diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,6624 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 3.0, + "eval_steps": 100, + "global_step": 4707, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 1.0615711252653927e-09, + "logits/generated": -2.9876155853271484, + "logits/real": -2.9668800830841064, + "logps/generated": -327.02325439453125, + "logps/real": -350.8715515136719, + "loss": 0.6931, + "rewards/accuracies": 0.0, + "rewards/generated": 0.0, + "rewards/margins": 0.0, + "rewards/real": 0.0, + "step": 1 + }, + { + "epoch": 0.01, + "learning_rate": 1.0615711252653927e-08, + "logits/generated": -2.994011878967285, + "logits/real": -2.9896292686462402, + "logps/generated": -326.0218505859375, + "logps/real": -275.2265930175781, + "loss": 0.6924, + "rewards/accuracies": 0.5, + "rewards/generated": -0.0030316433403640985, + "rewards/margins": 0.0026864688843488693, + "rewards/real": -0.0003451745433267206, + "step": 10 + }, + { + "epoch": 0.01, + "learning_rate": 2.1231422505307853e-08, + "logits/generated": -3.005816698074341, + "logits/real": -3.010007619857788, + "logps/generated": -409.13092041015625, + "logps/real": -347.76190185546875, + "loss": 0.6862, + "rewards/accuracies": 0.48750001192092896, + "rewards/generated": 0.020123440772294998, + "rewards/margins": 0.005083871074020863, + "rewards/real": 0.025207316502928734, + "step": 20 + }, + { + "epoch": 0.02, + "learning_rate": 3.184713375796178e-08, + "logits/generated": -3.0247390270233154, + "logits/real": -3.0156521797180176, + "logps/generated": -356.543212890625, + "logps/real": -292.52398681640625, + "loss": 0.6665, + "rewards/accuracies": 0.7124999761581421, + "rewards/generated": 0.022770192474126816, + "rewards/margins": 0.06586791574954987, + "rewards/real": 0.08863811194896698, + "step": 30 + }, + { + "epoch": 0.03, + "learning_rate": 4.2462845010615706e-08, + "logits/generated": -2.9926440715789795, + "logits/real": -2.9965052604675293, + "logps/generated": -390.49578857421875, + "logps/real": -362.8460998535156, + "loss": 0.6278, + "rewards/accuracies": 0.824999988079071, + "rewards/generated": 0.05502362921833992, + "rewards/margins": 0.18842557072639465, + "rewards/real": 0.24344921112060547, + "step": 40 + }, + { + "epoch": 0.03, + "learning_rate": 5.307855626326964e-08, + "logits/generated": -2.9960060119628906, + "logits/real": -2.9780421257019043, + "logps/generated": -366.8619079589844, + "logps/real": -341.0626525878906, + "loss": 0.5635, + "rewards/accuracies": 0.887499988079071, + "rewards/generated": 0.08444865047931671, + "rewards/margins": 0.3664897084236145, + "rewards/real": 0.45093828439712524, + "step": 50 + }, + { + "epoch": 0.04, + "learning_rate": 6.369426751592356e-08, + "logits/generated": -3.010775089263916, + "logits/real": -2.9793758392333984, + "logps/generated": -400.67132568359375, + "logps/real": -350.34063720703125, + "loss": 0.5167, + "rewards/accuracies": 0.762499988079071, + "rewards/generated": 0.1801263839006424, + "rewards/margins": 0.4621562957763672, + "rewards/real": 0.6422826051712036, + "step": 60 + }, + { + "epoch": 0.04, + "learning_rate": 7.43099787685775e-08, + "logits/generated": -2.993703842163086, + "logits/real": -2.9817957878112793, + "logps/generated": -389.41156005859375, + "logps/real": -316.8497619628906, + "loss": 0.4843, + "rewards/accuracies": 0.8125, + "rewards/generated": 0.17475228011608124, + "rewards/margins": 0.5869024991989136, + "rewards/real": 0.7616549730300903, + "step": 70 + }, + { + "epoch": 0.05, + "learning_rate": 8.492569002123141e-08, + "logits/generated": -2.999689817428589, + "logits/real": -2.984774589538574, + "logps/generated": -391.240234375, + "logps/real": -323.81121826171875, + "loss": 0.4319, + "rewards/accuracies": 0.8500000238418579, + "rewards/generated": 0.033304307609796524, + "rewards/margins": 0.7789232730865479, + "rewards/real": 0.8122276067733765, + "step": 80 + }, + { + "epoch": 0.06, + "learning_rate": 9.554140127388536e-08, + "logits/generated": -3.0078229904174805, + "logits/real": -2.9905121326446533, + "logps/generated": -395.6484375, + "logps/real": -291.0595703125, + "loss": 0.3843, + "rewards/accuracies": 0.8500000238418579, + "rewards/generated": -0.04171593859791756, + "rewards/margins": 0.9879744648933411, + "rewards/real": 0.9462584257125854, + "step": 90 + }, + { + "epoch": 0.06, + "learning_rate": 1.0615711252653928e-07, + "logits/generated": -2.9578118324279785, + "logits/real": -2.9447882175445557, + "logps/generated": -351.89630126953125, + "logps/real": -282.14617919921875, + "loss": 0.33, + "rewards/accuracies": 0.8999999761581421, + "rewards/generated": -0.26103243231773376, + "rewards/margins": 1.1919409036636353, + "rewards/real": 0.9309085011482239, + "step": 100 + }, + { + "epoch": 0.07, + "learning_rate": 1.1677282377919321e-07, + "logits/generated": -2.94851016998291, + "logits/real": -2.952113151550293, + "logps/generated": -417.225341796875, + "logps/real": -356.7054138183594, + "loss": 0.3059, + "rewards/accuracies": 0.9375, + "rewards/generated": -0.5314805507659912, + "rewards/margins": 1.7437711954116821, + "rewards/real": 1.2122905254364014, + "step": 110 + }, + { + "epoch": 0.08, + "learning_rate": 1.2738853503184713e-07, + "logits/generated": -2.931594133377075, + "logits/real": -2.935559034347534, + "logps/generated": -349.97021484375, + "logps/real": -313.77239990234375, + "loss": 0.2915, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -0.5922194123268127, + "rewards/margins": 1.5014526844024658, + "rewards/real": 0.9092333912849426, + "step": 120 + }, + { + "epoch": 0.08, + "learning_rate": 1.3800424628450104e-07, + "logits/generated": -2.9275901317596436, + "logits/real": -2.9150493144989014, + "logps/generated": -347.41094970703125, + "logps/real": -317.4172668457031, + "loss": 0.2766, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -0.914065957069397, + "rewards/margins": 1.9470272064208984, + "rewards/real": 1.0329612493515015, + "step": 130 + }, + { + "epoch": 0.09, + "learning_rate": 1.48619957537155e-07, + "logits/generated": -2.924373149871826, + "logits/real": -2.92722487449646, + "logps/generated": -444.05999755859375, + "logps/real": -333.08721923828125, + "loss": 0.2223, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -1.682334542274475, + "rewards/margins": 2.734717845916748, + "rewards/real": 1.0523834228515625, + "step": 140 + }, + { + "epoch": 0.1, + "learning_rate": 1.592356687898089e-07, + "logits/generated": -2.9124810695648193, + "logits/real": -2.912565231323242, + "logps/generated": -479.41290283203125, + "logps/real": -348.48736572265625, + "loss": 0.182, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -1.9681695699691772, + "rewards/margins": 2.729921817779541, + "rewards/real": 0.7617522478103638, + "step": 150 + }, + { + "epoch": 0.1, + "learning_rate": 1.6985138004246283e-07, + "logits/generated": -2.929657459259033, + "logits/real": -2.91477108001709, + "logps/generated": -372.936279296875, + "logps/real": -321.5260925292969, + "loss": 0.2272, + "rewards/accuracies": 0.8999999761581421, + "rewards/generated": -2.243396759033203, + "rewards/margins": 2.916565418243408, + "rewards/real": 0.6731682419776917, + "step": 160 + }, + { + "epoch": 0.11, + "learning_rate": 1.8046709129511674e-07, + "logits/generated": -2.91805100440979, + "logits/real": -2.916639804840088, + "logps/generated": -383.2129821777344, + "logps/real": -289.7926025390625, + "loss": 0.1763, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -2.5758872032165527, + "rewards/margins": 3.038135051727295, + "rewards/real": 0.4622476100921631, + "step": 170 + }, + { + "epoch": 0.11, + "learning_rate": 1.9108280254777072e-07, + "logits/generated": -2.8736469745635986, + "logits/real": -2.8829410076141357, + "logps/generated": -401.41387939453125, + "logps/real": -314.1357421875, + "loss": 0.1723, + "rewards/accuracies": 0.8999999761581421, + "rewards/generated": -3.268184185028076, + "rewards/margins": 3.375507354736328, + "rewards/real": 0.10732327401638031, + "step": 180 + }, + { + "epoch": 0.12, + "learning_rate": 2.0169851380042463e-07, + "logits/generated": -2.873676300048828, + "logits/real": -2.8592188358306885, + "logps/generated": -417.1192932128906, + "logps/real": -314.2078857421875, + "loss": 0.1933, + "rewards/accuracies": 0.9375, + "rewards/generated": -3.342254161834717, + "rewards/margins": 3.372480869293213, + "rewards/real": 0.030226852744817734, + "step": 190 + }, + { + "epoch": 0.13, + "learning_rate": 2.1231422505307855e-07, + "logits/generated": -2.8424830436706543, + "logits/real": -2.8474855422973633, + "logps/generated": -422.8392028808594, + "logps/real": -305.7722473144531, + "loss": 0.1597, + "rewards/accuracies": 0.9375, + "rewards/generated": -3.6326286792755127, + "rewards/margins": 3.787651777267456, + "rewards/real": 0.155023455619812, + "step": 200 + }, + { + "epoch": 0.13, + "learning_rate": 2.2292993630573247e-07, + "logits/generated": -2.8330202102661133, + "logits/real": -2.8425629138946533, + "logps/generated": -459.2235412597656, + "logps/real": -349.0935974121094, + "loss": 0.179, + "rewards/accuracies": 0.8999999761581421, + "rewards/generated": -4.4774346351623535, + "rewards/margins": 3.782116413116455, + "rewards/real": -0.6953185796737671, + "step": 210 + }, + { + "epoch": 0.14, + "learning_rate": 2.3354564755838642e-07, + "logits/generated": -2.8761401176452637, + "logits/real": -2.8607993125915527, + "logps/generated": -411.73101806640625, + "logps/real": -320.45489501953125, + "loss": 0.1427, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -3.940058946609497, + "rewards/margins": 4.07708740234375, + "rewards/real": 0.13702815771102905, + "step": 220 + }, + { + "epoch": 0.15, + "learning_rate": 2.4416135881104033e-07, + "logits/generated": -2.8605635166168213, + "logits/real": -2.839541435241699, + "logps/generated": -371.6196594238281, + "logps/real": -329.2597351074219, + "loss": 0.175, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -3.8501219749450684, + "rewards/margins": 4.0448479652404785, + "rewards/real": 0.19472622871398926, + "step": 230 + }, + { + "epoch": 0.15, + "learning_rate": 2.5477707006369425e-07, + "logits/generated": -2.8058648109436035, + "logits/real": -2.7639129161834717, + "logps/generated": -424.64178466796875, + "logps/real": -354.20391845703125, + "loss": 0.1562, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -3.804414749145508, + "rewards/margins": 3.9232983589172363, + "rewards/real": 0.11888311803340912, + "step": 240 + }, + { + "epoch": 0.16, + "learning_rate": 2.6539278131634817e-07, + "logits/generated": -2.7883121967315674, + "logits/real": -2.741762161254883, + "logps/generated": -402.73724365234375, + "logps/real": -350.4950256347656, + "loss": 0.1242, + "rewards/accuracies": 0.925000011920929, + "rewards/generated": -3.8438148498535156, + "rewards/margins": 3.8519673347473145, + "rewards/real": 0.008152651600539684, + "step": 250 + }, + { + "epoch": 0.17, + "learning_rate": 2.760084925690021e-07, + "logits/generated": -2.759934663772583, + "logits/real": -2.7293505668640137, + "logps/generated": -453.0765075683594, + "logps/real": -317.9742126464844, + "loss": 0.1148, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -5.475944519042969, + "rewards/margins": 5.090882301330566, + "rewards/real": -0.385062038898468, + "step": 260 + }, + { + "epoch": 0.17, + "learning_rate": 2.86624203821656e-07, + "logits/generated": -2.7776904106140137, + "logits/real": -2.733156681060791, + "logps/generated": -386.65521240234375, + "logps/real": -285.1985778808594, + "loss": 0.1439, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -4.516385555267334, + "rewards/margins": 4.166232585906982, + "rewards/real": -0.3501525819301605, + "step": 270 + }, + { + "epoch": 0.18, + "learning_rate": 2.9723991507431e-07, + "logits/generated": -2.7636427879333496, + "logits/real": -2.687100410461426, + "logps/generated": -422.11431884765625, + "logps/real": -290.365234375, + "loss": 0.131, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -4.704317569732666, + "rewards/margins": 4.192616939544678, + "rewards/real": -0.5117009878158569, + "step": 280 + }, + { + "epoch": 0.18, + "learning_rate": 3.078556263269639e-07, + "logits/generated": -2.7474493980407715, + "logits/real": -2.6468071937561035, + "logps/generated": -467.47125244140625, + "logps/real": -346.7054443359375, + "loss": 0.1298, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -5.134922981262207, + "rewards/margins": 4.830273628234863, + "rewards/real": -0.30464985966682434, + "step": 290 + }, + { + "epoch": 0.19, + "learning_rate": 3.184713375796178e-07, + "logits/generated": -2.6696765422821045, + "logits/real": -2.6201140880584717, + "logps/generated": -401.53314208984375, + "logps/real": -329.912353515625, + "loss": 0.1334, + "rewards/accuracies": 0.925000011920929, + "rewards/generated": -5.16738748550415, + "rewards/margins": 4.69888973236084, + "rewards/real": -0.4684976041316986, + "step": 300 + }, + { + "epoch": 0.2, + "learning_rate": 3.2908704883227173e-07, + "logits/generated": -2.782980442047119, + "logits/real": -2.6915059089660645, + "logps/generated": -438.428466796875, + "logps/real": -361.34375, + "loss": 0.1072, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -5.9074907302856445, + "rewards/margins": 5.830879211425781, + "rewards/real": -0.07661163806915283, + "step": 310 + }, + { + "epoch": 0.2, + "learning_rate": 3.3970276008492565e-07, + "logits/generated": -2.7609965801239014, + "logits/real": -2.674025058746338, + "logps/generated": -447.04681396484375, + "logps/real": -331.8215026855469, + "loss": 0.0812, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -6.188752174377441, + "rewards/margins": 6.114731311798096, + "rewards/real": -0.07402036339044571, + "step": 320 + }, + { + "epoch": 0.21, + "learning_rate": 3.5031847133757957e-07, + "logits/generated": -2.7006940841674805, + "logits/real": -2.652888774871826, + "logps/generated": -482.1563415527344, + "logps/real": -371.65032958984375, + "loss": 0.1593, + "rewards/accuracies": 0.925000011920929, + "rewards/generated": -6.773266792297363, + "rewards/margins": 5.90455961227417, + "rewards/real": -0.8687070608139038, + "step": 330 + }, + { + "epoch": 0.22, + "learning_rate": 3.609341825902335e-07, + "logits/generated": -2.71394419670105, + "logits/real": -2.675811767578125, + "logps/generated": -477.96697998046875, + "logps/real": -414.2215270996094, + "loss": 0.1075, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -6.758129119873047, + "rewards/margins": 6.150029182434082, + "rewards/real": -0.608100950717926, + "step": 340 + }, + { + "epoch": 0.22, + "learning_rate": 3.715498938428874e-07, + "logits/generated": -2.646404504776001, + "logits/real": -2.6175625324249268, + "logps/generated": -421.092041015625, + "logps/real": -295.613037109375, + "loss": 0.1311, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -6.8629302978515625, + "rewards/margins": 5.784576416015625, + "rewards/real": -1.0783538818359375, + "step": 350 + }, + { + "epoch": 0.23, + "learning_rate": 3.8216560509554143e-07, + "logits/generated": -2.6539955139160156, + "logits/real": -2.5983927249908447, + "logps/generated": -430.94647216796875, + "logps/real": -341.7402038574219, + "loss": 0.0935, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -7.6072492599487305, + "rewards/margins": 6.59588623046875, + "rewards/real": -1.0113630294799805, + "step": 360 + }, + { + "epoch": 0.24, + "learning_rate": 3.9278131634819535e-07, + "logits/generated": -2.6768739223480225, + "logits/real": -2.563809871673584, + "logps/generated": -514.014404296875, + "logps/real": -382.5062561035156, + "loss": 0.1021, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -8.356344223022461, + "rewards/margins": 8.064131736755371, + "rewards/real": -0.2922106683254242, + "step": 370 + }, + { + "epoch": 0.24, + "learning_rate": 4.0339702760084927e-07, + "logits/generated": -2.5732269287109375, + "logits/real": -2.5259804725646973, + "logps/generated": -436.3158264160156, + "logps/real": -307.83123779296875, + "loss": 0.1245, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -7.608613014221191, + "rewards/margins": 6.9700751304626465, + "rewards/real": -0.6385382413864136, + "step": 380 + }, + { + "epoch": 0.25, + "learning_rate": 4.140127388535032e-07, + "logits/generated": -2.661893606185913, + "logits/real": -2.617318630218506, + "logps/generated": -438.4296875, + "logps/real": -311.55206298828125, + "loss": 0.1313, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -6.864607334136963, + "rewards/margins": 6.348482608795166, + "rewards/real": -0.5161241292953491, + "step": 390 + }, + { + "epoch": 0.25, + "learning_rate": 4.246284501061571e-07, + "logits/generated": -2.645270586013794, + "logits/real": -2.6025874614715576, + "logps/generated": -376.67596435546875, + "logps/real": -281.83917236328125, + "loss": 0.0961, + "rewards/accuracies": 0.9375, + "rewards/generated": -6.3483428955078125, + "rewards/margins": 6.382236480712891, + "rewards/real": 0.03389265388250351, + "step": 400 + }, + { + "epoch": 0.26, + "learning_rate": 4.35244161358811e-07, + "logits/generated": -2.6498265266418457, + "logits/real": -2.6190037727355957, + "logps/generated": -461.171142578125, + "logps/real": -365.7615661621094, + "loss": 0.1003, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -6.8866987228393555, + "rewards/margins": 6.904853820800781, + "rewards/real": 0.01815515197813511, + "step": 410 + }, + { + "epoch": 0.27, + "learning_rate": 4.4585987261146494e-07, + "logits/generated": -2.6873116493225098, + "logits/real": -2.5698437690734863, + "logps/generated": -466.5087890625, + "logps/real": -342.3390808105469, + "loss": 0.0974, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -7.907741546630859, + "rewards/margins": 7.305785179138184, + "rewards/real": -0.6019567847251892, + "step": 420 + }, + { + "epoch": 0.27, + "learning_rate": 4.5647558386411886e-07, + "logits/generated": -2.5971856117248535, + "logits/real": -2.5531811714172363, + "logps/generated": -424.07806396484375, + "logps/real": -327.88037109375, + "loss": 0.147, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -7.188799858093262, + "rewards/margins": 6.112451076507568, + "rewards/real": -1.0763494968414307, + "step": 430 + }, + { + "epoch": 0.28, + "learning_rate": 4.6709129511677283e-07, + "logits/generated": -2.5770835876464844, + "logits/real": -2.5735411643981934, + "logps/generated": -423.05889892578125, + "logps/real": -304.71649169921875, + "loss": 0.1152, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -7.318545341491699, + "rewards/margins": 6.307910919189453, + "rewards/real": -1.0106346607208252, + "step": 440 + }, + { + "epoch": 0.29, + "learning_rate": 4.777070063694267e-07, + "logits/generated": -2.6009068489074707, + "logits/real": -2.5000970363616943, + "logps/generated": -543.5272827148438, + "logps/real": -344.96783447265625, + "loss": 0.1038, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -9.782231330871582, + "rewards/margins": 8.415586471557617, + "rewards/real": -1.3666445016860962, + "step": 450 + }, + { + "epoch": 0.29, + "learning_rate": 4.883227176220807e-07, + "logits/generated": -2.539663553237915, + "logits/real": -2.4826877117156982, + "logps/generated": -464.3202209472656, + "logps/real": -398.4128112792969, + "loss": 0.1128, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -9.191587448120117, + "rewards/margins": 8.196949005126953, + "rewards/real": -0.9946379661560059, + "step": 460 + }, + { + "epoch": 0.3, + "learning_rate": 4.989384288747346e-07, + "logits/generated": -2.6102230548858643, + "logits/real": -2.5595173835754395, + "logps/generated": -505.9552307128906, + "logps/real": -345.22357177734375, + "loss": 0.0904, + "rewards/accuracies": 0.9375, + "rewards/generated": -9.173894882202148, + "rewards/margins": 8.157112121582031, + "rewards/real": -1.0167815685272217, + "step": 470 + }, + { + "epoch": 0.31, + "learning_rate": 4.989376770538244e-07, + "logits/generated": -2.714066743850708, + "logits/real": -2.6313273906707764, + "logps/generated": -487.18206787109375, + "logps/real": -347.70965576171875, + "loss": 0.0827, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -8.938846588134766, + "rewards/margins": 7.621697902679443, + "rewards/real": -1.3171485662460327, + "step": 480 + }, + { + "epoch": 0.31, + "learning_rate": 4.977573182247403e-07, + "logits/generated": -2.6655235290527344, + "logits/real": -2.632723569869995, + "logps/generated": -449.29071044921875, + "logps/real": -348.1429748535156, + "loss": 0.1336, + "rewards/accuracies": 0.887499988079071, + "rewards/generated": -7.1729841232299805, + "rewards/margins": 6.382136344909668, + "rewards/real": -0.7908482551574707, + "step": 490 + }, + { + "epoch": 0.32, + "learning_rate": 4.965769593956563e-07, + "logits/generated": -2.676328420639038, + "logits/real": -2.571633815765381, + "logps/generated": -491.0113830566406, + "logps/real": -365.57830810546875, + "loss": 0.1138, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -8.424352645874023, + "rewards/margins": 7.796865940093994, + "rewards/real": -0.6274875402450562, + "step": 500 + }, + { + "epoch": 0.33, + "learning_rate": 4.953966005665722e-07, + "logits/generated": -2.5785083770751953, + "logits/real": -2.528327465057373, + "logps/generated": -488.34222412109375, + "logps/real": -366.8738708496094, + "loss": 0.0863, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -8.961752891540527, + "rewards/margins": 7.542865753173828, + "rewards/real": -1.4188873767852783, + "step": 510 + }, + { + "epoch": 0.33, + "learning_rate": 4.942162417374882e-07, + "logits/generated": -2.562635898590088, + "logits/real": -2.4578804969787598, + "logps/generated": -472.02008056640625, + "logps/real": -336.67120361328125, + "loss": 0.1157, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -9.987321853637695, + "rewards/margins": 8.158964157104492, + "rewards/real": -1.8283584117889404, + "step": 520 + }, + { + "epoch": 0.34, + "learning_rate": 4.930358829084041e-07, + "logits/generated": -2.55583119392395, + "logits/real": -2.5184712409973145, + "logps/generated": -487.912109375, + "logps/real": -379.39935302734375, + "loss": 0.074, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -8.376862525939941, + "rewards/margins": 7.726510047912598, + "rewards/real": -0.6503523588180542, + "step": 530 + }, + { + "epoch": 0.34, + "learning_rate": 4.918555240793201e-07, + "logits/generated": -2.618487596511841, + "logits/real": -2.5379557609558105, + "logps/generated": -494.197998046875, + "logps/real": -361.53485107421875, + "loss": 0.0834, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -8.759679794311523, + "rewards/margins": 7.890118598937988, + "rewards/real": -0.8695621490478516, + "step": 540 + }, + { + "epoch": 0.35, + "learning_rate": 4.90675165250236e-07, + "logits/generated": -2.594207763671875, + "logits/real": -2.52410626411438, + "logps/generated": -475.1624450683594, + "logps/real": -352.5563049316406, + "loss": 0.1208, + "rewards/accuracies": 0.8999999761581421, + "rewards/generated": -8.052266120910645, + "rewards/margins": 7.429941654205322, + "rewards/real": -0.6223247647285461, + "step": 550 + }, + { + "epoch": 0.36, + "learning_rate": 4.89494806421152e-07, + "logits/generated": -2.555713176727295, + "logits/real": -2.5284132957458496, + "logps/generated": -445.69781494140625, + "logps/real": -344.16351318359375, + "loss": 0.0951, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -8.230698585510254, + "rewards/margins": 7.527130126953125, + "rewards/real": -0.703567624092102, + "step": 560 + }, + { + "epoch": 0.36, + "learning_rate": 4.883144475920679e-07, + "logits/generated": -2.6265387535095215, + "logits/real": -2.549365282058716, + "logps/generated": -505.12396240234375, + "logps/real": -377.36328125, + "loss": 0.1085, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -9.444196701049805, + "rewards/margins": 8.654594421386719, + "rewards/real": -0.7896023988723755, + "step": 570 + }, + { + "epoch": 0.37, + "learning_rate": 4.871340887629839e-07, + "logits/generated": -2.5271811485290527, + "logits/real": -2.543853759765625, + "logps/generated": -486.79571533203125, + "logps/real": -333.89434814453125, + "loss": 0.0946, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -8.724726676940918, + "rewards/margins": 7.599035739898682, + "rewards/real": -1.1256901025772095, + "step": 580 + }, + { + "epoch": 0.38, + "learning_rate": 4.859537299338998e-07, + "logits/generated": -2.4935667514801025, + "logits/real": -2.45051908493042, + "logps/generated": -492.9710998535156, + "logps/real": -351.0712585449219, + "loss": 0.0935, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -10.154314994812012, + "rewards/margins": 8.238804817199707, + "rewards/real": -1.915509819984436, + "step": 590 + }, + { + "epoch": 0.38, + "learning_rate": 4.847733711048158e-07, + "logits/generated": -2.6015708446502686, + "logits/real": -2.54630970954895, + "logps/generated": -517.557861328125, + "logps/real": -343.5694885253906, + "loss": 0.1067, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -9.994497299194336, + "rewards/margins": 8.133109092712402, + "rewards/real": -1.8613884449005127, + "step": 600 + }, + { + "epoch": 0.39, + "learning_rate": 4.835930122757317e-07, + "logits/generated": -2.555729389190674, + "logits/real": -2.503249168395996, + "logps/generated": -511.1414489746094, + "logps/real": -329.421142578125, + "loss": 0.0843, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -10.898144721984863, + "rewards/margins": 8.920883178710938, + "rewards/real": -1.9772602319717407, + "step": 610 + }, + { + "epoch": 0.4, + "learning_rate": 4.824126534466478e-07, + "logits/generated": -2.5740256309509277, + "logits/real": -2.5255706310272217, + "logps/generated": -493.5047912597656, + "logps/real": -368.59832763671875, + "loss": 0.0901, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -10.093526840209961, + "rewards/margins": 8.574692726135254, + "rewards/real": -1.5188337564468384, + "step": 620 + }, + { + "epoch": 0.4, + "learning_rate": 4.812322946175637e-07, + "logits/generated": -2.5761618614196777, + "logits/real": -2.5374348163604736, + "logps/generated": -521.7885131835938, + "logps/real": -363.25653076171875, + "loss": 0.0763, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -10.228660583496094, + "rewards/margins": 9.312238693237305, + "rewards/real": -0.9164208173751831, + "step": 630 + }, + { + "epoch": 0.41, + "learning_rate": 4.800519357884797e-07, + "logits/generated": -2.6331121921539307, + "logits/real": -2.5574283599853516, + "logps/generated": -499.1038513183594, + "logps/real": -343.503173828125, + "loss": 0.1142, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -10.20173168182373, + "rewards/margins": 8.458353996276855, + "rewards/real": -1.743377685546875, + "step": 640 + }, + { + "epoch": 0.41, + "learning_rate": 4.788715769593956e-07, + "logits/generated": -2.56269907951355, + "logits/real": -2.4889848232269287, + "logps/generated": -450.4479064941406, + "logps/real": -311.9686584472656, + "loss": 0.0953, + "rewards/accuracies": 1.0, + "rewards/generated": -9.332066535949707, + "rewards/margins": 8.472679138183594, + "rewards/real": -0.859386146068573, + "step": 650 + }, + { + "epoch": 0.42, + "learning_rate": 4.776912181303116e-07, + "logits/generated": -2.5874459743499756, + "logits/real": -2.4983246326446533, + "logps/generated": -525.3177490234375, + "logps/real": -414.58197021484375, + "loss": 0.0955, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -10.059281349182129, + "rewards/margins": 8.511804580688477, + "rewards/real": -1.5474772453308105, + "step": 660 + }, + { + "epoch": 0.43, + "learning_rate": 4.7651085930122754e-07, + "logits/generated": -2.509523868560791, + "logits/real": -2.488023042678833, + "logps/generated": -448.4249572753906, + "logps/real": -313.8059997558594, + "loss": 0.1313, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -10.903493881225586, + "rewards/margins": 8.945398330688477, + "rewards/real": -1.9580967426300049, + "step": 670 + }, + { + "epoch": 0.43, + "learning_rate": 4.753305004721435e-07, + "logits/generated": -2.5889456272125244, + "logits/real": -2.540140390396118, + "logps/generated": -477.789306640625, + "logps/real": -343.6940002441406, + "loss": 0.0754, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -8.771103858947754, + "rewards/margins": 7.136463165283203, + "rewards/real": -1.6346410512924194, + "step": 680 + }, + { + "epoch": 0.44, + "learning_rate": 4.7415014164305945e-07, + "logits/generated": -2.618513822555542, + "logits/real": -2.555964231491089, + "logps/generated": -537.254638671875, + "logps/real": -370.8758544921875, + "loss": 0.0822, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -9.57807445526123, + "rewards/margins": 8.652335166931152, + "rewards/real": -0.9257386922836304, + "step": 690 + }, + { + "epoch": 0.45, + "learning_rate": 4.729697828139754e-07, + "logits/generated": -2.5748181343078613, + "logits/real": -2.5802690982818604, + "logps/generated": -495.4300842285156, + "logps/real": -362.72967529296875, + "loss": 0.1164, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -8.938674926757812, + "rewards/margins": 8.537571907043457, + "rewards/real": -0.4011026918888092, + "step": 700 + }, + { + "epoch": 0.45, + "learning_rate": 4.717894239848914e-07, + "logits/generated": -2.5499825477600098, + "logits/real": -2.5337607860565186, + "logps/generated": -455.4185485839844, + "logps/real": -313.13494873046875, + "loss": 0.0998, + "rewards/accuracies": 0.925000011920929, + "rewards/generated": -9.539193153381348, + "rewards/margins": 7.804329872131348, + "rewards/real": -1.734863042831421, + "step": 710 + }, + { + "epoch": 0.46, + "learning_rate": 4.706090651558073e-07, + "logits/generated": -2.5695395469665527, + "logits/real": -2.5750317573547363, + "logps/generated": -471.41064453125, + "logps/real": -362.0666809082031, + "loss": 0.1361, + "rewards/accuracies": 0.9375, + "rewards/generated": -7.691230773925781, + "rewards/margins": 6.894305229187012, + "rewards/real": -0.7969250679016113, + "step": 720 + }, + { + "epoch": 0.47, + "learning_rate": 4.694287063267233e-07, + "logits/generated": -2.618682384490967, + "logits/real": -2.5621445178985596, + "logps/generated": -506.92755126953125, + "logps/real": -389.9350891113281, + "loss": 0.1201, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -9.84778118133545, + "rewards/margins": 8.695353507995605, + "rewards/real": -1.1524274349212646, + "step": 730 + }, + { + "epoch": 0.47, + "learning_rate": 4.6824834749763923e-07, + "logits/generated": -2.6112401485443115, + "logits/real": -2.5744032859802246, + "logps/generated": -495.9729919433594, + "logps/real": -358.03985595703125, + "loss": 0.0557, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -9.461344718933105, + "rewards/margins": 8.752626419067383, + "rewards/real": -0.7087193727493286, + "step": 740 + }, + { + "epoch": 0.48, + "learning_rate": 4.6706798866855524e-07, + "logits/generated": -2.4730710983276367, + "logits/real": -2.5215208530426025, + "logps/generated": -500.0975646972656, + "logps/real": -367.23406982421875, + "loss": 0.0949, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -9.66076374053955, + "rewards/margins": 7.486401557922363, + "rewards/real": -2.174362897872925, + "step": 750 + }, + { + "epoch": 0.48, + "learning_rate": 4.6588762983947114e-07, + "logits/generated": -2.487940549850464, + "logits/real": -2.448495388031006, + "logps/generated": -511.4491271972656, + "logps/real": -346.4087829589844, + "loss": 0.1014, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -11.974629402160645, + "rewards/margins": 9.291276931762695, + "rewards/real": -2.6833527088165283, + "step": 760 + }, + { + "epoch": 0.49, + "learning_rate": 4.6470727101038715e-07, + "logits/generated": -2.4769680500030518, + "logits/real": -2.4194235801696777, + "logps/generated": -474.8147888183594, + "logps/real": -362.15716552734375, + "loss": 0.138, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -10.335468292236328, + "rewards/margins": 8.036766052246094, + "rewards/real": -2.298701763153076, + "step": 770 + }, + { + "epoch": 0.5, + "learning_rate": 4.6352691218130305e-07, + "logits/generated": -2.5269980430603027, + "logits/real": -2.460627555847168, + "logps/generated": -476.22271728515625, + "logps/real": -337.52362060546875, + "loss": 0.0954, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -9.532716751098633, + "rewards/margins": 7.73496150970459, + "rewards/real": -1.7977555990219116, + "step": 780 + }, + { + "epoch": 0.5, + "learning_rate": 4.6234655335221906e-07, + "logits/generated": -2.452920436859131, + "logits/real": -2.397965669631958, + "logps/generated": -465.40869140625, + "logps/real": -334.29058837890625, + "loss": 0.096, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -11.273387908935547, + "rewards/margins": 9.32444953918457, + "rewards/real": -1.9489390850067139, + "step": 790 + }, + { + "epoch": 0.51, + "learning_rate": 4.6116619452313496e-07, + "logits/generated": -2.5272440910339355, + "logits/real": -2.490408182144165, + "logps/generated": -457.65283203125, + "logps/real": -360.3447265625, + "loss": 0.1199, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -7.943813323974609, + "rewards/margins": 7.986338138580322, + "rewards/real": 0.042525794357061386, + "step": 800 + }, + { + "epoch": 0.52, + "learning_rate": 4.5998583569405097e-07, + "logits/generated": -2.576383113861084, + "logits/real": -2.5489490032196045, + "logps/generated": -510.33013916015625, + "logps/real": -372.75982666015625, + "loss": 0.0783, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -9.628268241882324, + "rewards/margins": 8.580741882324219, + "rewards/real": -1.0475265979766846, + "step": 810 + }, + { + "epoch": 0.52, + "learning_rate": 4.58805476864967e-07, + "logits/generated": -2.577169895172119, + "logits/real": -2.541865348815918, + "logps/generated": -499.7286682128906, + "logps/real": -350.94915771484375, + "loss": 0.1226, + "rewards/accuracies": 0.9375, + "rewards/generated": -9.23862075805664, + "rewards/margins": 8.5203275680542, + "rewards/real": -0.7182931900024414, + "step": 820 + }, + { + "epoch": 0.53, + "learning_rate": 4.576251180358829e-07, + "logits/generated": -2.556372880935669, + "logits/real": -2.533060312271118, + "logps/generated": -481.6539611816406, + "logps/real": -363.9843444824219, + "loss": 0.0938, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -8.849392890930176, + "rewards/margins": 8.888776779174805, + "rewards/real": 0.03938591107726097, + "step": 830 + }, + { + "epoch": 0.54, + "learning_rate": 4.564447592067989e-07, + "logits/generated": -2.6314210891723633, + "logits/real": -2.5852346420288086, + "logps/generated": -509.93231201171875, + "logps/real": -393.7592468261719, + "loss": 0.0929, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -9.991052627563477, + "rewards/margins": 8.720523834228516, + "rewards/real": -1.2705291509628296, + "step": 840 + }, + { + "epoch": 0.54, + "learning_rate": 4.552644003777148e-07, + "logits/generated": -2.6035678386688232, + "logits/real": -2.567884922027588, + "logps/generated": -530.4461059570312, + "logps/real": -403.38238525390625, + "loss": 0.0656, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -10.836085319519043, + "rewards/margins": 9.035760879516602, + "rewards/real": -1.80032479763031, + "step": 850 + }, + { + "epoch": 0.55, + "learning_rate": 4.540840415486308e-07, + "logits/generated": -2.5661263465881348, + "logits/real": -2.5603199005126953, + "logps/generated": -506.3961486816406, + "logps/real": -368.25244140625, + "loss": 0.1355, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -10.33775520324707, + "rewards/margins": 8.915458679199219, + "rewards/real": -1.4222954511642456, + "step": 860 + }, + { + "epoch": 0.55, + "learning_rate": 4.529036827195467e-07, + "logits/generated": -2.552281379699707, + "logits/real": -2.5277304649353027, + "logps/generated": -490.99676513671875, + "logps/real": -356.09033203125, + "loss": 0.0662, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -10.086469650268555, + "rewards/margins": 9.019510269165039, + "rewards/real": -1.0669586658477783, + "step": 870 + }, + { + "epoch": 0.56, + "learning_rate": 4.517233238904627e-07, + "logits/generated": -2.5234899520874023, + "logits/real": -2.5082201957702637, + "logps/generated": -491.60516357421875, + "logps/real": -353.71380615234375, + "loss": 0.0527, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -11.373248100280762, + "rewards/margins": 9.473039627075195, + "rewards/real": -1.9002081155776978, + "step": 880 + }, + { + "epoch": 0.57, + "learning_rate": 4.505429650613786e-07, + "logits/generated": -2.4585375785827637, + "logits/real": -2.41652250289917, + "logps/generated": -543.7042846679688, + "logps/real": -383.6292419433594, + "loss": 0.1025, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -12.099660873413086, + "rewards/margins": 9.468574523925781, + "rewards/real": -2.6310875415802, + "step": 890 + }, + { + "epoch": 0.57, + "learning_rate": 4.4936260623229463e-07, + "logits/generated": -2.4282710552215576, + "logits/real": -2.3773722648620605, + "logps/generated": -484.602783203125, + "logps/real": -374.7772521972656, + "loss": 0.1613, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -10.475399017333984, + "rewards/margins": 8.948919296264648, + "rewards/real": -1.526479721069336, + "step": 900 + }, + { + "epoch": 0.58, + "learning_rate": 4.4818224740321053e-07, + "logits/generated": -2.39573335647583, + "logits/real": -2.372222423553467, + "logps/generated": -521.895751953125, + "logps/real": -361.3406677246094, + "loss": 0.1039, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -10.608845710754395, + "rewards/margins": 9.889520645141602, + "rewards/real": -0.7193254232406616, + "step": 910 + }, + { + "epoch": 0.59, + "learning_rate": 4.4700188857412654e-07, + "logits/generated": -2.39178729057312, + "logits/real": -2.330113410949707, + "logps/generated": -548.0231323242188, + "logps/real": -376.58148193359375, + "loss": 0.1294, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -10.68170166015625, + "rewards/margins": 8.753602981567383, + "rewards/real": -1.9280989170074463, + "step": 920 + }, + { + "epoch": 0.59, + "learning_rate": 4.4582152974504244e-07, + "logits/generated": -2.447739601135254, + "logits/real": -2.3855857849121094, + "logps/generated": -476.3338928222656, + "logps/real": -323.8772888183594, + "loss": 0.1041, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -10.131559371948242, + "rewards/margins": 8.769943237304688, + "rewards/real": -1.3616163730621338, + "step": 930 + }, + { + "epoch": 0.6, + "learning_rate": 4.4464117091595845e-07, + "logits/generated": -2.531923770904541, + "logits/real": -2.493612289428711, + "logps/generated": -516.0556030273438, + "logps/real": -335.58062744140625, + "loss": 0.0783, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -9.325234413146973, + "rewards/margins": 8.22175407409668, + "rewards/real": -1.103480339050293, + "step": 940 + }, + { + "epoch": 0.61, + "learning_rate": 4.434608120868744e-07, + "logits/generated": -2.5285017490386963, + "logits/real": -2.549332857131958, + "logps/generated": -485.10382080078125, + "logps/real": -340.98431396484375, + "loss": 0.1074, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -10.05161190032959, + "rewards/margins": 8.83989429473877, + "rewards/real": -1.211718201637268, + "step": 950 + }, + { + "epoch": 0.61, + "learning_rate": 4.4228045325779036e-07, + "logits/generated": -2.531179428100586, + "logits/real": -2.527667999267578, + "logps/generated": -523.0577392578125, + "logps/real": -361.6462707519531, + "loss": 0.1249, + "rewards/accuracies": 0.9375, + "rewards/generated": -9.758930206298828, + "rewards/margins": 8.803415298461914, + "rewards/real": -0.9555162191390991, + "step": 960 + }, + { + "epoch": 0.62, + "learning_rate": 4.411000944287063e-07, + "logits/generated": -2.5421040058135986, + "logits/real": -2.4952003955841064, + "logps/generated": -483.36016845703125, + "logps/real": -328.30133056640625, + "loss": 0.0891, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -9.811933517456055, + "rewards/margins": 7.996350288391113, + "rewards/real": -1.8155813217163086, + "step": 970 + }, + { + "epoch": 0.62, + "learning_rate": 4.399197355996223e-07, + "logits/generated": -2.4519076347351074, + "logits/real": -2.4325168132781982, + "logps/generated": -447.27044677734375, + "logps/real": -328.2595520019531, + "loss": 0.0891, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -10.702995300292969, + "rewards/margins": 8.69007682800293, + "rewards/real": -2.0129170417785645, + "step": 980 + }, + { + "epoch": 0.63, + "learning_rate": 4.3873937677053823e-07, + "logits/generated": -2.464677572250366, + "logits/real": -2.429642915725708, + "logps/generated": -535.955322265625, + "logps/real": -362.7886047363281, + "loss": 0.1042, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -11.932709693908691, + "rewards/margins": 9.375015258789062, + "rewards/real": -2.55769419670105, + "step": 990 + }, + { + "epoch": 0.64, + "learning_rate": 4.375590179414542e-07, + "logits/generated": -2.4844415187835693, + "logits/real": -2.4864368438720703, + "logps/generated": -470.1790466308594, + "logps/real": -351.67919921875, + "loss": 0.1009, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -11.673440933227539, + "rewards/margins": 9.220005989074707, + "rewards/real": -2.4534342288970947, + "step": 1000 + }, + { + "epoch": 0.64, + "learning_rate": 4.3637865911237014e-07, + "logits/generated": -2.587498188018799, + "logits/real": -2.519575357437134, + "logps/generated": -594.9563598632812, + "logps/real": -462.4481506347656, + "loss": 0.0699, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -12.1700439453125, + "rewards/margins": 9.917440414428711, + "rewards/real": -2.2526051998138428, + "step": 1010 + }, + { + "epoch": 0.65, + "learning_rate": 4.351983002832861e-07, + "logits/generated": -2.4400439262390137, + "logits/real": -2.36588191986084, + "logps/generated": -474.21923828125, + "logps/real": -338.58502197265625, + "loss": 0.0642, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -12.955146789550781, + "rewards/margins": 9.449407577514648, + "rewards/real": -3.505737781524658, + "step": 1020 + }, + { + "epoch": 0.66, + "learning_rate": 4.3401794145420206e-07, + "logits/generated": -2.422441244125366, + "logits/real": -2.35774564743042, + "logps/generated": -552.1537475585938, + "logps/real": -395.7776794433594, + "loss": 0.0942, + "rewards/accuracies": 0.925000011920929, + "rewards/generated": -13.045738220214844, + "rewards/margins": 9.8902587890625, + "rewards/real": -3.1554789543151855, + "step": 1030 + }, + { + "epoch": 0.66, + "learning_rate": 4.32837582625118e-07, + "logits/generated": -2.4148480892181396, + "logits/real": -2.364844560623169, + "logps/generated": -487.69091796875, + "logps/real": -355.5132751464844, + "loss": 0.1098, + "rewards/accuracies": 0.8999999761581421, + "rewards/generated": -11.74626350402832, + "rewards/margins": 9.030174255371094, + "rewards/real": -2.716087579727173, + "step": 1040 + }, + { + "epoch": 0.67, + "learning_rate": 4.3165722379603397e-07, + "logits/generated": -2.3918330669403076, + "logits/real": -2.304607629776001, + "logps/generated": -527.9197387695312, + "logps/real": -414.79827880859375, + "loss": 0.109, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -11.830816268920898, + "rewards/margins": 9.040700912475586, + "rewards/real": -2.7901158332824707, + "step": 1050 + }, + { + "epoch": 0.68, + "learning_rate": 4.3047686496695e-07, + "logits/generated": -2.4523630142211914, + "logits/real": -2.3416523933410645, + "logps/generated": -499.56036376953125, + "logps/real": -389.0717468261719, + "loss": 0.067, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -11.348186492919922, + "rewards/margins": 9.180564880371094, + "rewards/real": -2.167621374130249, + "step": 1060 + }, + { + "epoch": 0.68, + "learning_rate": 4.292965061378659e-07, + "logits/generated": -2.3324472904205322, + "logits/real": -2.3067541122436523, + "logps/generated": -453.66253662109375, + "logps/real": -330.24847412109375, + "loss": 0.1026, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -11.371870040893555, + "rewards/margins": 9.505800247192383, + "rewards/real": -1.8660688400268555, + "step": 1070 + }, + { + "epoch": 0.69, + "learning_rate": 4.281161473087819e-07, + "logits/generated": -2.377768039703369, + "logits/real": -2.3538269996643066, + "logps/generated": -598.4312744140625, + "logps/real": -377.34722900390625, + "loss": 0.0877, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -12.7194185256958, + "rewards/margins": 10.486376762390137, + "rewards/real": -2.2330410480499268, + "step": 1080 + }, + { + "epoch": 0.69, + "learning_rate": 4.269357884796978e-07, + "logits/generated": -2.3940305709838867, + "logits/real": -2.341378688812256, + "logps/generated": -509.93939208984375, + "logps/real": -415.87054443359375, + "loss": 0.07, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -12.057973861694336, + "rewards/margins": 10.21934700012207, + "rewards/real": -1.8386268615722656, + "step": 1090 + }, + { + "epoch": 0.7, + "learning_rate": 4.257554296506138e-07, + "logits/generated": -2.4021096229553223, + "logits/real": -2.378485918045044, + "logps/generated": -532.3782348632812, + "logps/real": -388.20849609375, + "loss": 0.0826, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -11.92936897277832, + "rewards/margins": 9.567831993103027, + "rewards/real": -2.361536741256714, + "step": 1100 + }, + { + "epoch": 0.71, + "learning_rate": 4.245750708215297e-07, + "logits/generated": -2.425649404525757, + "logits/real": -2.3520121574401855, + "logps/generated": -543.6455688476562, + "logps/real": -410.24151611328125, + "loss": 0.1049, + "rewards/accuracies": 0.9375, + "rewards/generated": -13.359861373901367, + "rewards/margins": 10.351499557495117, + "rewards/real": -3.008361339569092, + "step": 1110 + }, + { + "epoch": 0.71, + "learning_rate": 4.233947119924457e-07, + "logits/generated": -2.330458164215088, + "logits/real": -2.2426085472106934, + "logps/generated": -511.33966064453125, + "logps/real": -382.2315979003906, + "loss": 0.0748, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -12.757667541503906, + "rewards/margins": 9.397706985473633, + "rewards/real": -3.3599605560302734, + "step": 1120 + }, + { + "epoch": 0.72, + "learning_rate": 4.222143531633616e-07, + "logits/generated": -2.274463415145874, + "logits/real": -2.251805543899536, + "logps/generated": -476.9012756347656, + "logps/real": -370.0158386230469, + "loss": 0.1536, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -11.846714973449707, + "rewards/margins": 9.258337020874023, + "rewards/real": -2.588376522064209, + "step": 1130 + }, + { + "epoch": 0.73, + "learning_rate": 4.210339943342776e-07, + "logits/generated": -2.376113176345825, + "logits/real": -2.3587307929992676, + "logps/generated": -542.9996337890625, + "logps/real": -387.203369140625, + "loss": 0.103, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -11.417093276977539, + "rewards/margins": 9.077038764953613, + "rewards/real": -2.340054750442505, + "step": 1140 + }, + { + "epoch": 0.73, + "learning_rate": 4.198536355051935e-07, + "logits/generated": -2.301286220550537, + "logits/real": -2.3263728618621826, + "logps/generated": -497.89385986328125, + "logps/real": -370.63092041015625, + "loss": 0.0409, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -12.582290649414062, + "rewards/margins": 9.770307540893555, + "rewards/real": -2.8119823932647705, + "step": 1150 + }, + { + "epoch": 0.74, + "learning_rate": 4.1867327667610954e-07, + "logits/generated": -2.279690980911255, + "logits/real": -2.3232409954071045, + "logps/generated": -544.8948364257812, + "logps/real": -373.1127624511719, + "loss": 0.0714, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -12.474981307983398, + "rewards/margins": 10.276716232299805, + "rewards/real": -2.1982650756835938, + "step": 1160 + }, + { + "epoch": 0.75, + "learning_rate": 4.1749291784702544e-07, + "logits/generated": -2.2274796962738037, + "logits/real": -2.2414534091949463, + "logps/generated": -513.9471435546875, + "logps/real": -335.01593017578125, + "loss": 0.0996, + "rewards/accuracies": 0.925000011920929, + "rewards/generated": -10.97945785522461, + "rewards/margins": 9.205892562866211, + "rewards/real": -1.773565649986267, + "step": 1170 + }, + { + "epoch": 0.75, + "learning_rate": 4.1631255901794145e-07, + "logits/generated": -2.1475322246551514, + "logits/real": -2.15641450881958, + "logps/generated": -538.5257568359375, + "logps/real": -377.38824462890625, + "loss": 0.122, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -11.927660942077637, + "rewards/margins": 9.073155403137207, + "rewards/real": -2.8545069694519043, + "step": 1180 + }, + { + "epoch": 0.76, + "learning_rate": 4.151322001888574e-07, + "logits/generated": -2.1769227981567383, + "logits/real": -2.220935344696045, + "logps/generated": -494.00042724609375, + "logps/real": -360.2157897949219, + "loss": 0.1054, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -11.777848243713379, + "rewards/margins": 9.796289443969727, + "rewards/real": -1.9815574884414673, + "step": 1190 + }, + { + "epoch": 0.76, + "learning_rate": 4.1395184135977336e-07, + "logits/generated": -2.154736280441284, + "logits/real": -2.184788942337036, + "logps/generated": -518.4021606445312, + "logps/real": -343.5872497558594, + "loss": 0.0786, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -11.488130569458008, + "rewards/margins": 10.474836349487305, + "rewards/real": -1.013293981552124, + "step": 1200 + }, + { + "epoch": 0.77, + "learning_rate": 4.127714825306893e-07, + "logits/generated": -2.1962056159973145, + "logits/real": -2.2149767875671387, + "logps/generated": -469.21575927734375, + "logps/real": -340.6158142089844, + "loss": 0.0544, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -11.18450927734375, + "rewards/margins": 9.35173511505127, + "rewards/real": -1.8327751159667969, + "step": 1210 + }, + { + "epoch": 0.78, + "learning_rate": 4.1159112370160527e-07, + "logits/generated": -2.1156868934631348, + "logits/real": -2.1794323921203613, + "logps/generated": -560.2537231445312, + "logps/real": -343.10504150390625, + "loss": 0.0903, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -12.63599967956543, + "rewards/margins": 11.78030014038086, + "rewards/real": -0.8556989431381226, + "step": 1220 + }, + { + "epoch": 0.78, + "learning_rate": 4.104107648725212e-07, + "logits/generated": -2.178110361099243, + "logits/real": -2.210237741470337, + "logps/generated": -521.1041870117188, + "logps/real": -365.5341796875, + "loss": 0.0645, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -12.666133880615234, + "rewards/margins": 10.79378604888916, + "rewards/real": -1.8723472356796265, + "step": 1230 + }, + { + "epoch": 0.79, + "learning_rate": 4.092304060434372e-07, + "logits/generated": -2.1800122261047363, + "logits/real": -2.181222915649414, + "logps/generated": -522.7543334960938, + "logps/real": -391.3867492675781, + "loss": 0.0929, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -11.76060676574707, + "rewards/margins": 10.676319122314453, + "rewards/real": -1.0842875242233276, + "step": 1240 + }, + { + "epoch": 0.8, + "learning_rate": 4.0805004721435314e-07, + "logits/generated": -2.255359411239624, + "logits/real": -2.257216691970825, + "logps/generated": -529.2927856445312, + "logps/real": -321.31768798828125, + "loss": 0.0908, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -12.798482894897461, + "rewards/margins": 10.624815940856934, + "rewards/real": -2.1736674308776855, + "step": 1250 + }, + { + "epoch": 0.8, + "learning_rate": 4.068696883852691e-07, + "logits/generated": -2.29473614692688, + "logits/real": -2.302818775177002, + "logps/generated": -484.41351318359375, + "logps/real": -350.274658203125, + "loss": 0.1146, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -13.613574028015137, + "rewards/margins": 10.635141372680664, + "rewards/real": -2.9784321784973145, + "step": 1260 + }, + { + "epoch": 0.81, + "learning_rate": 4.0568932955618505e-07, + "logits/generated": -2.273796796798706, + "logits/real": -2.318351984024048, + "logps/generated": -511.91717529296875, + "logps/real": -357.89654541015625, + "loss": 0.1316, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -13.401461601257324, + "rewards/margins": 11.316085815429688, + "rewards/real": -2.085376262664795, + "step": 1270 + }, + { + "epoch": 0.82, + "learning_rate": 4.04508970727101e-07, + "logits/generated": -2.2378830909729004, + "logits/real": -2.2756216526031494, + "logps/generated": -428.2696838378906, + "logps/real": -322.50909423828125, + "loss": 0.0892, + "rewards/accuracies": 0.925000011920929, + "rewards/generated": -10.981972694396973, + "rewards/margins": 9.416523933410645, + "rewards/real": -1.565448522567749, + "step": 1280 + }, + { + "epoch": 0.82, + "learning_rate": 4.0332861189801696e-07, + "logits/generated": -2.15289306640625, + "logits/real": -2.1541621685028076, + "logps/generated": -524.1163330078125, + "logps/real": -345.48388671875, + "loss": 0.1114, + "rewards/accuracies": 0.9375, + "rewards/generated": -12.033792495727539, + "rewards/margins": 10.111471176147461, + "rewards/real": -1.9223216772079468, + "step": 1290 + }, + { + "epoch": 0.83, + "learning_rate": 4.0214825306893297e-07, + "logits/generated": -2.1317696571350098, + "logits/real": -2.206805467605591, + "logps/generated": -545.2672119140625, + "logps/real": -363.6857604980469, + "loss": 0.1072, + "rewards/accuracies": 0.9125000238418579, + "rewards/generated": -12.273852348327637, + "rewards/margins": 10.354440689086914, + "rewards/real": -1.9194109439849854, + "step": 1300 + }, + { + "epoch": 0.83, + "learning_rate": 4.009678942398489e-07, + "logits/generated": -2.221585512161255, + "logits/real": -2.2462613582611084, + "logps/generated": -493.2377014160156, + "logps/real": -307.10418701171875, + "loss": 0.0874, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -11.128523826599121, + "rewards/margins": 9.498746871948242, + "rewards/real": -1.6297763586044312, + "step": 1310 + }, + { + "epoch": 0.84, + "learning_rate": 3.997875354107649e-07, + "logits/generated": -2.2427799701690674, + "logits/real": -2.3131985664367676, + "logps/generated": -482.2207946777344, + "logps/real": -330.1937561035156, + "loss": 0.0941, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -11.602561950683594, + "rewards/margins": 9.614253044128418, + "rewards/real": -1.988308310508728, + "step": 1320 + }, + { + "epoch": 0.85, + "learning_rate": 3.986071765816808e-07, + "logits/generated": -2.188753843307495, + "logits/real": -2.1746668815612793, + "logps/generated": -566.9933471679688, + "logps/real": -335.02728271484375, + "loss": 0.0908, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -13.266576766967773, + "rewards/margins": 11.424365043640137, + "rewards/real": -1.8422130346298218, + "step": 1330 + }, + { + "epoch": 0.85, + "learning_rate": 3.974268177525968e-07, + "logits/generated": -2.1205272674560547, + "logits/real": -2.1614651679992676, + "logps/generated": -532.3944091796875, + "logps/real": -377.13995361328125, + "loss": 0.0777, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -12.711483001708984, + "rewards/margins": 10.040422439575195, + "rewards/real": -2.671060085296631, + "step": 1340 + }, + { + "epoch": 0.86, + "learning_rate": 3.962464589235127e-07, + "logits/generated": -2.126450777053833, + "logits/real": -2.170285224914551, + "logps/generated": -533.8922119140625, + "logps/real": -419.98968505859375, + "loss": 0.0847, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -12.28015422821045, + "rewards/margins": 10.587915420532227, + "rewards/real": -1.692238450050354, + "step": 1350 + }, + { + "epoch": 0.87, + "learning_rate": 3.950661000944287e-07, + "logits/generated": -2.0494632720947266, + "logits/real": -2.1037888526916504, + "logps/generated": -505.348876953125, + "logps/real": -348.7581787109375, + "loss": 0.0844, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -12.927388191223145, + "rewards/margins": 11.16401481628418, + "rewards/real": -1.763375997543335, + "step": 1360 + }, + { + "epoch": 0.87, + "learning_rate": 3.938857412653446e-07, + "logits/generated": -2.027092218399048, + "logits/real": -2.065883159637451, + "logps/generated": -519.960205078125, + "logps/real": -368.42034912109375, + "loss": 0.0775, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -12.688058853149414, + "rewards/margins": 10.453657150268555, + "rewards/real": -2.234400749206543, + "step": 1370 + }, + { + "epoch": 0.88, + "learning_rate": 3.927053824362606e-07, + "logits/generated": -2.0650784969329834, + "logits/real": -2.023521900177002, + "logps/generated": -507.860107421875, + "logps/real": -372.95318603515625, + "loss": 0.0831, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -12.687623977661133, + "rewards/margins": 10.405519485473633, + "rewards/real": -2.2821052074432373, + "step": 1380 + }, + { + "epoch": 0.89, + "learning_rate": 3.915250236071765e-07, + "logits/generated": -1.972839593887329, + "logits/real": -2.0179429054260254, + "logps/generated": -596.0494384765625, + "logps/real": -409.5896301269531, + "loss": 0.0596, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -15.517305374145508, + "rewards/margins": 12.9534273147583, + "rewards/real": -2.563877820968628, + "step": 1390 + }, + { + "epoch": 0.89, + "learning_rate": 3.9034466477809253e-07, + "logits/generated": -2.026308298110962, + "logits/real": -2.08756685256958, + "logps/generated": -575.8211669921875, + "logps/real": -407.56158447265625, + "loss": 0.0884, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -14.804544448852539, + "rewards/margins": 12.167125701904297, + "rewards/real": -2.6374189853668213, + "step": 1400 + }, + { + "epoch": 0.9, + "learning_rate": 3.8916430594900843e-07, + "logits/generated": -2.1134438514709473, + "logits/real": -2.2039880752563477, + "logps/generated": -533.747802734375, + "logps/real": -355.21270751953125, + "loss": 0.0692, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -12.54589557647705, + "rewards/margins": 10.795358657836914, + "rewards/real": -1.7505359649658203, + "step": 1410 + }, + { + "epoch": 0.91, + "learning_rate": 3.8798394711992444e-07, + "logits/generated": -2.170846462249756, + "logits/real": -2.224730968475342, + "logps/generated": -542.9186401367188, + "logps/real": -356.8793029785156, + "loss": 0.0913, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -13.97801685333252, + "rewards/margins": 11.115743637084961, + "rewards/real": -2.8622727394104004, + "step": 1420 + }, + { + "epoch": 0.91, + "learning_rate": 3.868035882908404e-07, + "logits/generated": -2.0684473514556885, + "logits/real": -2.125868797302246, + "logps/generated": -511.23980712890625, + "logps/real": -367.94195556640625, + "loss": 0.1073, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -12.94532585144043, + "rewards/margins": 10.817112922668457, + "rewards/real": -2.1282143592834473, + "step": 1430 + }, + { + "epoch": 0.92, + "learning_rate": 3.8562322946175635e-07, + "logits/generated": -2.142906904220581, + "logits/real": -2.139702081680298, + "logps/generated": -521.8377685546875, + "logps/real": -378.81817626953125, + "loss": 0.0861, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -11.904378890991211, + "rewards/margins": 10.70759105682373, + "rewards/real": -1.1967872381210327, + "step": 1440 + }, + { + "epoch": 0.92, + "learning_rate": 3.844428706326723e-07, + "logits/generated": -2.103635311126709, + "logits/real": -2.134916067123413, + "logps/generated": -553.8816528320312, + "logps/real": -373.2740173339844, + "loss": 0.1037, + "rewards/accuracies": 0.9375, + "rewards/generated": -13.051103591918945, + "rewards/margins": 10.889741897583008, + "rewards/real": -2.1613619327545166, + "step": 1450 + }, + { + "epoch": 0.93, + "learning_rate": 3.8326251180358827e-07, + "logits/generated": -1.9052108526229858, + "logits/real": -2.009723424911499, + "logps/generated": -487.1270446777344, + "logps/real": -318.29754638671875, + "loss": 0.1107, + "rewards/accuracies": 0.9375, + "rewards/generated": -13.303110122680664, + "rewards/margins": 11.442071914672852, + "rewards/real": -1.8610382080078125, + "step": 1460 + }, + { + "epoch": 0.94, + "learning_rate": 3.820821529745042e-07, + "logits/generated": -1.9846614599227905, + "logits/real": -2.000155448913574, + "logps/generated": -479.4563903808594, + "logps/real": -330.0511169433594, + "loss": 0.1041, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -13.340143203735352, + "rewards/margins": 10.538009643554688, + "rewards/real": -2.8021321296691895, + "step": 1470 + }, + { + "epoch": 0.94, + "learning_rate": 3.809017941454202e-07, + "logits/generated": -2.0992157459259033, + "logits/real": -2.083374500274658, + "logps/generated": -474.174560546875, + "logps/real": -354.2595520019531, + "loss": 0.1071, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -12.097639083862305, + "rewards/margins": 9.328176498413086, + "rewards/real": -2.7694618701934814, + "step": 1480 + }, + { + "epoch": 0.95, + "learning_rate": 3.7972143531633613e-07, + "logits/generated": -2.041928291320801, + "logits/real": -2.0762786865234375, + "logps/generated": -516.2319946289062, + "logps/real": -333.8430480957031, + "loss": 0.1031, + "rewards/accuracies": 0.9375, + "rewards/generated": -14.001957893371582, + "rewards/margins": 12.395853042602539, + "rewards/real": -1.606106162071228, + "step": 1490 + }, + { + "epoch": 0.96, + "learning_rate": 3.785410764872521e-07, + "logits/generated": -2.175471067428589, + "logits/real": -2.2105460166931152, + "logps/generated": -477.33837890625, + "logps/real": -313.8334045410156, + "loss": 0.0936, + "rewards/accuracies": 0.9375, + "rewards/generated": -12.010794639587402, + "rewards/margins": 10.985909461975098, + "rewards/real": -1.0248849391937256, + "step": 1500 + }, + { + "epoch": 0.96, + "learning_rate": 3.7736071765816804e-07, + "logits/generated": -2.162140369415283, + "logits/real": -2.205298900604248, + "logps/generated": -500.3495178222656, + "logps/real": -340.25115966796875, + "loss": 0.0896, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -12.862123489379883, + "rewards/margins": 11.332796096801758, + "rewards/real": -1.52932608127594, + "step": 1510 + }, + { + "epoch": 0.97, + "learning_rate": 3.76180358829084e-07, + "logits/generated": -2.2106504440307617, + "logits/real": -2.1732020378112793, + "logps/generated": -520.9046630859375, + "logps/real": -356.8975524902344, + "loss": 0.0971, + "rewards/accuracies": 0.9375, + "rewards/generated": -11.427778244018555, + "rewards/margins": 10.071785926818848, + "rewards/real": -1.355992078781128, + "step": 1520 + }, + { + "epoch": 0.98, + "learning_rate": 3.75e-07, + "logits/generated": -2.2059266567230225, + "logits/real": -2.2535414695739746, + "logps/generated": -502.62493896484375, + "logps/real": -333.29962158203125, + "loss": 0.0531, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -13.523884773254395, + "rewards/margins": 12.068760871887207, + "rewards/real": -1.4551236629486084, + "step": 1530 + }, + { + "epoch": 0.98, + "learning_rate": 3.7381964117091597e-07, + "logits/generated": -2.231109142303467, + "logits/real": -2.280176877975464, + "logps/generated": -543.9066772460938, + "logps/real": -333.4288635253906, + "loss": 0.0802, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -13.546087265014648, + "rewards/margins": 11.472119331359863, + "rewards/real": -2.0739691257476807, + "step": 1540 + }, + { + "epoch": 0.99, + "learning_rate": 3.726392823418319e-07, + "logits/generated": -2.319192409515381, + "logits/real": -2.395427703857422, + "logps/generated": -592.6727905273438, + "logps/real": -443.85687255859375, + "loss": 0.0941, + "rewards/accuracies": 0.9624999761581421, + "rewards/generated": -14.170260429382324, + "rewards/margins": 12.117624282836914, + "rewards/real": -2.052638292312622, + "step": 1550 + }, + { + "epoch": 0.99, + "learning_rate": 3.714589235127479e-07, + "logits/generated": -2.3090901374816895, + "logits/real": -2.3818156719207764, + "logps/generated": -558.4683837890625, + "logps/real": -386.15997314453125, + "loss": 0.0723, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -13.98528003692627, + "rewards/margins": 11.203457832336426, + "rewards/real": -2.7818214893341064, + "step": 1560 + }, + { + "epoch": 1.0, + "learning_rate": 3.7027856468366383e-07, + "logits/generated": -2.309932231903076, + "logits/real": -2.3563430309295654, + "logps/generated": -478.1116638183594, + "logps/real": -304.4076232910156, + "loss": 0.0816, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -14.32862377166748, + "rewards/margins": 12.117714881896973, + "rewards/real": -2.2109079360961914, + "step": 1570 + }, + { + "epoch": 1.01, + "learning_rate": 3.690982058545798e-07, + "logits/generated": -2.3237366676330566, + "logits/real": -2.4287846088409424, + "logps/generated": -506.93408203125, + "logps/real": -335.35552978515625, + "loss": 0.0147, + "rewards/accuracies": 1.0, + "rewards/generated": -15.469167709350586, + "rewards/margins": 13.284278869628906, + "rewards/real": -2.1848883628845215, + "step": 1580 + }, + { + "epoch": 1.01, + "learning_rate": 3.6791784702549574e-07, + "logits/generated": -2.2916648387908936, + "logits/real": -2.3626348972320557, + "logps/generated": -554.8782348632812, + "logps/real": -362.2654724121094, + "loss": 0.0051, + "rewards/accuracies": 1.0, + "rewards/generated": -15.250271797180176, + "rewards/margins": 13.737541198730469, + "rewards/real": -1.512730598449707, + "step": 1590 + }, + { + "epoch": 1.02, + "learning_rate": 3.667374881964117e-07, + "logits/generated": -2.365065813064575, + "logits/real": -2.3905742168426514, + "logps/generated": -602.1806030273438, + "logps/real": -405.0635986328125, + "loss": 0.0116, + "rewards/accuracies": 1.0, + "rewards/generated": -17.20393943786621, + "rewards/margins": 14.85174560546875, + "rewards/real": -2.3521931171417236, + "step": 1600 + }, + { + "epoch": 1.03, + "learning_rate": 3.6555712936732766e-07, + "logits/generated": -2.3196659088134766, + "logits/real": -2.3548758029937744, + "logps/generated": -508.1708984375, + "logps/real": -333.3728332519531, + "loss": 0.0283, + "rewards/accuracies": 1.0, + "rewards/generated": -14.976804733276367, + "rewards/margins": 12.516164779663086, + "rewards/real": -2.460641860961914, + "step": 1610 + }, + { + "epoch": 1.03, + "learning_rate": 3.643767705382436e-07, + "logits/generated": -2.3346681594848633, + "logits/real": -2.395854949951172, + "logps/generated": -540.6458129882812, + "logps/real": -331.14361572265625, + "loss": 0.0168, + "rewards/accuracies": 1.0, + "rewards/generated": -16.316436767578125, + "rewards/margins": 13.759660720825195, + "rewards/real": -2.5567734241485596, + "step": 1620 + }, + { + "epoch": 1.04, + "learning_rate": 3.6319641170915957e-07, + "logits/generated": -2.2885191440582275, + "logits/real": -2.3124356269836426, + "logps/generated": -607.8283081054688, + "logps/real": -372.39837646484375, + "loss": 0.0079, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -15.951924324035645, + "rewards/margins": 15.02081298828125, + "rewards/real": -0.9311105012893677, + "step": 1630 + }, + { + "epoch": 1.05, + "learning_rate": 3.620160528800755e-07, + "logits/generated": -2.2756428718566895, + "logits/real": -2.3602356910705566, + "logps/generated": -568.5755615234375, + "logps/real": -425.0919494628906, + "loss": 0.0048, + "rewards/accuracies": 1.0, + "rewards/generated": -14.275492668151855, + "rewards/margins": 13.160967826843262, + "rewards/real": -1.1145250797271729, + "step": 1640 + }, + { + "epoch": 1.05, + "learning_rate": 3.6083569405099153e-07, + "logits/generated": -2.1741244792938232, + "logits/real": -2.2903761863708496, + "logps/generated": -554.7332153320312, + "logps/real": -389.9463806152344, + "loss": 0.0078, + "rewards/accuracies": 1.0, + "rewards/generated": -16.888017654418945, + "rewards/margins": 15.324864387512207, + "rewards/real": -1.563153624534607, + "step": 1650 + }, + { + "epoch": 1.06, + "learning_rate": 3.5965533522190744e-07, + "logits/generated": -2.2743141651153564, + "logits/real": -2.3127028942108154, + "logps/generated": -573.2037353515625, + "logps/real": -409.07598876953125, + "loss": 0.0068, + "rewards/accuracies": 1.0, + "rewards/generated": -15.042158126831055, + "rewards/margins": 13.179901123046875, + "rewards/real": -1.8622581958770752, + "step": 1660 + }, + { + "epoch": 1.06, + "learning_rate": 3.5847497639282345e-07, + "logits/generated": -2.2046515941619873, + "logits/real": -2.2793116569519043, + "logps/generated": -557.3995361328125, + "logps/real": -374.20367431640625, + "loss": 0.0063, + "rewards/accuracies": 1.0, + "rewards/generated": -15.566205024719238, + "rewards/margins": 14.624272346496582, + "rewards/real": -0.941932201385498, + "step": 1670 + }, + { + "epoch": 1.07, + "learning_rate": 3.5729461756373935e-07, + "logits/generated": -2.17391300201416, + "logits/real": -2.274019956588745, + "logps/generated": -548.1880493164062, + "logps/real": -366.0269470214844, + "loss": 0.0131, + "rewards/accuracies": 1.0, + "rewards/generated": -17.253692626953125, + "rewards/margins": 14.757375717163086, + "rewards/real": -2.4963152408599854, + "step": 1680 + }, + { + "epoch": 1.08, + "learning_rate": 3.5611425873465536e-07, + "logits/generated": -2.179938554763794, + "logits/real": -2.291468381881714, + "logps/generated": -537.0164794921875, + "logps/real": -336.8132019042969, + "loss": 0.0168, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -18.351551055908203, + "rewards/margins": 16.063148498535156, + "rewards/real": -2.2884011268615723, + "step": 1690 + }, + { + "epoch": 1.08, + "learning_rate": 3.5493389990557126e-07, + "logits/generated": -2.1702983379364014, + "logits/real": -2.2541327476501465, + "logps/generated": -561.18408203125, + "logps/real": -365.71142578125, + "loss": 0.0165, + "rewards/accuracies": 1.0, + "rewards/generated": -17.675691604614258, + "rewards/margins": 14.446200370788574, + "rewards/real": -3.229489803314209, + "step": 1700 + }, + { + "epoch": 1.09, + "learning_rate": 3.5375354107648727e-07, + "logits/generated": -2.2532219886779785, + "logits/real": -2.311969757080078, + "logps/generated": -564.5615234375, + "logps/real": -366.0008850097656, + "loss": 0.014, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -17.055387496948242, + "rewards/margins": 14.41291332244873, + "rewards/real": -2.642474412918091, + "step": 1710 + }, + { + "epoch": 1.1, + "learning_rate": 3.5257318224740317e-07, + "logits/generated": -2.1806800365448, + "logits/real": -2.2332406044006348, + "logps/generated": -507.8981018066406, + "logps/real": -316.1936950683594, + "loss": 0.0097, + "rewards/accuracies": 1.0, + "rewards/generated": -17.727888107299805, + "rewards/margins": 14.673624038696289, + "rewards/real": -3.0542654991149902, + "step": 1720 + }, + { + "epoch": 1.1, + "learning_rate": 3.513928234183192e-07, + "logits/generated": -2.171751022338867, + "logits/real": -2.261719226837158, + "logps/generated": -583.4220581054688, + "logps/real": -418.9441833496094, + "loss": 0.0118, + "rewards/accuracies": 1.0, + "rewards/generated": -17.582807540893555, + "rewards/margins": 13.78455638885498, + "rewards/real": -3.7982494831085205, + "step": 1730 + }, + { + "epoch": 1.11, + "learning_rate": 3.502124645892351e-07, + "logits/generated": -2.162670612335205, + "logits/real": -2.232741594314575, + "logps/generated": -612.6056518554688, + "logps/real": -364.3959045410156, + "loss": 0.0029, + "rewards/accuracies": 1.0, + "rewards/generated": -20.081586837768555, + "rewards/margins": 16.691913604736328, + "rewards/real": -3.3896732330322266, + "step": 1740 + }, + { + "epoch": 1.12, + "learning_rate": 3.490321057601511e-07, + "logits/generated": -2.165930986404419, + "logits/real": -2.2348618507385254, + "logps/generated": -586.1246948242188, + "logps/real": -364.8479309082031, + "loss": 0.0054, + "rewards/accuracies": 1.0, + "rewards/generated": -18.28532600402832, + "rewards/margins": 14.780363082885742, + "rewards/real": -3.5049636363983154, + "step": 1750 + }, + { + "epoch": 1.12, + "learning_rate": 3.47851746931067e-07, + "logits/generated": -2.192909002304077, + "logits/real": -2.237509250640869, + "logps/generated": -622.2453002929688, + "logps/real": -379.9577331542969, + "loss": 0.0057, + "rewards/accuracies": 1.0, + "rewards/generated": -19.904523849487305, + "rewards/margins": 16.164608001708984, + "rewards/real": -3.739915370941162, + "step": 1760 + }, + { + "epoch": 1.13, + "learning_rate": 3.46671388101983e-07, + "logits/generated": -2.1582560539245605, + "logits/real": -2.2041397094726562, + "logps/generated": -670.0623779296875, + "logps/real": -407.6207580566406, + "loss": 0.0068, + "rewards/accuracies": 1.0, + "rewards/generated": -23.308486938476562, + "rewards/margins": 18.176908493041992, + "rewards/real": -5.131577491760254, + "step": 1770 + }, + { + "epoch": 1.13, + "learning_rate": 3.4549102927289896e-07, + "logits/generated": -2.099989175796509, + "logits/real": -2.2095441818237305, + "logps/generated": -562.9180908203125, + "logps/real": -396.07818603515625, + "loss": 0.0033, + "rewards/accuracies": 1.0, + "rewards/generated": -19.229347229003906, + "rewards/margins": 14.39805793762207, + "rewards/real": -4.8312883377075195, + "step": 1780 + }, + { + "epoch": 1.14, + "learning_rate": 3.443106704438149e-07, + "logits/generated": -2.0601720809936523, + "logits/real": -2.246066093444824, + "logps/generated": -591.3522338867188, + "logps/real": -406.21142578125, + "loss": 0.0032, + "rewards/accuracies": 1.0, + "rewards/generated": -20.999242782592773, + "rewards/margins": 16.740962982177734, + "rewards/real": -4.2582783699035645, + "step": 1790 + }, + { + "epoch": 1.15, + "learning_rate": 3.4313031161473087e-07, + "logits/generated": -2.1487627029418945, + "logits/real": -2.170689344406128, + "logps/generated": -536.7785034179688, + "logps/real": -329.46221923828125, + "loss": 0.0184, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -16.940834045410156, + "rewards/margins": 14.779701232910156, + "rewards/real": -2.16113018989563, + "step": 1800 + }, + { + "epoch": 1.15, + "learning_rate": 3.4194995278564683e-07, + "logits/generated": -2.1991372108459473, + "logits/real": -2.237712860107422, + "logps/generated": -565.8905029296875, + "logps/real": -313.5826721191406, + "loss": 0.0089, + "rewards/accuracies": 1.0, + "rewards/generated": -18.22218894958496, + "rewards/margins": 16.043153762817383, + "rewards/real": -2.1790366172790527, + "step": 1810 + }, + { + "epoch": 1.16, + "learning_rate": 3.407695939565628e-07, + "logits/generated": -2.154999017715454, + "logits/real": -2.261317491531372, + "logps/generated": -587.2596435546875, + "logps/real": -403.8623962402344, + "loss": 0.0031, + "rewards/accuracies": 1.0, + "rewards/generated": -18.34090232849121, + "rewards/margins": 16.827770233154297, + "rewards/real": -1.5131312608718872, + "step": 1820 + }, + { + "epoch": 1.17, + "learning_rate": 3.3958923512747874e-07, + "logits/generated": -2.1115026473999023, + "logits/real": -2.202688217163086, + "logps/generated": -557.8204956054688, + "logps/real": -408.7705078125, + "loss": 0.0085, + "rewards/accuracies": 1.0, + "rewards/generated": -19.18920135498047, + "rewards/margins": 15.727496147155762, + "rewards/real": -3.461704730987549, + "step": 1830 + }, + { + "epoch": 1.17, + "learning_rate": 3.384088762983947e-07, + "logits/generated": -2.1163887977600098, + "logits/real": -2.207763671875, + "logps/generated": -566.0260009765625, + "logps/real": -372.63385009765625, + "loss": 0.018, + "rewards/accuracies": 0.949999988079071, + "rewards/generated": -17.137704849243164, + "rewards/margins": 14.465890884399414, + "rewards/real": -2.6718127727508545, + "step": 1840 + }, + { + "epoch": 1.18, + "learning_rate": 3.3722851746931065e-07, + "logits/generated": -2.2660069465637207, + "logits/real": -2.287060260772705, + "logps/generated": -570.5521240234375, + "logps/real": -391.91729736328125, + "loss": 0.0066, + "rewards/accuracies": 1.0, + "rewards/generated": -16.603046417236328, + "rewards/margins": 13.50883674621582, + "rewards/real": -3.0942091941833496, + "step": 1850 + }, + { + "epoch": 1.19, + "learning_rate": 3.360481586402266e-07, + "logits/generated": -2.1887662410736084, + "logits/real": -2.2266957759857178, + "logps/generated": -573.5321044921875, + "logps/real": -351.8019714355469, + "loss": 0.0112, + "rewards/accuracies": 1.0, + "rewards/generated": -18.973896026611328, + "rewards/margins": 15.386941909790039, + "rewards/real": -3.5869545936584473, + "step": 1860 + }, + { + "epoch": 1.19, + "learning_rate": 3.3486779981114256e-07, + "logits/generated": -2.1134400367736816, + "logits/real": -2.1982038021087646, + "logps/generated": -598.865478515625, + "logps/real": -384.9998779296875, + "loss": 0.004, + "rewards/accuracies": 1.0, + "rewards/generated": -20.413084030151367, + "rewards/margins": 15.823282241821289, + "rewards/real": -4.5898027420043945, + "step": 1870 + }, + { + "epoch": 1.2, + "learning_rate": 3.336874409820585e-07, + "logits/generated": -2.155069351196289, + "logits/real": -2.215625047683716, + "logps/generated": -595.2161865234375, + "logps/real": -345.4395446777344, + "loss": 0.0273, + "rewards/accuracies": 1.0, + "rewards/generated": -19.925769805908203, + "rewards/margins": 16.561832427978516, + "rewards/real": -3.3639378547668457, + "step": 1880 + }, + { + "epoch": 1.2, + "learning_rate": 3.3250708215297453e-07, + "logits/generated": -2.244886636734009, + "logits/real": -2.229398488998413, + "logps/generated": -601.0575561523438, + "logps/real": -380.1138610839844, + "loss": 0.0056, + "rewards/accuracies": 1.0, + "rewards/generated": -17.558795928955078, + "rewards/margins": 15.04124927520752, + "rewards/real": -2.5175468921661377, + "step": 1890 + }, + { + "epoch": 1.21, + "learning_rate": 3.3132672332389043e-07, + "logits/generated": -2.1024043560028076, + "logits/real": -2.2166638374328613, + "logps/generated": -597.3502197265625, + "logps/real": -345.0755920410156, + "loss": 0.0064, + "rewards/accuracies": 1.0, + "rewards/generated": -18.900390625, + "rewards/margins": 16.444355010986328, + "rewards/real": -2.4560346603393555, + "step": 1900 + }, + { + "epoch": 1.22, + "learning_rate": 3.3014636449480644e-07, + "logits/generated": -2.1653573513031006, + "logits/real": -2.2361793518066406, + "logps/generated": -589.8990478515625, + "logps/real": -388.7892150878906, + "loss": 0.0112, + "rewards/accuracies": 1.0, + "rewards/generated": -17.51367950439453, + "rewards/margins": 16.049463272094727, + "rewards/real": -1.4642179012298584, + "step": 1910 + }, + { + "epoch": 1.22, + "learning_rate": 3.2896600566572234e-07, + "logits/generated": -2.1428253650665283, + "logits/real": -2.228580951690674, + "logps/generated": -557.1248168945312, + "logps/real": -397.64764404296875, + "loss": 0.0111, + "rewards/accuracies": 1.0, + "rewards/generated": -16.451763153076172, + "rewards/margins": 14.995409965515137, + "rewards/real": -1.4563549757003784, + "step": 1920 + }, + { + "epoch": 1.23, + "learning_rate": 3.2778564683663835e-07, + "logits/generated": -2.129307270050049, + "logits/real": -2.211942195892334, + "logps/generated": -569.16552734375, + "logps/real": -344.17657470703125, + "loss": 0.0251, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -15.866914749145508, + "rewards/margins": 13.207151412963867, + "rewards/real": -2.659763813018799, + "step": 1930 + }, + { + "epoch": 1.24, + "learning_rate": 3.2660528800755425e-07, + "logits/generated": -2.1618289947509766, + "logits/real": -2.1872544288635254, + "logps/generated": -536.9007568359375, + "logps/real": -360.4764404296875, + "loss": 0.0106, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -16.128765106201172, + "rewards/margins": 13.156270027160645, + "rewards/real": -2.9724948406219482, + "step": 1940 + }, + { + "epoch": 1.24, + "learning_rate": 3.2542492917847026e-07, + "logits/generated": -2.081109046936035, + "logits/real": -2.1807050704956055, + "logps/generated": -552.4083251953125, + "logps/real": -381.96075439453125, + "loss": 0.0107, + "rewards/accuracies": 1.0, + "rewards/generated": -17.499576568603516, + "rewards/margins": 14.36053466796875, + "rewards/real": -3.139042615890503, + "step": 1950 + }, + { + "epoch": 1.25, + "learning_rate": 3.2424457034938617e-07, + "logits/generated": -2.1019864082336426, + "logits/real": -2.1223390102386475, + "logps/generated": -617.0413818359375, + "logps/real": -371.85540771484375, + "loss": 0.0103, + "rewards/accuracies": 1.0, + "rewards/generated": -19.166786193847656, + "rewards/margins": 15.655584335327148, + "rewards/real": -3.5111992359161377, + "step": 1960 + }, + { + "epoch": 1.26, + "learning_rate": 3.230642115203022e-07, + "logits/generated": -2.1129393577575684, + "logits/real": -2.099522590637207, + "logps/generated": -567.219482421875, + "logps/real": -398.39215087890625, + "loss": 0.0069, + "rewards/accuracies": 1.0, + "rewards/generated": -18.29250144958496, + "rewards/margins": 15.135157585144043, + "rewards/real": -3.1573448181152344, + "step": 1970 + }, + { + "epoch": 1.26, + "learning_rate": 3.218838526912181e-07, + "logits/generated": -2.039750099182129, + "logits/real": -2.1424005031585693, + "logps/generated": -564.7282104492188, + "logps/real": -382.9680480957031, + "loss": 0.0047, + "rewards/accuracies": 1.0, + "rewards/generated": -19.77239990234375, + "rewards/margins": 16.35171890258789, + "rewards/real": -3.4206786155700684, + "step": 1980 + }, + { + "epoch": 1.27, + "learning_rate": 3.207034938621341e-07, + "logits/generated": -2.060238838195801, + "logits/real": -2.039731025695801, + "logps/generated": -593.91748046875, + "logps/real": -395.67071533203125, + "loss": 0.0166, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -20.168869018554688, + "rewards/margins": 16.086454391479492, + "rewards/real": -4.0824151039123535, + "step": 1990 + }, + { + "epoch": 1.27, + "learning_rate": 3.1952313503305e-07, + "logits/generated": -1.9739450216293335, + "logits/real": -2.0953528881073, + "logps/generated": -588.6085205078125, + "logps/real": -397.8580627441406, + "loss": 0.007, + "rewards/accuracies": 1.0, + "rewards/generated": -20.304332733154297, + "rewards/margins": 16.184040069580078, + "rewards/real": -4.12029504776001, + "step": 2000 + }, + { + "epoch": 1.28, + "learning_rate": 3.18342776203966e-07, + "logits/generated": -2.025446653366089, + "logits/real": -2.06404709815979, + "logps/generated": -509.741455078125, + "logps/real": -320.9604797363281, + "loss": 0.0178, + "rewards/accuracies": 1.0, + "rewards/generated": -18.078128814697266, + "rewards/margins": 14.200398445129395, + "rewards/real": -3.877730131149292, + "step": 2010 + }, + { + "epoch": 1.29, + "learning_rate": 3.1716241737488195e-07, + "logits/generated": -2.0446219444274902, + "logits/real": -2.102245807647705, + "logps/generated": -581.5309448242188, + "logps/real": -345.5025939941406, + "loss": 0.0079, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -20.858938217163086, + "rewards/margins": 17.1157283782959, + "rewards/real": -3.743211030960083, + "step": 2020 + }, + { + "epoch": 1.29, + "learning_rate": 3.159820585457979e-07, + "logits/generated": -1.9954283237457275, + "logits/real": -2.0560336112976074, + "logps/generated": -556.6575927734375, + "logps/real": -380.7696838378906, + "loss": 0.006, + "rewards/accuracies": 1.0, + "rewards/generated": -19.783016204833984, + "rewards/margins": 16.61421012878418, + "rewards/real": -3.168807029724121, + "step": 2030 + }, + { + "epoch": 1.3, + "learning_rate": 3.1480169971671387e-07, + "logits/generated": -2.0975687503814697, + "logits/real": -2.21252703666687, + "logps/generated": -594.607421875, + "logps/real": -369.9338684082031, + "loss": 0.0457, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -18.788341522216797, + "rewards/margins": 15.700846672058105, + "rewards/real": -3.0874946117401123, + "step": 2040 + }, + { + "epoch": 1.31, + "learning_rate": 3.136213408876298e-07, + "logits/generated": -1.937974214553833, + "logits/real": -2.0436413288116455, + "logps/generated": -551.8988037109375, + "logps/real": -400.60943603515625, + "loss": 0.0275, + "rewards/accuracies": 1.0, + "rewards/generated": -16.716506958007812, + "rewards/margins": 15.352325439453125, + "rewards/real": -1.3641842603683472, + "step": 2050 + }, + { + "epoch": 1.31, + "learning_rate": 3.124409820585458e-07, + "logits/generated": -1.9299554824829102, + "logits/real": -1.977850317955017, + "logps/generated": -577.6395263671875, + "logps/real": -386.54742431640625, + "loss": 0.0125, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -14.422266960144043, + "rewards/margins": 12.841423034667969, + "rewards/real": -1.5808441638946533, + "step": 2060 + }, + { + "epoch": 1.32, + "learning_rate": 3.1126062322946173e-07, + "logits/generated": -2.025808572769165, + "logits/real": -2.0407521724700928, + "logps/generated": -581.5086669921875, + "logps/real": -378.8692626953125, + "loss": 0.0198, + "rewards/accuracies": 1.0, + "rewards/generated": -14.943647384643555, + "rewards/margins": 15.0796537399292, + "rewards/real": 0.13600602746009827, + "step": 2070 + }, + { + "epoch": 1.33, + "learning_rate": 3.100802644003777e-07, + "logits/generated": -2.064063787460327, + "logits/real": -2.0676605701446533, + "logps/generated": -546.7476196289062, + "logps/real": -347.3872375488281, + "loss": 0.0074, + "rewards/accuracies": 1.0, + "rewards/generated": -15.180923461914062, + "rewards/margins": 13.453030586242676, + "rewards/real": -1.7278907299041748, + "step": 2080 + }, + { + "epoch": 1.33, + "learning_rate": 3.0889990557129365e-07, + "logits/generated": -1.910861611366272, + "logits/real": -1.9376237392425537, + "logps/generated": -584.0259399414062, + "logps/real": -344.9956970214844, + "loss": 0.012, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -17.09406280517578, + "rewards/margins": 14.956217765808105, + "rewards/real": -2.137843608856201, + "step": 2090 + }, + { + "epoch": 1.34, + "learning_rate": 3.077195467422096e-07, + "logits/generated": -1.9489418268203735, + "logits/real": -1.971780776977539, + "logps/generated": -551.9901123046875, + "logps/real": -364.4021301269531, + "loss": 0.019, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -16.30406379699707, + "rewards/margins": 13.739787101745605, + "rewards/real": -2.5642757415771484, + "step": 2100 + }, + { + "epoch": 1.34, + "learning_rate": 3.0653918791312556e-07, + "logits/generated": -1.9933067560195923, + "logits/real": -2.0058422088623047, + "logps/generated": -586.0560302734375, + "logps/real": -378.6280517578125, + "loss": 0.0146, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -15.956277847290039, + "rewards/margins": 14.133363723754883, + "rewards/real": -1.8229141235351562, + "step": 2110 + }, + { + "epoch": 1.35, + "learning_rate": 3.053588290840415e-07, + "logits/generated": -1.9493356943130493, + "logits/real": -1.9630330801010132, + "logps/generated": -555.2738037109375, + "logps/real": -400.0673828125, + "loss": 0.015, + "rewards/accuracies": 1.0, + "rewards/generated": -17.071149826049805, + "rewards/margins": 14.732151985168457, + "rewards/real": -2.3389980792999268, + "step": 2120 + }, + { + "epoch": 1.36, + "learning_rate": 3.041784702549575e-07, + "logits/generated": -1.9965765476226807, + "logits/real": -2.017127752304077, + "logps/generated": -514.8785400390625, + "logps/real": -356.6257019042969, + "loss": 0.037, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -16.99004364013672, + "rewards/margins": 14.245416641235352, + "rewards/real": -2.744624137878418, + "step": 2130 + }, + { + "epoch": 1.36, + "learning_rate": 3.029981114258734e-07, + "logits/generated": -2.1022512912750244, + "logits/real": -2.1027119159698486, + "logps/generated": -604.0387573242188, + "logps/real": -385.0242614746094, + "loss": 0.0106, + "rewards/accuracies": 1.0, + "rewards/generated": -18.965845108032227, + "rewards/margins": 15.505331039428711, + "rewards/real": -3.4605114459991455, + "step": 2140 + }, + { + "epoch": 1.37, + "learning_rate": 3.0181775259678943e-07, + "logits/generated": -2.040330648422241, + "logits/real": -2.043025493621826, + "logps/generated": -526.4854125976562, + "logps/real": -329.0558166503906, + "loss": 0.0099, + "rewards/accuracies": 1.0, + "rewards/generated": -18.533267974853516, + "rewards/margins": 14.560073852539062, + "rewards/real": -3.9731955528259277, + "step": 2150 + }, + { + "epoch": 1.38, + "learning_rate": 3.0063739376770534e-07, + "logits/generated": -1.9047149419784546, + "logits/real": -1.9725589752197266, + "logps/generated": -566.95703125, + "logps/real": -349.62225341796875, + "loss": 0.0082, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -17.994510650634766, + "rewards/margins": 13.828272819519043, + "rewards/real": -4.16623592376709, + "step": 2160 + }, + { + "epoch": 1.38, + "learning_rate": 2.9945703493862135e-07, + "logits/generated": -1.948716402053833, + "logits/real": -1.977604866027832, + "logps/generated": -559.7679443359375, + "logps/real": -383.2321472167969, + "loss": 0.01, + "rewards/accuracies": 1.0, + "rewards/generated": -18.997352600097656, + "rewards/margins": 15.017451286315918, + "rewards/real": -3.9799041748046875, + "step": 2170 + }, + { + "epoch": 1.39, + "learning_rate": 2.9827667610953725e-07, + "logits/generated": -2.073700189590454, + "logits/real": -2.0532214641571045, + "logps/generated": -609.56640625, + "logps/real": -405.2546691894531, + "loss": 0.0061, + "rewards/accuracies": 1.0, + "rewards/generated": -20.087238311767578, + "rewards/margins": 15.188919067382812, + "rewards/real": -4.898318290710449, + "step": 2180 + }, + { + "epoch": 1.4, + "learning_rate": 2.9709631728045326e-07, + "logits/generated": -1.952592134475708, + "logits/real": -1.9506418704986572, + "logps/generated": -640.0557861328125, + "logps/real": -387.53521728515625, + "loss": 0.0065, + "rewards/accuracies": 1.0, + "rewards/generated": -23.34592056274414, + "rewards/margins": 18.07444190979004, + "rewards/real": -5.271478652954102, + "step": 2190 + }, + { + "epoch": 1.4, + "learning_rate": 2.9591595845136916e-07, + "logits/generated": -1.850500464439392, + "logits/real": -1.9458564519882202, + "logps/generated": -637.2237548828125, + "logps/real": -384.0931701660156, + "loss": 0.0058, + "rewards/accuracies": 1.0, + "rewards/generated": -23.331235885620117, + "rewards/margins": 17.88479995727539, + "rewards/real": -5.446437835693359, + "step": 2200 + }, + { + "epoch": 1.41, + "learning_rate": 2.9473559962228517e-07, + "logits/generated": -1.9568474292755127, + "logits/real": -1.942103624343872, + "logps/generated": -567.5823974609375, + "logps/real": -360.9796142578125, + "loss": 0.0112, + "rewards/accuracies": 1.0, + "rewards/generated": -21.4537410736084, + "rewards/margins": 16.80596923828125, + "rewards/real": -4.647771835327148, + "step": 2210 + }, + { + "epoch": 1.41, + "learning_rate": 2.9355524079320107e-07, + "logits/generated": -1.9142744541168213, + "logits/real": -1.9671341180801392, + "logps/generated": -533.4711303710938, + "logps/real": -363.37579345703125, + "loss": 0.0074, + "rewards/accuracies": 1.0, + "rewards/generated": -18.878368377685547, + "rewards/margins": 15.44507884979248, + "rewards/real": -3.433288097381592, + "step": 2220 + }, + { + "epoch": 1.42, + "learning_rate": 2.923748819641171e-07, + "logits/generated": -1.9527578353881836, + "logits/real": -2.004321336746216, + "logps/generated": -551.7341918945312, + "logps/real": -358.7703552246094, + "loss": 0.0166, + "rewards/accuracies": 1.0, + "rewards/generated": -17.526161193847656, + "rewards/margins": 15.12028694152832, + "rewards/real": -2.405874729156494, + "step": 2230 + }, + { + "epoch": 1.43, + "learning_rate": 2.91194523135033e-07, + "logits/generated": -1.9039808511734009, + "logits/real": -1.980120062828064, + "logps/generated": -543.2921752929688, + "logps/real": -353.02783203125, + "loss": 0.0111, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -19.195070266723633, + "rewards/margins": 16.527559280395508, + "rewards/real": -2.6675097942352295, + "step": 2240 + }, + { + "epoch": 1.43, + "learning_rate": 2.90014164305949e-07, + "logits/generated": -1.847700834274292, + "logits/real": -1.9540469646453857, + "logps/generated": -593.4288330078125, + "logps/real": -375.37640380859375, + "loss": 0.0038, + "rewards/accuracies": 1.0, + "rewards/generated": -18.847530364990234, + "rewards/margins": 15.369224548339844, + "rewards/real": -3.478306531906128, + "step": 2250 + }, + { + "epoch": 1.44, + "learning_rate": 2.88833805476865e-07, + "logits/generated": -1.789811372756958, + "logits/real": -1.8600037097930908, + "logps/generated": -596.4209594726562, + "logps/real": -343.8986511230469, + "loss": 0.0122, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -20.957176208496094, + "rewards/margins": 17.40493392944336, + "rewards/real": -3.552243709564209, + "step": 2260 + }, + { + "epoch": 1.45, + "learning_rate": 2.876534466477809e-07, + "logits/generated": -1.9649999141693115, + "logits/real": -1.9614098072052002, + "logps/generated": -623.26416015625, + "logps/real": -385.95208740234375, + "loss": 0.0114, + "rewards/accuracies": 1.0, + "rewards/generated": -18.63286018371582, + "rewards/margins": 15.91157054901123, + "rewards/real": -2.7212867736816406, + "step": 2270 + }, + { + "epoch": 1.45, + "learning_rate": 2.864730878186969e-07, + "logits/generated": -1.9861758947372437, + "logits/real": -2.0576374530792236, + "logps/generated": -531.93994140625, + "logps/real": -346.9412536621094, + "loss": 0.0105, + "rewards/accuracies": 1.0, + "rewards/generated": -16.093090057373047, + "rewards/margins": 14.241249084472656, + "rewards/real": -1.8518438339233398, + "step": 2280 + }, + { + "epoch": 1.46, + "learning_rate": 2.852927289896128e-07, + "logits/generated": -1.969087839126587, + "logits/real": -2.045712471008301, + "logps/generated": -526.2496948242188, + "logps/real": -293.7600402832031, + "loss": 0.0086, + "rewards/accuracies": 1.0, + "rewards/generated": -17.938573837280273, + "rewards/margins": 15.072761535644531, + "rewards/real": -2.8658127784729004, + "step": 2290 + }, + { + "epoch": 1.47, + "learning_rate": 2.841123701605288e-07, + "logits/generated": -1.9406864643096924, + "logits/real": -1.9671666622161865, + "logps/generated": -586.916015625, + "logps/real": -350.1014709472656, + "loss": 0.0183, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -19.514951705932617, + "rewards/margins": 16.66866683959961, + "rewards/real": -2.8462815284729004, + "step": 2300 + }, + { + "epoch": 1.47, + "learning_rate": 2.8293201133144473e-07, + "logits/generated": -1.8336292505264282, + "logits/real": -1.9111759662628174, + "logps/generated": -579.3409423828125, + "logps/real": -409.47247314453125, + "loss": 0.0091, + "rewards/accuracies": 1.0, + "rewards/generated": -17.272781372070312, + "rewards/margins": 14.239641189575195, + "rewards/real": -3.033140182495117, + "step": 2310 + }, + { + "epoch": 1.48, + "learning_rate": 2.8175165250236074e-07, + "logits/generated": -1.9630630016326904, + "logits/real": -2.0040669441223145, + "logps/generated": -627.9946899414062, + "logps/real": -402.39947509765625, + "loss": 0.0311, + "rewards/accuracies": 1.0, + "rewards/generated": -18.852710723876953, + "rewards/margins": 16.41490936279297, + "rewards/real": -2.4378018379211426, + "step": 2320 + }, + { + "epoch": 1.49, + "learning_rate": 2.8057129367327664e-07, + "logits/generated": -1.8958536386489868, + "logits/real": -1.966052770614624, + "logps/generated": -556.4025268554688, + "logps/real": -369.4639892578125, + "loss": 0.0116, + "rewards/accuracies": 1.0, + "rewards/generated": -18.53133773803711, + "rewards/margins": 15.089123725891113, + "rewards/real": -3.442213535308838, + "step": 2330 + }, + { + "epoch": 1.49, + "learning_rate": 2.7939093484419265e-07, + "logits/generated": -1.7555195093154907, + "logits/real": -1.8204386234283447, + "logps/generated": -621.0537109375, + "logps/real": -434.33575439453125, + "loss": 0.007, + "rewards/accuracies": 1.0, + "rewards/generated": -17.336116790771484, + "rewards/margins": 14.26280403137207, + "rewards/real": -3.073312997817993, + "step": 2340 + }, + { + "epoch": 1.5, + "learning_rate": 2.7821057601510855e-07, + "logits/generated": -1.9087238311767578, + "logits/real": -1.864392638206482, + "logps/generated": -592.5103759765625, + "logps/real": -371.4455871582031, + "loss": 0.0102, + "rewards/accuracies": 1.0, + "rewards/generated": -18.54368782043457, + "rewards/margins": 15.724161148071289, + "rewards/real": -2.8195271492004395, + "step": 2350 + }, + { + "epoch": 1.5, + "learning_rate": 2.7703021718602456e-07, + "logits/generated": -1.7295973300933838, + "logits/real": -1.8111556768417358, + "logps/generated": -549.3383178710938, + "logps/real": -381.1587829589844, + "loss": 0.0124, + "rewards/accuracies": 1.0, + "rewards/generated": -17.521865844726562, + "rewards/margins": 14.329907417297363, + "rewards/real": -3.1919589042663574, + "step": 2360 + }, + { + "epoch": 1.51, + "learning_rate": 2.758498583569405e-07, + "logits/generated": -1.9111597537994385, + "logits/real": -1.915789246559143, + "logps/generated": -589.12841796875, + "logps/real": -392.4088439941406, + "loss": 0.0082, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -17.989315032958984, + "rewards/margins": 14.640668869018555, + "rewards/real": -3.348646879196167, + "step": 2370 + }, + { + "epoch": 1.52, + "learning_rate": 2.7466949952785647e-07, + "logits/generated": -1.8938426971435547, + "logits/real": -1.9094927310943604, + "logps/generated": -577.1150512695312, + "logps/real": -367.19818115234375, + "loss": 0.0257, + "rewards/accuracies": 1.0, + "rewards/generated": -18.56988525390625, + "rewards/margins": 15.400243759155273, + "rewards/real": -3.169642448425293, + "step": 2380 + }, + { + "epoch": 1.52, + "learning_rate": 2.7348914069877243e-07, + "logits/generated": -1.8998384475708008, + "logits/real": -1.9750938415527344, + "logps/generated": -590.9421997070312, + "logps/real": -358.05084228515625, + "loss": 0.0101, + "rewards/accuracies": 1.0, + "rewards/generated": -17.235286712646484, + "rewards/margins": 15.2969388961792, + "rewards/real": -1.9383504390716553, + "step": 2390 + }, + { + "epoch": 1.53, + "learning_rate": 2.723087818696884e-07, + "logits/generated": -1.9660714864730835, + "logits/real": -2.039141893386841, + "logps/generated": -641.7316284179688, + "logps/real": -438.95068359375, + "loss": 0.0118, + "rewards/accuracies": 1.0, + "rewards/generated": -19.690990447998047, + "rewards/margins": 17.07895851135254, + "rewards/real": -2.612032413482666, + "step": 2400 + }, + { + "epoch": 1.54, + "learning_rate": 2.7112842304060434e-07, + "logits/generated": -1.8549489974975586, + "logits/real": -1.8724877834320068, + "logps/generated": -581.1431884765625, + "logps/real": -346.097900390625, + "loss": 0.0206, + "rewards/accuracies": 1.0, + "rewards/generated": -19.278120040893555, + "rewards/margins": 14.678240776062012, + "rewards/real": -4.599879264831543, + "step": 2410 + }, + { + "epoch": 1.54, + "learning_rate": 2.699480642115203e-07, + "logits/generated": -1.8291066884994507, + "logits/real": -1.9074589014053345, + "logps/generated": -524.2640380859375, + "logps/real": -363.2029113769531, + "loss": 0.0117, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -20.019777297973633, + "rewards/margins": 15.589925765991211, + "rewards/real": -4.429852485656738, + "step": 2420 + }, + { + "epoch": 1.55, + "learning_rate": 2.6876770538243625e-07, + "logits/generated": -1.8376376628875732, + "logits/real": -1.9334462881088257, + "logps/generated": -592.1920776367188, + "logps/real": -365.0409240722656, + "loss": 0.0029, + "rewards/accuracies": 1.0, + "rewards/generated": -19.884618759155273, + "rewards/margins": 16.311355590820312, + "rewards/real": -3.573265790939331, + "step": 2430 + }, + { + "epoch": 1.56, + "learning_rate": 2.675873465533522e-07, + "logits/generated": -1.8783462047576904, + "logits/real": -1.946720838546753, + "logps/generated": -627.4698486328125, + "logps/real": -336.26483154296875, + "loss": 0.009, + "rewards/accuracies": 1.0, + "rewards/generated": -20.798580169677734, + "rewards/margins": 15.276145935058594, + "rewards/real": -5.522434711456299, + "step": 2440 + }, + { + "epoch": 1.56, + "learning_rate": 2.6640698772426816e-07, + "logits/generated": -1.7473777532577515, + "logits/real": -1.8735145330429077, + "logps/generated": -690.8787231445312, + "logps/real": -410.44305419921875, + "loss": 0.0054, + "rewards/accuracies": 1.0, + "rewards/generated": -23.989131927490234, + "rewards/margins": 17.854623794555664, + "rewards/real": -6.134505271911621, + "step": 2450 + }, + { + "epoch": 1.57, + "learning_rate": 2.652266288951841e-07, + "logits/generated": -1.7989393472671509, + "logits/real": -1.896427869796753, + "logps/generated": -634.1585083007812, + "logps/real": -387.42999267578125, + "loss": 0.0144, + "rewards/accuracies": 1.0, + "rewards/generated": -21.825700759887695, + "rewards/margins": 18.100568771362305, + "rewards/real": -3.725130796432495, + "step": 2460 + }, + { + "epoch": 1.57, + "learning_rate": 2.640462700661001e-07, + "logits/generated": -1.7635921239852905, + "logits/real": -1.9182155132293701, + "logps/generated": -603.0245361328125, + "logps/real": -398.66705322265625, + "loss": 0.0074, + "rewards/accuracies": 1.0, + "rewards/generated": -22.45081901550293, + "rewards/margins": 17.299501419067383, + "rewards/real": -5.151317596435547, + "step": 2470 + }, + { + "epoch": 1.58, + "learning_rate": 2.6286591123701603e-07, + "logits/generated": -1.7605371475219727, + "logits/real": -1.8389637470245361, + "logps/generated": -564.5905151367188, + "logps/real": -408.32208251953125, + "loss": 0.0279, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -18.519445419311523, + "rewards/margins": 15.005685806274414, + "rewards/real": -3.5137619972229004, + "step": 2480 + }, + { + "epoch": 1.59, + "learning_rate": 2.61685552407932e-07, + "logits/generated": -1.9637407064437866, + "logits/real": -2.0421011447906494, + "logps/generated": -548.9338989257812, + "logps/real": -365.4721374511719, + "loss": 0.0364, + "rewards/accuracies": 1.0, + "rewards/generated": -16.821117401123047, + "rewards/margins": 14.631574630737305, + "rewards/real": -2.189544677734375, + "step": 2490 + }, + { + "epoch": 1.59, + "learning_rate": 2.60505193578848e-07, + "logits/generated": -1.9351282119750977, + "logits/real": -2.0200695991516113, + "logps/generated": -547.0625610351562, + "logps/real": -368.0826416015625, + "loss": 0.0142, + "rewards/accuracies": 1.0, + "rewards/generated": -16.905696868896484, + "rewards/margins": 14.632412910461426, + "rewards/real": -2.273285388946533, + "step": 2500 + }, + { + "epoch": 1.6, + "learning_rate": 2.593248347497639e-07, + "logits/generated": -2.01572847366333, + "logits/real": -2.033661365509033, + "logps/generated": -558.5911254882812, + "logps/real": -337.11077880859375, + "loss": 0.0142, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -17.953121185302734, + "rewards/margins": 14.896084785461426, + "rewards/real": -3.057035207748413, + "step": 2510 + }, + { + "epoch": 1.61, + "learning_rate": 2.581444759206799e-07, + "logits/generated": -1.9725624322891235, + "logits/real": -2.0928730964660645, + "logps/generated": -521.43798828125, + "logps/real": -358.08575439453125, + "loss": 0.0134, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -15.997856140136719, + "rewards/margins": 14.111856460571289, + "rewards/real": -1.8859999179840088, + "step": 2520 + }, + { + "epoch": 1.61, + "learning_rate": 2.569641170915958e-07, + "logits/generated": -2.042390823364258, + "logits/real": -2.118346691131592, + "logps/generated": -624.4253540039062, + "logps/real": -406.1341247558594, + "loss": 0.0083, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -17.157394409179688, + "rewards/margins": 14.666315078735352, + "rewards/real": -2.4910807609558105, + "step": 2530 + }, + { + "epoch": 1.62, + "learning_rate": 2.557837582625118e-07, + "logits/generated": -1.96627676486969, + "logits/real": -2.15256929397583, + "logps/generated": -572.9945068359375, + "logps/real": -338.84124755859375, + "loss": 0.0063, + "rewards/accuracies": 1.0, + "rewards/generated": -18.74167251586914, + "rewards/margins": 15.573163986206055, + "rewards/real": -3.168510913848877, + "step": 2540 + }, + { + "epoch": 1.63, + "learning_rate": 2.546033994334277e-07, + "logits/generated": -2.035891532897949, + "logits/real": -2.0882015228271484, + "logps/generated": -559.4519653320312, + "logps/real": -347.10888671875, + "loss": 0.016, + "rewards/accuracies": 1.0, + "rewards/generated": -17.2998046875, + "rewards/margins": 14.863592147827148, + "rewards/real": -2.4362099170684814, + "step": 2550 + }, + { + "epoch": 1.63, + "learning_rate": 2.5342304060434373e-07, + "logits/generated": -2.0236713886260986, + "logits/real": -2.1474721431732178, + "logps/generated": -529.9071655273438, + "logps/real": -397.75146484375, + "loss": 0.0041, + "rewards/accuracies": 1.0, + "rewards/generated": -16.087312698364258, + "rewards/margins": 13.304840087890625, + "rewards/real": -2.782471179962158, + "step": 2560 + }, + { + "epoch": 1.64, + "learning_rate": 2.5224268177525964e-07, + "logits/generated": -2.025327205657959, + "logits/real": -2.1291823387145996, + "logps/generated": -604.8438110351562, + "logps/real": -394.15179443359375, + "loss": 0.0065, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -19.648643493652344, + "rewards/margins": 17.793973922729492, + "rewards/real": -1.8546674251556396, + "step": 2570 + }, + { + "epoch": 1.64, + "learning_rate": 2.5106232294617564e-07, + "logits/generated": -2.023483991622925, + "logits/real": -2.128129482269287, + "logps/generated": -517.6430053710938, + "logps/real": -315.3306884765625, + "loss": 0.0174, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -17.537776947021484, + "rewards/margins": 15.272656440734863, + "rewards/real": -2.2651212215423584, + "step": 2580 + }, + { + "epoch": 1.65, + "learning_rate": 2.498819641170916e-07, + "logits/generated": -2.151948928833008, + "logits/real": -2.201842784881592, + "logps/generated": -526.9323120117188, + "logps/real": -372.4267883300781, + "loss": 0.0045, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -16.658296585083008, + "rewards/margins": 15.20775032043457, + "rewards/real": -1.4505455493927002, + "step": 2590 + }, + { + "epoch": 1.66, + "learning_rate": 2.4870160528800756e-07, + "logits/generated": -2.1339409351348877, + "logits/real": -2.200197219848633, + "logps/generated": -643.9109497070312, + "logps/real": -348.0181579589844, + "loss": 0.0049, + "rewards/accuracies": 1.0, + "rewards/generated": -18.10948944091797, + "rewards/margins": 16.632869720458984, + "rewards/real": -1.476617455482483, + "step": 2600 + }, + { + "epoch": 1.66, + "learning_rate": 2.475212464589235e-07, + "logits/generated": -2.1192688941955566, + "logits/real": -2.1132771968841553, + "logps/generated": -591.93505859375, + "logps/real": -353.3289794921875, + "loss": 0.0119, + "rewards/accuracies": 1.0, + "rewards/generated": -17.125486373901367, + "rewards/margins": 15.0489501953125, + "rewards/real": -2.076535701751709, + "step": 2610 + }, + { + "epoch": 1.67, + "learning_rate": 2.4634088762983947e-07, + "logits/generated": -2.056791067123413, + "logits/real": -2.1385152339935303, + "logps/generated": -536.34423828125, + "logps/real": -291.66473388671875, + "loss": 0.0077, + "rewards/accuracies": 1.0, + "rewards/generated": -18.626489639282227, + "rewards/margins": 16.677309036254883, + "rewards/real": -1.9491809606552124, + "step": 2620 + }, + { + "epoch": 1.68, + "learning_rate": 2.451605288007554e-07, + "logits/generated": -2.140279531478882, + "logits/real": -2.2342429161071777, + "logps/generated": -601.8104248046875, + "logps/real": -410.847900390625, + "loss": 0.0083, + "rewards/accuracies": 1.0, + "rewards/generated": -19.37628746032715, + "rewards/margins": 16.318693161010742, + "rewards/real": -3.057591438293457, + "step": 2630 + }, + { + "epoch": 1.68, + "learning_rate": 2.439801699716714e-07, + "logits/generated": -2.1028506755828857, + "logits/real": -2.23150372505188, + "logps/generated": -608.1312255859375, + "logps/real": -395.540771484375, + "loss": 0.0103, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -19.10951042175293, + "rewards/margins": 17.4754581451416, + "rewards/real": -1.6340545415878296, + "step": 2640 + }, + { + "epoch": 1.69, + "learning_rate": 2.4279981114258734e-07, + "logits/generated": -2.1094822883605957, + "logits/real": -2.1931681632995605, + "logps/generated": -532.2767333984375, + "logps/real": -341.1546630859375, + "loss": 0.0077, + "rewards/accuracies": 1.0, + "rewards/generated": -17.928924560546875, + "rewards/margins": 14.880243301391602, + "rewards/real": -3.0486807823181152, + "step": 2650 + }, + { + "epoch": 1.7, + "learning_rate": 2.416194523135033e-07, + "logits/generated": -2.069281578063965, + "logits/real": -2.193920612335205, + "logps/generated": -533.3090209960938, + "logps/real": -331.2969970703125, + "loss": 0.003, + "rewards/accuracies": 1.0, + "rewards/generated": -20.43006706237793, + "rewards/margins": 17.58050537109375, + "rewards/real": -2.849562883377075, + "step": 2660 + }, + { + "epoch": 1.7, + "learning_rate": 2.4043909348441925e-07, + "logits/generated": -2.1191678047180176, + "logits/real": -2.225715398788452, + "logps/generated": -576.5401000976562, + "logps/real": -431.4150390625, + "loss": 0.0034, + "rewards/accuracies": 1.0, + "rewards/generated": -19.92889404296875, + "rewards/margins": 16.879291534423828, + "rewards/real": -3.0496022701263428, + "step": 2670 + }, + { + "epoch": 1.71, + "learning_rate": 2.392587346553352e-07, + "logits/generated": -1.9661951065063477, + "logits/real": -2.105461359024048, + "logps/generated": -558.3134155273438, + "logps/real": -346.55206298828125, + "loss": 0.0112, + "rewards/accuracies": 1.0, + "rewards/generated": -20.488914489746094, + "rewards/margins": 17.84439468383789, + "rewards/real": -2.6445188522338867, + "step": 2680 + }, + { + "epoch": 1.71, + "learning_rate": 2.3807837582625119e-07, + "logits/generated": -1.9735558032989502, + "logits/real": -2.1056549549102783, + "logps/generated": -551.8961181640625, + "logps/real": -357.57305908203125, + "loss": 0.0107, + "rewards/accuracies": 1.0, + "rewards/generated": -18.85560417175293, + "rewards/margins": 15.564756393432617, + "rewards/real": -3.2908477783203125, + "step": 2690 + }, + { + "epoch": 1.72, + "learning_rate": 2.3689801699716714e-07, + "logits/generated": -1.9503015279769897, + "logits/real": -2.0592892169952393, + "logps/generated": -657.7816772460938, + "logps/real": -359.5049133300781, + "loss": 0.024, + "rewards/accuracies": 1.0, + "rewards/generated": -20.253496170043945, + "rewards/margins": 17.200815200805664, + "rewards/real": -3.052682399749756, + "step": 2700 + }, + { + "epoch": 1.73, + "learning_rate": 2.357176581680831e-07, + "logits/generated": -1.9928719997406006, + "logits/real": -2.099074363708496, + "logps/generated": -668.011962890625, + "logps/real": -421.2786560058594, + "loss": 0.0065, + "rewards/accuracies": 1.0, + "rewards/generated": -19.69591522216797, + "rewards/margins": 15.926002502441406, + "rewards/real": -3.7699134349823, + "step": 2710 + }, + { + "epoch": 1.73, + "learning_rate": 2.3453729933899905e-07, + "logits/generated": -1.8637622594833374, + "logits/real": -2.0058817863464355, + "logps/generated": -565.0535888671875, + "logps/real": -371.9683532714844, + "loss": 0.0139, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -19.948787689208984, + "rewards/margins": 17.805095672607422, + "rewards/real": -2.143691062927246, + "step": 2720 + }, + { + "epoch": 1.74, + "learning_rate": 2.33356940509915e-07, + "logits/generated": -1.8944871425628662, + "logits/real": -1.973111867904663, + "logps/generated": -562.8475341796875, + "logps/real": -369.47393798828125, + "loss": 0.0096, + "rewards/accuracies": 1.0, + "rewards/generated": -20.202678680419922, + "rewards/margins": 16.770105361938477, + "rewards/real": -3.4325757026672363, + "step": 2730 + }, + { + "epoch": 1.75, + "learning_rate": 2.3217658168083097e-07, + "logits/generated": -1.9282621145248413, + "logits/real": -2.04433012008667, + "logps/generated": -611.9664306640625, + "logps/real": -364.351806640625, + "loss": 0.0085, + "rewards/accuracies": 1.0, + "rewards/generated": -19.971899032592773, + "rewards/margins": 16.668487548828125, + "rewards/real": -3.3034119606018066, + "step": 2740 + }, + { + "epoch": 1.75, + "learning_rate": 2.3099622285174692e-07, + "logits/generated": -1.973746657371521, + "logits/real": -1.9992039203643799, + "logps/generated": -582.0021362304688, + "logps/real": -388.60870361328125, + "loss": 0.0078, + "rewards/accuracies": 1.0, + "rewards/generated": -18.697378158569336, + "rewards/margins": 15.425535202026367, + "rewards/real": -3.2718424797058105, + "step": 2750 + }, + { + "epoch": 1.76, + "learning_rate": 2.2981586402266288e-07, + "logits/generated": -1.84421706199646, + "logits/real": -1.938523530960083, + "logps/generated": -631.9494018554688, + "logps/real": -414.2872009277344, + "loss": 0.004, + "rewards/accuracies": 1.0, + "rewards/generated": -22.86662483215332, + "rewards/margins": 18.36226463317871, + "rewards/real": -4.504361152648926, + "step": 2760 + }, + { + "epoch": 1.77, + "learning_rate": 2.2863550519357883e-07, + "logits/generated": -1.7387750148773193, + "logits/real": -1.866440773010254, + "logps/generated": -559.9677734375, + "logps/real": -360.2195739746094, + "loss": 0.0087, + "rewards/accuracies": 1.0, + "rewards/generated": -20.104677200317383, + "rewards/margins": 15.611299514770508, + "rewards/real": -4.493377208709717, + "step": 2770 + }, + { + "epoch": 1.77, + "learning_rate": 2.2745514636449482e-07, + "logits/generated": -1.7847436666488647, + "logits/real": -1.8568544387817383, + "logps/generated": -645.9898071289062, + "logps/real": -373.1492614746094, + "loss": 0.0179, + "rewards/accuracies": 1.0, + "rewards/generated": -22.015918731689453, + "rewards/margins": 17.876659393310547, + "rewards/real": -4.1392598152160645, + "step": 2780 + }, + { + "epoch": 1.78, + "learning_rate": 2.2627478753541077e-07, + "logits/generated": -1.7664238214492798, + "logits/real": -1.8905664682388306, + "logps/generated": -617.681884765625, + "logps/real": -382.783447265625, + "loss": 0.0072, + "rewards/accuracies": 1.0, + "rewards/generated": -24.042709350585938, + "rewards/margins": 19.182605743408203, + "rewards/real": -4.860101699829102, + "step": 2790 + }, + { + "epoch": 1.78, + "learning_rate": 2.2509442870632673e-07, + "logits/generated": -1.707985520362854, + "logits/real": -1.8283040523529053, + "logps/generated": -597.2868041992188, + "logps/real": -339.0484924316406, + "loss": 0.0156, + "rewards/accuracies": 1.0, + "rewards/generated": -20.454498291015625, + "rewards/margins": 16.770410537719727, + "rewards/real": -3.6840851306915283, + "step": 2800 + }, + { + "epoch": 1.79, + "learning_rate": 2.2391406987724268e-07, + "logits/generated": -1.7828060388565063, + "logits/real": -1.9414622783660889, + "logps/generated": -601.31884765625, + "logps/real": -395.41790771484375, + "loss": 0.01, + "rewards/accuracies": 1.0, + "rewards/generated": -19.245166778564453, + "rewards/margins": 16.227514266967773, + "rewards/real": -3.017648220062256, + "step": 2810 + }, + { + "epoch": 1.8, + "learning_rate": 2.2273371104815864e-07, + "logits/generated": -1.7189252376556396, + "logits/real": -1.8330001831054688, + "logps/generated": -625.9137573242188, + "logps/real": -384.0910339355469, + "loss": 0.0109, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -18.94672966003418, + "rewards/margins": 16.3648681640625, + "rewards/real": -2.5818610191345215, + "step": 2820 + }, + { + "epoch": 1.8, + "learning_rate": 2.215533522190746e-07, + "logits/generated": -1.792650580406189, + "logits/real": -1.8975502252578735, + "logps/generated": -573.1437377929688, + "logps/real": -371.3572692871094, + "loss": 0.0194, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -19.24198341369629, + "rewards/margins": 16.90559196472168, + "rewards/real": -2.3363890647888184, + "step": 2830 + }, + { + "epoch": 1.81, + "learning_rate": 2.2037299338999055e-07, + "logits/generated": -1.8636354207992554, + "logits/real": -1.993896484375, + "logps/generated": -632.2186279296875, + "logps/real": -440.17010498046875, + "loss": 0.0085, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -18.489797592163086, + "rewards/margins": 15.86462116241455, + "rewards/real": -2.625178813934326, + "step": 2840 + }, + { + "epoch": 1.82, + "learning_rate": 2.191926345609065e-07, + "logits/generated": -1.8625421524047852, + "logits/real": -1.9541743993759155, + "logps/generated": -533.5960083007812, + "logps/real": -303.5115966796875, + "loss": 0.0068, + "rewards/accuracies": 1.0, + "rewards/generated": -19.34178924560547, + "rewards/margins": 16.277996063232422, + "rewards/real": -3.063790798187256, + "step": 2850 + }, + { + "epoch": 1.82, + "learning_rate": 2.1801227573182246e-07, + "logits/generated": -1.8944851160049438, + "logits/real": -2.0513484477996826, + "logps/generated": -523.5350341796875, + "logps/real": -364.8716735839844, + "loss": 0.02, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -18.236291885375977, + "rewards/margins": 15.601948738098145, + "rewards/real": -2.6343421936035156, + "step": 2860 + }, + { + "epoch": 1.83, + "learning_rate": 2.1683191690273842e-07, + "logits/generated": -1.926628828048706, + "logits/real": -2.017603635787964, + "logps/generated": -615.1983642578125, + "logps/real": -387.6255798339844, + "loss": 0.0043, + "rewards/accuracies": 1.0, + "rewards/generated": -20.679590225219727, + "rewards/margins": 16.500484466552734, + "rewards/real": -4.179107189178467, + "step": 2870 + }, + { + "epoch": 1.84, + "learning_rate": 2.1565155807365437e-07, + "logits/generated": -1.883113145828247, + "logits/real": -1.9269574880599976, + "logps/generated": -589.8172607421875, + "logps/real": -377.12188720703125, + "loss": 0.0117, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -20.098003387451172, + "rewards/margins": 16.679973602294922, + "rewards/real": -3.418027400970459, + "step": 2880 + }, + { + "epoch": 1.84, + "learning_rate": 2.1447119924457033e-07, + "logits/generated": -1.853742241859436, + "logits/real": -1.9918594360351562, + "logps/generated": -625.268798828125, + "logps/real": -399.7982177734375, + "loss": 0.0122, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -21.647125244140625, + "rewards/margins": 18.090368270874023, + "rewards/real": -3.556757688522339, + "step": 2890 + }, + { + "epoch": 1.85, + "learning_rate": 2.132908404154863e-07, + "logits/generated": -1.8341211080551147, + "logits/real": -2.004746437072754, + "logps/generated": -591.0967407226562, + "logps/real": -395.78594970703125, + "loss": 0.0106, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -20.23607635498047, + "rewards/margins": 14.956598281860352, + "rewards/real": -5.279480934143066, + "step": 2900 + }, + { + "epoch": 1.85, + "learning_rate": 2.1211048158640227e-07, + "logits/generated": -1.8232825994491577, + "logits/real": -1.9642972946166992, + "logps/generated": -627.2144775390625, + "logps/real": -407.8313293457031, + "loss": 0.0116, + "rewards/accuracies": 1.0, + "rewards/generated": -20.575244903564453, + "rewards/margins": 16.956144332885742, + "rewards/real": -3.6191020011901855, + "step": 2910 + }, + { + "epoch": 1.86, + "learning_rate": 2.1093012275731822e-07, + "logits/generated": -1.8161029815673828, + "logits/real": -1.9360853433609009, + "logps/generated": -556.29736328125, + "logps/real": -382.58642578125, + "loss": 0.0131, + "rewards/accuracies": 1.0, + "rewards/generated": -20.463085174560547, + "rewards/margins": 16.687808990478516, + "rewards/real": -3.7752761840820312, + "step": 2920 + }, + { + "epoch": 1.87, + "learning_rate": 2.0974976392823418e-07, + "logits/generated": -1.8562109470367432, + "logits/real": -1.9310420751571655, + "logps/generated": -535.4537963867188, + "logps/real": -308.462646484375, + "loss": 0.0134, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -19.243331909179688, + "rewards/margins": 16.052427291870117, + "rewards/real": -3.190903902053833, + "step": 2930 + }, + { + "epoch": 1.87, + "learning_rate": 2.0856940509915014e-07, + "logits/generated": -1.78777277469635, + "logits/real": -1.9053926467895508, + "logps/generated": -657.8607177734375, + "logps/real": -399.79998779296875, + "loss": 0.0073, + "rewards/accuracies": 1.0, + "rewards/generated": -22.9168643951416, + "rewards/margins": 18.889469146728516, + "rewards/real": -4.027396202087402, + "step": 2940 + }, + { + "epoch": 1.88, + "learning_rate": 2.073890462700661e-07, + "logits/generated": -1.8248649835586548, + "logits/real": -1.8879339694976807, + "logps/generated": -646.277587890625, + "logps/real": -382.84844970703125, + "loss": 0.0037, + "rewards/accuracies": 1.0, + "rewards/generated": -20.766597747802734, + "rewards/margins": 16.174793243408203, + "rewards/real": -4.591801643371582, + "step": 2950 + }, + { + "epoch": 1.89, + "learning_rate": 2.0620868744098205e-07, + "logits/generated": -1.8774172067642212, + "logits/real": -1.9616762399673462, + "logps/generated": -581.1622924804688, + "logps/real": -339.016845703125, + "loss": 0.0053, + "rewards/accuracies": 1.0, + "rewards/generated": -18.804542541503906, + "rewards/margins": 14.824602127075195, + "rewards/real": -3.9799411296844482, + "step": 2960 + }, + { + "epoch": 1.89, + "learning_rate": 2.05028328611898e-07, + "logits/generated": -1.8371098041534424, + "logits/real": -1.9512956142425537, + "logps/generated": -515.0367431640625, + "logps/real": -297.47882080078125, + "loss": 0.0102, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -18.159709930419922, + "rewards/margins": 15.079408645629883, + "rewards/real": -3.0803000926971436, + "step": 2970 + }, + { + "epoch": 1.9, + "learning_rate": 2.0384796978281396e-07, + "logits/generated": -1.8390519618988037, + "logits/real": -1.981048583984375, + "logps/generated": -567.8472290039062, + "logps/real": -380.55535888671875, + "loss": 0.025, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -19.843801498413086, + "rewards/margins": 16.086761474609375, + "rewards/real": -3.7570385932922363, + "step": 2980 + }, + { + "epoch": 1.91, + "learning_rate": 2.0266761095372992e-07, + "logits/generated": -1.6872667074203491, + "logits/real": -1.8214622735977173, + "logps/generated": -674.7174682617188, + "logps/real": -419.5072326660156, + "loss": 0.0068, + "rewards/accuracies": 1.0, + "rewards/generated": -24.660266876220703, + "rewards/margins": 19.672096252441406, + "rewards/real": -4.988171100616455, + "step": 2990 + }, + { + "epoch": 1.91, + "learning_rate": 2.0148725212464587e-07, + "logits/generated": -1.6084331274032593, + "logits/real": -1.6493154764175415, + "logps/generated": -598.4984130859375, + "logps/real": -410.1678771972656, + "loss": 0.0152, + "rewards/accuracies": 1.0, + "rewards/generated": -21.55122947692871, + "rewards/margins": 17.81758689880371, + "rewards/real": -3.7336437702178955, + "step": 3000 + }, + { + "epoch": 1.92, + "learning_rate": 2.0030689329556183e-07, + "logits/generated": -1.6727161407470703, + "logits/real": -1.6407667398452759, + "logps/generated": -575.52783203125, + "logps/real": -379.488525390625, + "loss": 0.0199, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -19.098388671875, + "rewards/margins": 15.673646926879883, + "rewards/real": -3.424741744995117, + "step": 3010 + }, + { + "epoch": 1.92, + "learning_rate": 1.991265344664778e-07, + "logits/generated": -1.6170536279678345, + "logits/real": -1.7253925800323486, + "logps/generated": -529.9498901367188, + "logps/real": -295.4236755371094, + "loss": 0.0086, + "rewards/accuracies": 1.0, + "rewards/generated": -19.003612518310547, + "rewards/margins": 17.109363555908203, + "rewards/real": -1.8942499160766602, + "step": 3020 + }, + { + "epoch": 1.93, + "learning_rate": 1.9794617563739377e-07, + "logits/generated": -1.6006145477294922, + "logits/real": -1.683121919631958, + "logps/generated": -560.7699584960938, + "logps/real": -333.6227111816406, + "loss": 0.005, + "rewards/accuracies": 1.0, + "rewards/generated": -18.92214012145996, + "rewards/margins": 16.359729766845703, + "rewards/real": -2.5624115467071533, + "step": 3030 + }, + { + "epoch": 1.94, + "learning_rate": 1.9676581680830972e-07, + "logits/generated": -1.5276119709014893, + "logits/real": -1.5233627557754517, + "logps/generated": -562.1134033203125, + "logps/real": -383.27764892578125, + "loss": 0.0053, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -18.839946746826172, + "rewards/margins": 15.850112915039062, + "rewards/real": -2.9898345470428467, + "step": 3040 + }, + { + "epoch": 1.94, + "learning_rate": 1.9558545797922568e-07, + "logits/generated": -1.6427018642425537, + "logits/real": -1.6606805324554443, + "logps/generated": -593.7340087890625, + "logps/real": -376.98114013671875, + "loss": 0.0171, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -19.36456871032715, + "rewards/margins": 16.161014556884766, + "rewards/real": -3.203557252883911, + "step": 3050 + }, + { + "epoch": 1.95, + "learning_rate": 1.9440509915014163e-07, + "logits/generated": -1.6294114589691162, + "logits/real": -1.676039457321167, + "logps/generated": -590.101806640625, + "logps/real": -395.4278869628906, + "loss": 0.003, + "rewards/accuracies": 1.0, + "rewards/generated": -19.253379821777344, + "rewards/margins": 16.33585548400879, + "rewards/real": -2.9175257682800293, + "step": 3060 + }, + { + "epoch": 1.96, + "learning_rate": 1.932247403210576e-07, + "logits/generated": -1.4732720851898193, + "logits/real": -1.5254461765289307, + "logps/generated": -615.4378662109375, + "logps/real": -439.57177734375, + "loss": 0.0109, + "rewards/accuracies": 1.0, + "rewards/generated": -21.267192840576172, + "rewards/margins": 17.48326873779297, + "rewards/real": -3.783924102783203, + "step": 3070 + }, + { + "epoch": 1.96, + "learning_rate": 1.9204438149197355e-07, + "logits/generated": -1.4804540872573853, + "logits/real": -1.543918251991272, + "logps/generated": -621.9572143554688, + "logps/real": -386.83258056640625, + "loss": 0.0068, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -20.246889114379883, + "rewards/margins": 15.34863567352295, + "rewards/real": -4.898252487182617, + "step": 3080 + }, + { + "epoch": 1.97, + "learning_rate": 1.908640226628895e-07, + "logits/generated": -1.5399271249771118, + "logits/real": -1.601269006729126, + "logps/generated": -600.8512573242188, + "logps/real": -388.85052490234375, + "loss": 0.0118, + "rewards/accuracies": 1.0, + "rewards/generated": -23.299779891967773, + "rewards/margins": 18.00295066833496, + "rewards/real": -5.296830654144287, + "step": 3090 + }, + { + "epoch": 1.98, + "learning_rate": 1.8968366383380546e-07, + "logits/generated": -1.5507320165634155, + "logits/real": -1.6016521453857422, + "logps/generated": -587.0738525390625, + "logps/real": -345.40924072265625, + "loss": 0.0043, + "rewards/accuracies": 1.0, + "rewards/generated": -20.77097511291504, + "rewards/margins": 16.8800106048584, + "rewards/real": -3.890961170196533, + "step": 3100 + }, + { + "epoch": 1.98, + "learning_rate": 1.885033050047214e-07, + "logits/generated": -1.5871801376342773, + "logits/real": -1.5961694717407227, + "logps/generated": -598.388427734375, + "logps/real": -406.2129821777344, + "loss": 0.0033, + "rewards/accuracies": 1.0, + "rewards/generated": -20.48702049255371, + "rewards/margins": 16.04343032836914, + "rewards/real": -4.443593502044678, + "step": 3110 + }, + { + "epoch": 1.99, + "learning_rate": 1.8732294617563737e-07, + "logits/generated": -1.5982401371002197, + "logits/real": -1.6817703247070312, + "logps/generated": -575.6530151367188, + "logps/real": -383.60491943359375, + "loss": 0.014, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -20.538284301757812, + "rewards/margins": 16.78091049194336, + "rewards/real": -3.7573750019073486, + "step": 3120 + }, + { + "epoch": 1.99, + "learning_rate": 1.8614258734655332e-07, + "logits/generated": -1.6962788105010986, + "logits/real": -1.7089637517929077, + "logps/generated": -561.3473510742188, + "logps/real": -402.51416015625, + "loss": 0.0165, + "rewards/accuracies": 1.0, + "rewards/generated": -18.493038177490234, + "rewards/margins": 16.046886444091797, + "rewards/real": -2.4461536407470703, + "step": 3130 + }, + { + "epoch": 2.0, + "learning_rate": 1.849622285174693e-07, + "logits/generated": -1.6082890033721924, + "logits/real": -1.6420183181762695, + "logps/generated": -594.0357666015625, + "logps/real": -368.7581787109375, + "loss": 0.0133, + "rewards/accuracies": 1.0, + "rewards/generated": -21.077438354492188, + "rewards/margins": 17.73508071899414, + "rewards/real": -3.3423619270324707, + "step": 3140 + }, + { + "epoch": 2.01, + "learning_rate": 1.8378186968838526e-07, + "logits/generated": -1.55757737159729, + "logits/real": -1.649776816368103, + "logps/generated": -638.5458984375, + "logps/real": -389.14056396484375, + "loss": 0.0036, + "rewards/accuracies": 1.0, + "rewards/generated": -20.582530975341797, + "rewards/margins": 17.621540069580078, + "rewards/real": -2.960991382598877, + "step": 3150 + }, + { + "epoch": 2.01, + "learning_rate": 1.8260151085930122e-07, + "logits/generated": -1.556721806526184, + "logits/real": -1.5956058502197266, + "logps/generated": -659.8116455078125, + "logps/real": -379.24163818359375, + "loss": 0.0009, + "rewards/accuracies": 1.0, + "rewards/generated": -22.081661224365234, + "rewards/margins": 17.532869338989258, + "rewards/real": -4.548790454864502, + "step": 3160 + }, + { + "epoch": 2.02, + "learning_rate": 1.8142115203021718e-07, + "logits/generated": -1.61980402469635, + "logits/real": -1.636548638343811, + "logps/generated": -565.48876953125, + "logps/real": -327.365966796875, + "loss": 0.0025, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -20.394580841064453, + "rewards/margins": 16.083187103271484, + "rewards/real": -4.311388969421387, + "step": 3170 + }, + { + "epoch": 2.03, + "learning_rate": 1.8024079320113313e-07, + "logits/generated": -1.5890557765960693, + "logits/real": -1.6496063470840454, + "logps/generated": -626.4884033203125, + "logps/real": -376.7876892089844, + "loss": 0.0006, + "rewards/accuracies": 1.0, + "rewards/generated": -22.2101993560791, + "rewards/margins": 17.99839973449707, + "rewards/real": -4.211801052093506, + "step": 3180 + }, + { + "epoch": 2.03, + "learning_rate": 1.790604343720491e-07, + "logits/generated": -1.5914881229400635, + "logits/real": -1.5750452280044556, + "logps/generated": -601.8194580078125, + "logps/real": -403.5359191894531, + "loss": 0.0046, + "rewards/accuracies": 1.0, + "rewards/generated": -21.1069393157959, + "rewards/margins": 17.50069236755371, + "rewards/real": -3.606245517730713, + "step": 3190 + }, + { + "epoch": 2.04, + "learning_rate": 1.7788007554296504e-07, + "logits/generated": -1.5610685348510742, + "logits/real": -1.653970718383789, + "logps/generated": -659.7487182617188, + "logps/real": -400.736328125, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -22.223270416259766, + "rewards/margins": 18.803062438964844, + "rewards/real": -3.4202098846435547, + "step": 3200 + }, + { + "epoch": 2.05, + "learning_rate": 1.76699716713881e-07, + "logits/generated": -1.4919041395187378, + "logits/real": -1.5235627889633179, + "logps/generated": -651.9771728515625, + "logps/real": -388.6980285644531, + "loss": 0.003, + "rewards/accuracies": 1.0, + "rewards/generated": -22.809450149536133, + "rewards/margins": 18.30077362060547, + "rewards/real": -4.5086774826049805, + "step": 3210 + }, + { + "epoch": 2.05, + "learning_rate": 1.7551935788479695e-07, + "logits/generated": -1.5555975437164307, + "logits/real": -1.6326204538345337, + "logps/generated": -603.9024658203125, + "logps/real": -368.42681884765625, + "loss": 0.0032, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -22.483081817626953, + "rewards/margins": 18.959680557250977, + "rewards/real": -3.523402690887451, + "step": 3220 + }, + { + "epoch": 2.06, + "learning_rate": 1.743389990557129e-07, + "logits/generated": -1.463686227798462, + "logits/real": -1.558170199394226, + "logps/generated": -581.8925170898438, + "logps/real": -385.3419494628906, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -20.905954360961914, + "rewards/margins": 17.427875518798828, + "rewards/real": -3.478073835372925, + "step": 3230 + }, + { + "epoch": 2.07, + "learning_rate": 1.7315864022662887e-07, + "logits/generated": -1.5275118350982666, + "logits/real": -1.4879990816116333, + "logps/generated": -588.8587646484375, + "logps/real": -352.18804931640625, + "loss": 0.0027, + "rewards/accuracies": 1.0, + "rewards/generated": -21.36737060546875, + "rewards/margins": 17.114416122436523, + "rewards/real": -4.252957344055176, + "step": 3240 + }, + { + "epoch": 2.07, + "learning_rate": 1.7197828139754482e-07, + "logits/generated": -1.441959023475647, + "logits/real": -1.5623648166656494, + "logps/generated": -665.1011352539062, + "logps/real": -377.1506042480469, + "loss": 0.0011, + "rewards/accuracies": 1.0, + "rewards/generated": -23.579158782958984, + "rewards/margins": 19.479379653930664, + "rewards/real": -4.0997748374938965, + "step": 3250 + }, + { + "epoch": 2.08, + "learning_rate": 1.707979225684608e-07, + "logits/generated": -1.5857994556427002, + "logits/real": -1.50002920627594, + "logps/generated": -615.2679443359375, + "logps/real": -382.7515869140625, + "loss": 0.0034, + "rewards/accuracies": 1.0, + "rewards/generated": -23.19659996032715, + "rewards/margins": 18.741558074951172, + "rewards/real": -4.455041408538818, + "step": 3260 + }, + { + "epoch": 2.08, + "learning_rate": 1.6961756373937676e-07, + "logits/generated": -1.5039589405059814, + "logits/real": -1.5329926013946533, + "logps/generated": -639.8225708007812, + "logps/real": -447.75030517578125, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -23.96138572692871, + "rewards/margins": 19.722862243652344, + "rewards/real": -4.238521099090576, + "step": 3270 + }, + { + "epoch": 2.09, + "learning_rate": 1.6843720491029272e-07, + "logits/generated": -1.4386045932769775, + "logits/real": -1.5476876497268677, + "logps/generated": -651.0532836914062, + "logps/real": -361.032470703125, + "loss": 0.0013, + "rewards/accuracies": 1.0, + "rewards/generated": -24.863887786865234, + "rewards/margins": 19.566587448120117, + "rewards/real": -5.297298431396484, + "step": 3280 + }, + { + "epoch": 2.1, + "learning_rate": 1.672568460812087e-07, + "logits/generated": -1.462738275527954, + "logits/real": -1.4480798244476318, + "logps/generated": -601.6398315429688, + "logps/real": -390.4088134765625, + "loss": 0.0058, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -23.60664939880371, + "rewards/margins": 17.876808166503906, + "rewards/real": -5.7298407554626465, + "step": 3290 + }, + { + "epoch": 2.1, + "learning_rate": 1.6607648725212465e-07, + "logits/generated": -1.468587040901184, + "logits/real": -1.5072485208511353, + "logps/generated": -646.3826293945312, + "logps/real": -409.9957580566406, + "loss": 0.0006, + "rewards/accuracies": 1.0, + "rewards/generated": -23.434160232543945, + "rewards/margins": 17.710805892944336, + "rewards/real": -5.72335147857666, + "step": 3300 + }, + { + "epoch": 2.11, + "learning_rate": 1.648961284230406e-07, + "logits/generated": -1.5678973197937012, + "logits/real": -1.5635255575180054, + "logps/generated": -629.82421875, + "logps/real": -408.1054992675781, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -23.272071838378906, + "rewards/margins": 19.15908432006836, + "rewards/real": -4.112987995147705, + "step": 3310 + }, + { + "epoch": 2.12, + "learning_rate": 1.6371576959395657e-07, + "logits/generated": -1.572665810585022, + "logits/real": -1.5948829650878906, + "logps/generated": -575.0339965820312, + "logps/real": -335.0082092285156, + "loss": 0.0004, + "rewards/accuracies": 1.0, + "rewards/generated": -22.599546432495117, + "rewards/margins": 19.142099380493164, + "rewards/real": -3.457448959350586, + "step": 3320 + }, + { + "epoch": 2.12, + "learning_rate": 1.6253541076487252e-07, + "logits/generated": -1.4921143054962158, + "logits/real": -1.5882599353790283, + "logps/generated": -622.8916625976562, + "logps/real": -317.2669372558594, + "loss": 0.0025, + "rewards/accuracies": 1.0, + "rewards/generated": -22.83989143371582, + "rewards/margins": 18.200075149536133, + "rewards/real": -4.639816761016846, + "step": 3330 + }, + { + "epoch": 2.13, + "learning_rate": 1.6135505193578848e-07, + "logits/generated": -1.5064975023269653, + "logits/real": -1.5329867601394653, + "logps/generated": -616.3176879882812, + "logps/real": -362.6773681640625, + "loss": 0.0012, + "rewards/accuracies": 1.0, + "rewards/generated": -24.10775375366211, + "rewards/margins": 19.66227149963379, + "rewards/real": -4.445483207702637, + "step": 3340 + }, + { + "epoch": 2.14, + "learning_rate": 1.6017469310670443e-07, + "logits/generated": -1.4593093395233154, + "logits/real": -1.536402940750122, + "logps/generated": -615.1842041015625, + "logps/real": -401.3619689941406, + "loss": 0.0007, + "rewards/accuracies": 1.0, + "rewards/generated": -23.014205932617188, + "rewards/margins": 17.37249755859375, + "rewards/real": -5.641705513000488, + "step": 3350 + }, + { + "epoch": 2.14, + "learning_rate": 1.589943342776204e-07, + "logits/generated": -1.3944981098175049, + "logits/real": -1.4975627660751343, + "logps/generated": -619.4702758789062, + "logps/real": -325.2210693359375, + "loss": 0.0032, + "rewards/accuracies": 1.0, + "rewards/generated": -25.038524627685547, + "rewards/margins": 19.82014274597168, + "rewards/real": -5.218380928039551, + "step": 3360 + }, + { + "epoch": 2.15, + "learning_rate": 1.5781397544853635e-07, + "logits/generated": -1.4509702920913696, + "logits/real": -1.4926021099090576, + "logps/generated": -636.9576416015625, + "logps/real": -405.04443359375, + "loss": 0.0006, + "rewards/accuracies": 1.0, + "rewards/generated": -24.981292724609375, + "rewards/margins": 20.111270904541016, + "rewards/real": -4.870023250579834, + "step": 3370 + }, + { + "epoch": 2.15, + "learning_rate": 1.5663361661945233e-07, + "logits/generated": -1.463219165802002, + "logits/real": -1.5502787828445435, + "logps/generated": -580.4637451171875, + "logps/real": -379.69622802734375, + "loss": 0.0046, + "rewards/accuracies": 0.9750000238418579, + "rewards/generated": -22.16978645324707, + "rewards/margins": 16.70908546447754, + "rewards/real": -5.460699558258057, + "step": 3380 + }, + { + "epoch": 2.16, + "learning_rate": 1.5545325779036828e-07, + "logits/generated": -1.48343825340271, + "logits/real": -1.5905817747116089, + "logps/generated": -616.1544189453125, + "logps/real": -369.7923583984375, + "loss": 0.01, + "rewards/accuracies": 1.0, + "rewards/generated": -24.254405975341797, + "rewards/margins": 19.30912971496582, + "rewards/real": -4.945278644561768, + "step": 3390 + }, + { + "epoch": 2.17, + "learning_rate": 1.5427289896128424e-07, + "logits/generated": -1.6332385540008545, + "logits/real": -1.6335346698760986, + "logps/generated": -636.2514038085938, + "logps/real": -393.36273193359375, + "loss": 0.0007, + "rewards/accuracies": 1.0, + "rewards/generated": -21.752424240112305, + "rewards/margins": 18.011730194091797, + "rewards/real": -3.740694761276245, + "step": 3400 + }, + { + "epoch": 2.17, + "learning_rate": 1.530925401322002e-07, + "logits/generated": -1.5605157613754272, + "logits/real": -1.5932276248931885, + "logps/generated": -629.7593383789062, + "logps/real": -403.24603271484375, + "loss": 0.0008, + "rewards/accuracies": 1.0, + "rewards/generated": -22.148426055908203, + "rewards/margins": 18.18630599975586, + "rewards/real": -3.9621195793151855, + "step": 3410 + }, + { + "epoch": 2.18, + "learning_rate": 1.5191218130311615e-07, + "logits/generated": -1.6432632207870483, + "logits/real": -1.7228806018829346, + "logps/generated": -613.7503662109375, + "logps/real": -395.9375, + "loss": 0.0033, + "rewards/accuracies": 1.0, + "rewards/generated": -22.290279388427734, + "rewards/margins": 18.721607208251953, + "rewards/real": -3.5686745643615723, + "step": 3420 + }, + { + "epoch": 2.19, + "learning_rate": 1.507318224740321e-07, + "logits/generated": -1.6993591785430908, + "logits/real": -1.7430530786514282, + "logps/generated": -580.5427856445312, + "logps/real": -355.5113220214844, + "loss": 0.0004, + "rewards/accuracies": 1.0, + "rewards/generated": -21.60049819946289, + "rewards/margins": 16.75869369506836, + "rewards/real": -4.841803550720215, + "step": 3430 + }, + { + "epoch": 2.19, + "learning_rate": 1.4955146364494806e-07, + "logits/generated": -1.6865381002426147, + "logits/real": -1.7329572439193726, + "logps/generated": -640.44189453125, + "logps/real": -342.9864807128906, + "loss": 0.0027, + "rewards/accuracies": 1.0, + "rewards/generated": -24.40448760986328, + "rewards/margins": 19.22942352294922, + "rewards/real": -5.1750640869140625, + "step": 3440 + }, + { + "epoch": 2.2, + "learning_rate": 1.4837110481586402e-07, + "logits/generated": -1.6941635608673096, + "logits/real": -1.7331607341766357, + "logps/generated": -718.8358154296875, + "logps/real": -405.1249084472656, + "loss": 0.0012, + "rewards/accuracies": 1.0, + "rewards/generated": -25.554664611816406, + "rewards/margins": 21.22394561767578, + "rewards/real": -4.330716133117676, + "step": 3450 + }, + { + "epoch": 2.21, + "learning_rate": 1.4719074598677998e-07, + "logits/generated": -1.550095558166504, + "logits/real": -1.715970754623413, + "logps/generated": -643.9942626953125, + "logps/real": -379.50189208984375, + "loss": 0.0016, + "rewards/accuracies": 1.0, + "rewards/generated": -22.543724060058594, + "rewards/margins": 18.525737762451172, + "rewards/real": -4.017988204956055, + "step": 3460 + }, + { + "epoch": 2.21, + "learning_rate": 1.4601038715769593e-07, + "logits/generated": -1.5843255519866943, + "logits/real": -1.574999213218689, + "logps/generated": -696.6734619140625, + "logps/real": -435.64349365234375, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -25.31515884399414, + "rewards/margins": 19.46282386779785, + "rewards/real": -5.8523335456848145, + "step": 3470 + }, + { + "epoch": 2.22, + "learning_rate": 1.448300283286119e-07, + "logits/generated": -1.6632843017578125, + "logits/real": -1.7208706140518188, + "logps/generated": -604.4093017578125, + "logps/real": -381.2565002441406, + "loss": 0.0015, + "rewards/accuracies": 1.0, + "rewards/generated": -22.31944465637207, + "rewards/margins": 17.110166549682617, + "rewards/real": -5.209275245666504, + "step": 3480 + }, + { + "epoch": 2.22, + "learning_rate": 1.4364966949952784e-07, + "logits/generated": -1.5533571243286133, + "logits/real": -1.669923186302185, + "logps/generated": -590.7406005859375, + "logps/real": -359.255126953125, + "loss": 0.0057, + "rewards/accuracies": 1.0, + "rewards/generated": -24.880474090576172, + "rewards/margins": 19.988914489746094, + "rewards/real": -4.891559600830078, + "step": 3490 + }, + { + "epoch": 2.23, + "learning_rate": 1.4246931067044383e-07, + "logits/generated": -1.6246362924575806, + "logits/real": -1.7729676961898804, + "logps/generated": -617.2137451171875, + "logps/real": -352.3486328125, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -23.405954360961914, + "rewards/margins": 19.063854217529297, + "rewards/real": -4.342098712921143, + "step": 3500 + }, + { + "epoch": 2.24, + "learning_rate": 1.4128895184135978e-07, + "logits/generated": -1.6617110967636108, + "logits/real": -1.6978448629379272, + "logps/generated": -621.179931640625, + "logps/real": -390.19232177734375, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -24.1503849029541, + "rewards/margins": 19.014019012451172, + "rewards/real": -5.136368751525879, + "step": 3510 + }, + { + "epoch": 2.24, + "learning_rate": 1.4010859301227574e-07, + "logits/generated": -1.605457067489624, + "logits/real": -1.6876481771469116, + "logps/generated": -614.6359252929688, + "logps/real": -370.97711181640625, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -24.024070739746094, + "rewards/margins": 19.143352508544922, + "rewards/real": -4.880719184875488, + "step": 3520 + }, + { + "epoch": 2.25, + "learning_rate": 1.389282341831917e-07, + "logits/generated": -1.6278917789459229, + "logits/real": -1.6513782739639282, + "logps/generated": -624.6473999023438, + "logps/real": -374.70361328125, + "loss": 0.0048, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -24.308168411254883, + "rewards/margins": 19.365467071533203, + "rewards/real": -4.942699909210205, + "step": 3530 + }, + { + "epoch": 2.26, + "learning_rate": 1.3774787535410765e-07, + "logits/generated": -1.6541423797607422, + "logits/real": -1.7135674953460693, + "logps/generated": -666.3045654296875, + "logps/real": -358.337158203125, + "loss": 0.0013, + "rewards/accuracies": 1.0, + "rewards/generated": -26.521636962890625, + "rewards/margins": 21.59090805053711, + "rewards/real": -4.930731773376465, + "step": 3540 + }, + { + "epoch": 2.26, + "learning_rate": 1.365675165250236e-07, + "logits/generated": -1.6046949625015259, + "logits/real": -1.7014013528823853, + "logps/generated": -548.9305419921875, + "logps/real": -328.26922607421875, + "loss": 0.0007, + "rewards/accuracies": 1.0, + "rewards/generated": -21.43295669555664, + "rewards/margins": 16.715219497680664, + "rewards/real": -4.717737674713135, + "step": 3550 + }, + { + "epoch": 2.27, + "learning_rate": 1.3538715769593956e-07, + "logits/generated": -1.6616226434707642, + "logits/real": -1.7068140506744385, + "logps/generated": -678.4255981445312, + "logps/real": -416.6632385253906, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -26.132476806640625, + "rewards/margins": 20.578107833862305, + "rewards/real": -5.554368019104004, + "step": 3560 + }, + { + "epoch": 2.28, + "learning_rate": 1.3420679886685552e-07, + "logits/generated": -1.5858196020126343, + "logits/real": -1.6896979808807373, + "logps/generated": -626.3059692382812, + "logps/real": -338.3504943847656, + "loss": 0.0024, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -26.809459686279297, + "rewards/margins": 21.306217193603516, + "rewards/real": -5.503243446350098, + "step": 3570 + }, + { + "epoch": 2.28, + "learning_rate": 1.3302644003777147e-07, + "logits/generated": -1.5780128240585327, + "logits/real": -1.718907356262207, + "logps/generated": -679.6033325195312, + "logps/real": -400.218994140625, + "loss": 0.0023, + "rewards/accuracies": 1.0, + "rewards/generated": -26.130970001220703, + "rewards/margins": 20.593902587890625, + "rewards/real": -5.537067413330078, + "step": 3580 + }, + { + "epoch": 2.29, + "learning_rate": 1.3184608120868743e-07, + "logits/generated": -1.4745320081710815, + "logits/real": -1.6113662719726562, + "logps/generated": -617.8749389648438, + "logps/real": -392.05157470703125, + "loss": 0.0007, + "rewards/accuracies": 1.0, + "rewards/generated": -27.33835792541504, + "rewards/margins": 22.41128921508789, + "rewards/real": -4.927067279815674, + "step": 3590 + }, + { + "epoch": 2.29, + "learning_rate": 1.3066572237960338e-07, + "logits/generated": -1.5372802019119263, + "logits/real": -1.6040500402450562, + "logps/generated": -633.8583984375, + "logps/real": -359.6305236816406, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -25.859004974365234, + "rewards/margins": 21.345258712768555, + "rewards/real": -4.513743877410889, + "step": 3600 + }, + { + "epoch": 2.3, + "learning_rate": 1.2948536355051934e-07, + "logits/generated": -1.516701102256775, + "logits/real": -1.6452713012695312, + "logps/generated": -673.6900634765625, + "logps/real": -413.9730529785156, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -26.489360809326172, + "rewards/margins": 20.552282333374023, + "rewards/real": -5.937077522277832, + "step": 3610 + }, + { + "epoch": 2.31, + "learning_rate": 1.2830500472143532e-07, + "logits/generated": -1.5738109350204468, + "logits/real": -1.7013378143310547, + "logps/generated": -649.1641845703125, + "logps/real": -382.46624755859375, + "loss": 0.0038, + "rewards/accuracies": 1.0, + "rewards/generated": -24.782451629638672, + "rewards/margins": 19.909442901611328, + "rewards/real": -4.87300968170166, + "step": 3620 + }, + { + "epoch": 2.31, + "learning_rate": 1.2712464589235128e-07, + "logits/generated": -1.5686719417572021, + "logits/real": -1.6392751932144165, + "logps/generated": -659.6578979492188, + "logps/real": -429.4422302246094, + "loss": 0.0009, + "rewards/accuracies": 1.0, + "rewards/generated": -25.061580657958984, + "rewards/margins": 20.066463470458984, + "rewards/real": -4.995121002197266, + "step": 3630 + }, + { + "epoch": 2.32, + "learning_rate": 1.2594428706326723e-07, + "logits/generated": -1.544968605041504, + "logits/real": -1.6382554769515991, + "logps/generated": -632.112060546875, + "logps/real": -406.7085876464844, + "loss": 0.0023, + "rewards/accuracies": 1.0, + "rewards/generated": -25.972497940063477, + "rewards/margins": 20.0604248046875, + "rewards/real": -5.912076473236084, + "step": 3640 + }, + { + "epoch": 2.33, + "learning_rate": 1.247639282341832e-07, + "logits/generated": -1.6376492977142334, + "logits/real": -1.6503185033798218, + "logps/generated": -645.3873291015625, + "logps/real": -362.284912109375, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -26.817455291748047, + "rewards/margins": 20.737594604492188, + "rewards/real": -6.079861164093018, + "step": 3650 + }, + { + "epoch": 2.33, + "learning_rate": 1.2358356940509915e-07, + "logits/generated": -1.3051881790161133, + "logits/real": -1.5069706439971924, + "logps/generated": -676.6127319335938, + "logps/real": -395.44873046875, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -27.064083099365234, + "rewards/margins": 20.38376808166504, + "rewards/real": -6.6803178787231445, + "step": 3660 + }, + { + "epoch": 2.34, + "learning_rate": 1.224032105760151e-07, + "logits/generated": -1.4905213117599487, + "logits/real": -1.5628612041473389, + "logps/generated": -657.2412719726562, + "logps/real": -405.6756591796875, + "loss": 0.0071, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -26.98478126525879, + "rewards/margins": 21.09181785583496, + "rewards/real": -5.89295768737793, + "step": 3670 + }, + { + "epoch": 2.35, + "learning_rate": 1.2122285174693106e-07, + "logits/generated": -1.4171074628829956, + "logits/real": -1.4756571054458618, + "logps/generated": -624.70947265625, + "logps/real": -317.23272705078125, + "loss": 0.0016, + "rewards/accuracies": 1.0, + "rewards/generated": -28.16691017150879, + "rewards/margins": 22.441673278808594, + "rewards/real": -5.72523832321167, + "step": 3680 + }, + { + "epoch": 2.35, + "learning_rate": 1.2004249291784701e-07, + "logits/generated": -1.458076000213623, + "logits/real": -1.5287890434265137, + "logps/generated": -647.94873046875, + "logps/real": -419.01220703125, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -27.16118812561035, + "rewards/margins": 21.156476974487305, + "rewards/real": -6.004707336425781, + "step": 3690 + }, + { + "epoch": 2.36, + "learning_rate": 1.1886213408876297e-07, + "logits/generated": -1.4430928230285645, + "logits/real": -1.510324478149414, + "logps/generated": -650.9915161132812, + "logps/real": -400.11614990234375, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -27.097997665405273, + "rewards/margins": 21.253087997436523, + "rewards/real": -5.844912528991699, + "step": 3700 + }, + { + "epoch": 2.36, + "learning_rate": 1.1768177525967894e-07, + "logits/generated": -1.4635891914367676, + "logits/real": -1.5101027488708496, + "logps/generated": -603.7237548828125, + "logps/real": -335.04071044921875, + "loss": 0.0045, + "rewards/accuracies": 1.0, + "rewards/generated": -27.055862426757812, + "rewards/margins": 21.230571746826172, + "rewards/real": -5.825291633605957, + "step": 3710 + }, + { + "epoch": 2.37, + "learning_rate": 1.165014164305949e-07, + "logits/generated": -1.4326387643814087, + "logits/real": -1.3925200700759888, + "logps/generated": -700.2026977539062, + "logps/real": -428.26824951171875, + "loss": 0.0008, + "rewards/accuracies": 1.0, + "rewards/generated": -26.33407211303711, + "rewards/margins": 20.085529327392578, + "rewards/real": -6.24854040145874, + "step": 3720 + }, + { + "epoch": 2.38, + "learning_rate": 1.1532105760151085e-07, + "logits/generated": -1.3747960329055786, + "logits/real": -1.3340098857879639, + "logps/generated": -666.5633544921875, + "logps/real": -413.79241943359375, + "loss": 0.0007, + "rewards/accuracies": 1.0, + "rewards/generated": -29.475093841552734, + "rewards/margins": 22.57372283935547, + "rewards/real": -6.901371002197266, + "step": 3730 + }, + { + "epoch": 2.38, + "learning_rate": 1.1414069877242681e-07, + "logits/generated": -1.3411725759506226, + "logits/real": -1.359246850013733, + "logps/generated": -705.9603881835938, + "logps/real": -446.3271484375, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -29.915241241455078, + "rewards/margins": 22.39255714416504, + "rewards/real": -7.522684574127197, + "step": 3740 + }, + { + "epoch": 2.39, + "learning_rate": 1.1296033994334276e-07, + "logits/generated": -1.43299400806427, + "logits/real": -1.4390532970428467, + "logps/generated": -666.9552001953125, + "logps/real": -449.28692626953125, + "loss": 0.0004, + "rewards/accuracies": 1.0, + "rewards/generated": -26.734521865844727, + "rewards/margins": 20.729856491088867, + "rewards/real": -6.004665851593018, + "step": 3750 + }, + { + "epoch": 2.4, + "learning_rate": 1.1177998111425872e-07, + "logits/generated": -1.373038649559021, + "logits/real": -1.4165340662002563, + "logps/generated": -697.279541015625, + "logps/real": -429.6863708496094, + "loss": 0.0004, + "rewards/accuracies": 1.0, + "rewards/generated": -26.920108795166016, + "rewards/margins": 20.705747604370117, + "rewards/real": -6.214361190795898, + "step": 3760 + }, + { + "epoch": 2.4, + "learning_rate": 1.1059962228517469e-07, + "logits/generated": -1.3377251625061035, + "logits/real": -1.4376966953277588, + "logps/generated": -732.7567138671875, + "logps/real": -394.92376708984375, + "loss": 0.0001, + "rewards/accuracies": 1.0, + "rewards/generated": -30.918285369873047, + "rewards/margins": 23.17978286743164, + "rewards/real": -7.73850154876709, + "step": 3770 + }, + { + "epoch": 2.41, + "learning_rate": 1.0941926345609064e-07, + "logits/generated": -1.3250409364700317, + "logits/real": -1.3560101985931396, + "logps/generated": -723.0950317382812, + "logps/real": -436.11846923828125, + "loss": 0.0023, + "rewards/accuracies": 1.0, + "rewards/generated": -31.01926612854004, + "rewards/margins": 23.410831451416016, + "rewards/real": -7.608432769775391, + "step": 3780 + }, + { + "epoch": 2.42, + "learning_rate": 1.082389046270066e-07, + "logits/generated": -1.473495602607727, + "logits/real": -1.461737871170044, + "logps/generated": -634.8446655273438, + "logps/real": -423.19110107421875, + "loss": 0.0007, + "rewards/accuracies": 1.0, + "rewards/generated": -27.666391372680664, + "rewards/margins": 20.940290451049805, + "rewards/real": -6.726095676422119, + "step": 3790 + }, + { + "epoch": 2.42, + "learning_rate": 1.0705854579792256e-07, + "logits/generated": -1.382064700126648, + "logits/real": -1.5347161293029785, + "logps/generated": -639.7772216796875, + "logps/real": -394.82501220703125, + "loss": 0.0027, + "rewards/accuracies": 1.0, + "rewards/generated": -27.145679473876953, + "rewards/margins": 20.91800308227539, + "rewards/real": -6.227672576904297, + "step": 3800 + }, + { + "epoch": 2.43, + "learning_rate": 1.0587818696883851e-07, + "logits/generated": -1.467881441116333, + "logits/real": -1.5597871541976929, + "logps/generated": -654.2734985351562, + "logps/real": -431.6123046875, + "loss": 0.0026, + "rewards/accuracies": 1.0, + "rewards/generated": -28.531848907470703, + "rewards/margins": 20.634830474853516, + "rewards/real": -7.897017478942871, + "step": 3810 + }, + { + "epoch": 2.43, + "learning_rate": 1.0469782813975447e-07, + "logits/generated": -1.5202605724334717, + "logits/real": -1.516615629196167, + "logps/generated": -619.8519287109375, + "logps/real": -372.96759033203125, + "loss": 0.0011, + "rewards/accuracies": 1.0, + "rewards/generated": -27.353952407836914, + "rewards/margins": 20.2352294921875, + "rewards/real": -7.118723392486572, + "step": 3820 + }, + { + "epoch": 2.44, + "learning_rate": 1.0351746931067045e-07, + "logits/generated": -1.5924253463745117, + "logits/real": -1.5501093864440918, + "logps/generated": -634.8673095703125, + "logps/real": -399.2364501953125, + "loss": 0.0403, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -24.814449310302734, + "rewards/margins": 19.313993453979492, + "rewards/real": -5.500450134277344, + "step": 3830 + }, + { + "epoch": 2.45, + "learning_rate": 1.023371104815864e-07, + "logits/generated": -1.6328630447387695, + "logits/real": -1.6498903036117554, + "logps/generated": -654.9214477539062, + "logps/real": -399.1449279785156, + "loss": 0.0008, + "rewards/accuracies": 1.0, + "rewards/generated": -22.60084342956543, + "rewards/margins": 19.071792602539062, + "rewards/real": -3.529052257537842, + "step": 3840 + }, + { + "epoch": 2.45, + "learning_rate": 1.0115675165250236e-07, + "logits/generated": -1.6831458806991577, + "logits/real": -1.6596988439559937, + "logps/generated": -626.7855224609375, + "logps/real": -393.76715087890625, + "loss": 0.0026, + "rewards/accuracies": 1.0, + "rewards/generated": -21.234188079833984, + "rewards/margins": 18.330078125, + "rewards/real": -2.904111385345459, + "step": 3850 + }, + { + "epoch": 2.46, + "learning_rate": 9.997639282341832e-08, + "logits/generated": -1.555281400680542, + "logits/real": -1.5789473056793213, + "logps/generated": -580.3716430664062, + "logps/real": -386.8173828125, + "loss": 0.0047, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -21.891408920288086, + "rewards/margins": 17.281095504760742, + "rewards/real": -4.610313892364502, + "step": 3860 + }, + { + "epoch": 2.47, + "learning_rate": 9.879603399433427e-08, + "logits/generated": -1.6273943185806274, + "logits/real": -1.6304123401641846, + "logps/generated": -605.4176025390625, + "logps/real": -396.30804443359375, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -23.583742141723633, + "rewards/margins": 19.86680030822754, + "rewards/real": -3.7169413566589355, + "step": 3870 + }, + { + "epoch": 2.47, + "learning_rate": 9.761567516525023e-08, + "logits/generated": -1.6533687114715576, + "logits/real": -1.6083656549453735, + "logps/generated": -624.6300048828125, + "logps/real": -387.25042724609375, + "loss": 0.001, + "rewards/accuracies": 1.0, + "rewards/generated": -21.59857940673828, + "rewards/margins": 18.309185028076172, + "rewards/real": -3.289391040802002, + "step": 3880 + }, + { + "epoch": 2.48, + "learning_rate": 9.64353163361662e-08, + "logits/generated": -1.5607434511184692, + "logits/real": -1.6188228130340576, + "logps/generated": -551.50048828125, + "logps/real": -353.45330810546875, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -21.521461486816406, + "rewards/margins": 17.681446075439453, + "rewards/real": -3.8400161266326904, + "step": 3890 + }, + { + "epoch": 2.49, + "learning_rate": 9.525495750708215e-08, + "logits/generated": -1.5396366119384766, + "logits/real": -1.5993788242340088, + "logps/generated": -589.0250244140625, + "logps/real": -373.02056884765625, + "loss": 0.0008, + "rewards/accuracies": 1.0, + "rewards/generated": -23.210857391357422, + "rewards/margins": 19.39897346496582, + "rewards/real": -3.8118834495544434, + "step": 3900 + }, + { + "epoch": 2.49, + "learning_rate": 9.407459867799811e-08, + "logits/generated": -1.627367615699768, + "logits/real": -1.6631829738616943, + "logps/generated": -690.0119018554688, + "logps/real": -412.15679931640625, + "loss": 0.0008, + "rewards/accuracies": 1.0, + "rewards/generated": -25.02149772644043, + "rewards/margins": 21.027681350708008, + "rewards/real": -3.99381685256958, + "step": 3910 + }, + { + "epoch": 2.5, + "learning_rate": 9.289423984891407e-08, + "logits/generated": -1.595618724822998, + "logits/real": -1.614874243736267, + "logps/generated": -612.2095947265625, + "logps/real": -379.32867431640625, + "loss": 0.0007, + "rewards/accuracies": 1.0, + "rewards/generated": -23.731725692749023, + "rewards/margins": 19.470792770385742, + "rewards/real": -4.260932922363281, + "step": 3920 + }, + { + "epoch": 2.5, + "learning_rate": 9.171388101983002e-08, + "logits/generated": -1.6799474954605103, + "logits/real": -1.6550372838974, + "logps/generated": -603.3154296875, + "logps/real": -359.45684814453125, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -22.279056549072266, + "rewards/margins": 19.095932006835938, + "rewards/real": -3.1831231117248535, + "step": 3930 + }, + { + "epoch": 2.51, + "learning_rate": 9.053352219074599e-08, + "logits/generated": -1.5561491250991821, + "logits/real": -1.617343544960022, + "logps/generated": -629.9618530273438, + "logps/real": -401.78436279296875, + "loss": 0.0008, + "rewards/accuracies": 1.0, + "rewards/generated": -21.963420867919922, + "rewards/margins": 17.68639373779297, + "rewards/real": -4.2770280838012695, + "step": 3940 + }, + { + "epoch": 2.52, + "learning_rate": 8.935316336166195e-08, + "logits/generated": -1.5067987442016602, + "logits/real": -1.601683259010315, + "logps/generated": -580.4235229492188, + "logps/real": -347.91461181640625, + "loss": 0.005, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -22.338470458984375, + "rewards/margins": 18.845163345336914, + "rewards/real": -3.4933090209960938, + "step": 3950 + }, + { + "epoch": 2.52, + "learning_rate": 8.81728045325779e-08, + "logits/generated": -1.582057237625122, + "logits/real": -1.6376173496246338, + "logps/generated": -605.1909790039062, + "logps/real": -352.7174987792969, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -21.965229034423828, + "rewards/margins": 18.8221378326416, + "rewards/real": -3.143094539642334, + "step": 3960 + }, + { + "epoch": 2.53, + "learning_rate": 8.699244570349386e-08, + "logits/generated": -1.510329008102417, + "logits/real": -1.5715265274047852, + "logps/generated": -610.292236328125, + "logps/real": -368.6882019042969, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -22.84933090209961, + "rewards/margins": 18.83030891418457, + "rewards/real": -4.019021987915039, + "step": 3970 + }, + { + "epoch": 2.54, + "learning_rate": 8.581208687440982e-08, + "logits/generated": -1.584003210067749, + "logits/real": -1.6206289529800415, + "logps/generated": -668.8029174804688, + "logps/real": -396.7430725097656, + "loss": 0.0025, + "rewards/accuracies": 1.0, + "rewards/generated": -23.319042205810547, + "rewards/margins": 19.196847915649414, + "rewards/real": -4.122194766998291, + "step": 3980 + }, + { + "epoch": 2.54, + "learning_rate": 8.463172804532577e-08, + "logits/generated": -1.6870428323745728, + "logits/real": -1.6492443084716797, + "logps/generated": -625.905029296875, + "logps/real": -402.93719482421875, + "loss": 0.0046, + "rewards/accuracies": 1.0, + "rewards/generated": -21.225175857543945, + "rewards/margins": 17.057008743286133, + "rewards/real": -4.1681647300720215, + "step": 3990 + }, + { + "epoch": 2.55, + "learning_rate": 8.345136921624174e-08, + "logits/generated": -1.5616772174835205, + "logits/real": -1.6335293054580688, + "logps/generated": -627.5584106445312, + "logps/real": -398.01043701171875, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -23.357421875, + "rewards/margins": 18.800556182861328, + "rewards/real": -4.556865692138672, + "step": 4000 + }, + { + "epoch": 2.56, + "learning_rate": 8.22710103871577e-08, + "logits/generated": -1.555558681488037, + "logits/real": -1.588215947151184, + "logps/generated": -573.5257568359375, + "logps/real": -370.97723388671875, + "loss": 0.0004, + "rewards/accuracies": 1.0, + "rewards/generated": -21.65182113647461, + "rewards/margins": 18.165281295776367, + "rewards/real": -3.486539363861084, + "step": 4010 + }, + { + "epoch": 2.56, + "learning_rate": 8.109065155807365e-08, + "logits/generated": -1.6217458248138428, + "logits/real": -1.666949987411499, + "logps/generated": -578.7002563476562, + "logps/real": -362.3473205566406, + "loss": 0.0026, + "rewards/accuracies": 1.0, + "rewards/generated": -23.200794219970703, + "rewards/margins": 19.177310943603516, + "rewards/real": -4.023482799530029, + "step": 4020 + }, + { + "epoch": 2.57, + "learning_rate": 7.991029272898961e-08, + "logits/generated": -1.5723766088485718, + "logits/real": -1.6782785654067993, + "logps/generated": -653.6397094726562, + "logps/real": -356.05908203125, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -24.347869873046875, + "rewards/margins": 19.576274871826172, + "rewards/real": -4.771597862243652, + "step": 4030 + }, + { + "epoch": 2.57, + "learning_rate": 7.872993389990556e-08, + "logits/generated": -1.412308931350708, + "logits/real": -1.623708724975586, + "logps/generated": -692.9271240234375, + "logps/real": -373.7893981933594, + "loss": 0.0028, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -25.27395248413086, + "rewards/margins": 21.84659767150879, + "rewards/real": -3.427356243133545, + "step": 4040 + }, + { + "epoch": 2.58, + "learning_rate": 7.754957507082152e-08, + "logits/generated": -1.5342824459075928, + "logits/real": -1.5639487504959106, + "logps/generated": -622.4161376953125, + "logps/real": -397.85369873046875, + "loss": 0.0014, + "rewards/accuracies": 1.0, + "rewards/generated": -22.821455001831055, + "rewards/margins": 18.687612533569336, + "rewards/real": -4.133843421936035, + "step": 4050 + }, + { + "epoch": 2.59, + "learning_rate": 7.636921624173749e-08, + "logits/generated": -1.5000661611557007, + "logits/real": -1.5538852214813232, + "logps/generated": -623.6600341796875, + "logps/real": -346.08697509765625, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -26.10746192932129, + "rewards/margins": 21.521560668945312, + "rewards/real": -4.585900783538818, + "step": 4060 + }, + { + "epoch": 2.59, + "learning_rate": 7.518885741265344e-08, + "logits/generated": -1.4869574308395386, + "logits/real": -1.5690340995788574, + "logps/generated": -683.4605712890625, + "logps/real": -465.76605224609375, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -23.629016876220703, + "rewards/margins": 20.043920516967773, + "rewards/real": -3.5850958824157715, + "step": 4070 + }, + { + "epoch": 2.6, + "learning_rate": 7.40084985835694e-08, + "logits/generated": -1.5702314376831055, + "logits/real": -1.6110095977783203, + "logps/generated": -599.2610473632812, + "logps/real": -338.958251953125, + "loss": 0.0004, + "rewards/accuracies": 1.0, + "rewards/generated": -23.275604248046875, + "rewards/margins": 19.017620086669922, + "rewards/real": -4.257984638214111, + "step": 4080 + }, + { + "epoch": 2.61, + "learning_rate": 7.282813975448536e-08, + "logits/generated": -1.6213353872299194, + "logits/real": -1.7156651020050049, + "logps/generated": -650.2491455078125, + "logps/real": -398.2097473144531, + "loss": 0.0007, + "rewards/accuracies": 1.0, + "rewards/generated": -23.314720153808594, + "rewards/margins": 18.578779220581055, + "rewards/real": -4.7359395027160645, + "step": 4090 + }, + { + "epoch": 2.61, + "learning_rate": 7.164778092540131e-08, + "logits/generated": -1.4870989322662354, + "logits/real": -1.5456396341323853, + "logps/generated": -615.18408203125, + "logps/real": -416.35107421875, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -23.070531845092773, + "rewards/margins": 18.07291030883789, + "rewards/real": -4.997621059417725, + "step": 4100 + }, + { + "epoch": 2.62, + "learning_rate": 7.046742209631727e-08, + "logits/generated": -1.5412216186523438, + "logits/real": -1.575585126876831, + "logps/generated": -670.0528564453125, + "logps/real": -402.521484375, + "loss": 0.0025, + "rewards/accuracies": 1.0, + "rewards/generated": -25.783315658569336, + "rewards/margins": 21.93747329711914, + "rewards/real": -3.84584379196167, + "step": 4110 + }, + { + "epoch": 2.63, + "learning_rate": 6.928706326723324e-08, + "logits/generated": -1.5150539875030518, + "logits/real": -1.4853129386901855, + "logps/generated": -647.4837646484375, + "logps/real": -402.5469970703125, + "loss": 0.0006, + "rewards/accuracies": 1.0, + "rewards/generated": -24.628934860229492, + "rewards/margins": 19.664955139160156, + "rewards/real": -4.963980197906494, + "step": 4120 + }, + { + "epoch": 2.63, + "learning_rate": 6.81067044381492e-08, + "logits/generated": -1.4701521396636963, + "logits/real": -1.5523631572723389, + "logps/generated": -656.4703369140625, + "logps/real": -443.06622314453125, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -24.65596580505371, + "rewards/margins": 20.46110725402832, + "rewards/real": -4.194857597351074, + "step": 4130 + }, + { + "epoch": 2.64, + "learning_rate": 6.692634560906515e-08, + "logits/generated": -1.4741718769073486, + "logits/real": -1.519181728363037, + "logps/generated": -587.7735595703125, + "logps/real": -379.3127746582031, + "loss": 0.0008, + "rewards/accuracies": 1.0, + "rewards/generated": -23.39375877380371, + "rewards/margins": 18.563289642333984, + "rewards/real": -4.830471992492676, + "step": 4140 + }, + { + "epoch": 2.64, + "learning_rate": 6.57459867799811e-08, + "logits/generated": -1.5772595405578613, + "logits/real": -1.6052191257476807, + "logps/generated": -680.3468017578125, + "logps/real": -413.01190185546875, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -26.16423988342285, + "rewards/margins": 21.689952850341797, + "rewards/real": -4.474282741546631, + "step": 4150 + }, + { + "epoch": 2.65, + "learning_rate": 6.456562795089706e-08, + "logits/generated": -1.4749724864959717, + "logits/real": -1.5260889530181885, + "logps/generated": -613.30712890625, + "logps/real": -381.1894836425781, + "loss": 0.0009, + "rewards/accuracies": 1.0, + "rewards/generated": -23.512374877929688, + "rewards/margins": 19.124862670898438, + "rewards/real": -4.387513637542725, + "step": 4160 + }, + { + "epoch": 2.66, + "learning_rate": 6.338526912181302e-08, + "logits/generated": -1.4718973636627197, + "logits/real": -1.5110886096954346, + "logps/generated": -670.8785400390625, + "logps/real": -423.2325134277344, + "loss": 0.0025, + "rewards/accuracies": 1.0, + "rewards/generated": -22.933095932006836, + "rewards/margins": 18.731773376464844, + "rewards/real": -4.201323509216309, + "step": 4170 + }, + { + "epoch": 2.66, + "learning_rate": 6.220491029272899e-08, + "logits/generated": -1.415259838104248, + "logits/real": -1.4739396572113037, + "logps/generated": -584.6718139648438, + "logps/real": -385.3739929199219, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -23.22675895690918, + "rewards/margins": 18.408578872680664, + "rewards/real": -4.818178176879883, + "step": 4180 + }, + { + "epoch": 2.67, + "learning_rate": 6.102455146364494e-08, + "logits/generated": -1.4643930196762085, + "logits/real": -1.5111385583877563, + "logps/generated": -668.3934936523438, + "logps/real": -370.7761535644531, + "loss": 0.0024, + "rewards/accuracies": 1.0, + "rewards/generated": -24.76473617553711, + "rewards/margins": 19.840856552124023, + "rewards/real": -4.923879146575928, + "step": 4190 + }, + { + "epoch": 2.68, + "learning_rate": 5.984419263456091e-08, + "logits/generated": -1.3747190237045288, + "logits/real": -1.498532772064209, + "logps/generated": -603.4751586914062, + "logps/real": -357.296142578125, + "loss": 0.0038, + "rewards/accuracies": 1.0, + "rewards/generated": -24.800933837890625, + "rewards/margins": 19.628721237182617, + "rewards/real": -5.172211170196533, + "step": 4200 + }, + { + "epoch": 2.68, + "learning_rate": 5.866383380547686e-08, + "logits/generated": -1.5236282348632812, + "logits/real": -1.5463355779647827, + "logps/generated": -656.5723266601562, + "logps/real": -391.4309997558594, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -24.792877197265625, + "rewards/margins": 20.05835723876953, + "rewards/real": -4.734522342681885, + "step": 4210 + }, + { + "epoch": 2.69, + "learning_rate": 5.7483474976392816e-08, + "logits/generated": -1.4797223806381226, + "logits/real": -1.6046321392059326, + "logps/generated": -641.0149536132812, + "logps/real": -397.97991943359375, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -25.294628143310547, + "rewards/margins": 22.011356353759766, + "rewards/real": -3.2832741737365723, + "step": 4220 + }, + { + "epoch": 2.7, + "learning_rate": 5.630311614730878e-08, + "logits/generated": -1.4964040517807007, + "logits/real": -1.5269901752471924, + "logps/generated": -593.4207763671875, + "logps/real": -348.24481201171875, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -24.442432403564453, + "rewards/margins": 19.568811416625977, + "rewards/real": -4.873621463775635, + "step": 4230 + }, + { + "epoch": 2.7, + "learning_rate": 5.5122757318224735e-08, + "logits/generated": -1.5052317380905151, + "logits/real": -1.5962156057357788, + "logps/generated": -593.2183837890625, + "logps/real": -393.0603332519531, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -24.73797035217285, + "rewards/margins": 18.78076171875, + "rewards/real": -5.9572062492370605, + "step": 4240 + }, + { + "epoch": 2.71, + "learning_rate": 5.394239848914069e-08, + "logits/generated": -1.5205810070037842, + "logits/real": -1.5325863361358643, + "logps/generated": -676.7809448242188, + "logps/real": -422.16741943359375, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -27.194992065429688, + "rewards/margins": 22.257848739624023, + "rewards/real": -4.937142848968506, + "step": 4250 + }, + { + "epoch": 2.72, + "learning_rate": 5.276203966005665e-08, + "logits/generated": -1.4951252937316895, + "logits/real": -1.6003971099853516, + "logps/generated": -677.518310546875, + "logps/real": -391.42681884765625, + "loss": 0.0024, + "rewards/accuracies": 1.0, + "rewards/generated": -26.136348724365234, + "rewards/margins": 20.942659378051758, + "rewards/real": -5.19368839263916, + "step": 4260 + }, + { + "epoch": 2.72, + "learning_rate": 5.1581680830972616e-08, + "logits/generated": -1.474792718887329, + "logits/real": -1.5504586696624756, + "logps/generated": -652.3521728515625, + "logps/real": -390.1879577636719, + "loss": 0.0026, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -25.78470230102539, + "rewards/margins": 20.17237091064453, + "rewards/real": -5.61233377456665, + "step": 4270 + }, + { + "epoch": 2.73, + "learning_rate": 5.040132200188857e-08, + "logits/generated": -1.4497756958007812, + "logits/real": -1.47810959815979, + "logps/generated": -628.8690795898438, + "logps/real": -378.2281188964844, + "loss": 0.0004, + "rewards/accuracies": 1.0, + "rewards/generated": -26.356597900390625, + "rewards/margins": 22.217060089111328, + "rewards/real": -4.139541149139404, + "step": 4280 + }, + { + "epoch": 2.73, + "learning_rate": 4.9220963172804534e-08, + "logits/generated": -1.5309412479400635, + "logits/real": -1.4622726440429688, + "logps/generated": -634.1995849609375, + "logps/real": -382.15045166015625, + "loss": 0.0001, + "rewards/accuracies": 1.0, + "rewards/generated": -25.850317001342773, + "rewards/margins": 20.804723739624023, + "rewards/real": -5.045591354370117, + "step": 4290 + }, + { + "epoch": 2.74, + "learning_rate": 4.804060434372049e-08, + "logits/generated": -1.4767143726348877, + "logits/real": -1.5489951372146606, + "logps/generated": -627.1965942382812, + "logps/real": -376.2416687011719, + "loss": 0.0023, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -25.66696548461914, + "rewards/margins": 20.39335060119629, + "rewards/real": -5.273613929748535, + "step": 4300 + }, + { + "epoch": 2.75, + "learning_rate": 4.6860245514636446e-08, + "logits/generated": -1.5038503408432007, + "logits/real": -1.5427448749542236, + "logps/generated": -619.7374267578125, + "logps/real": -401.5597229003906, + "loss": 0.0006, + "rewards/accuracies": 1.0, + "rewards/generated": -24.61331558227539, + "rewards/margins": 19.254152297973633, + "rewards/real": -5.359164714813232, + "step": 4310 + }, + { + "epoch": 2.75, + "learning_rate": 4.567988668555241e-08, + "logits/generated": -1.5452930927276611, + "logits/real": -1.548268437385559, + "logps/generated": -641.5293579101562, + "logps/real": -369.5228271484375, + "loss": 0.0004, + "rewards/accuracies": 1.0, + "rewards/generated": -25.234844207763672, + "rewards/margins": 20.726560592651367, + "rewards/real": -4.5082855224609375, + "step": 4320 + }, + { + "epoch": 2.76, + "learning_rate": 4.4499527856468364e-08, + "logits/generated": -1.4813705682754517, + "logits/real": -1.489142656326294, + "logps/generated": -660.6444091796875, + "logps/real": -415.00311279296875, + "loss": 0.0022, + "rewards/accuracies": 1.0, + "rewards/generated": -26.411270141601562, + "rewards/margins": 20.429658889770508, + "rewards/real": -5.981614112854004, + "step": 4330 + }, + { + "epoch": 2.77, + "learning_rate": 4.331916902738432e-08, + "logits/generated": -1.3553807735443115, + "logits/real": -1.4226939678192139, + "logps/generated": -643.2559814453125, + "logps/real": -412.3113708496094, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -25.227434158325195, + "rewards/margins": 20.927614212036133, + "rewards/real": -4.2998175621032715, + "step": 4340 + }, + { + "epoch": 2.77, + "learning_rate": 4.213881019830028e-08, + "logits/generated": -1.592546820640564, + "logits/real": -1.5294647216796875, + "logps/generated": -730.6166381835938, + "logps/real": -410.7853088378906, + "loss": 0.0001, + "rewards/accuracies": 1.0, + "rewards/generated": -30.744155883789062, + "rewards/margins": 24.723114013671875, + "rewards/real": -6.021040916442871, + "step": 4350 + }, + { + "epoch": 2.78, + "learning_rate": 4.095845136921624e-08, + "logits/generated": -1.4481804370880127, + "logits/real": -1.5656325817108154, + "logps/generated": -591.4080200195312, + "logps/real": -345.214599609375, + "loss": 0.0019, + "rewards/accuracies": 1.0, + "rewards/generated": -25.179399490356445, + "rewards/margins": 20.560068130493164, + "rewards/real": -4.619330406188965, + "step": 4360 + }, + { + "epoch": 2.79, + "learning_rate": 3.9778092540132195e-08, + "logits/generated": -1.4652435779571533, + "logits/real": -1.5738567113876343, + "logps/generated": -691.9963989257812, + "logps/real": -423.44525146484375, + "loss": 0.0001, + "rewards/accuracies": 1.0, + "rewards/generated": -25.55663299560547, + "rewards/margins": 20.305315017700195, + "rewards/real": -5.251320838928223, + "step": 4370 + }, + { + "epoch": 2.79, + "learning_rate": 3.859773371104816e-08, + "logits/generated": -1.568688988685608, + "logits/real": -1.6783746480941772, + "logps/generated": -658.6432495117188, + "logps/real": -403.0459899902344, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -24.652185440063477, + "rewards/margins": 19.244029998779297, + "rewards/real": -5.408158302307129, + "step": 4380 + }, + { + "epoch": 2.8, + "learning_rate": 3.741737488196411e-08, + "logits/generated": -1.4557690620422363, + "logits/real": -1.5428321361541748, + "logps/generated": -659.35791015625, + "logps/real": -394.1236267089844, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -24.819717407226562, + "rewards/margins": 20.5244083404541, + "rewards/real": -4.295307636260986, + "step": 4390 + }, + { + "epoch": 2.8, + "learning_rate": 3.623701605288007e-08, + "logits/generated": -1.4206634759902954, + "logits/real": -1.5938780307769775, + "logps/generated": -744.5172119140625, + "logps/real": -398.4449157714844, + "loss": 0.0008, + "rewards/accuracies": 1.0, + "rewards/generated": -27.102832794189453, + "rewards/margins": 21.66973114013672, + "rewards/real": -5.433098793029785, + "step": 4400 + }, + { + "epoch": 2.81, + "learning_rate": 3.505665722379603e-08, + "logits/generated": -1.4969748258590698, + "logits/real": -1.5142086744308472, + "logps/generated": -655.7028198242188, + "logps/real": -380.55670166015625, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -26.31386375427246, + "rewards/margins": 21.87497329711914, + "rewards/real": -4.4388885498046875, + "step": 4410 + }, + { + "epoch": 2.82, + "learning_rate": 3.387629839471199e-08, + "logits/generated": -1.5558000802993774, + "logits/real": -1.6792634725570679, + "logps/generated": -681.4622192382812, + "logps/real": -384.5983581542969, + "loss": 0.0025, + "rewards/accuracies": 1.0, + "rewards/generated": -28.737651824951172, + "rewards/margins": 23.08871078491211, + "rewards/real": -5.6489434242248535, + "step": 4420 + }, + { + "epoch": 2.82, + "learning_rate": 3.2695939565627943e-08, + "logits/generated": -1.561899185180664, + "logits/real": -1.5363354682922363, + "logps/generated": -615.0111083984375, + "logps/real": -354.4222106933594, + "loss": 0.0036, + "rewards/accuracies": 1.0, + "rewards/generated": -26.08943748474121, + "rewards/margins": 20.943613052368164, + "rewards/real": -5.145823001861572, + "step": 4430 + }, + { + "epoch": 2.83, + "learning_rate": 3.1515580736543906e-08, + "logits/generated": -1.5919640064239502, + "logits/real": -1.5486515760421753, + "logps/generated": -611.601806640625, + "logps/real": -390.8034362792969, + "loss": 0.0019, + "rewards/accuracies": 1.0, + "rewards/generated": -21.80211639404297, + "rewards/margins": 16.953441619873047, + "rewards/real": -4.848675727844238, + "step": 4440 + }, + { + "epoch": 2.84, + "learning_rate": 3.033522190745987e-08, + "logits/generated": -1.593080759048462, + "logits/real": -1.5874629020690918, + "logps/generated": -685.1858520507812, + "logps/real": -476.7594299316406, + "loss": 0.0019, + "rewards/accuracies": 1.0, + "rewards/generated": -25.43196678161621, + "rewards/margins": 20.063030242919922, + "rewards/real": -5.368937015533447, + "step": 4450 + }, + { + "epoch": 2.84, + "learning_rate": 2.9154863078375824e-08, + "logits/generated": -1.4821428060531616, + "logits/real": -1.4978411197662354, + "logps/generated": -577.1400756835938, + "logps/real": -349.43096923828125, + "loss": 0.0011, + "rewards/accuracies": 1.0, + "rewards/generated": -24.762447357177734, + "rewards/margins": 19.5300350189209, + "rewards/real": -5.2324113845825195, + "step": 4460 + }, + { + "epoch": 2.85, + "learning_rate": 2.7974504249291784e-08, + "logits/generated": -1.5699541568756104, + "logits/real": -1.5277783870697021, + "logps/generated": -654.4290161132812, + "logps/real": -397.21343994140625, + "loss": 0.0004, + "rewards/accuracies": 1.0, + "rewards/generated": -25.094036102294922, + "rewards/margins": 21.21283721923828, + "rewards/real": -3.881199598312378, + "step": 4470 + }, + { + "epoch": 2.86, + "learning_rate": 2.679414542020774e-08, + "logits/generated": -1.6046009063720703, + "logits/real": -1.6073520183563232, + "logps/generated": -639.8587646484375, + "logps/real": -397.7523498535156, + "loss": 0.0001, + "rewards/accuracies": 1.0, + "rewards/generated": -25.259946823120117, + "rewards/margins": 20.08840560913086, + "rewards/real": -5.171541690826416, + "step": 4480 + }, + { + "epoch": 2.86, + "learning_rate": 2.5613786591123702e-08, + "logits/generated": -1.5873267650604248, + "logits/real": -1.6481456756591797, + "logps/generated": -605.360595703125, + "logps/real": -397.13616943359375, + "loss": 0.0024, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -24.51711082458496, + "rewards/margins": 19.769981384277344, + "rewards/real": -4.747128963470459, + "step": 4490 + }, + { + "epoch": 2.87, + "learning_rate": 2.443342776203966e-08, + "logits/generated": -1.5405027866363525, + "logits/real": -1.6537116765975952, + "logps/generated": -606.9188232421875, + "logps/real": -349.00897216796875, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -25.013578414916992, + "rewards/margins": 20.005685806274414, + "rewards/real": -5.007894992828369, + "step": 4500 + }, + { + "epoch": 2.87, + "learning_rate": 2.3253068932955617e-08, + "logits/generated": -1.5307514667510986, + "logits/real": -1.514164924621582, + "logps/generated": -650.7203369140625, + "logps/real": -403.4606018066406, + "loss": 0.0045, + "rewards/accuracies": 1.0, + "rewards/generated": -24.99856948852539, + "rewards/margins": 19.83810043334961, + "rewards/real": -5.160470008850098, + "step": 4510 + }, + { + "epoch": 2.88, + "learning_rate": 2.2072710103871576e-08, + "logits/generated": -1.578589677810669, + "logits/real": -1.5841494798660278, + "logps/generated": -567.2638549804688, + "logps/real": -315.01458740234375, + "loss": 0.0026, + "rewards/accuracies": 1.0, + "rewards/generated": -25.07675552368164, + "rewards/margins": 20.00092315673828, + "rewards/real": -5.075830936431885, + "step": 4520 + }, + { + "epoch": 2.89, + "learning_rate": 2.0892351274787535e-08, + "logits/generated": -1.454241394996643, + "logits/real": -1.5308386087417603, + "logps/generated": -691.4742431640625, + "logps/real": -369.84930419921875, + "loss": 0.0026, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -26.64725112915039, + "rewards/margins": 21.223133087158203, + "rewards/real": -5.4241156578063965, + "step": 4530 + }, + { + "epoch": 2.89, + "learning_rate": 1.971199244570349e-08, + "logits/generated": -1.5576417446136475, + "logits/real": -1.6084697246551514, + "logps/generated": -677.3316040039062, + "logps/real": -415.7643127441406, + "loss": 0.0003, + "rewards/accuracies": 1.0, + "rewards/generated": -25.05765724182129, + "rewards/margins": 19.37668228149414, + "rewards/real": -5.680974006652832, + "step": 4540 + }, + { + "epoch": 2.9, + "learning_rate": 1.853163361661945e-08, + "logits/generated": -1.583449125289917, + "logits/real": -1.6323738098144531, + "logps/generated": -624.1783447265625, + "logps/real": -387.78582763671875, + "loss": 0.0014, + "rewards/accuracies": 1.0, + "rewards/generated": -25.389423370361328, + "rewards/margins": 21.189250946044922, + "rewards/real": -4.200170993804932, + "step": 4550 + }, + { + "epoch": 2.91, + "learning_rate": 1.735127478753541e-08, + "logits/generated": -1.5298831462860107, + "logits/real": -1.4958879947662354, + "logps/generated": -633.8932495117188, + "logps/real": -334.78802490234375, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -23.955232620239258, + "rewards/margins": 20.1116886138916, + "rewards/real": -3.843545436859131, + "step": 4560 + }, + { + "epoch": 2.91, + "learning_rate": 1.6170915958451366e-08, + "logits/generated": -1.4841558933258057, + "logits/real": -1.6201757192611694, + "logps/generated": -677.6199951171875, + "logps/real": -358.0763244628906, + "loss": 0.0012, + "rewards/accuracies": 1.0, + "rewards/generated": -26.705102920532227, + "rewards/margins": 22.642436981201172, + "rewards/real": -4.062666416168213, + "step": 4570 + }, + { + "epoch": 2.92, + "learning_rate": 1.4990557129367328e-08, + "logits/generated": -1.5609204769134521, + "logits/real": -1.5243724584579468, + "logps/generated": -603.8170776367188, + "logps/real": -408.506103515625, + "loss": 0.0048, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -22.919692993164062, + "rewards/margins": 17.796039581298828, + "rewards/real": -5.123648643493652, + "step": 4580 + }, + { + "epoch": 2.93, + "learning_rate": 1.3810198300283284e-08, + "logits/generated": -1.441948652267456, + "logits/real": -1.487939715385437, + "logps/generated": -646.0272827148438, + "logps/real": -363.57220458984375, + "loss": 0.0002, + "rewards/accuracies": 1.0, + "rewards/generated": -25.057811737060547, + "rewards/margins": 20.18503189086914, + "rewards/real": -4.872781276702881, + "step": 4590 + }, + { + "epoch": 2.93, + "learning_rate": 1.2629839471199245e-08, + "logits/generated": -1.593481421470642, + "logits/real": -1.6024043560028076, + "logps/generated": -612.1333618164062, + "logps/real": -421.7381896972656, + "loss": 0.0001, + "rewards/accuracies": 1.0, + "rewards/generated": -25.70052719116211, + "rewards/margins": 20.35653305053711, + "rewards/real": -5.343997478485107, + "step": 4600 + }, + { + "epoch": 2.94, + "learning_rate": 1.1449480642115203e-08, + "logits/generated": -1.5912842750549316, + "logits/real": -1.7060985565185547, + "logps/generated": -702.9200439453125, + "logps/real": -399.92633056640625, + "loss": 0.0023, + "rewards/accuracies": 1.0, + "rewards/generated": -27.616954803466797, + "rewards/margins": 21.881147384643555, + "rewards/real": -5.735805988311768, + "step": 4610 + }, + { + "epoch": 2.94, + "learning_rate": 1.0269121813031162e-08, + "logits/generated": -1.5111281871795654, + "logits/real": -1.5352246761322021, + "logps/generated": -724.6343994140625, + "logps/real": -447.47088623046875, + "loss": 0.0007, + "rewards/accuracies": 1.0, + "rewards/generated": -26.329986572265625, + "rewards/margins": 22.345134735107422, + "rewards/real": -3.984851121902466, + "step": 4620 + }, + { + "epoch": 2.95, + "learning_rate": 9.08876298394712e-09, + "logits/generated": -1.5755963325500488, + "logits/real": -1.541659951210022, + "logps/generated": -633.173828125, + "logps/real": -371.69757080078125, + "loss": 0.0045, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -26.723129272460938, + "rewards/margins": 21.624584197998047, + "rewards/real": -5.098546981811523, + "step": 4630 + }, + { + "epoch": 2.96, + "learning_rate": 7.908404154863077e-09, + "logits/generated": -1.5423119068145752, + "logits/real": -1.6387542486190796, + "logps/generated": -665.977294921875, + "logps/real": -400.87957763671875, + "loss": 0.0024, + "rewards/accuracies": 0.987500011920929, + "rewards/generated": -25.161529541015625, + "rewards/margins": 20.474363327026367, + "rewards/real": -4.687165260314941, + "step": 4640 + }, + { + "epoch": 2.96, + "learning_rate": 6.728045325779036e-09, + "logits/generated": -1.5407065153121948, + "logits/real": -1.4502336978912354, + "logps/generated": -633.8583374023438, + "logps/real": -405.2549743652344, + "loss": 0.0007, + "rewards/accuracies": 1.0, + "rewards/generated": -24.977598190307617, + "rewards/margins": 19.61507797241211, + "rewards/real": -5.362523078918457, + "step": 4650 + }, + { + "epoch": 2.97, + "learning_rate": 5.547686496694995e-09, + "logits/generated": -1.383454442024231, + "logits/real": -1.4603092670440674, + "logps/generated": -629.0911254882812, + "logps/real": -358.1429748535156, + "loss": 0.001, + "rewards/accuracies": 1.0, + "rewards/generated": -27.214752197265625, + "rewards/margins": 22.154844284057617, + "rewards/real": -5.059906005859375, + "step": 4660 + }, + { + "epoch": 2.98, + "learning_rate": 4.367327667610953e-09, + "logits/generated": -1.5550544261932373, + "logits/real": -1.5835330486297607, + "logps/generated": -690.2367553710938, + "logps/real": -397.4266052246094, + "loss": 0.0001, + "rewards/accuracies": 1.0, + "rewards/generated": -27.849355697631836, + "rewards/margins": 22.347171783447266, + "rewards/real": -5.502187252044678, + "step": 4670 + }, + { + "epoch": 2.98, + "learning_rate": 3.186968838526912e-09, + "logits/generated": -1.512865424156189, + "logits/real": -1.5152509212493896, + "logps/generated": -613.0997314453125, + "logps/real": -387.0373840332031, + "loss": 0.0001, + "rewards/accuracies": 1.0, + "rewards/generated": -24.694095611572266, + "rewards/margins": 19.704021453857422, + "rewards/real": -4.9900712966918945, + "step": 4680 + }, + { + "epoch": 2.99, + "learning_rate": 2.0066100094428705e-09, + "logits/generated": -1.5903544425964355, + "logits/real": -1.532593011856079, + "logps/generated": -613.1495361328125, + "logps/real": -355.979248046875, + "loss": 0.0005, + "rewards/accuracies": 1.0, + "rewards/generated": -24.96600914001465, + "rewards/margins": 20.51987648010254, + "rewards/real": -4.446131706237793, + "step": 4690 + }, + { + "epoch": 3.0, + "learning_rate": 8.26251180358829e-10, + "logits/generated": -1.6342672109603882, + "logits/real": -1.5792245864868164, + "logps/generated": -651.708740234375, + "logps/real": -397.2961730957031, + "loss": 0.0023, + "rewards/accuracies": 1.0, + "rewards/generated": -26.37361717224121, + "rewards/margins": 21.51944923400879, + "rewards/real": -4.854170799255371, + "step": 4700 + }, + { + "epoch": 3.0, + "step": 4707, + "total_flos": 0.0, + "train_loss": 0.049856398493659054, + "train_runtime": 39539.5403, + "train_samples_per_second": 3.809, + "train_steps_per_second": 0.119 + } + ], + "logging_steps": 10, + "max_steps": 4707, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 500, + "total_flos": 0.0, + "train_batch_size": 8, + "trial_name": null, + "trial_params": null +}