diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,21 +1,21 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9996190476190476,
+  "epoch": 1.996400719856029,
   "eval_steps": 500,
-  "global_step": 1312,
+  "global_step": 832,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 3.787878787878788e-08,
-      "logits/chosen": 0.17224127054214478,
-      "logits/rejected": 0.18124699592590332,
-      "logps/chosen": -379.32623291015625,
-      "logps/rejected": -349.5926208496094,
-      "loss": 0.2902,
+      "learning_rate": 5.952380952380953e-08,
+      "logits/chosen": 0.11703574657440186,
+      "logits/rejected": 0.3661181330680847,
+      "logps/chosen": -218.64993286132812,
+      "logps/rejected": -191.34808349609375,
+      "loss": 0.3408,
       "rewards/accuracies": 0.0,
       "rewards/chosen": 0.0,
       "rewards/margins": 0.0,
@@ -23,1853 +23,1181 @@
       "step": 1
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.787878787878788e-07,
-      "logits/chosen": 0.07856506109237671,
-      "logits/rejected": 0.2510358691215515,
-      "logps/chosen": -334.9958190917969,
-      "logps/rejected": -283.54034423828125,
-      "loss": 0.3745,
-      "rewards/accuracies": 0.4375,
-      "rewards/chosen": 1.0113296411873307e-05,
-      "rewards/margins": 5.3244151786202565e-05,
-      "rewards/rejected": -4.313084718887694e-05,
+      "epoch": 0.02,
+      "learning_rate": 5.952380952380953e-07,
+      "logits/chosen": 0.10404814779758453,
+      "logits/rejected": 0.23778128623962402,
+      "logps/chosen": -401.4896240234375,
+      "logps/rejected": -345.9862976074219,
+      "loss": 0.3642,
+      "rewards/accuracies": 0.4791666567325592,
+      "rewards/chosen": 0.0004916194593533874,
+      "rewards/margins": 0.0005594216636382043,
+      "rewards/rejected": -6.780229159630835e-05,
       "step": 10
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.575757575757576e-07,
-      "logits/chosen": 0.07835443317890167,
-      "logits/rejected": 0.23690445721149445,
-      "logps/chosen": -343.5411682128906,
-      "logps/rejected": -300.49774169921875,
-      "loss": 0.345,
-      "rewards/accuracies": 0.518750011920929,
-      "rewards/chosen": 9.991687193178223e-07,
-      "rewards/margins": 5.578011041507125e-05,
-      "rewards/rejected": -5.478093953570351e-05,
+      "epoch": 0.05,
+      "learning_rate": 1.1904761904761906e-06,
+      "logits/chosen": 0.13218173384666443,
+      "logits/rejected": 0.20688870549201965,
+      "logps/chosen": -336.506591796875,
+      "logps/rejected": -319.3189392089844,
+      "loss": 0.3689,
+      "rewards/accuracies": 0.4124999940395355,
+      "rewards/chosen": 0.00020826223772019148,
+      "rewards/margins": -0.000311180017888546,
+      "rewards/rejected": 0.0005194421974010766,
       "step": 20
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 1.1363636363636364e-06,
-      "logits/chosen": 0.09362699836492538,
-      "logits/rejected": 0.24969105422496796,
-      "logps/chosen": -384.11199951171875,
-      "logps/rejected": -290.873779296875,
-      "loss": 0.3667,
-      "rewards/accuracies": 0.53125,
-      "rewards/chosen": 0.00015662855003029108,
-      "rewards/margins": 9.987165867642034e-06,
-      "rewards/rejected": 0.0001466413668822497,
+      "epoch": 0.07,
+      "learning_rate": 1.7857142857142859e-06,
+      "logits/chosen": 0.11459924280643463,
+      "logits/rejected": 0.1922653764486313,
+      "logps/chosen": -342.02569580078125,
+      "logps/rejected": -324.1275939941406,
+      "loss": 0.3786,
+      "rewards/accuracies": 0.46875,
+      "rewards/chosen": -0.0006439354037865996,
+      "rewards/margins": 0.0004738263669423759,
+      "rewards/rejected": -0.0011177618289366364,
       "step": 30
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.5151515151515152e-06,
-      "logits/chosen": 0.08275317400693893,
-      "logits/rejected": 0.20892572402954102,
-      "logps/chosen": -361.24462890625,
-      "logps/rejected": -296.53094482421875,
-      "loss": 0.3012,
-      "rewards/accuracies": 0.512499988079071,
-      "rewards/chosen": 0.00019404105842113495,
-      "rewards/margins": 0.000132537359604612,
-      "rewards/rejected": 6.150371336843818e-05,
+      "epoch": 0.1,
+      "learning_rate": 2.380952380952381e-06,
+      "logits/chosen": 0.13577614724636078,
+      "logits/rejected": 0.17847472429275513,
+      "logps/chosen": -298.6214294433594,
+      "logps/rejected": -289.40850830078125,
+      "loss": 0.3689,
+      "rewards/accuracies": 0.543749988079071,
+      "rewards/chosen": 0.0008146329782903194,
+      "rewards/margins": 0.0024004268925637007,
+      "rewards/rejected": -0.001585794030688703,
       "step": 40
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.8939393939393941e-06,
-      "logits/chosen": 0.14996236562728882,
-      "logits/rejected": 0.2148081511259079,
-      "logps/chosen": -339.32916259765625,
-      "logps/rejected": -292.900146484375,
-      "loss": 0.3391,
-      "rewards/accuracies": 0.518750011920929,
-      "rewards/chosen": 0.0002820778754539788,
-      "rewards/margins": 0.00035016084439121187,
-      "rewards/rejected": -6.80829762131907e-05,
+      "epoch": 0.12,
+      "learning_rate": 2.9761904761904763e-06,
+      "logits/chosen": 0.10261678695678711,
+      "logits/rejected": 0.20306341350078583,
+      "logps/chosen": -351.93572998046875,
+      "logps/rejected": -362.153564453125,
+      "loss": 0.3692,
+      "rewards/accuracies": 0.6000000238418579,
+      "rewards/chosen": 0.0009010158246383071,
+      "rewards/margins": 0.004100508522242308,
+      "rewards/rejected": -0.003199493046849966,
       "step": 50
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.2727272727272728e-06,
-      "logits/chosen": 0.09292325377464294,
-      "logits/rejected": 0.25342074036598206,
-      "logps/chosen": -357.79498291015625,
-      "logps/rejected": -279.56475830078125,
-      "loss": 0.3523,
-      "rewards/accuracies": 0.643750011920929,
-      "rewards/chosen": 0.0005591081571765244,
-      "rewards/margins": 0.0007476316532120109,
-      "rewards/rejected": -0.00018852358334697783,
+      "epoch": 0.14,
+      "learning_rate": 3.5714285714285718e-06,
+      "logits/chosen": 0.13770776987075806,
+      "logits/rejected": 0.2188442498445511,
+      "logps/chosen": -349.51690673828125,
+      "logps/rejected": -351.1549377441406,
+      "loss": 0.3655,
+      "rewards/accuracies": 0.606249988079071,
+      "rewards/chosen": 0.003258631331846118,
+      "rewards/margins": 0.007584023289382458,
+      "rewards/rejected": -0.004325392190366983,
       "step": 60
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 2.6515151515151514e-06,
-      "logits/chosen": 0.1520170122385025,
-      "logits/rejected": 0.23838527500629425,
-      "logps/chosen": -344.97381591796875,
-      "logps/rejected": -272.6285400390625,
-      "loss": 0.333,
-      "rewards/accuracies": 0.637499988079071,
-      "rewards/chosen": 0.000808493874501437,
-      "rewards/margins": 0.0009522804175503552,
-      "rewards/rejected": -0.0001437865139450878,
+      "epoch": 0.17,
+      "learning_rate": 4.166666666666667e-06,
+      "logits/chosen": 0.1271902620792389,
+      "logits/rejected": 0.23070549964904785,
+      "logps/chosen": -378.33843994140625,
+      "logps/rejected": -350.60662841796875,
+      "loss": 0.3586,
+      "rewards/accuracies": 0.6812499761581421,
+      "rewards/chosen": 0.006277147680521011,
+      "rewards/margins": 0.015207210555672646,
+      "rewards/rejected": -0.00893006194382906,
       "step": 70
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 3.0303030303030305e-06,
-      "logits/chosen": 0.10541319847106934,
-      "logits/rejected": 0.2519112229347229,
-      "logps/chosen": -337.99932861328125,
-      "logps/rejected": -285.3942565917969,
-      "loss": 0.3276,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": 0.0017709163948893547,
-      "rewards/margins": 0.0019623935222625732,
-      "rewards/rejected": -0.00019147712737321854,
+      "epoch": 0.19,
+      "learning_rate": 4.761904761904762e-06,
+      "logits/chosen": 0.08625562489032745,
+      "logits/rejected": 0.12316304445266724,
+      "logps/chosen": -307.9439697265625,
+      "logps/rejected": -335.3281555175781,
+      "loss": 0.3489,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/chosen": 0.013669237494468689,
+      "rewards/margins": 0.02226843498647213,
+      "rewards/rejected": -0.008599198423326015,
       "step": 80
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 3.409090909090909e-06,
-      "logits/chosen": 0.08055099099874496,
-      "logits/rejected": 0.23055055737495422,
-      "logps/chosen": -358.501220703125,
-      "logps/rejected": -301.07373046875,
-      "loss": 0.325,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": 0.0031805038452148438,
-      "rewards/margins": 0.0035764030180871487,
-      "rewards/rejected": -0.0003958995803259313,
+      "epoch": 0.22,
+      "learning_rate": 4.9992062457191005e-06,
+      "logits/chosen": 0.137899249792099,
+      "logits/rejected": 0.2165641039609909,
+      "logps/chosen": -355.6449890136719,
+      "logps/rejected": -338.1387634277344,
+      "loss": 0.3229,
+      "rewards/accuracies": 0.6812499761581421,
+      "rewards/chosen": 0.013719858601689339,
+      "rewards/margins": 0.042457275092601776,
+      "rewards/rejected": -0.028737416490912437,
       "step": 90
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 3.7878787878787882e-06,
-      "logits/chosen": 0.1381727159023285,
-      "logits/rejected": 0.36128586530685425,
-      "logps/chosen": -408.2177734375,
-      "logps/rejected": -295.70391845703125,
-      "loss": 0.3551,
-      "rewards/accuracies": 0.6875,
-      "rewards/chosen": 0.005324442870914936,
-      "rewards/margins": 0.006567128002643585,
-      "rewards/rejected": -0.00124268583022058,
+      "epoch": 0.24,
+      "learning_rate": 4.994357350311441e-06,
+      "logits/chosen": 0.14011432230472565,
+      "logits/rejected": 0.21795734763145447,
+      "logps/chosen": -360.2173156738281,
+      "logps/rejected": -358.1722717285156,
+      "loss": 0.3043,
+      "rewards/accuracies": 0.6187499761581421,
+      "rewards/chosen": -0.001885895850136876,
+      "rewards/margins": 0.06035756319761276,
+      "rewards/rejected": -0.06224345415830612,
       "step": 100
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 4.166666666666667e-06,
-      "logits/chosen": 0.11063258349895477,
-      "logits/rejected": 0.3397473692893982,
-      "logps/chosen": -365.5294494628906,
-      "logps/rejected": -296.68212890625,
-      "loss": 0.3176,
-      "rewards/accuracies": 0.7124999761581421,
-      "rewards/chosen": 0.0065561323426663876,
-      "rewards/margins": 0.009804257191717625,
-      "rewards/rejected": -0.0032481239177286625,
+      "epoch": 0.26,
+      "learning_rate": 4.98510907587894e-06,
+      "logits/chosen": 0.13077042996883392,
+      "logits/rejected": 0.21840377151966095,
+      "logps/chosen": -356.6605224609375,
+      "logps/rejected": -348.19476318359375,
+      "loss": 0.3169,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/chosen": -0.008259604685008526,
+      "rewards/margins": 0.08336080610752106,
+      "rewards/rejected": -0.09162042289972305,
       "step": 110
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 4.5454545454545455e-06,
-      "logits/chosen": 0.10760724544525146,
-      "logits/rejected": 0.3315739035606384,
-      "logps/chosen": -369.8674011230469,
-      "logps/rejected": -290.553955078125,
-      "loss": 0.2862,
-      "rewards/accuracies": 0.7562500238418579,
-      "rewards/chosen": 0.009247648529708385,
-      "rewards/margins": 0.017721759155392647,
-      "rewards/rejected": -0.008474110625684261,
+      "epoch": 0.29,
+      "learning_rate": 4.97147773390341e-06,
+      "logits/chosen": 0.14791826903820038,
+      "logits/rejected": 0.1786331683397293,
+      "logps/chosen": -320.29608154296875,
+      "logps/rejected": -337.16864013671875,
+      "loss": 0.2861,
+      "rewards/accuracies": 0.6499999761581421,
+      "rewards/chosen": 0.008191597647964954,
+      "rewards/margins": 0.09656454622745514,
+      "rewards/rejected": -0.08837294578552246,
       "step": 120
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 4.924242424242425e-06,
-      "logits/chosen": 0.19570864737033844,
-      "logits/rejected": 0.29995661973953247,
-      "logps/chosen": -357.7752685546875,
-      "logps/rejected": -298.80865478515625,
-      "loss": 0.3091,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": 0.0016540288925170898,
-      "rewards/margins": 0.030770784243941307,
-      "rewards/rejected": -0.029116755351424217,
+      "epoch": 0.31,
+      "learning_rate": 4.953487366425163e-06,
+      "logits/chosen": 0.12249626964330673,
+      "logits/rejected": 0.16907112300395966,
+      "logps/chosen": -342.0648498535156,
+      "logps/rejected": -363.51031494140625,
+      "loss": 0.3175,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/chosen": 0.0017940097022801638,
+      "rewards/margins": 0.07947574555873871,
+      "rewards/rejected": -0.07768173515796661,
       "step": 130
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.999432965739786e-06,
-      "logits/chosen": 0.18279646337032318,
-      "logits/rejected": 0.24193449318408966,
-      "logps/chosen": -326.20623779296875,
-      "logps/rejected": -340.66204833984375,
-      "loss": 0.3146,
-      "rewards/accuracies": 0.675000011920929,
-      "rewards/chosen": -0.02705274522304535,
-      "rewards/margins": 0.038378529250621796,
-      "rewards/rejected": -0.06543128192424774,
+      "epoch": 0.34,
+      "learning_rate": 4.931169703639282e-06,
+      "logits/chosen": 0.0919104740023613,
+      "logits/rejected": 0.18652714788913727,
+      "logps/chosen": -337.65374755859375,
+      "logps/rejected": -364.11199951171875,
+      "loss": 0.2828,
+      "rewards/accuracies": 0.7562500238418579,
+      "rewards/chosen": 0.03137553483247757,
+      "rewards/margins": 0.12489553540945053,
+      "rewards/rejected": -0.09352000057697296,
       "step": 140
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 4.997129829895409e-06,
-      "logits/chosen": 0.14213022589683533,
-      "logits/rejected": 0.2512076199054718,
-      "logps/chosen": -408.73541259765625,
-      "logps/rejected": -432.90771484375,
-      "loss": 0.2979,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": -0.0731552243232727,
-      "rewards/margins": 0.08651129901409149,
-      "rewards/rejected": -0.159666508436203,
+      "epoch": 0.36,
+      "learning_rate": 4.904564107932048e-06,
+      "logits/chosen": 0.13001379370689392,
+      "logits/rejected": 0.20237913727760315,
+      "logps/chosen": -351.857421875,
+      "logps/rejected": -336.6232604980469,
+      "loss": 0.2899,
+      "rewards/accuracies": 0.6812499761581421,
+      "rewards/chosen": -0.0018621661001816392,
+      "rewards/margins": 0.10416339337825775,
+      "rewards/rejected": -0.10602555423974991,
       "step": 150
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 4.9930567839810125e-06,
-      "logits/chosen": 0.17261534929275513,
-      "logits/rejected": 0.27892106771469116,
-      "logps/chosen": -483.6229553222656,
-      "logps/rejected": -527.606201171875,
-      "loss": 0.2727,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": -0.121568962931633,
-      "rewards/margins": 0.10641299188137054,
-      "rewards/rejected": -0.22798196971416473,
+      "epoch": 0.38,
+      "learning_rate": 4.873717504456219e-06,
+      "logits/chosen": 0.06932858377695084,
+      "logits/rejected": 0.15127311646938324,
+      "logps/chosen": -345.0473937988281,
+      "logps/rejected": -363.4601745605469,
+      "loss": 0.2889,
+      "rewards/accuracies": 0.65625,
+      "rewards/chosen": -0.01222093403339386,
+      "rewards/margins": 0.11404307186603546,
+      "rewards/rejected": -0.12626400589942932,
       "step": 160
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 4.987216714880929e-06,
-      "logits/chosen": 0.19240622222423553,
-      "logits/rejected": 0.22908082604408264,
-      "logps/chosen": -516.4906005859375,
-      "logps/rejected": -531.3421020507812,
-      "loss": 0.278,
-      "rewards/accuracies": 0.6812499761581421,
-      "rewards/chosen": -0.17584487795829773,
-      "rewards/margins": 0.10210248082876205,
-      "rewards/rejected": -0.27794739603996277,
+      "epoch": 0.41,
+      "learning_rate": 4.838684298367616e-06,
+      "logits/chosen": 0.16357803344726562,
+      "logits/rejected": 0.23174886405467987,
+      "logps/chosen": -357.15289306640625,
+      "logps/rejected": -358.61065673828125,
+      "loss": 0.2884,
+      "rewards/accuracies": 0.6625000238418579,
+      "rewards/chosen": 0.017432499676942825,
+      "rewards/margins": 0.11707814782857895,
+      "rewards/rejected": -0.09964564442634583,
       "step": 170
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.979613761906212e-06,
-      "logits/chosen": 0.12470928579568863,
-      "logits/rejected": 0.2596343755722046,
-      "logps/chosen": -551.5357666015625,
-      "logps/rejected": -654.6400756835938,
-      "loss": 0.2664,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": -0.21089038252830505,
-      "rewards/margins": 0.15304332971572876,
-      "rewards/rejected": -0.3639337420463562,
+      "epoch": 0.43,
+      "learning_rate": 4.7995262788689865e-06,
+      "logits/chosen": 0.16258656978607178,
+      "logits/rejected": 0.2536885738372803,
+      "logps/chosen": -337.7535705566406,
+      "logps/rejected": -346.13470458984375,
+      "loss": 0.2789,
+      "rewards/accuracies": 0.6937500238418579,
+      "rewards/chosen": 0.02853301540017128,
+      "rewards/margins": 0.1016291156411171,
+      "rewards/rejected": -0.07309609651565552,
       "step": 180
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 4.970253313860788e-06,
-      "logits/chosen": 0.1833106130361557,
-      "logits/rejected": 0.2698153257369995,
-      "logps/chosen": -562.19677734375,
-      "logps/rejected": -634.8347778320312,
-      "loss": 0.2832,
-      "rewards/accuracies": 0.65625,
-      "rewards/chosen": -0.2445872575044632,
-      "rewards/margins": 0.11998845636844635,
-      "rewards/rejected": -0.36457571387290955,
+      "epoch": 0.46,
+      "learning_rate": 4.756312510230377e-06,
+      "logits/chosen": 0.14243337512016296,
+      "logits/rejected": 0.24410876631736755,
+      "logps/chosen": -376.64599609375,
+      "logps/rejected": -363.4615478515625,
+      "loss": 0.2828,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": 0.03516390174627304,
+      "rewards/margins": 0.12421919405460358,
+      "rewards/rejected": -0.08905528485774994,
       "step": 190
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 4.959142005221991e-06,
-      "logits/chosen": 0.13080090284347534,
-      "logits/rejected": 0.21319513022899628,
-      "logps/chosen": -603.390380859375,
-      "logps/rejected": -718.3425903320312,
-      "loss": 0.2595,
-      "rewards/accuracies": 0.768750011920929,
-      "rewards/chosen": -0.25654521584510803,
-      "rewards/margins": 0.16353540122509003,
-      "rewards/rejected": -0.42008060216903687,
+      "epoch": 0.48,
+      "learning_rate": 4.709119209978242e-06,
+      "logits/chosen": 0.17320121824741364,
+      "logits/rejected": 0.2264091521501541,
+      "logps/chosen": -362.0121765136719,
+      "logps/rejected": -352.7041931152344,
+      "loss": 0.283,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/chosen": 0.039128489792346954,
+      "rewards/margins": 0.11708054691553116,
+      "rewards/rejected": -0.07795204222202301,
       "step": 200
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 4.94628771143819e-06,
-      "logits/chosen": 0.17947080731391907,
-      "logits/rejected": 0.2654314935207367,
-      "logps/chosen": -653.5524291992188,
-      "logps/rejected": -726.80126953125,
-      "loss": 0.3026,
-      "rewards/accuracies": 0.6875,
-      "rewards/chosen": -0.2890639305114746,
-      "rewards/margins": 0.12797169387340546,
-      "rewards/rejected": -0.41703566908836365,
+      "epoch": 0.5,
+      "learning_rate": 4.6580296144681155e-06,
+      "logits/chosen": 0.1604190617799759,
+      "logits/rejected": 0.17792078852653503,
+      "logps/chosen": -315.1614074707031,
+      "logps/rejected": -340.53619384765625,
+      "loss": 0.2754,
+      "rewards/accuracies": 0.768750011920929,
+      "rewards/chosen": 0.05531097203493118,
+      "rewards/margins": 0.15012916922569275,
+      "rewards/rejected": -0.09481821954250336,
       "step": 210
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 4.931699543346854e-06,
-      "logits/chosen": 0.1152615174651146,
-      "logits/rejected": 0.25124651193618774,
-      "logps/chosen": -588.23046875,
-      "logps/rejected": -704.0632934570312,
-      "loss": 0.2673,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.2535329759120941,
-      "rewards/margins": 0.17576026916503906,
-      "rewards/rejected": -0.4292932152748108,
+      "epoch": 0.53,
+      "learning_rate": 4.603133832077953e-06,
+      "logits/chosen": 0.11915634572505951,
+      "logits/rejected": 0.15653367340564728,
+      "logps/chosen": -351.16986083984375,
+      "logps/rejected": -354.53607177734375,
+      "loss": 0.2738,
+      "rewards/accuracies": 0.793749988079071,
+      "rewards/chosen": 0.06388933956623077,
+      "rewards/margins": 0.1507207453250885,
+      "rewards/rejected": -0.08683140575885773,
       "step": 220
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.9153878407169815e-06,
-      "logits/chosen": 0.13138779997825623,
-      "logits/rejected": 0.1749398410320282,
-      "logps/chosen": -536.669921875,
-      "logps/rejected": -619.0616455078125,
-      "loss": 0.2609,
-      "rewards/accuracies": 0.668749988079071,
-      "rewards/chosen": -0.2183808982372284,
-      "rewards/margins": 0.12902414798736572,
-      "rewards/rejected": -0.3474050462245941,
+      "epoch": 0.55,
+      "learning_rate": 4.544528684281056e-06,
+      "logits/chosen": 0.09443524479866028,
+      "logits/rejected": 0.1415812075138092,
+      "logps/chosen": -355.2025451660156,
+      "logps/rejected": -349.1300354003906,
+      "loss": 0.276,
+      "rewards/accuracies": 0.625,
+      "rewards/chosen": 0.021877283230423927,
+      "rewards/margins": 0.1259470283985138,
+      "rewards/rejected": -0.10406973212957382,
       "step": 230
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 4.897364164920515e-06,
-      "logits/chosen": 0.14722837507724762,
-      "logits/rejected": 0.2979207932949066,
-      "logps/chosen": -644.4813232421875,
-      "logps/rejected": -710.7215576171875,
-      "loss": 0.2364,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": -0.2495686262845993,
-      "rewards/margins": 0.14611390233039856,
-      "rewards/rejected": -0.39568251371383667,
+      "epoch": 0.58,
+      "learning_rate": 4.482317534878901e-06,
+      "logits/chosen": 0.08314280211925507,
+      "logits/rejected": 0.11439633369445801,
+      "logps/chosen": -333.59295654296875,
+      "logps/rejected": -341.5171203613281,
+      "loss": 0.2668,
+      "rewards/accuracies": 0.668749988079071,
+      "rewards/chosen": 0.016557829454541206,
+      "rewards/margins": 0.11629464477300644,
+      "rewards/rejected": -0.09973680973052979,
       "step": 240
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 4.8776412907378845e-06,
-      "logits/chosen": 0.14406076073646545,
-      "logits/rejected": 0.24922068417072296,
-      "logps/chosen": -602.2944946289062,
-      "logps/rejected": -689.9060668945312,
-      "loss": 0.2803,
-      "rewards/accuracies": 0.6937500238418579,
-      "rewards/chosen": -0.25432199239730835,
-      "rewards/margins": 0.14366620779037476,
-      "rewards/rejected": -0.3979881703853607,
+      "epoch": 0.6,
+      "learning_rate": 4.416610107695043e-06,
+      "logits/chosen": 0.11690554767847061,
+      "logits/rejected": 0.06475332379341125,
+      "logps/chosen": -331.7200012207031,
+      "logps/rejected": -341.45245361328125,
+      "loss": 0.2819,
+      "rewards/accuracies": 0.668749988079071,
+      "rewards/chosen": -0.025893816724419594,
+      "rewards/margins": 0.13103850185871124,
+      "rewards/rejected": -0.15693232417106628,
       "step": 250
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 4.8562331973035396e-06,
-      "logits/chosen": 0.14575393497943878,
-      "logits/rejected": 0.28581660985946655,
-      "logps/chosen": -567.1799926757812,
-      "logps/rejected": -649.4906005859375,
-      "loss": 0.2541,
-      "rewards/accuracies": 0.7562500238418579,
-      "rewards/chosen": -0.21804265677928925,
-      "rewards/margins": 0.13621987402439117,
-      "rewards/rejected": -0.3542625308036804,
+      "epoch": 0.62,
+      "learning_rate": 4.3475222930516484e-06,
+      "logits/chosen": 0.08940346539020538,
+      "logits/rejected": 0.12766343355178833,
+      "logps/chosen": -333.33343505859375,
+      "logps/rejected": -372.55755615234375,
+      "loss": 0.2833,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": -0.026656050235033035,
+      "rewards/margins": 0.16600963473320007,
+      "rewards/rejected": -0.19266566634178162,
       "step": 260
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.833155058197842e-06,
-      "logits/chosen": 0.20403075218200684,
-      "logits/rejected": 0.32390326261520386,
-      "logps/chosen": -611.0584106445312,
-      "logps/rejected": -643.7525634765625,
-      "loss": 0.2827,
+      "epoch": 0.65,
+      "learning_rate": 4.2751759433699745e-06,
+      "logits/chosen": 0.04847298935055733,
+      "logits/rejected": 0.11083607375621796,
+      "logps/chosen": -342.9352722167969,
+      "logps/rejected": -357.6617736816406,
+      "loss": 0.274,
       "rewards/accuracies": 0.6937500238418579,
-      "rewards/chosen": -0.23078171908855438,
-      "rewards/margins": 0.12669074535369873,
-      "rewards/rejected": -0.3574724495410919,
+      "rewards/chosen": -0.021627375856041908,
+      "rewards/margins": 0.12919363379478455,
+      "rewards/rejected": -0.1508210003376007,
       "step": 270
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 4.808423230692374e-06,
-      "logits/chosen": 0.16620513796806335,
-      "logits/rejected": 0.3157016634941101,
-      "logps/chosen": -557.9627075195312,
-      "logps/rejected": -638.0758666992188,
-      "loss": 0.2304,
-      "rewards/accuracies": 0.768750011920929,
-      "rewards/chosen": -0.1864185631275177,
-      "rewards/margins": 0.15307244658470154,
-      "rewards/rejected": -0.33949097990989685,
+      "epoch": 0.67,
+      "learning_rate": 4.199698658255298e-06,
+      "logits/chosen": 0.056878913193941116,
+      "logits/rejected": 0.14858202636241913,
+      "logps/chosen": -370.22637939453125,
+      "logps/rejected": -398.57159423828125,
+      "loss": 0.2715,
+      "rewards/accuracies": 0.6937500238418579,
+      "rewards/chosen": -0.021515587344765663,
+      "rewards/margins": 0.1492767035961151,
+      "rewards/rejected": -0.17079228162765503,
       "step": 280
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 4.7820552441562625e-06,
-      "logits/chosen": 0.18919572234153748,
-      "logits/rejected": 0.24673417210578918,
-      "logps/chosen": -532.9969482421875,
-      "logps/rejected": -591.529296875,
-      "loss": 0.2764,
+      "epoch": 0.7,
+      "learning_rate": 4.121223559445343e-06,
+      "logits/chosen": 0.03415738046169281,
+      "logits/rejected": 0.12577436864376068,
+      "logps/chosen": -352.68072509765625,
+      "logps/rejected": -383.16204833984375,
+      "loss": 0.264,
       "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": -0.19002854824066162,
-      "rewards/margins": 0.12101318687200546,
-      "rewards/rejected": -0.3110417425632477,
+      "rewards/chosen": -0.03958406671881676,
+      "rewards/margins": 0.1690487265586853,
+      "rewards/rejected": -0.20863279700279236,
       "step": 290
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 4.754069787631761e-06,
-      "logits/chosen": 0.15965518355369568,
-      "logits/rejected": 0.3048693537712097,
-      "logps/chosen": -550.864990234375,
-      "logps/rejected": -679.41357421875,
-      "loss": 0.25,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": -0.21647436916828156,
-      "rewards/margins": 0.16393651068210602,
-      "rewards/rejected": -0.3804108500480652,
+      "epoch": 0.72,
+      "learning_rate": 4.039889056019159e-06,
+      "logits/chosen": 0.02515377476811409,
+      "logits/rejected": 0.10390216112136841,
+      "logps/chosen": -353.2736511230469,
+      "logps/rejected": -353.888671875,
+      "loss": 0.2461,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/chosen": -0.031048249453306198,
+      "rewards/margins": 0.1348181664943695,
+      "rewards/rejected": -0.1658664047718048,
       "step": 300
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.724486696587862e-06,
-      "logits/chosen": 0.17140202224254608,
-      "logits/rejected": 0.21409063041210175,
-      "logps/chosen": -635.85791015625,
-      "logps/rejected": -747.188232421875,
-      "loss": 0.258,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": -0.29053109884262085,
-      "rewards/margins": 0.15401865541934967,
-      "rewards/rejected": -0.4445497393608093,
+      "epoch": 0.74,
+      "learning_rate": 3.955838600280535e-06,
+      "logits/chosen": 0.025213222950696945,
+      "logits/rejected": 0.1410323679447174,
+      "logps/chosen": -387.21856689453125,
+      "logps/rejected": -373.70355224609375,
+      "loss": 0.2703,
+      "rewards/accuracies": 0.7749999761581421,
+      "rewards/chosen": 0.010617800056934357,
+      "rewards/margins": 0.19538867473602295,
+      "rewards/rejected": -0.184770867228508,
       "step": 310
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 4.693326938861367e-06,
-      "logits/chosen": 0.19796046614646912,
-      "logits/rejected": 0.25672799348831177,
-      "logps/chosen": -643.25537109375,
-      "logps/rejected": -801.0643310546875,
-      "loss": 0.2427,
+      "epoch": 0.77,
+      "learning_rate": 3.869220434746509e-06,
+      "logits/chosen": 0.06151404231786728,
+      "logits/rejected": 0.1290605366230011,
+      "logps/chosen": -345.41571044921875,
+      "logps/rejected": -370.25592041015625,
+      "loss": 0.2703,
       "rewards/accuracies": 0.75,
-      "rewards/chosen": -0.31196296215057373,
-      "rewards/margins": 0.19956240057945251,
-      "rewards/rejected": -0.5115253925323486,
+      "rewards/chosen": -0.019938651472330093,
+      "rewards/margins": 0.16865777969360352,
+      "rewards/rejected": -0.1885964572429657,
       "step": 320
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 4.660612599795343e-06,
-      "logits/chosen": 0.11408114433288574,
-      "logits/rejected": 0.22174029052257538,
-      "logps/chosen": -647.49658203125,
-      "logps/rejected": -756.5814208984375,
-      "loss": 0.2472,
-      "rewards/accuracies": 0.699999988079071,
-      "rewards/chosen": -0.29143720865249634,
-      "rewards/margins": 0.1790235936641693,
-      "rewards/rejected": -0.47046083211898804,
+      "epoch": 0.79,
+      "learning_rate": 3.7801873306872315e-06,
+      "logits/chosen": 0.06525089591741562,
+      "logits/rejected": 0.12144273519515991,
+      "logps/chosen": -340.03277587890625,
+      "logps/rejected": -371.6439514160156,
+      "loss": 0.2577,
+      "rewards/accuracies": 0.7562500238418579,
+      "rewards/chosen": 0.02320241369307041,
+      "rewards/margins": 0.17125853896141052,
+      "rewards/rejected": -0.14805614948272705,
       "step": 330
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 4.626366866585528e-06,
-      "logits/chosen": 0.08651003241539001,
-      "logits/rejected": 0.29140934348106384,
-      "logps/chosen": -647.8692626953125,
-      "logps/rejected": -736.3836669921875,
-      "loss": 0.2449,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": -0.28671354055404663,
-      "rewards/margins": 0.16735221445560455,
-      "rewards/rejected": -0.45406574010849,
+      "epoch": 0.82,
+      "learning_rate": 3.688896318678322e-06,
+      "logits/chosen": 0.055392809212207794,
+      "logits/rejected": 0.12697988748550415,
+      "logps/chosen": -349.14556884765625,
+      "logps/rejected": -333.9625549316406,
+      "loss": 0.2748,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": 0.016882654279470444,
+      "rewards/margins": 0.16859912872314453,
+      "rewards/rejected": -0.1517164707183838,
       "step": 340
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.590614011845758e-06,
-      "logits/chosen": 0.14404548704624176,
-      "logits/rejected": 0.26181453466415405,
-      "logps/chosen": -638.6263427734375,
-      "logps/rejected": -730.7149658203125,
-      "loss": 0.2148,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": -0.2571524977684021,
-      "rewards/margins": 0.1780269742012024,
-      "rewards/rejected": -0.4351794719696045,
+      "epoch": 0.84,
+      "learning_rate": 3.5955084116409382e-06,
+      "logits/chosen": 0.08919240534305573,
+      "logits/rejected": 0.1610582321882248,
+      "logps/chosen": -367.30621337890625,
+      "logps/rejected": -346.13873291015625,
+      "loss": 0.2664,
+      "rewards/accuracies": 0.668749988079071,
+      "rewards/chosen": -0.04106982424855232,
+      "rewards/margins": 0.14082172513008118,
+      "rewards/rejected": -0.1818915605545044,
       "step": 350
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 4.553379376404085e-06,
-      "logits/chosen": 0.17719906568527222,
-      "logits/rejected": 0.17130860686302185,
-      "logps/chosen": -565.2915649414062,
-      "logps/rejected": -659.3453369140625,
-      "loss": 0.252,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": -0.23078814148902893,
-      "rewards/margins": 0.16517826914787292,
-      "rewards/rejected": -0.39596638083457947,
+      "epoch": 0.86,
+      "learning_rate": 3.5001883208580668e-06,
+      "logits/chosen": 0.056862883269786835,
+      "logits/rejected": 0.14601710438728333,
+      "logps/chosen": -383.3697204589844,
+      "logps/rejected": -388.45147705078125,
+      "loss": 0.2359,
+      "rewards/accuracies": 0.7562500238418579,
+      "rewards/chosen": -0.005547699984163046,
+      "rewards/margins": 0.20355132222175598,
+      "rewards/rejected": -0.20909900963306427,
       "step": 360
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 4.514689351341751e-06,
-      "logits/chosen": 0.18946215510368347,
-      "logits/rejected": 0.2520686089992523,
-      "logps/chosen": -707.220458984375,
-      "logps/rejected": -799.5474853515625,
-      "loss": 0.2296,
+      "epoch": 0.89,
+      "learning_rate": 3.403104165467883e-06,
+      "logits/chosen": 0.047759585082530975,
+      "logits/rejected": 0.1289873570203781,
+      "logps/chosen": -363.989990234375,
+      "logps/rejected": -361.4288330078125,
+      "loss": 0.2491,
       "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": -0.29390206933021545,
-      "rewards/margins": 0.1939956247806549,
-      "rewards/rejected": -0.48789769411087036,
+      "rewards/chosen": -0.030249441042542458,
+      "rewards/margins": 0.1802445650100708,
+      "rewards/rejected": -0.2104939967393875,
       "step": 370
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 4.474571359287791e-06,
-      "logits/chosen": 0.1794353723526001,
-      "logits/rejected": 0.20889122784137726,
-      "logps/chosen": -659.88720703125,
-      "logps/rejected": -792.5078125,
-      "loss": 0.2767,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": -0.32972732186317444,
-      "rewards/margins": 0.16797736287117004,
-      "rewards/rejected": -0.4977046847343445,
+      "epoch": 0.91,
+      "learning_rate": 3.30442717594657e-06,
+      "logits/chosen": 0.06461011618375778,
+      "logits/rejected": 0.14733566343784332,
+      "logps/chosen": -350.331298828125,
+      "logps/rejected": -334.6890563964844,
+      "loss": 0.2754,
+      "rewards/accuracies": 0.6625000238418579,
+      "rewards/chosen": -0.04477550461888313,
+      "rewards/margins": 0.12478353828191757,
+      "rewards/rejected": -0.1695590317249298,
       "step": 380
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 4.4330538349824684e-06,
-      "logits/chosen": 0.15636876225471497,
-      "logits/rejected": 0.2390608787536621,
-      "logps/chosen": -675.450927734375,
-      "logps/rejected": -797.5559692382812,
-      "loss": 0.2478,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": -0.3252645432949066,
-      "rewards/margins": 0.17893439531326294,
-      "rewards/rejected": -0.5041989684104919,
+      "epoch": 0.94,
+      "learning_rate": 3.2043313921035747e-06,
+      "logits/chosen": 0.07650026679039001,
+      "logits/rejected": 0.10351625829935074,
+      "logps/chosen": -319.55328369140625,
+      "logps/rejected": -328.97625732421875,
+      "loss": 0.2601,
+      "rewards/accuracies": 0.6812499761581421,
+      "rewards/chosen": -0.017551960423588753,
+      "rewards/margins": 0.1492632031440735,
+      "rewards/rejected": -0.1668151617050171,
       "step": 390
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 4.3901662051233755e-06,
-      "logits/chosen": 0.1320812702178955,
-      "logits/rejected": 0.25596413016319275,
-      "logps/chosen": -722.6912841796875,
-      "logps/rejected": -806.4666748046875,
-      "loss": 0.2426,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": -0.33014723658561707,
-      "rewards/margins": 0.17146170139312744,
-      "rewards/rejected": -0.5016089677810669,
+      "epoch": 0.96,
+      "learning_rate": 3.102993356121938e-06,
+      "logits/chosen": 0.045068711042404175,
+      "logits/rejected": 0.133053719997406,
+      "logps/chosen": -376.1606750488281,
+      "logps/rejected": -360.3962097167969,
+      "loss": 0.2547,
+      "rewards/accuracies": 0.731249988079071,
+      "rewards/chosen": -0.012314733117818832,
+      "rewards/margins": 0.18502004444599152,
+      "rewards/rejected": -0.19733479619026184,
       "step": 400
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 4.345938867508439e-06,
-      "logits/chosen": 0.15700757503509521,
-      "logits/rejected": 0.2542612552642822,
-      "logps/chosen": -715.8280029296875,
-      "logps/rejected": -812.6696166992188,
-      "loss": 0.2483,
-      "rewards/accuracies": 0.6937500238418579,
-      "rewards/chosen": -0.34785327315330505,
-      "rewards/margins": 0.17259207367897034,
-      "rewards/rejected": -0.5204453468322754,
+      "epoch": 0.98,
+      "learning_rate": 3.0005918011851245e-06,
+      "logits/chosen": 0.03985997289419174,
+      "logits/rejected": 0.1656588464975357,
+      "logps/chosen": -379.48199462890625,
+      "logps/rejected": -362.08380126953125,
+      "loss": 0.273,
+      "rewards/accuracies": 0.699999988079071,
+      "rewards/chosen": 0.00483871391043067,
+      "rewards/margins": 0.1729108989238739,
+      "rewards/rejected": -0.16807220876216888,
       "step": 410
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 4.30040316949064e-06,
-      "logits/chosen": 0.12117477506399155,
-      "logits/rejected": 0.21865728497505188,
-      "logps/chosen": -640.8743896484375,
-      "logps/rejected": -720.25439453125,
-      "loss": 0.2724,
+      "epoch": 1.01,
+      "learning_rate": 2.8973073362395e-06,
+      "logits/chosen": 0.06932957470417023,
+      "logits/rejected": 0.11695323139429092,
+      "logps/chosen": -350.8485107421875,
+      "logps/rejected": -359.5559387207031,
+      "loss": 0.2562,
       "rewards/accuracies": 0.6812499761581421,
-      "rewards/chosen": -0.30353325605392456,
-      "rewards/margins": 0.136855810880661,
-      "rewards/rejected": -0.44038906693458557,
+      "rewards/chosen": -0.015226135030388832,
+      "rewards/margins": 0.13259340822696686,
+      "rewards/rejected": -0.14781954884529114,
       "step": 420
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 4.253591385759705e-06,
-      "logits/chosen": 0.1048569455742836,
-      "logits/rejected": 0.2262450009584427,
-      "logps/chosen": -654.50634765625,
-      "logps/rejected": -762.971435546875,
-      "loss": 0.2437,
-      "rewards/accuracies": 0.7562500238418579,
-      "rewards/chosen": -0.30736270546913147,
-      "rewards/margins": 0.18053661286830902,
-      "rewards/rejected": -0.48789939284324646,
+      "epoch": 1.03,
+      "learning_rate": 2.7933221274484725e-06,
+      "logits/chosen": 0.022776301950216293,
+      "logits/rejected": 0.1463911086320877,
+      "logps/chosen": -344.72900390625,
+      "logps/rejected": -374.57110595703125,
+      "loss": 0.2546,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/chosen": 0.01165957935154438,
+      "rewards/margins": 0.17338308691978455,
+      "rewards/rejected": -0.1617235392332077,
       "step": 430
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 4.205536695466524e-06,
-      "logits/chosen": 0.1525915265083313,
-      "logits/rejected": 0.27372902631759644,
-      "logps/chosen": -654.9791259765625,
-      "logps/rejected": -811.6696166992188,
-      "loss": 0.2447,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": -0.319768488407135,
-      "rewards/margins": 0.19004443287849426,
-      "rewards/rejected": -0.5098129510879517,
+      "epoch": 1.06,
+      "learning_rate": 2.6888195769001147e-06,
+      "logits/chosen": 0.011232647113502026,
+      "logits/rejected": 0.08440439403057098,
+      "logps/chosen": -315.56158447265625,
+      "logps/rejected": -370.6732177734375,
+      "loss": 0.2635,
+      "rewards/accuracies": 0.7437499761581421,
+      "rewards/chosen": -0.0027101226150989532,
+      "rewards/margins": 0.18474070727825165,
+      "rewards/rejected": -0.1874508261680603,
       "step": 440
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 4.15627315870651e-06,
-      "logits/chosen": 0.136866956949234,
-      "logits/rejected": 0.21996262669563293,
-      "logps/chosen": -676.5054931640625,
-      "logps/rejected": -751.3809204101562,
-      "loss": 0.2517,
-      "rewards/accuracies": 0.699999988079071,
-      "rewards/chosen": -0.3240048289299011,
-      "rewards/margins": 0.1569635421037674,
-      "rewards/rejected": -0.48096832633018494,
+      "epoch": 1.08,
+      "learning_rate": 2.583983999134951e-06,
+      "logits/chosen": 0.033940933644771576,
+      "logits/rejected": 0.12383987754583359,
+      "logps/chosen": -353.528076171875,
+      "logps/rejected": -358.25433349609375,
+      "loss": 0.2647,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -0.01790205016732216,
+      "rewards/margins": 0.16596254706382751,
+      "rewards/rejected": -0.18386459350585938,
       "step": 450
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 4.105835692378557e-06,
-      "logits/chosen": 0.17371432483196259,
-      "logits/rejected": 0.2416466474533081,
-      "logps/chosen": -675.8497314453125,
-      "logps/rejected": -758.9337158203125,
-      "loss": 0.2008,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": -0.28624922037124634,
-      "rewards/margins": 0.17619088292121887,
-      "rewards/rejected": -0.4624401032924652,
+      "epoch": 1.1,
+      "learning_rate": 2.479000296064417e-06,
+      "logits/chosen": 0.03699932247400284,
+      "logits/rejected": 0.13089559972286224,
+      "logps/chosen": -375.724609375,
+      "logps/rejected": -400.3955383300781,
+      "loss": 0.2481,
+      "rewards/accuracies": 0.71875,
+      "rewards/chosen": -0.041518934071063995,
+      "rewards/margins": 0.1739250123500824,
+      "rewards/rejected": -0.21544396877288818,
       "step": 460
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 4.05426004543672e-06,
-      "logits/chosen": 0.12746404111385345,
-      "logits/rejected": 0.2692243754863739,
-      "logps/chosen": -697.2105712890625,
-      "logps/rejected": -788.6505126953125,
-      "loss": 0.2271,
-      "rewards/accuracies": 0.768750011920929,
-      "rewards/chosen": -0.3282613158226013,
-      "rewards/margins": 0.18242862820625305,
-      "rewards/rejected": -0.510689914226532,
+      "epoch": 1.13,
+      "learning_rate": 2.374053630853358e-06,
+      "logits/chosen": 0.07867871224880219,
+      "logits/rejected": 0.0793570876121521,
+      "logps/chosen": -392.0462646484375,
+      "logps/rejected": -398.4570617675781,
+      "loss": 0.2589,
+      "rewards/accuracies": 0.7437499761581421,
+      "rewards/chosen": -0.03614164516329765,
+      "rewards/margins": 0.18701379001140594,
+      "rewards/rejected": -0.2231554538011551,
       "step": 470
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 4.001582773552153e-06,
-      "logits/chosen": 0.19310589134693146,
-      "logits/rejected": 0.24649909138679504,
-      "logps/chosen": -676.9974365234375,
-      "logps/rejected": -746.8987426757812,
-      "loss": 0.2236,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": -0.31384724378585815,
-      "rewards/margins": 0.16469937562942505,
-      "rewards/rejected": -0.478546679019928,
+      "epoch": 1.15,
+      "learning_rate": 2.269329101341745e-06,
+      "logits/chosen": 0.04767027124762535,
+      "logits/rejected": 0.10338594764471054,
+      "logps/chosen": -311.9175109863281,
+      "logps/rejected": -353.84375,
+      "loss": 0.253,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/chosen": 0.022265803068876266,
+      "rewards/margins": 0.21186105906963348,
+      "rewards/rejected": -0.18959525227546692,
       "step": 480
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 3.947841213203262e-06,
-      "logits/chosen": 0.15690350532531738,
-      "logits/rejected": 0.300692081451416,
-      "logps/chosen": -735.6814575195312,
-      "logps/rejected": -848.1085815429688,
-      "loss": 0.2125,
-      "rewards/accuracies": 0.7562500238418579,
-      "rewards/chosen": -0.34843936562538147,
-      "rewards/margins": 0.20286861062049866,
-      "rewards/rejected": -0.5513080358505249,
+      "epoch": 1.18,
+      "learning_rate": 2.1650114135816052e-06,
+      "logits/chosen": 0.04343586042523384,
+      "logits/rejected": 0.14493630826473236,
+      "logps/chosen": -368.74066162109375,
+      "logps/rejected": -401.21746826171875,
+      "loss": 0.254,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -0.0024279176723212004,
+      "rewards/margins": 0.1874578297138214,
+      "rewards/rejected": -0.18988573551177979,
       "step": 490
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 3.893073455212438e-06,
-      "logits/chosen": 0.18581806123256683,
-      "logits/rejected": 0.28662142157554626,
-      "logps/chosen": -699.9567260742188,
-      "logps/rejected": -875.0812377929688,
-      "loss": 0.2029,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": -0.35302621126174927,
-      "rewards/margins": 0.2367512434720993,
-      "rewards/rejected": -0.589777410030365,
+      "epoch": 1.2,
+      "learning_rate": 2.06128455606496e-06,
+      "logits/chosen": 0.04143913835287094,
+      "logits/rejected": 0.06632859259843826,
+      "logps/chosen": -320.82281494140625,
+      "logps/rejected": -348.89923095703125,
+      "loss": 0.2438,
+      "rewards/accuracies": 0.699999988079071,
+      "rewards/chosen": -0.0026562472339719534,
+      "rewards/margins": 0.18748678267002106,
+      "rewards/rejected": -0.19014303386211395,
       "step": 500
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 3.837318317748134e-06,
-      "logits/chosen": 0.2250034064054489,
-      "logits/rejected": 0.25657814741134644,
-      "logps/chosen": -629.2000732421875,
-      "logps/rejected": -729.8438720703125,
-      "loss": 0.2881,
-      "rewards/accuracies": 0.699999988079071,
-      "rewards/chosen": -0.2759498357772827,
-      "rewards/margins": 0.16293799877166748,
-      "rewards/rejected": -0.43888789415359497,
+      "epoch": 1.22,
+      "learning_rate": 1.958331475217357e-06,
+      "logits/chosen": 0.03532598540186882,
+      "logits/rejected": 0.07111676037311554,
+      "logps/chosen": -345.3083801269531,
+      "logps/rejected": -391.5373840332031,
+      "loss": 0.2428,
+      "rewards/accuracies": 0.7437499761581421,
+      "rewards/chosen": -0.011091398075222969,
+      "rewards/margins": 0.18126052618026733,
+      "rewards/rejected": -0.19235190749168396,
       "step": 510
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 3.7806153188114027e-06,
-      "logits/chosen": 0.15463793277740479,
-      "logits/rejected": 0.26033270359039307,
-      "logps/chosen": -536.32861328125,
-      "logps/rejected": -633.5609130859375,
-      "loss": 0.2673,
-      "rewards/accuracies": 0.6937500238418579,
-      "rewards/chosen": -0.219068244099617,
-      "rewards/margins": 0.13708043098449707,
-      "rewards/rejected": -0.3561486601829529,
+      "epoch": 1.25,
+      "learning_rate": 1.856333752729311e-06,
+      "logits/chosen": 0.06463773548603058,
+      "logits/rejected": 0.07833746820688248,
+      "logps/chosen": -303.89508056640625,
+      "logps/rejected": -328.54095458984375,
+      "loss": 0.2549,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": -0.030626490712165833,
+      "rewards/margins": 0.14131976664066315,
+      "rewards/rejected": -0.17194625735282898,
       "step": 520
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 3.7230046482264256e-06,
-      "logits/chosen": 0.1772613823413849,
-      "logits/rejected": 0.3070564270019531,
-      "logps/chosen": -638.0252685546875,
-      "logps/rejected": -699.7626953125,
-      "loss": 0.2509,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.24965958297252655,
-      "rewards/margins": 0.15243235230445862,
-      "rewards/rejected": -0.402091920375824,
+      "epoch": 1.27,
+      "learning_rate": 1.7554712852947915e-06,
+      "logits/chosen": 0.017867419868707657,
+      "logits/rejected": 0.13077208399772644,
+      "logps/chosen": -354.83990478515625,
+      "logps/rejected": -369.40447998046875,
+      "loss": 0.2688,
+      "rewards/accuracies": 0.6812499761581421,
+      "rewards/chosen": -0.022668231278657913,
+      "rewards/margins": 0.164995938539505,
+      "rewards/rejected": -0.1876641809940338,
       "step": 530
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 3.6645271391548542e-06,
-      "logits/chosen": 0.1249006986618042,
-      "logits/rejected": 0.20080497860908508,
-      "logps/chosen": -620.0784912109375,
-      "logps/rejected": -716.3678588867188,
-      "loss": 0.2308,
-      "rewards/accuracies": 0.668749988079071,
-      "rewards/chosen": -0.2955653667449951,
-      "rewards/margins": 0.15287812054157257,
-      "rewards/rejected": -0.4484435021877289,
+      "epoch": 1.3,
+      "learning_rate": 1.6559219673215784e-06,
+      "logits/chosen": 0.07014649361371994,
+      "logits/rejected": 0.11957643926143646,
+      "logps/chosen": -341.1030578613281,
+      "logps/rejected": -360.0315246582031,
+      "loss": 0.2559,
+      "rewards/accuracies": 0.7250000238418579,
+      "rewards/chosen": 0.0031127408146858215,
+      "rewards/margins": 0.17289015650749207,
+      "rewards/rejected": -0.16977740824222565,
       "step": 540
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 3.6052242391541746e-06,
-      "logits/chosen": 0.1353389322757721,
-      "logits/rejected": 0.22389094531536102,
-      "logps/chosen": -614.9147338867188,
-      "logps/rejected": -771.8558349609375,
-      "loss": 0.2102,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.2764993906021118,
-      "rewards/margins": 0.18298561871051788,
-      "rewards/rejected": -0.4594849944114685,
+      "epoch": 1.32,
+      "learning_rate": 1.5578613771731214e-06,
+      "logits/chosen": 0.044239241629838943,
+      "logits/rejected": 0.11994221061468124,
+      "logps/chosen": -347.32757568359375,
+      "logps/rejected": -388.6127624511719,
+      "loss": 0.244,
+      "rewards/accuracies": 0.7562500238418579,
+      "rewards/chosen": -0.0042419894598424435,
+      "rewards/margins": 0.21681733429431915,
+      "rewards/rejected": -0.22105932235717773,
       "step": 550
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.5451379808006014e-06,
-      "logits/chosen": 0.17011868953704834,
-      "logits/rejected": 0.27556750178337097,
-      "logps/chosen": -658.0120849609375,
-      "logps/rejected": -755.5858764648438,
-      "loss": 0.2345,
-      "rewards/accuracies": 0.7124999761581421,
-      "rewards/chosen": -0.29901498556137085,
-      "rewards/margins": 0.18377789855003357,
-      "rewards/rejected": -0.48279285430908203,
+      "epoch": 1.34,
+      "learning_rate": 1.4614624674952843e-06,
+      "logits/chosen": 0.07131338119506836,
+      "logits/rejected": 0.14118310809135437,
+      "logps/chosen": -381.21112060546875,
+      "logps/rejected": -375.3702087402344,
+      "loss": 0.2594,
+      "rewards/accuracies": 0.737500011920929,
+      "rewards/chosen": -0.01365007646381855,
+      "rewards/margins": 0.16313722729682922,
+      "rewards/rejected": -0.17678730189800262,
       "step": 560
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 3.484310951897323e-06,
-      "logits/chosen": 0.1761491745710373,
-      "logits/rejected": 0.30945947766304016,
-      "logps/chosen": -717.3760986328125,
-      "logps/rejected": -787.5486450195312,
-      "loss": 0.2483,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": -0.32107868790626526,
-      "rewards/margins": 0.1693280041217804,
-      "rewards/rejected": -0.49040669202804565,
+      "epoch": 1.37,
+      "learning_rate": 1.3668952601741442e-06,
+      "logits/chosen": 0.019948173314332962,
+      "logits/rejected": 0.14301837980747223,
+      "logps/chosen": -359.31829833984375,
+      "logps/rejected": -386.3388366699219,
+      "loss": 0.2421,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/chosen": -0.003145938040688634,
+      "rewards/margins": 0.17547301948070526,
+      "rewards/rejected": -0.17861898243427277,
       "step": 570
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 3.4227862652892106e-06,
-      "logits/chosen": 0.20576810836791992,
-      "logits/rejected": 0.26615187525749207,
-      "logps/chosen": -668.9432983398438,
-      "logps/rejected": -773.5794067382812,
-      "loss": 0.2502,
-      "rewards/accuracies": 0.6937500238418579,
-      "rewards/chosen": -0.31081390380859375,
-      "rewards/margins": 0.15438678860664368,
-      "rewards/rejected": -0.46520066261291504,
+      "epoch": 1.39,
+      "learning_rate": 1.2743265464628787e-06,
+      "logits/chosen": 0.04147445410490036,
+      "logits/rejected": 0.07641445100307465,
+      "logps/chosen": -358.9191589355469,
+      "logps/rejected": -354.82989501953125,
+      "loss": 0.2574,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": -0.03237663954496384,
+      "rewards/margins": 0.14051951467990875,
+      "rewards/rejected": -0.17289616167545319,
       "step": 580
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 3.3606075283054005e-06,
-      "logits/chosen": 0.18598072230815887,
-      "logits/rejected": 0.2837832570075989,
-      "logps/chosen": -603.8995361328125,
-      "logps/rejected": -735.7412719726562,
-      "loss": 0.208,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": -0.2771075367927551,
-      "rewards/margins": 0.1824244260787964,
-      "rewards/rejected": -0.4595320224761963,
+      "epoch": 1.42,
+      "learning_rate": 1.1839195928066101e-06,
+      "logits/chosen": 0.010291008278727531,
+      "logits/rejected": 0.08601720631122589,
+      "logps/chosen": -338.0829162597656,
+      "logps/rejected": -349.2616882324219,
+      "loss": 0.2504,
+      "rewards/accuracies": 0.7437499761581421,
+      "rewards/chosen": -0.012054244987666607,
+      "rewards/margins": 0.18035855889320374,
+      "rewards/rejected": -0.19241279363632202,
       "step": 590
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 3.2978188118513814e-06,
-      "logits/chosen": 0.1718396097421646,
-      "logits/rejected": 0.3636724054813385,
-      "logps/chosen": -653.9473266601562,
-      "logps/rejected": -759.1709594726562,
-      "loss": 0.2263,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.28921762108802795,
-      "rewards/margins": 0.19078542292118073,
-      "rewards/rejected": -0.4800030589103699,
+      "epoch": 1.44,
+      "learning_rate": 1.0958338528840893e-06,
+      "logits/chosen": 0.07830692082643509,
+      "logits/rejected": 0.1112513542175293,
+      "logps/chosen": -318.32928466796875,
+      "logps/rejected": -351.01531982421875,
+      "loss": 0.2642,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": -0.012662211433053017,
+      "rewards/margins": 0.15172497928142548,
+      "rewards/rejected": -0.16438719630241394,
       "step": 600
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 3.234464619172522e-06,
-      "logits/chosen": 0.1497826725244522,
-      "logits/rejected": 0.2777649164199829,
-      "logps/chosen": -638.6043090820312,
-      "logps/rejected": -779.666748046875,
-      "loss": 0.223,
-      "rewards/accuracies": 0.7875000238418579,
-      "rewards/chosen": -0.28172770142555237,
-      "rewards/margins": 0.22779683768749237,
-      "rewards/rejected": -0.5095245242118835,
+      "epoch": 1.46,
+      "learning_rate": 1.0102246863740498e-06,
+      "logits/chosen": 0.013798505067825317,
+      "logits/rejected": 0.13072696328163147,
+      "logps/chosen": -326.76336669921875,
+      "logps/rejected": -380.63458251953125,
+      "loss": 0.2398,
+      "rewards/accuracies": 0.737500011920929,
+      "rewards/chosen": -0.0045418571680784225,
+      "rewards/margins": 0.19731177389621735,
+      "rewards/rejected": -0.20185360312461853,
       "step": 610
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 3.1705898543111576e-06,
-      "logits/chosen": 0.13684847950935364,
-      "logits/rejected": 0.24817728996276855,
-      "logps/chosen": -634.4373779296875,
-      "logps/rejected": -796.2052612304688,
-      "loss": 0.2257,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": -0.2933855652809143,
-      "rewards/margins": 0.20933008193969727,
-      "rewards/rejected": -0.5027156472206116,
+      "epoch": 1.49,
+      "learning_rate": 9.272430849423175e-07,
+      "logits/chosen": 0.041550200432538986,
+      "logits/rejected": 0.12003109604120255,
+      "logps/chosen": -350.9006652832031,
+      "logps/rejected": -404.7802734375,
+      "loss": 0.2245,
+      "rewards/accuracies": 0.800000011920929,
+      "rewards/chosen": 0.015362398698925972,
+      "rewards/margins": 0.22952251136302948,
+      "rewards/rejected": -0.21416012942790985,
       "step": 620
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 3.106239790279606e-06,
-      "logits/chosen": 0.16658630967140198,
-      "logits/rejected": 0.2954631447792053,
-      "logps/chosen": -654.8861083984375,
-      "logps/rejected": -780.149169921875,
-      "loss": 0.2097,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.3087119460105896,
-      "rewards/margins": 0.1822930872440338,
-      "rewards/rejected": -0.4910050928592682,
+      "epoch": 1.51,
+      "learning_rate": 8.470354059328919e-07,
+      "logits/chosen": 0.104413703083992,
+      "logits/rejected": 0.11118074506521225,
+      "logps/chosen": -336.5838928222656,
+      "logps/rejected": -373.56085205078125,
+      "loss": 0.2452,
+      "rewards/accuracies": 0.768750011920929,
+      "rewards/chosen": 0.010279458947479725,
+      "rewards/margins": 0.2295042723417282,
+      "rewards/rejected": -0.21922484040260315,
       "step": 630
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 3.041460036971664e-06,
-      "logits/chosen": 0.13499195873737335,
-      "logits/rejected": 0.27665549516677856,
-      "logps/chosen": -655.3276977539062,
-      "logps/rejected": -844.3450927734375,
-      "loss": 0.2251,
-      "rewards/accuracies": 0.78125,
-      "rewards/chosen": -0.3263702094554901,
-      "rewards/margins": 0.22413411736488342,
-      "rewards/rejected": -0.5505043268203735,
+      "epoch": 1.54,
+      "learning_rate": 7.697431142327633e-07,
+      "logits/chosen": 0.07976067811250687,
+      "logits/rejected": 0.12730778753757477,
+      "logps/chosen": -348.73443603515625,
+      "logps/rejected": -358.34088134765625,
+      "loss": 0.2338,
+      "rewards/accuracies": 0.6499999761581421,
+      "rewards/chosen": -0.030282145366072655,
+      "rewards/margins": 0.16269627213478088,
+      "rewards/rejected": -0.1929783970117569,
       "step": 640
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.976296508835326e-06,
-      "logits/chosen": 0.16133855283260345,
-      "logits/rejected": 0.256599485874176,
-      "logps/chosen": -660.0172119140625,
-      "logps/rejected": -769.0188598632812,
-      "loss": 0.2228,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": -0.3122148811817169,
-      "rewards/margins": 0.18627096712589264,
-      "rewards/rejected": -0.49848586320877075,
+      "epoch": 1.56,
+      "learning_rate": 6.955025327656839e-07,
+      "logits/chosen": 0.04196876287460327,
+      "logits/rejected": 0.11756552755832672,
+      "logps/chosen": -327.8496398925781,
+      "logps/rejected": -355.4369201660156,
+      "loss": 0.2558,
+      "rewards/accuracies": 0.706250011920929,
+      "rewards/chosen": 0.0020761913619935513,
+      "rewards/margins": 0.17507974803447723,
+      "rewards/rejected": -0.17300358414649963,
       "step": 650
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 2.910795392329649e-06,
-      "logits/chosen": 0.1639477163553238,
-      "logits/rejected": 0.24110262095928192,
-      "logps/chosen": -739.8809204101562,
-      "logps/rejected": -895.7755126953125,
-      "loss": 0.2189,
-      "rewards/accuracies": 0.78125,
-      "rewards/chosen": -0.33884376287460327,
-      "rewards/margins": 0.23572292923927307,
-      "rewards/rejected": -0.5745667219161987,
+      "epoch": 1.58,
+      "learning_rate": 6.244446020550182e-07,
+      "logits/chosen": 0.05316174030303955,
+      "logits/rejected": 0.10895484685897827,
+      "logps/chosen": -354.5049133300781,
+      "logps/rejected": -411.59765625,
+      "loss": 0.2319,
+      "rewards/accuracies": 0.7562500238418579,
+      "rewards/chosen": -0.0010157767683267593,
+      "rewards/margins": 0.21365651488304138,
+      "rewards/rejected": -0.2146722972393036,
       "step": 660
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 2.8450031131888147e-06,
-      "logits/chosen": 0.13471753895282745,
-      "logits/rejected": 0.27453264594078064,
-      "logps/chosen": -660.6944580078125,
-      "logps/rejected": -760.9178466796875,
-      "loss": 0.2316,
-      "rewards/accuracies": 0.6625000238418579,
-      "rewards/chosen": -0.32461482286453247,
-      "rewards/margins": 0.18133333325386047,
-      "rewards/rejected": -0.5059481859207153,
+      "epoch": 1.61,
+      "learning_rate": 5.566946492796766e-07,
+      "logits/chosen": 0.07230822741985321,
+      "logits/rejected": 0.09754084050655365,
+      "logps/chosen": -368.22802734375,
+      "logps/rejected": -368.54974365234375,
+      "loss": 0.2451,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": -0.02271811105310917,
+      "rewards/margins": 0.14353466033935547,
+      "rewards/rejected": -0.16625277698040009,
       "step": 670
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 2.7789663035166035e-06,
-      "logits/chosen": 0.07190994918346405,
-      "logits/rejected": 0.22703304886817932,
-      "logps/chosen": -670.5394287109375,
-      "logps/rejected": -823.9978637695312,
-      "loss": 0.2036,
+      "epoch": 1.63,
+      "learning_rate": 4.923721672305148e-07,
+      "logits/chosen": 0.04747115820646286,
+      "logits/rejected": 0.10951533168554306,
+      "logps/chosen": -373.25653076171875,
+      "logps/rejected": -403.66619873046875,
+      "loss": 0.262,
       "rewards/accuracies": 0.7562500238418579,
-      "rewards/chosen": -0.3125961720943451,
-      "rewards/margins": 0.2316323220729828,
-      "rewards/rejected": -0.5442285537719727,
+      "rewards/chosen": -3.2638385164318606e-05,
+      "rewards/margins": 0.20511355996131897,
+      "rewards/rejected": -0.2051461637020111,
       "step": 680
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 2.7127317687345973e-06,
-      "logits/chosen": 0.07092462480068207,
-      "logits/rejected": 0.26120471954345703,
-      "logps/chosen": -671.69189453125,
-      "logps/rejected": -811.6243286132812,
-      "loss": 0.2018,
-      "rewards/accuracies": 0.800000011920929,
-      "rewards/chosen": -0.30282872915267944,
-      "rewards/margins": 0.21179743111133575,
-      "rewards/rejected": -0.514626145362854,
+      "epoch": 1.66,
+      "learning_rate": 4.3159060355700943e-07,
+      "logits/chosen": 0.007146243005990982,
+      "logits/rejected": 0.15595687925815582,
+      "logps/chosen": -360.5429382324219,
+      "logps/rejected": -360.84271240234375,
+      "loss": 0.2528,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -0.026043469086289406,
+      "rewards/margins": 0.19069012999534607,
+      "rewards/rejected": -0.21673360466957092,
       "step": 690
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 2.6463464544075344e-06,
-      "logits/chosen": 0.14573340117931366,
-      "logits/rejected": 0.23916473984718323,
-      "logps/chosen": -684.8336181640625,
-      "logps/rejected": -777.0270385742188,
-      "loss": 0.1987,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": -0.3061198592185974,
-      "rewards/margins": 0.19952335953712463,
-      "rewards/rejected": -0.5056431889533997,
+      "epoch": 1.68,
+      "learning_rate": 3.7445716067596506e-07,
+      "logits/chosen": -0.016133427619934082,
+      "logits/rejected": 0.06616418063640594,
+      "logps/chosen": -315.7747497558594,
+      "logps/rejected": -344.2303771972656,
+      "loss": 0.242,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": 0.012198897078633308,
+      "rewards/margins": 0.2178380936384201,
+      "rewards/rejected": -0.20563916862010956,
       "step": 700
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 2.579857412969345e-06,
-      "logits/chosen": 0.11962984502315521,
-      "logits/rejected": 0.23462708294391632,
-      "logps/chosen": -730.8359375,
-      "logps/rejected": -856.9847412109375,
-      "loss": 0.237,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": -0.37815654277801514,
-      "rewards/margins": 0.18362250924110413,
-      "rewards/rejected": -0.5617790818214417,
+      "epoch": 1.7,
+      "learning_rate": 3.2107260669512334e-07,
+      "logits/chosen": 0.06611919403076172,
+      "logits/rejected": 0.08203423768281937,
+      "logps/chosen": -342.01263427734375,
+      "logps/rejected": -353.5125427246094,
+      "loss": 0.2461,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": -0.01212338637560606,
+      "rewards/margins": 0.17198148369789124,
+      "rewards/rejected": -0.18410487473011017,
       "step": 710
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 2.513311770373421e-06,
-      "logits/chosen": 0.11839403957128525,
-      "logits/rejected": 0.16228660941123962,
-      "logps/chosen": -644.6649780273438,
-      "logps/rejected": -824.513671875,
-      "loss": 0.2276,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": -0.32936322689056396,
-      "rewards/margins": 0.2137618511915207,
-      "rewards/rejected": -0.5431250929832458,
+      "epoch": 1.73,
+      "learning_rate": 2.7153109768518926e-07,
+      "logits/chosen": 0.05342602729797363,
+      "logits/rejected": 0.11405602842569351,
+      "logps/chosen": -393.02593994140625,
+      "logps/rejected": -416.9335021972656,
+      "loss": 0.244,
+      "rewards/accuracies": 0.7437499761581421,
+      "rewards/chosen": -0.015018805861473083,
+      "rewards/margins": 0.2232932150363922,
+      "rewards/rejected": -0.2383120059967041,
       "step": 720
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 2.446756692690804e-06,
-      "logits/chosen": 0.13511911034584045,
-      "logits/rejected": 0.2399536371231079,
-      "logps/chosen": -717.3846435546875,
-      "logps/rejected": -858.6282958984375,
-      "loss": 0.2365,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": -0.3688200116157532,
-      "rewards/margins": 0.2047378122806549,
-      "rewards/rejected": -0.5735577940940857,
+      "epoch": 1.75,
+      "learning_rate": 2.2592001161370392e-07,
+      "logits/chosen": 0.059743158519268036,
+      "logits/rejected": 0.08855228126049042,
+      "logps/chosen": -365.6115417480469,
+      "logps/rejected": -373.24310302734375,
+      "loss": 0.2413,
+      "rewards/accuracies": 0.731249988079071,
+      "rewards/chosen": -0.007994825020432472,
+      "rewards/margins": 0.19029465317726135,
+      "rewards/rejected": -0.19828948378562927,
       "step": 730
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 2.380239352679908e-06,
-      "logits/chosen": 0.14800789952278137,
-      "logits/rejected": 0.2702687382698059,
-      "logps/chosen": -674.1021728515625,
-      "logps/rejected": -830.0198974609375,
-      "loss": 0.2284,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": -0.31319013237953186,
-      "rewards/margins": 0.2306814193725586,
-      "rewards/rejected": -0.5438715219497681,
+      "epoch": 1.78,
+      "learning_rate": 1.8431979423369607e-07,
+      "logits/chosen": 0.01501550804823637,
+      "logits/rejected": 0.09877587854862213,
+      "logps/chosen": -335.7201232910156,
+      "logps/rejected": -356.1680603027344,
+      "loss": 0.2601,
+      "rewards/accuracies": 0.65625,
+      "rewards/chosen": -0.013049180619418621,
+      "rewards/margins": 0.1567631959915161,
+      "rewards/rejected": -0.16981235146522522,
       "step": 740
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 2.313806896351529e-06,
-      "logits/chosen": 0.07657043635845184,
-      "logits/rejected": 0.2402833253145218,
-      "logps/chosen": -611.2683715820312,
-      "logps/rejected": -762.6300048828125,
-      "loss": 0.2543,
-      "rewards/accuracies": 0.7124999761581421,
-      "rewards/chosen": -0.3043960928916931,
-      "rewards/margins": 0.188668355345726,
-      "rewards/rejected": -0.49306440353393555,
+      "epoch": 1.8,
+      "learning_rate": 1.468038171988881e-07,
+      "logits/chosen": -0.008327131159603596,
+      "logits/rejected": 0.04639572650194168,
+      "logps/chosen": -354.1353759765625,
+      "logps/rejected": -387.98297119140625,
+      "loss": 0.2595,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/chosen": -0.02448558434844017,
+      "rewards/margins": 0.1901397705078125,
+      "rewards/rejected": -0.21462532877922058,
       "step": 750
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 2.247506409552795e-06,
-      "logits/chosen": 0.11987291276454926,
-      "logits/rejected": 0.1521257907152176,
-      "logps/chosen": -654.2996826171875,
-      "logps/rejected": -776.6150512695312,
-      "loss": 0.248,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": -0.2802131772041321,
-      "rewards/margins": 0.1828833967447281,
-      "rewards/rejected": -0.463096559047699,
+      "epoch": 1.82,
+      "learning_rate": 1.1343824865573422e-07,
+      "logits/chosen": 0.01856027916073799,
+      "logits/rejected": 0.07309429347515106,
+      "logps/chosen": -321.44903564453125,
+      "logps/rejected": -341.5816955566406,
+      "loss": 0.2495,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": -0.022448932752013206,
+      "rewards/margins": 0.17198805510997772,
+      "rewards/rejected": -0.19443701207637787,
       "step": 760
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 2.1813848845937695e-06,
-      "logits/chosen": 0.07216247916221619,
-      "logits/rejected": 0.16098852455615997,
-      "logps/chosen": -612.78759765625,
-      "logps/rejected": -761.1871948242188,
-      "loss": 0.2486,
-      "rewards/accuracies": 0.675000011920929,
-      "rewards/chosen": -0.31463655829429626,
-      "rewards/margins": 0.16312028467655182,
-      "rewards/rejected": -0.4777568280696869,
+      "epoch": 1.85,
+      "learning_rate": 8.428193654051036e-08,
+      "logits/chosen": 0.04589134082198143,
+      "logits/rejected": 0.10319966077804565,
+      "logps/chosen": -388.9933776855469,
+      "logps/rejected": -376.8731994628906,
+      "loss": 0.2475,
+      "rewards/accuracies": 0.793749988079071,
+      "rewards/chosen": 0.008324380032718182,
+      "rewards/margins": 0.20527882874011993,
+      "rewards/rejected": -0.19695445895195007,
       "step": 770
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 2.1154891869403436e-06,
-      "logits/chosen": 0.056468479335308075,
-      "logits/rejected": 0.21281781792640686,
-      "logps/chosen": -676.2210693359375,
-      "logps/rejected": -826.8810424804688,
-      "loss": 0.202,
-      "rewards/accuracies": 0.793749988079071,
-      "rewards/chosen": -0.31608521938323975,
-      "rewards/margins": 0.22052684426307678,
-      "rewards/rejected": -0.5366120934486389,
+      "epoch": 1.87,
+      "learning_rate": 5.9386304787299175e-08,
+      "logits/chosen": 0.03318192437291145,
+      "logits/rejected": 0.1395682990550995,
+      "logps/chosen": -377.56622314453125,
+      "logps/rejected": -377.5900573730469,
+      "loss": 0.2477,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/chosen": -0.0049881902523338795,
+      "rewards/margins": 0.2095176726579666,
+      "rewards/rejected": -0.2145058661699295,
       "step": 780
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 2.0498660219970395e-06,
-      "logits/chosen": 0.14304831624031067,
-      "logits/rejected": 0.2865908741950989,
-      "logps/chosen": -720.0322875976562,
-      "logps/rejected": -834.4865112304688,
-      "loss": 0.2143,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": -0.3310641646385193,
-      "rewards/margins": 0.21627302467823029,
-      "rewards/rejected": -0.5473372340202332,
+      "epoch": 1.9,
+      "learning_rate": 3.8795262629929e-08,
+      "logits/chosen": 0.03711915761232376,
+      "logits/rejected": 0.07861719280481339,
+      "logps/chosen": -311.10015869140625,
+      "logps/rejected": -340.22918701171875,
+      "loss": 0.2288,
+      "rewards/accuracies": 0.78125,
+      "rewards/chosen": 0.007546453736722469,
+      "rewards/margins": 0.215033620595932,
+      "rewards/rejected": -0.20748718082904816,
       "step": 790
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.9845619020032552e-06,
-      "logits/chosen": 0.0913710817694664,
-      "logits/rejected": 0.2621229588985443,
-      "logps/chosen": -681.4385986328125,
-      "logps/rejected": -793.8607788085938,
-      "loss": 0.2232,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": -0.33381250500679016,
-      "rewards/margins": 0.18326430022716522,
-      "rewards/rejected": -0.5170767903327942,
+      "epoch": 1.92,
+      "learning_rate": 2.2545127157831416e-08,
+      "logits/chosen": 0.06011080741882324,
+      "logits/rejected": 0.08075010776519775,
+      "logps/chosen": -342.993408203125,
+      "logps/rejected": -338.7896728515625,
+      "loss": 0.252,
+      "rewards/accuracies": 0.737500011920929,
+      "rewards/chosen": -0.03219890594482422,
+      "rewards/margins": 0.15845921635627747,
+      "rewards/rejected": -0.1906580924987793,
       "step": 800
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.9196231130664282e-06,
-      "logits/chosen": 0.12347328662872314,
-      "logits/rejected": 0.22142863273620605,
-      "logps/chosen": -656.6223754882812,
-      "logps/rejected": -849.30908203125,
-      "loss": 0.2207,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.32625263929367065,
-      "rewards/margins": 0.2373245507478714,
-      "rewards/rejected": -0.5635771751403809,
+      "epoch": 1.94,
+      "learning_rate": 1.0664559262413831e-08,
+      "logits/chosen": 0.06324592232704163,
+      "logits/rejected": 0.15417756140232086,
+      "logps/chosen": -383.63238525390625,
+      "logps/rejected": -373.19720458984375,
+      "loss": 0.2445,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/chosen": -0.013102272525429726,
+      "rewards/margins": 0.21051840484142303,
+      "rewards/rejected": -0.2236206978559494,
       "step": 810
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.8550956823554708e-06,
-      "logits/chosen": 0.12958547472953796,
-      "logits/rejected": 0.20247094333171844,
-      "logps/chosen": -676.3146362304688,
-      "logps/rejected": -852.1516723632812,
-      "loss": 0.2231,
-      "rewards/accuracies": 0.793749988079071,
-      "rewards/chosen": -0.3258567452430725,
-      "rewards/margins": 0.22926822304725647,
-      "rewards/rejected": -0.5551249384880066,
+      "epoch": 1.97,
+      "learning_rate": 3.1745130869123564e-09,
+      "logits/chosen": 0.02718031406402588,
+      "logits/rejected": 0.09324290603399277,
+      "logps/chosen": -342.188232421875,
+      "logps/rejected": -382.42657470703125,
+      "loss": 0.2445,
+      "rewards/accuracies": 0.6937500238418579,
+      "rewards/chosen": -0.02895962819457054,
+      "rewards/margins": 0.1746593415737152,
+      "rewards/rejected": -0.20361897349357605,
       "step": 820
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.7910253454777346e-06,
-      "logits/chosen": 0.1100487932562828,
-      "logits/rejected": 0.178089901804924,
-      "logps/chosen": -628.2322387695312,
-      "logps/rejected": -760.2506103515625,
-      "loss": 0.2364,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": -0.3000151813030243,
-      "rewards/margins": 0.20657257735729218,
-      "rewards/rejected": -0.5065878033638,
-      "step": 830
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 1.7274575140626318e-06,
-      "logits/chosen": 0.06159307807683945,
-      "logits/rejected": 0.21589604020118713,
-      "logps/chosen": -658.613525390625,
-      "logps/rejected": -839.2703247070312,
-      "loss": 0.2066,
-      "rewards/accuracies": 0.762499988079071,
-      "rewards/chosen": -0.3004804253578186,
-      "rewards/margins": 0.23831859230995178,
-      "rewards/rejected": -0.5387989282608032,
-      "step": 840
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 1.6644372435748823e-06,
-      "logits/chosen": 0.13025906682014465,
-      "logits/rejected": 0.25792089104652405,
-      "logps/chosen": -657.7048950195312,
-      "logps/rejected": -741.805908203125,
-      "loss": 0.2056,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": -0.3026004731655121,
-      "rewards/margins": 0.18632087111473083,
-      "rewards/rejected": -0.4889214038848877,
-      "step": 850
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.6020092013802002e-06,
-      "logits/chosen": 0.13131192326545715,
-      "logits/rejected": 0.2267313450574875,
-      "logps/chosen": -609.1980590820312,
-      "logps/rejected": -762.2578735351562,
-      "loss": 0.2173,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": -0.2911318838596344,
-      "rewards/margins": 0.19050107896327972,
-      "rewards/rejected": -0.48163294792175293,
-      "step": 860
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 1.5402176350860653e-06,
-      "logits/chosen": 0.1238790899515152,
-      "logits/rejected": 0.215298131108284,
-      "logps/chosen": -654.3424072265625,
-      "logps/rejected": -766.5152587890625,
-      "loss": 0.2276,
-      "rewards/accuracies": 0.6875,
-      "rewards/chosen": -0.311137855052948,
-      "rewards/margins": 0.18526899814605713,
-      "rewards/rejected": -0.4964068531990051,
-      "step": 870
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 1.4791063411799938e-06,
-      "logits/chosen": 0.07573570311069489,
-      "logits/rejected": 0.16694195568561554,
-      "logps/chosen": -611.1826782226562,
-      "logps/rejected": -747.8428955078125,
-      "loss": 0.225,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.28111016750335693,
-      "rewards/margins": 0.19318345189094543,
-      "rewards/rejected": -0.47429361939430237,
-      "step": 880
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 1.4187186339875697e-06,
-      "logits/chosen": 0.14150698482990265,
-      "logits/rejected": 0.19218984246253967,
-      "logps/chosen": -653.740478515625,
-      "logps/rejected": -790.3258666992188,
-      "loss": 0.1959,
-      "rewards/accuracies": 0.768750011920929,
-      "rewards/chosen": -0.2978840172290802,
-      "rewards/margins": 0.19249173998832703,
-      "rewards/rejected": -0.4903757572174072,
-      "step": 890
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.3590973149722103e-06,
-      "logits/chosen": 0.07968850433826447,
-      "logits/rejected": 0.24394066631793976,
-      "logps/chosen": -660.4170532226562,
-      "logps/rejected": -787.5786743164062,
-      "loss": 0.2189,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": -0.3150777518749237,
-      "rewards/margins": 0.19490757584571838,
-      "rewards/rejected": -0.5099853277206421,
-      "step": 900
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 1.300284642398445e-06,
-      "logits/chosen": 0.1339327096939087,
-      "logits/rejected": 0.19529958069324493,
-      "logps/chosen": -637.8862915039062,
-      "logps/rejected": -804.20263671875,
-      "loss": 0.2179,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": -0.30193617939949036,
-      "rewards/margins": 0.2071986198425293,
-      "rewards/rejected": -0.5091347694396973,
-      "step": 910
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 1.2423223013801946e-06,
-      "logits/chosen": 0.09140697866678238,
-      "logits/rejected": 0.23906917870044708,
-      "logps/chosen": -615.2650146484375,
-      "logps/rejected": -794.8267211914062,
-      "loss": 0.2345,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.290291965007782,
-      "rewards/margins": 0.21678297221660614,
-      "rewards/rejected": -0.5070749521255493,
-      "step": 920
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 1.1852513743352886e-06,
-      "logits/chosen": 0.09063401818275452,
-      "logits/rejected": 0.18860659003257751,
-      "logps/chosen": -603.1500854492188,
-      "logps/rejected": -759.2166748046875,
-      "loss": 0.1973,
-      "rewards/accuracies": 0.762499988079071,
-      "rewards/chosen": -0.26861757040023804,
-      "rewards/margins": 0.2037418633699417,
-      "rewards/rejected": -0.47235947847366333,
-      "step": 930
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.1291123118671665e-06,
-      "logits/chosen": 0.082930788397789,
-      "logits/rejected": 0.177327498793602,
-      "logps/chosen": -685.0097045898438,
-      "logps/rejected": -753.899658203125,
-      "loss": 0.2698,
-      "rewards/accuracies": 0.668749988079071,
-      "rewards/chosen": -0.31101205945014954,
-      "rewards/margins": 0.16297343373298645,
-      "rewards/rejected": -0.473985493183136,
-      "step": 940
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 1.073944904094385e-06,
-      "logits/chosen": 0.12176795303821564,
-      "logits/rejected": 0.24055452644824982,
-      "logps/chosen": -727.9547729492188,
-      "logps/rejected": -807.3270874023438,
-      "loss": 0.2284,
+      "epoch": 1.99,
+      "learning_rate": 8.819906889168117e-11,
+      "logits/chosen": 0.07415173202753067,
+      "logits/rejected": 0.12375295162200928,
+      "logps/chosen": -362.17572021484375,
+      "logps/rejected": -372.21044921875,
+      "loss": 0.2579,
       "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": -0.3418019115924835,
-      "rewards/margins": 0.1853446215391159,
-      "rewards/rejected": -0.5271465182304382,
-      "step": 950
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 1.019788252448267e-06,
-      "logits/chosen": 0.1278046816587448,
-      "logits/rejected": 0.18657834827899933,
-      "logps/chosen": -629.8650512695312,
-      "logps/rejected": -775.0892333984375,
-      "loss": 0.2073,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": -0.31090590357780457,
-      "rewards/margins": 0.19522252678871155,
-      "rewards/rejected": -0.5061284899711609,
-      "step": 960
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 9.66680741958685e-07,
-      "logits/chosen": 0.06174594908952713,
-      "logits/rejected": 0.21714496612548828,
-      "logps/chosen": -706.654052734375,
-      "logps/rejected": -845.1712646484375,
-      "loss": 0.2177,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.33396342396736145,
-      "rewards/margins": 0.21265073120594025,
-      "rewards/rejected": -0.5466141700744629,
-      "step": 970
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 9.146600140475945e-07,
-      "logits/chosen": 0.09592024236917496,
-      "logits/rejected": 0.1995583027601242,
-      "logps/chosen": -628.862548828125,
-      "logps/rejected": -779.5205688476562,
-      "loss": 0.2233,
-      "rewards/accuracies": 0.7124999761581421,
-      "rewards/chosen": -0.3146194815635681,
-      "rewards/margins": 0.17806106805801392,
-      "rewards/rejected": -0.49268054962158203,
-      "step": 980
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 8.637629398496378e-07,
-      "logits/chosen": 0.06787069141864777,
-      "logits/rejected": 0.19926394522190094,
-      "logps/chosen": -675.4822387695312,
-      "logps/rejected": -792.4046630859375,
-      "loss": 0.2406,
-      "rewards/accuracies": 0.7562500238418579,
-      "rewards/chosen": -0.31674814224243164,
-      "rewards/margins": 0.19142326712608337,
-      "rewards/rejected": -0.5081714391708374,
-      "step": 990
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 8.140255940787059e-07,
-      "logits/chosen": 0.14530155062675476,
-      "logits/rejected": 0.2198048084974289,
-      "logps/chosen": -646.3609008789062,
-      "logps/rejected": -774.7366333007812,
-      "loss": 0.2234,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.30126315355300903,
-      "rewards/margins": 0.1865108758211136,
-      "rewards/rejected": -0.4877740442752838,
-      "step": 1000
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 7.654832294589776e-07,
-      "logits/chosen": 0.14776812493801117,
-      "logits/rejected": 0.18824756145477295,
-      "logps/chosen": -676.3685302734375,
-      "logps/rejected": -820.2717895507812,
-      "loss": 0.242,
-      "rewards/accuracies": 0.78125,
-      "rewards/chosen": -0.3261755406856537,
-      "rewards/margins": 0.19091561436653137,
-      "rewards/rejected": -0.5170911550521851,
-      "step": 1010
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 7.181702517385789e-07,
-      "logits/chosen": 0.15526942908763885,
-      "logits/rejected": 0.20596864819526672,
-      "logps/chosen": -720.7452392578125,
-      "logps/rejected": -820.1512451171875,
-      "loss": 0.2423,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": -0.33318132162094116,
-      "rewards/margins": 0.173573836684227,
-      "rewards/rejected": -0.5067551732063293,
-      "step": 1020
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.721201953035511e-07,
-      "logits/chosen": 0.09918368607759476,
-      "logits/rejected": 0.23762516677379608,
-      "logps/chosen": -690.7906494140625,
-      "logps/rejected": -799.6129150390625,
-      "loss": 0.222,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": -0.31717661023139954,
-      "rewards/margins": 0.19311536848545074,
-      "rewards/rejected": -0.5102919340133667,
-      "step": 1030
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.273656994094232e-07,
-      "logits/chosen": 0.08452818542718887,
-      "logits/rejected": 0.15428626537322998,
-      "logps/chosen": -674.3331298828125,
-      "logps/rejected": -852.9098510742188,
-      "loss": 0.212,
-      "rewards/accuracies": 0.793749988079071,
-      "rewards/chosen": -0.3417903482913971,
-      "rewards/margins": 0.20718708634376526,
-      "rewards/rejected": -0.5489774942398071,
-      "step": 1040
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 5.839384850472359e-07,
-      "logits/chosen": 0.11137109994888306,
-      "logits/rejected": 0.23986658453941345,
-      "logps/chosen": -681.7805786132812,
-      "logps/rejected": -840.8401489257812,
-      "loss": 0.2013,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": -0.33553558588027954,
-      "rewards/margins": 0.22832973301410675,
-      "rewards/rejected": -0.5638653039932251,
-      "step": 1050
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 5.418693324604082e-07,
-      "logits/chosen": 0.054320525377988815,
-      "logits/rejected": 0.2136838734149933,
-      "logps/chosen": -713.266357421875,
-      "logps/rejected": -872.5158081054688,
-      "loss": 0.1902,
-      "rewards/accuracies": 0.800000011920929,
-      "rewards/chosen": -0.3621819019317627,
-      "rewards/margins": 0.21344491839408875,
-      "rewards/rejected": -0.5756268501281738,
-      "step": 1060
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 5.01188059328386e-07,
-      "logits/chosen": 0.10655899345874786,
-      "logits/rejected": 0.20958073437213898,
-      "logps/chosen": -671.4863891601562,
-      "logps/rejected": -796.0863647460938,
-      "loss": 0.1935,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": -0.31292790174484253,
-      "rewards/margins": 0.21162652969360352,
-      "rewards/rejected": -0.5245543718338013,
-      "step": 1070
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.619234996325314e-07,
-      "logits/chosen": 0.1319715678691864,
-      "logits/rejected": 0.21460673213005066,
-      "logps/chosen": -709.2652587890625,
-      "logps/rejected": -860.7151489257812,
-      "loss": 0.2324,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": -0.3620796799659729,
-      "rewards/margins": 0.1911153942346573,
-      "rewards/rejected": -0.5531951189041138,
-      "step": 1080
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 4.241034832192434e-07,
-      "logits/chosen": 0.10131983458995819,
-      "logits/rejected": 0.1885637789964676,
-      "logps/chosen": -677.9981079101562,
-      "logps/rejected": -864.3063354492188,
-      "loss": 0.2159,
-      "rewards/accuracies": 0.793749988079071,
-      "rewards/chosen": -0.3420425057411194,
-      "rewards/margins": 0.23228974640369415,
-      "rewards/rejected": -0.5743322372436523,
-      "step": 1090
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 3.877548160747768e-07,
-      "logits/chosen": 0.11466535180807114,
-      "logits/rejected": 0.24565303325653076,
-      "logps/chosen": -680.638671875,
-      "logps/rejected": -798.11376953125,
-      "loss": 0.1963,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": -0.30832600593566895,
-      "rewards/margins": 0.23151281476020813,
-      "rewards/rejected": -0.5398387908935547,
-      "step": 1100
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.529032613257574e-07,
-      "logits/chosen": 0.10837472975254059,
-      "logits/rejected": 0.17577466368675232,
-      "logps/chosen": -682.72216796875,
-      "logps/rejected": -828.9464111328125,
-      "loss": 0.2376,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": -0.3374750316143036,
-      "rewards/margins": 0.20901791751384735,
-      "rewards/rejected": -0.5464929342269897,
-      "step": 1110
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 3.195735209788528e-07,
-      "logits/chosen": 0.1015692800283432,
-      "logits/rejected": 0.1785646677017212,
-      "logps/chosen": -639.4000244140625,
-      "logps/rejected": -781.9521484375,
-      "loss": 0.2455,
-      "rewards/accuracies": 0.6812499761581421,
-      "rewards/chosen": -0.33476072549819946,
-      "rewards/margins": 0.18203067779541016,
-      "rewards/rejected": -0.5167914032936096,
-      "step": 1120
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 2.8778921841253774e-07,
-      "logits/chosen": 0.06415721774101257,
-      "logits/rejected": 0.2145998477935791,
-      "logps/chosen": -690.4867553710938,
-      "logps/rejected": -865.8243408203125,
-      "loss": 0.1726,
-      "rewards/accuracies": 0.8125,
-      "rewards/chosen": -0.3243727684020996,
-      "rewards/margins": 0.251103937625885,
-      "rewards/rejected": -0.5754767656326294,
-      "step": 1130
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 2.5757288163336806e-07,
-      "logits/chosen": 0.08277393132448196,
-      "logits/rejected": 0.20556513965129852,
-      "logps/chosen": -701.6412353515625,
-      "logps/rejected": -884.2093505859375,
-      "loss": 0.2016,
-      "rewards/accuracies": 0.8062499761581421,
-      "rewards/chosen": -0.34318509697914124,
-      "rewards/margins": 0.23703384399414062,
-      "rewards/rejected": -0.5802189707756042,
-      "step": 1140
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.2894592730863336e-07,
-      "logits/chosen": 0.08898656070232391,
-      "logits/rejected": 0.19494621455669403,
-      "logps/chosen": -663.3425903320312,
-      "logps/rejected": -838.2100830078125,
-      "loss": 0.2081,
-      "rewards/accuracies": 0.7875000238418579,
-      "rewards/chosen": -0.3233157992362976,
-      "rewards/margins": 0.22612051665782928,
-      "rewards/rejected": -0.5494363903999329,
-      "step": 1150
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 2.019286455866981e-07,
-      "logits/chosen": 0.09930244833230972,
-      "logits/rejected": 0.25030988454818726,
-      "logps/chosen": -665.492919921875,
-      "logps/rejected": -791.5203857421875,
-      "loss": 0.2154,
-      "rewards/accuracies": 0.706250011920929,
-      "rewards/chosen": -0.3297829031944275,
-      "rewards/margins": 0.19049373269081116,
-      "rewards/rejected": -0.520276665687561,
-      "step": 1160
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 1.7654018571579557e-07,
-      "logits/chosen": 0.07714973390102386,
-      "logits/rejected": 0.1515274941921234,
-      "logps/chosen": -710.2567749023438,
-      "logps/rejected": -828.8133544921875,
-      "loss": 0.2217,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": -0.3351573944091797,
-      "rewards/margins": 0.20143434405326843,
-      "rewards/rejected": -0.5365917086601257,
-      "step": 1170
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 1.5279854247146703e-07,
-      "logits/chosen": 0.08655952662229538,
-      "logits/rejected": 0.18316319584846497,
-      "logps/chosen": -664.911376953125,
-      "logps/rejected": -804.0281982421875,
-      "loss": 0.2363,
-      "rewards/accuracies": 0.7124999761581421,
-      "rewards/chosen": -0.3494827151298523,
-      "rewards/margins": 0.19858424365520477,
-      "rewards/rejected": -0.5480669140815735,
-      "step": 1180
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.307205434022671e-07,
-      "logits/chosen": 0.0854581817984581,
-      "logits/rejected": 0.21030446887016296,
-      "logps/chosen": -680.3010864257812,
-      "logps/rejected": -877.068359375,
-      "loss": 0.1977,
-      "rewards/accuracies": 0.793749988079071,
-      "rewards/chosen": -0.33833056688308716,
-      "rewards/margins": 0.23077313601970673,
-      "rewards/rejected": -0.5691036581993103,
-      "step": 1190
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 1.1032183690276754e-07,
-      "logits/chosen": 0.07660888135433197,
-      "logits/rejected": 0.1921808272600174,
-      "logps/chosen": -673.0133056640625,
-      "logps/rejected": -855.6375732421875,
-      "loss": 0.2076,
-      "rewards/accuracies": 0.7875000238418579,
-      "rewards/chosen": -0.32487696409225464,
-      "rewards/margins": 0.2425091564655304,
-      "rewards/rejected": -0.5673861503601074,
-      "step": 1200
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 9.161688112232836e-08,
-      "logits/chosen": 0.10119873285293579,
-      "logits/rejected": 0.22447247803211212,
-      "logps/chosen": -711.1212158203125,
-      "logps/rejected": -857.6790771484375,
-      "loss": 0.2041,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": -0.35159099102020264,
-      "rewards/margins": 0.22090163826942444,
-      "rewards/rejected": -0.5724925994873047,
-      "step": 1210
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 7.46189337174788e-08,
-      "logits/chosen": 0.0487164705991745,
-      "logits/rejected": 0.1770685911178589,
-      "logps/chosen": -655.6456298828125,
-      "logps/rejected": -825.86572265625,
-      "loss": 0.1868,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.3402259647846222,
-      "rewards/margins": 0.2170572578907013,
-      "rewards/rejected": -0.5572832822799683,
-      "step": 1220
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.934004245518793e-08,
-      "logits/chosen": 0.09085109829902649,
-      "logits/rejected": 0.22920957207679749,
-      "logps/chosen": -656.9745483398438,
-      "logps/rejected": -807.7351684570312,
-      "loss": 0.2086,
-      "rewards/accuracies": 0.7562500238418579,
-      "rewards/chosen": -0.3091570734977722,
-      "rewards/margins": 0.2053556740283966,
-      "rewards/rejected": -0.5145127177238464,
-      "step": 1230
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 4.579103667367385e-08,
-      "logits/chosen": 0.10511846840381622,
-      "logits/rejected": 0.22097325325012207,
-      "logps/chosen": -677.5391235351562,
-      "logps/rejected": -815.5825805664062,
-      "loss": 0.2188,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": -0.3394906222820282,
-      "rewards/margins": 0.2081030309200287,
-      "rewards/rejected": -0.5475937128067017,
-      "step": 1240
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 3.398151960681162e-08,
-      "logits/chosen": 0.04612383991479874,
-      "logits/rejected": 0.16620075702667236,
-      "logps/chosen": -679.2454833984375,
-      "logps/rejected": -800.4119262695312,
-      "loss": 0.2539,
-      "rewards/accuracies": 0.6812499761581421,
-      "rewards/chosen": -0.35038450360298157,
-      "rewards/margins": 0.1713865101337433,
-      "rewards/rejected": -0.5217710733413696,
-      "step": 1250
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 2.3919861577572924e-08,
-      "logits/chosen": 0.11022261530160904,
-      "logits/rejected": 0.166605606675148,
-      "logps/chosen": -690.1370239257812,
-      "logps/rejected": -797.578369140625,
-      "loss": 0.2181,
-      "rewards/accuracies": 0.768750011920929,
-      "rewards/chosen": -0.32749998569488525,
-      "rewards/margins": 0.20777150988578796,
-      "rewards/rejected": -0.5352715253829956,
-      "step": 1260
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 1.5613194065327854e-08,
-      "logits/chosen": 0.052184127271175385,
-      "logits/rejected": 0.22973528504371643,
-      "logps/chosen": -629.1040649414062,
-      "logps/rejected": -766.4088134765625,
-      "loss": 0.1872,
-      "rewards/accuracies": 0.7562500238418579,
-      "rewards/chosen": -0.2912130355834961,
-      "rewards/margins": 0.22654838860034943,
-      "rewards/rejected": -0.5177614092826843,
-      "step": 1270
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 9.067404651211808e-09,
-      "logits/chosen": 0.11986621469259262,
-      "logits/rejected": 0.21088480949401855,
-      "logps/chosen": -687.1105346679688,
-      "logps/rejected": -877.3173828125,
-      "loss": 0.2028,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": -0.3289037048816681,
-      "rewards/margins": 0.23036351799964905,
-      "rewards/rejected": -0.5592672228813171,
-      "step": 1280
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.287132845137709e-09,
-      "logits/chosen": 0.08242613077163696,
-      "logits/rejected": 0.20786967873573303,
-      "logps/chosen": -682.5579223632812,
-      "logps/rejected": -803.6607666015625,
-      "loss": 0.2108,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": -0.32676413655281067,
-      "rewards/margins": 0.19439134001731873,
-      "rewards/rejected": -0.5211554765701294,
-      "step": 1290
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 1.2757667974155896e-09,
-      "logits/chosen": 0.03548216074705124,
-      "logits/rejected": 0.12074669450521469,
-      "logps/chosen": -665.2953491210938,
-      "logps/rejected": -846.1324462890625,
-      "loss": 0.2049,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": -0.3495555818080902,
-      "rewards/margins": 0.1981724500656128,
-      "rewards/rejected": -0.5477280020713806,
-      "step": 1300
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 3.544089730633804e-11,
-      "logits/chosen": 0.10499806702136993,
-      "logits/rejected": 0.2223556488752365,
-      "logps/chosen": -656.1048583984375,
-      "logps/rejected": -808.8086547851562,
-      "loss": 0.1796,
-      "rewards/accuracies": 0.71875,
-      "rewards/chosen": -0.3139723837375641,
-      "rewards/margins": 0.22497034072875977,
-      "rewards/rejected": -0.5389427542686462,
-      "step": 1310
+      "rewards/chosen": -0.023642729967832565,
+      "rewards/margins": 0.1827639937400818,
+      "rewards/rejected": -0.20640675723552704,
+      "step": 830
     },
     {
-      "epoch": 1.0,
-      "step": 1312,
+      "epoch": 2.0,
+      "step": 832,
       "total_flos": 0.0,
-      "train_loss": 0.24134081795175627,
-      "train_runtime": 12108.13,
-      "train_samples_per_second": 1.734,
-      "train_steps_per_second": 0.108
+      "train_loss": 0.27172684411589915,
+      "train_runtime": 11567.6763,
+      "train_samples_per_second": 3.458,
+      "train_steps_per_second": 0.072
     }
   ],
   "logging_steps": 10,
-  "max_steps": 1312,
+  "max_steps": 832,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
+  "num_train_epochs": 2,
   "save_steps": 100,
   "total_flos": 0.0,
   "train_batch_size": 4,