566406,
+      "loss": 0.1422,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -68.501220703125,
+      "rewards/margins": 50.839942932128906,
+      "rewards/real": -17.66128158569336,
+      "step": 880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.393314366998578e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -949.60546875,
+      "logps/real": -431.2792053222656,
+      "loss": 0.2069,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -65.57563781738281,
+      "rewards/margins": 47.400535583496094,
+      "rewards/real": -18.175098419189453,
+      "step": 890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.3577524893314365e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -958.1769409179688,
+      "logps/real": -407.71038818359375,
+      "loss": 0.0272,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -65.54978942871094,
+      "rewards/margins": 49.18589401245117,
+      "rewards/real": -16.363895416259766,
+      "step": 900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.322190611664296e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -854.2259521484375,
+      "logps/real": -461.548583984375,
+      "loss": 0.2207,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -57.03990936279297,
+      "rewards/margins": 37.42232131958008,
+      "rewards/real": -19.617586135864258,
+      "step": 910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.2866287339971549e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -913.2406005859375,
+      "logps/real": -424.36065673828125,
+      "loss": 0.1078,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -61.39838790893555,
+      "rewards/margins": 44.66621780395508,
+      "rewards/real": -16.732179641723633,
+      "step": 920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.251066856330014e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -903.1484375,
+      "logps/real": -405.7799987792969,
+      "loss": 0.2241,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -62.2490234375,
+      "rewards/margins": 45.68339920043945,
+      "rewards/real": -16.565624237060547,
+      "step": 930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.2155049786628733e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -916.8377075195312,
+      "logps/real": -432.683837890625,
+      "loss": 0.1258,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -61.457244873046875,
+      "rewards/margins": 44.813026428222656,
+      "rewards/real": -16.64422035217285,
+      "step": 940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1799431009957325e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -894.6036987304688,
+      "logps/real": -406.5244445800781,
+      "loss": 0.029,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -60.447288513183594,
+      "rewards/margins": 43.23609161376953,
+      "rewards/real": -17.211193084716797,
+      "step": 950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.1443812233285914e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -938.0587158203125,
+      "logps/real": -458.41143798828125,
+      "loss": 0.1849,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -64.76756286621094,
+      "rewards/margins": 45.54555892944336,
+      "rewards/real": -19.222000122070312,
+      "step": 960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.108819345661451e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -990.7427978515625,
+      "logps/real": -451.2483825683594,
+      "loss": 0.2407,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -68.94884490966797,
+      "rewards/margins": 48.71486282348633,
+      "rewards/real": -20.23398208618164,
+      "step": 970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0732574679943098e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -833.5616455078125,
+      "logps/real": -393.88153076171875,
+      "loss": 0.162,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -54.51671600341797,
+      "rewards/margins": 37.748043060302734,
+      "rewards/real": -16.768672943115234,
+      "step": 980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0376955903271693e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -875.89453125,
+      "logps/real": -475.2669982910156,
+      "loss": 0.1122,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -59.66436004638672,
+      "rewards/margins": 39.77927780151367,
+      "rewards/real": -19.88509178161621,
+      "step": 990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0021337126600283e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -917.2653198242188,
+      "logps/real": -410.7203674316406,
+      "loss": 0.0933,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -63.5474853515625,
+      "rewards/margins": 45.195716857910156,
+      "rewards/real": -18.35177230834961,
+      "step": 1000
+    },
+    {
+      "epoch": 0.64,
+      "eval_logits/generated": -Infinity,
+      "eval_logits/real": -Infinity,
+      "eval_logps/generated": -639.9600219726562,
+      "eval_logps/real": -298.9767761230469,
+      "eval_loss": 0.15978027880191803,
+      "eval_rewards/accuracies": 0.9609872698783875,
+      "eval_rewards/generated": -34.85248565673828,
+      "eval_rewards/margins": 30.193498611450195,
+      "eval_rewards/real": -4.658985614776611,
+      "eval_runtime": 516.5493,
+      "eval_samples_per_second": 9.68,
+      "eval_steps_per_second": 0.304,
+      "step": 1000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9665718349928875e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -931.1051025390625,
+      "logps/real": -400.90509033203125,
+      "loss": 0.1599,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -65.04185485839844,
+      "rewards/margins": 48.49779510498047,
+      "rewards/real": -16.544055938720703,
+      "step": 1010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.931009957325747e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -962.6014404296875,
+      "logps/real": -403.762939453125,
+      "loss": 0.1063,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -66.96389770507812,
+      "rewards/margins": 49.26046371459961,
+      "rewards/real": -17.703428268432617,
+      "step": 1020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.895448079658606e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -907.5885620117188,
+      "logps/real": -444.3199768066406,
+      "loss": 0.1852,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -62.3195915222168,
+      "rewards/margins": 43.62810134887695,
+      "rewards/real": -18.691490173339844,
+      "step": 1030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.859886201991465e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -943.9777221679688,
+      "logps/real": -450.1148986816406,
+      "loss": 0.237,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -65.29325866699219,
+      "rewards/margins": 46.713497161865234,
+      "rewards/real": -18.57975959777832,
+      "step": 1040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8243243243243243e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -988.4417724609375,
+      "logps/real": -427.38702392578125,
+      "loss": 0.0954,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -69.71871948242188,
+      "rewards/margins": 52.45946502685547,
+      "rewards/real": -17.259246826171875,
+      "step": 1050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7887624466571835e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -911.5084228515625,
+      "logps/real": -415.616455078125,
+      "loss": 0.1856,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -62.075843811035156,
+      "rewards/margins": 45.55692672729492,
+      "rewards/real": -16.51891326904297,
+      "step": 1060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7532005689900424e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -979.6092529296875,
+      "logps/real": -422.32720947265625,
+      "loss": 0.2408,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -68.75633239746094,
+      "rewards/margins": 51.622406005859375,
+      "rewards/real": -17.13392448425293,
+      "step": 1070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.717638691322902e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -931.5291748046875,
+      "logps/real": -425.2210388183594,
+      "loss": 0.2356,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -65.39581298828125,
+      "rewards/margins": 47.234867095947266,
+      "rewards/real": -18.160947799682617,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.6820768136557609e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -958.6076049804688,
+      "logps/real": -406.325927734375,
+      "loss": 0.0331,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -66.18185424804688,
+      "rewards/margins": 50.36635208129883,
+      "rewards/real": -15.815505981445312,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.64651493598862e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -857.2650146484375,
+      "logps/real": -447.1287536621094,
+      "loss": 0.1216,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -55.9595947265625,
+      "rewards/margins": 37.767311096191406,
+      "rewards/real": -18.192276000976562,
+      "step": 1100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.6109530583214793e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -849.4425048828125,
+      "logps/real": -399.19500732421875,
+      "loss": 0.1411,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -58.21696090698242,
+      "rewards/margins": 42.287139892578125,
+      "rewards/real": -15.929832458496094,
+      "step": 1110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5753911806543385e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -1003.6033325195312,
+      "logps/real": -437.3089294433594,
+      "loss": 0.2411,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -69.50968933105469,
+      "rewards/margins": 51.704078674316406,
+      "rewards/real": -17.805606842041016,
+      "step": 1120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5398293029871974e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -963.5474853515625,
+      "logps/real": -435.29095458984375,
+      "loss": 0.175,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -66.86241149902344,
+      "rewards/margins": 49.975059509277344,
+      "rewards/real": -16.887353897094727,
+      "step": 1130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.504267425320057e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -963.6954956054688,
+      "logps/real": -415.65960693359375,
+      "loss": 0.0674,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -67.89158630371094,
+      "rewards/margins": 50.22557830810547,
+      "rewards/real": -17.666006088256836,
+      "step": 1140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4687055476529158e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -963.4187622070312,
+      "logps/real": -441.03094482421875,
+      "loss": 0.256,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -65.62948608398438,
+      "rewards/margins": 47.720008850097656,
+      "rewards/real": -17.909475326538086,
+      "step": 1150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4331436699857753e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -948.861328125,
+      "logps/real": -444.0084533691406,
+      "loss": 0.2541,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -63.19414138793945,
+      "rewards/margins": 43.93315887451172,
+      "rewards/real": -19.26097869873047,
+      "step": 1160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3975817923186345e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -920.9754028320312,
+      "logps/real": -451.8453063964844,
+      "loss": 0.1545,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -62.874755859375,
+      "rewards/margins": 45.2825927734375,
+      "rewards/real": -17.592164993286133,
+      "step": 1170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.3620199146514935e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -962.2693481445312,
+      "logps/real": -381.82183837890625,
+      "loss": 0.1726,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -66.71905517578125,
+      "rewards/margins": 50.402278900146484,
+      "rewards/real": -16.3167667388916,
+      "step": 1180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.326458036984353e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -844.4295043945312,
+      "logps/real": -408.8478088378906,
+      "loss": 0.0521,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -55.36383819580078,
+      "rewards/margins": 38.550174713134766,
+      "rewards/real": -16.813655853271484,
+      "step": 1190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.290896159317212e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -1029.0372314453125,
+      "logps/real": -405.693359375,
+      "loss": 0.0738,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -73.27754211425781,
+      "rewards/margins": 56.71906661987305,
+      "rewards/real": -16.558481216430664,
+      "step": 1200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.255334281650071e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -955.4649658203125,
+      "logps/real": -442.18377685546875,
+      "loss": 0.1159,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -66.71927642822266,
+      "rewards/margins": 48.399192810058594,
+      "rewards/real": -18.320079803466797,
+      "step": 1210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2197724039829303e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -925.6458129882812,
+      "logps/real": -418.9791564941406,
+      "loss": 0.115,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -63.2381477355957,
+      "rewards/margins": 45.51228332519531,
+      "rewards/real": -17.725872039794922,
+      "step": 1220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1842105263157894e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -915.9041137695312,
+      "logps/real": -407.2160949707031,
+      "loss": 0.1614,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -62.57989501953125,
+      "rewards/margins": 44.657745361328125,
+      "rewards/real": -17.922143936157227,
+      "step": 1230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1486486486486487e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -895.3005981445312,
+      "logps/real": -435.64947509765625,
+      "loss": 0.1643,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -61.3538932800293,
+      "rewards/margins": 43.27713394165039,
+      "rewards/real": -18.076759338378906,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1130867709815078e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -926.37255859375,
+      "logps/real": -427.218505859375,
+      "loss": 0.1664,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -63.830535888671875,
+      "rewards/margins": 45.601234436035156,
+      "rewards/real": -18.229299545288086,
+      "step": 1250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.077524893314367e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -968.36279296875,
+      "logps/real": -382.15057373046875,
+      "loss": 0.1394,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -66.76223754882812,
+      "rewards/margins": 51.3797721862793,
+      "rewards/real": -15.382467269897461,
+      "step": 1260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0419630156472262e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -958.3225708007812,
+      "logps/real": -481.63055419921875,
+      "loss": 0.3112,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -67.70050048828125,
+      "rewards/margins": 46.94557189941406,
+      "rewards/real": -20.754926681518555,
+      "step": 1270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0064011379800854e-07,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -937.0823364257812,
+      "logps/real": -458.1944274902344,
+      "loss": 0.088,
+      "rewards/accuracies": 0.9375,
+      "rewards/generated": -64.70439910888672,
+      "rewards/margins": 45.66063690185547,
+      "rewards/real": -19.043764114379883,
+      "step": 1280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.708392603129445e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -929.6287841796875,
+      "logps/real": -412.03436279296875,
+      "loss": 0.1778,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -65.4493408203125,
+      "rewards/margins": 48.98451614379883,
+      "rewards/real": -16.46481704711914,
+      "step": 1290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.352773826458037e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -1027.0244140625,
+      "logps/real": -444.76092529296875,
+      "loss": 0.1457,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -72.01765441894531,
+      "rewards/margins": 53.2171745300293,
+      "rewards/real": -18.800477981567383,
+      "step": 1300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.997155049786629e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -937.8785400390625,
+      "logps/real": -400.3741760253906,
+      "loss": 0.0517,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -64.58556365966797,
+      "rewards/margins": 48.209327697753906,
+      "rewards/real": -16.37624168395996,
+      "step": 1310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.64153627311522e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -962.7506103515625,
+      "logps/real": -421.57952880859375,
+      "loss": 0.2135,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -67.5987777709961,
+      "rewards/margins": 49.87495803833008,
+      "rewards/real": -17.72382164001465,
+      "step": 1320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.285917496443812e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -929.7264404296875,
+      "logps/real": -431.411376953125,
+      "loss": 0.0276,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -63.279151916503906,
+      "rewards/margins": 45.76356506347656,
+      "rewards/real": -17.515583038330078,
+      "step": 1330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.930298719772404e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -960.4884033203125,
+      "logps/real": -449.7841796875,
+      "loss": 0.1428,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -67.2632827758789,
+      "rewards/margins": 49.2481689453125,
+      "rewards/real": -18.015111923217773,
+      "step": 1340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.574679943100994e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -918.400390625,
+      "logps/real": -389.68658447265625,
+      "loss": 0.2082,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -64.85960388183594,
+      "rewards/margins": 48.430328369140625,
+      "rewards/real": -16.429283142089844,
+      "step": 1350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.219061166429587e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -938.8449096679688,
+      "logps/real": -409.8709411621094,
+      "loss": 0.0844,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -66.37425994873047,
+      "rewards/margins": 50.338191986083984,
+      "rewards/real": -16.036067962646484,
+      "step": 1360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.863442389758179e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -961.57958984375,
+      "logps/real": -456.1361389160156,
+      "loss": 0.1468,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -67.77473449707031,
+      "rewards/margins": 49.818424224853516,
+      "rewards/real": -17.956310272216797,
+      "step": 1370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.507823613086771e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -1005.2136840820312,
+      "logps/real": -452.8905334472656,
+      "loss": 0.2335,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -70.96513366699219,
+      "rewards/margins": 52.77602005004883,
+      "rewards/real": -18.189117431640625,
+      "step": 1380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.152204836415363e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -981.8912963867188,
+      "logps/real": -449.5849609375,
+      "loss": 0.102,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -69.54399871826172,
+      "rewards/margins": 50.90169143676758,
+      "rewards/real": -18.642309188842773,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.796586059743954e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -939.12841796875,
+      "logps/real": -418.45855712890625,
+      "loss": 0.053,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -64.34689331054688,
+      "rewards/margins": 47.31360626220703,
+      "rewards/real": -17.03328514099121,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4409672830725456e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -996.6143798828125,
+      "logps/real": -419.6319885253906,
+      "loss": 0.1242,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -69.24372863769531,
+      "rewards/margins": 51.62445831298828,
+      "rewards/real": -17.619264602661133,
+      "step": 1410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.0853485064011376e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -1000.04052734375,
+      "logps/real": -417.8561096191406,
+      "loss": 0.0612,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -70.85627746582031,
+      "rewards/margins": 52.63084030151367,
+      "rewards/real": -18.22542953491211,
+      "step": 1420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.72972972972973e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -934.1195068359375,
+      "logps/real": -418.022705078125,
+      "loss": 0.2932,
+      "rewards/accuracies": 0.9125000238418579,
+      "rewards/generated": -66.13401794433594,
+      "rewards/margins": 50.10154342651367,
+      "rewards/real": -16.0324764251709,
+      "step": 1430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.374110953058322e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -913.4429931640625,
+      "logps/real": -439.560791015625,
+      "loss": 0.3329,
+      "rewards/accuracies": 0.925000011920929,
+      "rewards/generated": -63.869712829589844,
+      "rewards/margins": 44.8972053527832,
+      "rewards/real": -18.972515106201172,
+      "step": 1440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.018492176386913e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -911.0625,
+      "logps/real": -418.2806091308594,
+      "loss": 0.2745,
+      "rewards/accuracies": 0.949999988079071,
+      "rewards/generated": -61.86652755737305,
+      "rewards/margins": 44.91896438598633,
+      "rewards/real": -16.94756317138672,
+      "step": 1450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.6628733997155046e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -981.3883666992188,
+      "logps/real": -376.0027770996094,
+      "loss": 0.3734,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -68.52994537353516,
+      "rewards/margins": 52.24712371826172,
+      "rewards/real": -16.282825469970703,
+      "step": 1460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.3072546230440967e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -1051.210205078125,
+      "logps/real": -414.3362731933594,
+      "loss": 0.1266,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -75.56956481933594,
+      "rewards/margins": 58.789466857910156,
+      "rewards/real": -16.780101776123047,
+      "step": 1470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9516358463726884e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -964.0816650390625,
+      "logps/real": -437.93572998046875,
+      "loss": 0.1666,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -67.49874877929688,
+      "rewards/margins": 50.19408416748047,
+      "rewards/real": -17.304664611816406,
+      "step": 1480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.59601706970128e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -985.89404296875,
+      "logps/real": -457.82720947265625,
+      "loss": 0.2917,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -68.58557891845703,
+      "rewards/margins": 50.40437698364258,
+      "rewards/real": -18.181203842163086,
+      "step": 1490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.240398293029872e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -966.505859375,
+      "logps/real": -419.21337890625,
+      "loss": 0.2065,
+      "rewards/accuracies": 0.9624999761581421,
+      "rewards/generated": -67.9041519165039,
+      "rewards/margins": 51.61236572265625,
+      "rewards/real": -16.291778564453125,
+      "step": 1500
+    },
+    {
+      "epoch": 0.96,
+      "eval_logits/generated": -Infinity,
+      "eval_logits/real": -Infinity,
+      "eval_logps/generated": -659.6319580078125,
+      "eval_logps/real": -295.5523376464844,
+      "eval_loss": 0.1642562448978424,
+      "eval_rewards/accuracies": 0.962579607963562,
+      "eval_rewards/generated": -36.8196907043457,
+      "eval_rewards/margins": 32.503150939941406,
+      "eval_rewards/real": -4.316542148590088,
+      "eval_runtime": 514.2854,
+      "eval_samples_per_second": 9.722,
+      "eval_steps_per_second": 0.305,
+      "step": 1500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8847795163584636e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -906.3391723632812,
+      "logps/real": -424.758544921875,
+      "loss": 0.1254,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -62.673912048339844,
+      "rewards/margins": 45.69168472290039,
+      "rewards/real": -16.98223114013672,
+      "step": 1510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5291607396870554e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -930.1663208007812,
+      "logps/real": -418.9947204589844,
+      "loss": 0.1585,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -64.94093322753906,
+      "rewards/margins": 47.36656951904297,
+      "rewards/real": -17.574371337890625,
+      "step": 1520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1735419630156473e-08,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -991.5504760742188,
+      "logps/real": -432.77117919921875,
+      "loss": 0.0505,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -69.73775482177734,
+      "rewards/margins": 52.04203414916992,
+      "rewards/real": -17.69571876525879,
+      "step": 1530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.179231863442388e-09,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -957.3475341796875,
+      "logps/real": -448.859619140625,
+      "loss": 0.1641,
+      "rewards/accuracies": 0.9750000238418579,
+      "rewards/generated": -66.77298736572266,
+      "rewards/margins": 47.40740203857422,
+      "rewards/real": -19.36557960510254,
+      "step": 1540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.623044096728307e-09,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -999.8143310546875,
+      "logps/real": -424.55364990234375,
+      "loss": 0.0578,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/generated": -71.12496948242188,
+      "rewards/margins": 54.26820755004883,
+      "rewards/real": -16.85675621032715,
+      "step": 1550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0668563300142248e-09,
+      "logits/generated": -Infinity,
+      "logits/real": -Infinity,
+      "logps/generated": -885.8681640625,
+      "logps/real": -414.7955627441406,
+      "loss": 0.0539,
+      "rewards/accuracies": 1.0,
+      "rewards/generated": -60.05329513549805,
+      "rewards/margins": 43.077919006347656,
+      "rewards/real": -16.975379943847656,
+      "step": 1560
+    },
     {
       "epoch": 1.0,
-      "step": 782,
+      "step": 1563,
       "total_flos": 0.0,
-      "train_loss": 0.6284351689954493,
-      "train_runtime": 6560.8804,
-      "train_samples_per_second": 3.81,
-      "train_steps_per_second": 0.119
+      "train_loss": 0.42768371397759275,
+      "train_runtime": 16836.7545,
+      "train_samples_per_second": 2.97,
+      "train_steps_per_second": 0.093
     }
   ],
   "logging_steps": 10,
-  "max_steps": 782,
+  "max_steps": 1563,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 100,