{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 1274, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0007849293563579278, "grad_norm": 15.215933158653918, "learning_rate": 3.90625e-09, "logits/chosen": 5881.4375, "logits/rejected": 2834.66162109375, "logps/chosen": -257.5969543457031, "logps/rejected": -120.09489440917969, "loss": 0.6931, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.007849293563579277, "grad_norm": 16.865367987014004, "learning_rate": 3.9062499999999997e-08, "logits/chosen": 4946.0361328125, "logits/rejected": 4332.5419921875, "logps/chosen": -248.810546875, "logps/rejected": -214.26235961914062, "loss": 0.6931, "rewards/accuracies": 0.4629629850387573, "rewards/chosen": 0.005654487758874893, "rewards/margins": -0.024100713431835175, "rewards/rejected": 0.02975519560277462, "step": 10 }, { "epoch": 0.015698587127158554, "grad_norm": 13.985859452977783, "learning_rate": 7.812499999999999e-08, "logits/chosen": 6023.357421875, "logits/rejected": 4842.69091796875, "logps/chosen": -266.1932678222656, "logps/rejected": -232.408203125, "loss": 0.6933, "rewards/accuracies": 0.6166667342185974, "rewards/chosen": 0.12002797424793243, "rewards/margins": 0.07616584002971649, "rewards/rejected": 0.04386211186647415, "step": 20 }, { "epoch": 0.023547880690737835, "grad_norm": 13.790396141342175, "learning_rate": 1.1718749999999999e-07, "logits/chosen": 6032.6904296875, "logits/rejected": 5038.70263671875, "logps/chosen": -310.45391845703125, "logps/rejected": -262.88800048828125, "loss": 0.693, "rewards/accuracies": 0.5166667699813843, "rewards/chosen": 0.15539582073688507, "rewards/margins": 0.0068007828667759895, "rewards/rejected": 0.14859503507614136, "step": 30 }, { "epoch": 0.03139717425431711, "grad_norm": 14.620456190673327, "learning_rate": 1.5624999999999999e-07, "logits/chosen": 5242.1494140625, "logits/rejected": 4344.2333984375, "logps/chosen": -263.437255859375, "logps/rejected": -224.7820281982422, "loss": 0.6914, "rewards/accuracies": 0.6166666746139526, "rewards/chosen": 0.5830156803131104, "rewards/margins": 0.3010711073875427, "rewards/rejected": 0.28194451332092285, "step": 40 }, { "epoch": 0.03924646781789639, "grad_norm": 15.592551720677791, "learning_rate": 1.9531249999999998e-07, "logits/chosen": 6364.0400390625, "logits/rejected": 4975.20751953125, "logps/chosen": -324.3094177246094, "logps/rejected": -257.7867431640625, "loss": 0.6893, "rewards/accuracies": 0.6916666626930237, "rewards/chosen": 1.8001701831817627, "rewards/margins": 1.0371043682098389, "rewards/rejected": 0.7630658745765686, "step": 50 }, { "epoch": 0.04709576138147567, "grad_norm": 16.391367418358456, "learning_rate": 2.3437499999999998e-07, "logits/chosen": 5382.1494140625, "logits/rejected": 4504.00927734375, "logps/chosen": -268.0631103515625, "logps/rejected": -261.58746337890625, "loss": 0.6856, "rewards/accuracies": 0.6666666269302368, "rewards/chosen": 1.8639802932739258, "rewards/margins": 1.2605732679367065, "rewards/rejected": 0.6034069657325745, "step": 60 }, { "epoch": 0.054945054945054944, "grad_norm": 12.384941768151403, "learning_rate": 2.734375e-07, "logits/chosen": 5104.5263671875, "logits/rejected": 4838.5341796875, "logps/chosen": -231.60293579101562, "logps/rejected": -233.22048950195312, "loss": 0.6816, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": 1.6550413370132446, "rewards/margins": 2.222501039505005, "rewards/rejected": -0.5674598217010498, "step": 70 }, { "epoch": 0.06279434850863422, "grad_norm": 12.830970527103005, "learning_rate": 3.1249999999999997e-07, "logits/chosen": 5692.2880859375, "logits/rejected": 5162.015625, "logps/chosen": -257.05865478515625, "logps/rejected": -247.0496368408203, "loss": 0.6733, "rewards/accuracies": 0.6499999761581421, "rewards/chosen": -1.0802600383758545, "rewards/margins": 3.6570305824279785, "rewards/rejected": -4.737290859222412, "step": 80 }, { "epoch": 0.0706436420722135, "grad_norm": 15.007154171179435, "learning_rate": 3.5156249999999997e-07, "logits/chosen": 5876.7021484375, "logits/rejected": 5055.72802734375, "logps/chosen": -287.79583740234375, "logps/rejected": -253.888427734375, "loss": 0.6607, "rewards/accuracies": 0.73333340883255, "rewards/chosen": -2.6046829223632812, "rewards/margins": 7.8172926902771, "rewards/rejected": -10.421976089477539, "step": 90 }, { "epoch": 0.07849293563579278, "grad_norm": 15.679438946334914, "learning_rate": 3.9062499999999997e-07, "logits/chosen": 5773.1474609375, "logits/rejected": 5556.29931640625, "logps/chosen": -283.6567077636719, "logps/rejected": -283.3360595703125, "loss": 0.6563, "rewards/accuracies": 0.7000000476837158, "rewards/chosen": -5.3816046714782715, "rewards/margins": 9.766166687011719, "rewards/rejected": -15.147770881652832, "step": 100 }, { "epoch": 0.08634222919937205, "grad_norm": 20.9770832240082, "learning_rate": 4.2968749999999996e-07, "logits/chosen": 6215.5146484375, "logits/rejected": 5104.4287109375, "logps/chosen": -263.65704345703125, "logps/rejected": -264.6483459472656, "loss": 0.6301, "rewards/accuracies": 0.7166666388511658, "rewards/chosen": -6.519162654876709, "rewards/margins": 14.516267776489258, "rewards/rejected": -21.035430908203125, "step": 110 }, { "epoch": 0.09419152276295134, "grad_norm": 23.87457177383535, "learning_rate": 4.6874999999999996e-07, "logits/chosen": 6058.70361328125, "logits/rejected": 5066.8955078125, "logps/chosen": -306.99322509765625, "logps/rejected": -317.37359619140625, "loss": 0.6271, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -27.406360626220703, "rewards/margins": 20.36496925354004, "rewards/rejected": -47.771324157714844, "step": 120 }, { "epoch": 0.10204081632653061, "grad_norm": 16.324100765360704, "learning_rate": 4.999962424962166e-07, "logits/chosen": 6037.888671875, "logits/rejected": 5573.15625, "logps/chosen": -294.66815185546875, "logps/rejected": -291.427001953125, "loss": 0.6327, "rewards/accuracies": 0.6500000357627869, "rewards/chosen": -4.560878276824951, "rewards/margins": 13.435412406921387, "rewards/rejected": -17.996292114257812, "step": 130 }, { "epoch": 0.10989010989010989, "grad_norm": 16.748062027868876, "learning_rate": 4.998647417232375e-07, "logits/chosen": 5911.548828125, "logits/rejected": 5174.02490234375, "logps/chosen": -273.7682800292969, "logps/rejected": -279.62591552734375, "loss": 0.6321, "rewards/accuracies": 0.658333420753479, "rewards/chosen": -13.526052474975586, "rewards/margins": 15.743997573852539, "rewards/rejected": -29.270050048828125, "step": 140 }, { "epoch": 0.11773940345368916, "grad_norm": 31.023008490255297, "learning_rate": 4.995454786965036e-07, "logits/chosen": 6033.5908203125, "logits/rejected": 5010.5927734375, "logps/chosen": -298.99713134765625, "logps/rejected": -288.54974365234375, "loss": 0.6196, "rewards/accuracies": 0.6583333611488342, "rewards/chosen": -21.12509536743164, "rewards/margins": 25.535253524780273, "rewards/rejected": -46.66034698486328, "step": 150 }, { "epoch": 0.12558869701726844, "grad_norm": 21.722743760287116, "learning_rate": 4.990386933279972e-07, "logits/chosen": 6004.2900390625, "logits/rejected": 5323.96142578125, "logps/chosen": -311.3618469238281, "logps/rejected": -331.3314514160156, "loss": 0.6022, "rewards/accuracies": 0.6916667222976685, "rewards/chosen": -37.79763412475586, "rewards/margins": 27.400625228881836, "rewards/rejected": -65.1982650756836, "step": 160 }, { "epoch": 0.13343799058084774, "grad_norm": 25.52831096940018, "learning_rate": 4.983447664444096e-07, "logits/chosen": 6254.384765625, "logits/rejected": 5551.97216796875, "logps/chosen": -342.96783447265625, "logps/rejected": -353.68341064453125, "loss": 0.5992, "rewards/accuracies": 0.6916667222976685, "rewards/chosen": -53.17034912109375, "rewards/margins": 31.506967544555664, "rewards/rejected": -84.67731475830078, "step": 170 }, { "epoch": 0.141287284144427, "grad_norm": 28.89187294736775, "learning_rate": 4.97464219500968e-07, "logits/chosen": 5510.7216796875, "logits/rejected": 4774.7548828125, "logps/chosen": -340.0227355957031, "logps/rejected": -355.3163146972656, "loss": 0.6045, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -76.57838439941406, "rewards/margins": 31.470958709716797, "rewards/rejected": -108.04936218261719, "step": 180 }, { "epoch": 0.14913657770800628, "grad_norm": 42.522805209204186, "learning_rate": 4.963977141895843e-07, "logits/chosen": 5586.52490234375, "logits/rejected": 4761.33349609375, "logps/chosen": -362.53131103515625, "logps/rejected": -402.01214599609375, "loss": 0.5423, "rewards/accuracies": 0.75, "rewards/chosen": -75.24663543701172, "rewards/margins": 65.30410766601562, "rewards/rejected": -140.55075073242188, "step": 190 }, { "epoch": 0.15698587127158556, "grad_norm": 55.81951831225517, "learning_rate": 4.951460519416227e-07, "logits/chosen": 5544.2958984375, "logits/rejected": 5095.11865234375, "logps/chosen": -302.46087646484375, "logps/rejected": -370.82659912109375, "loss": 0.5473, "rewards/accuracies": 0.7333333492279053, "rewards/chosen": -51.2109260559082, "rewards/margins": 56.14204788208008, "rewards/rejected": -107.35298156738281, "step": 200 }, { "epoch": 0.16483516483516483, "grad_norm": 42.75948410341764, "learning_rate": 4.937101733256606e-07, "logits/chosen": 5037.6884765625, "logits/rejected": 4414.658203125, "logps/chosen": -292.40435791015625, "logps/rejected": -335.58416748046875, "loss": 0.5861, "rewards/accuracies": 0.6500000357627869, "rewards/chosen": -67.69142150878906, "rewards/margins": 45.4319953918457, "rewards/rejected": -113.1234130859375, "step": 210 }, { "epoch": 0.1726844583987441, "grad_norm": 30.060946954036435, "learning_rate": 4.920911573406924e-07, "logits/chosen": 6118.02734375, "logits/rejected": 5131.0458984375, "logps/chosen": -379.2674560546875, "logps/rejected": -385.7772521972656, "loss": 0.5771, "rewards/accuracies": 0.7416666746139526, "rewards/chosen": -96.63453674316406, "rewards/margins": 55.60221481323242, "rewards/rejected": -152.2367706298828, "step": 220 }, { "epoch": 0.18053375196232338, "grad_norm": 28.76465418976182, "learning_rate": 4.902902206053098e-07, "logits/chosen": 5579.58349609375, "logits/rejected": 5029.0703125, "logps/chosen": -366.4660339355469, "logps/rejected": -399.53497314453125, "loss": 0.5624, "rewards/accuracies": 0.6916667222976685, "rewards/chosen": -96.18879699707031, "rewards/margins": 45.323951721191406, "rewards/rejected": -141.5127410888672, "step": 230 }, { "epoch": 0.18838304552590268, "grad_norm": 54.13787502408357, "learning_rate": 4.883087164434672e-07, "logits/chosen": 5040.7802734375, "logits/rejected": 4004.92138671875, "logps/chosen": -309.6131591796875, "logps/rejected": -339.3407897949219, "loss": 0.5532, "rewards/accuracies": 0.7083333134651184, "rewards/chosen": -69.91960906982422, "rewards/margins": 56.494903564453125, "rewards/rejected": -126.41451263427734, "step": 240 }, { "epoch": 0.19623233908948196, "grad_norm": 62.60105204430168, "learning_rate": 4.861481338675183e-07, "logits/chosen": 5948.767578125, "logits/rejected": 5258.9033203125, "logps/chosen": -309.3854675292969, "logps/rejected": -391.39923095703125, "loss": 0.5565, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -73.7380599975586, "rewards/margins": 63.26903533935547, "rewards/rejected": -137.00711059570312, "step": 250 }, { "epoch": 0.20408163265306123, "grad_norm": 55.232751216903075, "learning_rate": 4.838100964592904e-07, "logits/chosen": 6140.47998046875, "logits/rejected": 4885.4326171875, "logps/chosen": -362.2947692871094, "logps/rejected": -366.9859619140625, "loss": 0.5789, "rewards/accuracies": 0.6833333373069763, "rewards/chosen": -77.54212188720703, "rewards/margins": 50.919708251953125, "rewards/rejected": -128.4618377685547, "step": 260 }, { "epoch": 0.2119309262166405, "grad_norm": 41.10014381283569, "learning_rate": 4.812963611500339e-07, "logits/chosen": 5998.28271484375, "logits/rejected": 5766.78271484375, "logps/chosen": -368.8971252441406, "logps/rejected": -409.7018127441406, "loss": 0.5292, "rewards/accuracies": 0.7083333730697632, "rewards/chosen": -94.38850402832031, "rewards/margins": 48.48493194580078, "rewards/rejected": -142.8734588623047, "step": 270 }, { "epoch": 0.21978021978021978, "grad_norm": 42.86021305584719, "learning_rate": 4.786088169001671e-07, "logits/chosen": 5132.68359375, "logits/rejected": 4420.81884765625, "logps/chosen": -316.2486267089844, "logps/rejected": -395.09698486328125, "loss": 0.5365, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -87.5333023071289, "rewards/margins": 73.15364074707031, "rewards/rejected": -160.68695068359375, "step": 280 }, { "epoch": 0.22762951334379905, "grad_norm": 39.61716076908588, "learning_rate": 4.7574948327980567e-07, "logits/chosen": 7161.69677734375, "logits/rejected": 5208.32666015625, "logps/chosen": -416.9876403808594, "logps/rejected": -432.7337951660156, "loss": 0.5232, "rewards/accuracies": 0.7833333015441895, "rewards/chosen": -92.19709014892578, "rewards/margins": 82.77031707763672, "rewards/rejected": -174.96739196777344, "step": 290 }, { "epoch": 0.23547880690737832, "grad_norm": 40.60123952997694, "learning_rate": 4.727205089511466e-07, "logits/chosen": 5186.611328125, "logits/rejected": 5092.3154296875, "logps/chosen": -315.5050048828125, "logps/rejected": -376.13775634765625, "loss": 0.544, "rewards/accuracies": 0.6916667222976685, "rewards/chosen": -78.76446533203125, "rewards/margins": 53.446327209472656, "rewards/rejected": -132.21078491210938, "step": 300 }, { "epoch": 0.24332810047095763, "grad_norm": 36.73605195716174, "learning_rate": 4.6952417005384247e-07, "logits/chosen": 5836.0244140625, "logits/rejected": 5190.3154296875, "logps/chosen": -357.38543701171875, "logps/rejected": -395.3680419921875, "loss": 0.5741, "rewards/accuracies": 0.6833333969116211, "rewards/chosen": -101.71684265136719, "rewards/margins": 51.98549270629883, "rewards/rejected": -153.7023162841797, "step": 310 }, { "epoch": 0.25117739403453687, "grad_norm": 27.35251673221429, "learning_rate": 4.661628684945851e-07, "logits/chosen": 5839.91357421875, "logits/rejected": 5018.45263671875, "logps/chosen": -380.8829040527344, "logps/rejected": -439.2705078125, "loss": 0.5526, "rewards/accuracies": 0.6833333969116211, "rewards/chosen": -103.59564208984375, "rewards/margins": 63.749732971191406, "rewards/rejected": -167.34536743164062, "step": 320 }, { "epoch": 0.25902668759811615, "grad_norm": 38.27537763521133, "learning_rate": 4.626391301421782e-07, "logits/chosen": 5626.90673828125, "logits/rejected": 5112.2119140625, "logps/chosen": -358.74237060546875, "logps/rejected": -378.30609130859375, "loss": 0.5696, "rewards/accuracies": 0.675000011920929, "rewards/chosen": -88.6442642211914, "rewards/margins": 40.269752502441406, "rewards/rejected": -128.9140167236328, "step": 330 }, { "epoch": 0.2668759811616955, "grad_norm": 26.767097280759927, "learning_rate": 4.5895560292945996e-07, "logits/chosen": 5947.69921875, "logits/rejected": 6022.0703125, "logps/chosen": -349.23016357421875, "logps/rejected": -436.1158142089844, "loss": 0.552, "rewards/accuracies": 0.7166666984558105, "rewards/chosen": -81.14701843261719, "rewards/margins": 53.08740234375, "rewards/rejected": -134.23443603515625, "step": 340 }, { "epoch": 0.27472527472527475, "grad_norm": 46.3857425794004, "learning_rate": 4.5511505486349865e-07, "logits/chosen": 6274.0380859375, "logits/rejected": 5629.50830078125, "logps/chosen": -357.3734436035156, "logps/rejected": -443.00396728515625, "loss": 0.5359, "rewards/accuracies": 0.73333340883255, "rewards/chosen": -84.46565246582031, "rewards/margins": 75.03978729248047, "rewards/rejected": -159.50546264648438, "step": 350 }, { "epoch": 0.282574568288854, "grad_norm": 40.05800464660584, "learning_rate": 4.5112037194555876e-07, "logits/chosen": 5809.412109375, "logits/rejected": 5615.33984375, "logps/chosen": -376.6376953125, "logps/rejected": -486.61651611328125, "loss": 0.5304, "rewards/accuracies": 0.7666666507720947, "rewards/chosen": -132.9527130126953, "rewards/margins": 79.69070434570312, "rewards/rejected": -212.6434326171875, "step": 360 }, { "epoch": 0.2904238618524333, "grad_norm": 55.02789805944333, "learning_rate": 4.4697455600239863e-07, "logits/chosen": 5257.5693359375, "logits/rejected": 4872.26904296875, "logps/chosen": -380.5321044921875, "logps/rejected": -417.4842224121094, "loss": 0.5587, "rewards/accuracies": 0.7000000476837158, "rewards/chosen": -121.62977600097656, "rewards/margins": 57.43682861328125, "rewards/rejected": -179.0665740966797, "step": 370 }, { "epoch": 0.29827315541601257, "grad_norm": 38.3663361971708, "learning_rate": 4.426807224305315e-07, "logits/chosen": 6374.603515625, "logits/rejected": 5175.81103515625, "logps/chosen": -397.3004150390625, "logps/rejected": -432.21746826171875, "loss": 0.5282, "rewards/accuracies": 0.7916666865348816, "rewards/chosen": -89.32344818115234, "rewards/margins": 86.98558807373047, "rewards/rejected": -176.3090362548828, "step": 380 }, { "epoch": 0.30612244897959184, "grad_norm": 49.57772214593223, "learning_rate": 4.3824209785514326e-07, "logits/chosen": 6473.6474609375, "logits/rejected": 4862.162109375, "logps/chosen": -374.4190673828125, "logps/rejected": -416.25091552734375, "loss": 0.5443, "rewards/accuracies": 0.7500000596046448, "rewards/chosen": -82.89260864257812, "rewards/margins": 87.8666000366211, "rewards/rejected": -170.7592010498047, "step": 390 }, { "epoch": 0.3139717425431711, "grad_norm": 58.149436511972006, "learning_rate": 4.3366201770542687e-07, "logits/chosen": 5480.921875, "logits/rejected": 5279.40869140625, "logps/chosen": -359.6849365234375, "logps/rejected": -448.46490478515625, "loss": 0.5345, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -85.0674819946289, "rewards/margins": 76.78694915771484, "rewards/rejected": -161.85443115234375, "step": 400 }, { "epoch": 0.3218210361067504, "grad_norm": 60.61370169105453, "learning_rate": 4.2894392370815567e-07, "logits/chosen": 5922.7822265625, "logits/rejected": 5269.4404296875, "logps/chosen": -375.8420104980469, "logps/rejected": -454.2471618652344, "loss": 0.4528, "rewards/accuracies": 0.7916666269302368, "rewards/chosen": -83.9731674194336, "rewards/margins": 87.64946746826172, "rewards/rejected": -171.62265014648438, "step": 410 }, { "epoch": 0.32967032967032966, "grad_norm": 66.71318683810037, "learning_rate": 4.2409136130137845e-07, "logits/chosen": 5581.8837890625, "logits/rejected": 5001.927734375, "logps/chosen": -398.95745849609375, "logps/rejected": -448.9306640625, "loss": 0.5291, "rewards/accuracies": 0.7249999642372131, "rewards/chosen": -113.06422424316406, "rewards/margins": 82.78407287597656, "rewards/rejected": -195.8482666015625, "step": 420 }, { "epoch": 0.33751962323390894, "grad_norm": 31.295866106788377, "learning_rate": 4.1910797697018017e-07, "logits/chosen": 5411.54052734375, "logits/rejected": 4497.97705078125, "logps/chosen": -369.30487060546875, "logps/rejected": -426.73895263671875, "loss": 0.515, "rewards/accuracies": 0.8250001072883606, "rewards/chosen": -112.50996398925781, "rewards/margins": 89.22474670410156, "rewards/rejected": -201.73471069335938, "step": 430 }, { "epoch": 0.3453689167974882, "grad_norm": 41.50008674395982, "learning_rate": 4.1399751550651084e-07, "logits/chosen": 5798.984375, "logits/rejected": 5733.89697265625, "logps/chosen": -350.151611328125, "logps/rejected": -420.47100830078125, "loss": 0.5273, "rewards/accuracies": 0.7083333730697632, "rewards/chosen": -99.38359832763672, "rewards/margins": 60.87189865112305, "rewards/rejected": -160.25550842285156, "step": 440 }, { "epoch": 0.3532182103610675, "grad_norm": 27.281230892660613, "learning_rate": 4.087638171951401e-07, "logits/chosen": 6737.7490234375, "logits/rejected": 4784.97607421875, "logps/chosen": -369.363037109375, "logps/rejected": -394.835205078125, "loss": 0.5192, "rewards/accuracies": 0.73333340883255, "rewards/chosen": -82.30615997314453, "rewards/margins": 86.52983093261719, "rewards/rejected": -168.8359832763672, "step": 450 }, { "epoch": 0.36106750392464676, "grad_norm": 26.52958051617833, "learning_rate": 4.034108149278543e-07, "logits/chosen": 6985.04931640625, "logits/rejected": 5294.892578125, "logps/chosen": -426.048095703125, "logps/rejected": -441.2728576660156, "loss": 0.4874, "rewards/accuracies": 0.783333420753479, "rewards/chosen": -93.33697509765625, "rewards/margins": 83.284423828125, "rewards/rejected": -176.6214141845703, "step": 460 }, { "epoch": 0.36891679748822603, "grad_norm": 67.6084653798285, "learning_rate": 3.979425312480629e-07, "logits/chosen": 5940.57568359375, "logits/rejected": 5153.30712890625, "logps/chosen": -428.033935546875, "logps/rejected": -501.040283203125, "loss": 0.5351, "rewards/accuracies": 0.7666666507720947, "rewards/chosen": -140.24819946289062, "rewards/margins": 82.69334411621094, "rewards/rejected": -222.9415283203125, "step": 470 }, { "epoch": 0.37676609105180536, "grad_norm": 29.50882408287276, "learning_rate": 3.923630753280357e-07, "logits/chosen": 6457.30322265625, "logits/rejected": 5457.03271484375, "logps/chosen": -391.68280029296875, "logps/rejected": -450.43585205078125, "loss": 0.5211, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -111.87620544433594, "rewards/margins": 91.76634216308594, "rewards/rejected": -203.6425323486328, "step": 480 }, { "epoch": 0.38461538461538464, "grad_norm": 38.10445969681201, "learning_rate": 3.866766398810424e-07, "logits/chosen": 6093.3935546875, "logits/rejected": 5703.35693359375, "logps/chosen": -315.85400390625, "logps/rejected": -447.4837341308594, "loss": 0.4873, "rewards/accuracies": 0.7916666269302368, "rewards/chosen": -74.91297912597656, "rewards/margins": 99.97135162353516, "rewards/rejected": -174.8843231201172, "step": 490 }, { "epoch": 0.3924646781789639, "grad_norm": 47.27132243249988, "learning_rate": 3.8088749801071496e-07, "logits/chosen": 6474.16748046875, "logits/rejected": 4981.7763671875, "logps/chosen": -439.0074157714844, "logps/rejected": -501.56842041015625, "loss": 0.4962, "rewards/accuracies": 0.7833333015441895, "rewards/chosen": -140.51797485351562, "rewards/margins": 96.61772155761719, "rewards/rejected": -237.1357421875, "step": 500 }, { "epoch": 0.4003139717425432, "grad_norm": 35.98604026243411, "learning_rate": 3.75e-07, "logits/chosen": 5234.4345703125, "logits/rejected": 4549.6767578125, "logps/chosen": -438.78631591796875, "logps/rejected": -498.96923828125, "loss": 0.5142, "rewards/accuracies": 0.7833333611488342, "rewards/chosen": -181.92649841308594, "rewards/margins": 87.92505645751953, "rewards/rejected": -269.8515625, "step": 510 }, { "epoch": 0.40816326530612246, "grad_norm": 39.174562124624565, "learning_rate": 3.6901857004211443e-07, "logits/chosen": 5512.3955078125, "logits/rejected": 5052.76416015625, "logps/chosen": -467.6012268066406, "logps/rejected": -543.6987915039062, "loss": 0.5675, "rewards/accuracies": 0.7083333730697632, "rewards/chosen": -201.4283447265625, "rewards/margins": 73.14409637451172, "rewards/rejected": -274.57244873046875, "step": 520 }, { "epoch": 0.41601255886970173, "grad_norm": 33.04601569313346, "learning_rate": 3.6294770291596076e-07, "logits/chosen": 6273.7939453125, "logits/rejected": 5080.54443359375, "logps/chosen": -442.8894958496094, "logps/rejected": -485.6820373535156, "loss": 0.4968, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -158.59738159179688, "rewards/margins": 63.013832092285156, "rewards/rejected": -221.6112060546875, "step": 530 }, { "epoch": 0.423861852433281, "grad_norm": 44.438216645790014, "learning_rate": 3.5679196060850034e-07, "logits/chosen": 5984.93310546875, "logits/rejected": 5265.84326171875, "logps/chosen": -416.12042236328125, "logps/rejected": -482.99188232421875, "loss": 0.5228, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -138.7367401123047, "rewards/margins": 88.85780334472656, "rewards/rejected": -227.5945281982422, "step": 540 }, { "epoch": 0.4317111459968603, "grad_norm": 60.7826664795441, "learning_rate": 3.505559688866229e-07, "logits/chosen": 5775.0068359375, "logits/rejected": 5291.79833984375, "logps/chosen": -403.44110107421875, "logps/rejected": -507.52532958984375, "loss": 0.5089, "rewards/accuracies": 0.7500000596046448, "rewards/chosen": -122.10447692871094, "rewards/margins": 94.2375717163086, "rewards/rejected": -216.342041015625, "step": 550 }, { "epoch": 0.43956043956043955, "grad_norm": 42.870075521718, "learning_rate": 3.4424441382108826e-07, "logits/chosen": 5771.8349609375, "logits/rejected": 5387.10400390625, "logps/chosen": -394.4440002441406, "logps/rejected": -449.600341796875, "loss": 0.5367, "rewards/accuracies": 0.7166666984558105, "rewards/chosen": -118.31639099121094, "rewards/margins": 64.5057373046875, "rewards/rejected": -182.82215881347656, "step": 560 }, { "epoch": 0.4474097331240188, "grad_norm": 55.59078774096673, "learning_rate": 3.378620382651523e-07, "logits/chosen": 6119.66943359375, "logits/rejected": 5629.6025390625, "logps/chosen": -428.170166015625, "logps/rejected": -488.3804626464844, "loss": 0.4851, "rewards/accuracies": 0.7833333611488342, "rewards/chosen": -113.60516357421875, "rewards/margins": 82.83430480957031, "rewards/rejected": -196.43948364257812, "step": 570 }, { "epoch": 0.4552590266875981, "grad_norm": 59.224759647083445, "learning_rate": 3.314136382905234e-07, "logits/chosen": 6144.49755859375, "logits/rejected": 5452.28662109375, "logps/chosen": -402.88995361328125, "logps/rejected": -519.4390258789062, "loss": 0.5047, "rewards/accuracies": 0.7666667103767395, "rewards/chosen": -115.64312744140625, "rewards/margins": 118.94549560546875, "rewards/rejected": -234.588623046875, "step": 580 }, { "epoch": 0.4631083202511774, "grad_norm": 39.79258234642554, "learning_rate": 3.249040595833274e-07, "logits/chosen": 6639.1748046875, "logits/rejected": 5518.78271484375, "logps/chosen": -443.51800537109375, "logps/rejected": -491.791748046875, "loss": 0.4812, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -137.3645782470703, "rewards/margins": 108.58697509765625, "rewards/rejected": -245.95156860351562, "step": 590 }, { "epoch": 0.47095761381475665, "grad_norm": 56.06606610855067, "learning_rate": 3.1833819380279023e-07, "logits/chosen": 6277.1806640625, "logits/rejected": 5263.185546875, "logps/chosen": -393.0625305175781, "logps/rejected": -485.69708251953125, "loss": 0.4737, "rewards/accuracies": 0.7833333611488342, "rewards/chosen": -140.57150268554688, "rewards/margins": 90.16180419921875, "rewards/rejected": -230.7333221435547, "step": 600 }, { "epoch": 0.478806907378336, "grad_norm": 31.75615082981101, "learning_rate": 3.11720974905373e-07, "logits/chosen": 6018.9443359375, "logits/rejected": 5180.486328125, "logps/chosen": -406.84033203125, "logps/rejected": -474.59161376953125, "loss": 0.4824, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -133.99887084960938, "rewards/margins": 90.10101318359375, "rewards/rejected": -224.0998992919922, "step": 610 }, { "epoch": 0.48665620094191525, "grad_norm": 35.42434649377237, "learning_rate": 3.0505737543712275e-07, "logits/chosen": 5074.7724609375, "logits/rejected": 4104.3525390625, "logps/chosen": -393.4937744140625, "logps/rejected": -432.28594970703125, "loss": 0.5006, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -145.6678009033203, "rewards/margins": 73.0408935546875, "rewards/rejected": -218.7086944580078, "step": 620 }, { "epoch": 0.4945054945054945, "grad_norm": 38.7031622569247, "learning_rate": 2.9835240279702513e-07, "logits/chosen": 6648.0712890625, "logits/rejected": 5553.2958984375, "logps/chosen": -431.44091796875, "logps/rejected": -483.741943359375, "loss": 0.4939, "rewards/accuracies": 0.7583333849906921, "rewards/chosen": -114.72941589355469, "rewards/margins": 102.97762298583984, "rewards/rejected": -217.70703125, "step": 630 }, { "epoch": 0.5023547880690737, "grad_norm": 40.47979948693865, "learning_rate": 2.9161109547416667e-07, "logits/chosen": 6354.0009765625, "logits/rejected": 5397.0810546875, "logps/chosen": -409.93115234375, "logps/rejected": -468.1148986816406, "loss": 0.5108, "rewards/accuracies": 0.6833333373069763, "rewards/chosen": -131.24069213867188, "rewards/margins": 59.033477783203125, "rewards/rejected": -190.27415466308594, "step": 640 }, { "epoch": 0.5102040816326531, "grad_norm": 38.0866483420765, "learning_rate": 2.848385192615339e-07, "logits/chosen": 5408.5224609375, "logits/rejected": 4339.13818359375, "logps/chosen": -387.74493408203125, "logps/rejected": -444.20135498046875, "loss": 0.4872, "rewards/accuracies": 0.7916666865348816, "rewards/chosen": -130.2064208984375, "rewards/margins": 86.89310455322266, "rewards/rejected": -217.09951782226562, "step": 650 }, { "epoch": 0.5180533751962323, "grad_norm": 44.33575742329842, "learning_rate": 2.780397634492949e-07, "logits/chosen": 6022.939453125, "logits/rejected": 4747.7412109375, "logps/chosen": -432.54583740234375, "logps/rejected": -515.0159912109375, "loss": 0.5238, "rewards/accuracies": 0.8166667222976685, "rewards/chosen": -142.22756958007812, "rewards/margins": 120.51753997802734, "rewards/rejected": -262.7451171875, "step": 660 }, { "epoch": 0.5259026687598116, "grad_norm": 50.41576405157563, "learning_rate": 2.71219937000424e-07, "logits/chosen": 6016.4833984375, "logits/rejected": 4903.53173828125, "logps/chosen": -411.1146545410156, "logps/rejected": -472.8374938964844, "loss": 0.5028, "rewards/accuracies": 0.7666666507720947, "rewards/chosen": -134.9041290283203, "rewards/margins": 86.84437561035156, "rewards/rejected": -221.74850463867188, "step": 670 }, { "epoch": 0.533751962323391, "grad_norm": 39.92012986823895, "learning_rate": 2.6438416471154273e-07, "logits/chosen": 5907.2734375, "logits/rejected": 4907.10400390625, "logps/chosen": -405.2810974121094, "logps/rejected": -446.21258544921875, "loss": 0.5068, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -127.45320892333984, "rewards/margins": 83.75382995605469, "rewards/rejected": -211.20703125, "step": 680 }, { "epoch": 0.5416012558869702, "grad_norm": 46.642471943141544, "learning_rate": 2.5753758336186326e-07, "logits/chosen": 5805.8095703125, "logits/rejected": 5259.81884765625, "logps/chosen": -395.00201416015625, "logps/rejected": -498.8514099121094, "loss": 0.4588, "rewards/accuracies": 0.824999988079071, "rewards/chosen": -119.91290283203125, "rewards/margins": 101.47969818115234, "rewards/rejected": -221.39260864257812, "step": 690 }, { "epoch": 0.5494505494505495, "grad_norm": 36.996180293304306, "learning_rate": 2.5068533785312666e-07, "logits/chosen": 5561.36474609375, "logits/rejected": 5299.3701171875, "logps/chosen": -388.0168151855469, "logps/rejected": -469.75836181640625, "loss": 0.5334, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -137.9005584716797, "rewards/margins": 86.40620422363281, "rewards/rejected": -224.30679321289062, "step": 700 }, { "epoch": 0.5572998430141287, "grad_norm": 34.412694237392124, "learning_rate": 2.4383257734343794e-07, "logits/chosen": 5518.5048828125, "logits/rejected": 5457.7568359375, "logps/chosen": -380.4613342285156, "logps/rejected": -476.30926513671875, "loss": 0.4833, "rewards/accuracies": 0.7333333492279053, "rewards/chosen": -131.21548461914062, "rewards/margins": 81.00456237792969, "rewards/rejected": -212.2200469970703, "step": 710 }, { "epoch": 0.565149136577708, "grad_norm": 38.789896986449385, "learning_rate": 2.3698445137790258e-07, "logits/chosen": 5958.36669921875, "logits/rejected": 5054.3583984375, "logps/chosen": -387.704833984375, "logps/rejected": -457.67449951171875, "loss": 0.4845, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -102.9400405883789, "rewards/margins": 96.6159896850586, "rewards/rejected": -199.55604553222656, "step": 720 }, { "epoch": 0.5729984301412873, "grad_norm": 37.6613549240828, "learning_rate": 2.3014610601897157e-07, "logits/chosen": 6442.0546875, "logits/rejected": 4870.72998046875, "logps/chosen": -422.08062744140625, "logps/rejected": -468.485107421875, "loss": 0.4897, "rewards/accuracies": 0.8083333969116211, "rewards/chosen": -127.0215072631836, "rewards/margins": 101.17884826660156, "rewards/rejected": -228.2003631591797, "step": 730 }, { "epoch": 0.5808477237048666, "grad_norm": 38.25050494607591, "learning_rate": 2.2332267997940513e-07, "logits/chosen": 5338.6640625, "logits/rejected": 4488.05859375, "logps/chosen": -369.0237731933594, "logps/rejected": -435.61993408203125, "loss": 0.4733, "rewards/accuracies": 0.783333420753479, "rewards/chosen": -121.0157241821289, "rewards/margins": 97.05131530761719, "rewards/rejected": -218.06704711914062, "step": 740 }, { "epoch": 0.5886970172684458, "grad_norm": 42.057029091125465, "learning_rate": 2.1651930076075723e-07, "logits/chosen": 5821.0546875, "logits/rejected": 5192.505859375, "logps/chosen": -394.3180847167969, "logps/rejected": -459.5777893066406, "loss": 0.5111, "rewards/accuracies": 0.7333333492279053, "rewards/chosen": -154.05587768554688, "rewards/margins": 82.97196197509766, "rewards/rejected": -237.02786254882812, "step": 750 }, { "epoch": 0.5965463108320251, "grad_norm": 45.569071414781824, "learning_rate": 2.0974108080028692e-07, "logits/chosen": 6099.1826171875, "logits/rejected": 4732.19921875, "logps/chosen": -412.697998046875, "logps/rejected": -476.2620544433594, "loss": 0.489, "rewards/accuracies": 0.7833333611488342, "rewards/chosen": -149.33741760253906, "rewards/margins": 89.0047836303711, "rewards/rejected": -238.3422088623047, "step": 760 }, { "epoch": 0.6043956043956044, "grad_norm": 42.174510698937965, "learning_rate": 2.0299311362918773e-07, "logits/chosen": 6329.58447265625, "logits/rejected": 5368.06640625, "logps/chosen": -438.70660400390625, "logps/rejected": -522.0857543945312, "loss": 0.5203, "rewards/accuracies": 0.7666666507720947, "rewards/chosen": -144.0616455078125, "rewards/margins": 93.22587585449219, "rewards/rejected": -237.28750610351562, "step": 770 }, { "epoch": 0.6122448979591837, "grad_norm": 37.379579846735666, "learning_rate": 1.962804700450265e-07, "logits/chosen": 6095.6416015625, "logits/rejected": 5715.27587890625, "logps/chosen": -425.42803955078125, "logps/rejected": -554.23095703125, "loss": 0.5009, "rewards/accuracies": 0.7750000357627869, "rewards/chosen": -142.29542541503906, "rewards/margins": 97.37234497070312, "rewards/rejected": -239.66775512695312, "step": 780 }, { "epoch": 0.6200941915227629, "grad_norm": 60.675692600558975, "learning_rate": 1.8960819430126334e-07, "logits/chosen": 5744.94677734375, "logits/rejected": 5007.6689453125, "logps/chosen": -397.6462097167969, "logps/rejected": -508.2320251464844, "loss": 0.4679, "rewards/accuracies": 0.7916666865348816, "rewards/chosen": -143.19480895996094, "rewards/margins": 109.6915054321289, "rewards/rejected": -252.8863067626953, "step": 790 }, { "epoch": 0.6279434850863422, "grad_norm": 48.67749888199865, "learning_rate": 1.8298130031671972e-07, "logits/chosen": 5731.26025390625, "logits/rejected": 4972.1796875, "logps/chosen": -406.46832275390625, "logps/rejected": -478.52679443359375, "loss": 0.5167, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -132.73187255859375, "rewards/margins": 79.27778625488281, "rewards/rejected": -212.00967407226562, "step": 800 }, { "epoch": 0.6357927786499215, "grad_norm": 45.40931204746602, "learning_rate": 1.7640476790784075e-07, "logits/chosen": 5235.833984375, "logits/rejected": 4690.0458984375, "logps/chosen": -396.5758361816406, "logps/rejected": -500.0785217285156, "loss": 0.4995, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -122.74595642089844, "rewards/margins": 103.55721282958984, "rewards/rejected": -226.3031768798828, "step": 810 }, { "epoch": 0.6436420722135008, "grad_norm": 41.494824288393794, "learning_rate": 1.6988353904658492e-07, "logits/chosen": 5789.142578125, "logits/rejected": 4402.1337890625, "logps/chosen": -401.8763122558594, "logps/rejected": -431.6974182128906, "loss": 0.4829, "rewards/accuracies": 0.8083333969116211, "rewards/chosen": -118.8035659790039, "rewards/margins": 97.06875610351562, "rewards/rejected": -215.87234497070312, "step": 820 }, { "epoch": 0.6514913657770801, "grad_norm": 101.22612353434789, "learning_rate": 1.634225141467513e-07, "logits/chosen": 5653.84228515625, "logits/rejected": 5008.7412109375, "logps/chosen": -396.14337158203125, "logps/rejected": -494.45855712890625, "loss": 0.4855, "rewards/accuracies": 0.7666667103767395, "rewards/chosen": -132.68887329101562, "rewards/margins": 113.98445129394531, "rewards/rejected": -246.67333984375, "step": 830 }, { "epoch": 0.6593406593406593, "grad_norm": 35.01950929926434, "learning_rate": 1.570265483815364e-07, "logits/chosen": 6262.73828125, "logits/rejected": 5047.0439453125, "logps/chosen": -412.3899841308594, "logps/rejected": -499.49151611328125, "loss": 0.4918, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -126.85672760009766, "rewards/margins": 98.43955993652344, "rewards/rejected": -225.2962646484375, "step": 840 }, { "epoch": 0.6671899529042387, "grad_norm": 40.333680668673054, "learning_rate": 1.5070044803508691e-07, "logits/chosen": 5750.6259765625, "logits/rejected": 5139.3857421875, "logps/chosen": -403.19586181640625, "logps/rejected": -486.945556640625, "loss": 0.4473, "rewards/accuracies": 0.7750000357627869, "rewards/chosen": -130.6537322998047, "rewards/margins": 110.33387756347656, "rewards/rejected": -240.9875946044922, "step": 850 }, { "epoch": 0.6750392464678179, "grad_norm": 56.30127916699102, "learning_rate": 1.444489668907914e-07, "logits/chosen": 6237.0947265625, "logits/rejected": 5255.3623046875, "logps/chosen": -422.582275390625, "logps/rejected": -470.31689453125, "loss": 0.537, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -119.71390533447266, "rewards/margins": 96.29790496826172, "rewards/rejected": -216.01181030273438, "step": 860 }, { "epoch": 0.6828885400313972, "grad_norm": 35.84649453580378, "learning_rate": 1.3827680265902232e-07, "logits/chosen": 6196.35009765625, "logits/rejected": 5068.6572265625, "logps/chosen": -387.5716247558594, "logps/rejected": -438.8368225097656, "loss": 0.4975, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -98.15916442871094, "rewards/margins": 86.46854400634766, "rewards/rejected": -184.62771606445312, "step": 870 }, { "epoch": 0.6907378335949764, "grad_norm": 44.18256141777563, "learning_rate": 1.3218859344701632e-07, "logits/chosen": 5469.056640625, "logits/rejected": 5134.06396484375, "logps/chosen": -357.2699890136719, "logps/rejected": -475.69195556640625, "loss": 0.4973, "rewards/accuracies": 0.7916666269302368, "rewards/chosen": -98.51728820800781, "rewards/margins": 88.86700439453125, "rewards/rejected": -187.38429260253906, "step": 880 }, { "epoch": 0.6985871271585558, "grad_norm": 47.87001613986282, "learning_rate": 1.2618891427354172e-07, "logits/chosen": 6415.8408203125, "logits/rejected": 5204.595703125, "logps/chosen": -414.93890380859375, "logps/rejected": -467.7093200683594, "loss": 0.5158, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -102.4104232788086, "rewards/margins": 103.2691650390625, "rewards/rejected": -205.67959594726562, "step": 890 }, { "epoch": 0.706436420722135, "grad_norm": 50.09799621871588, "learning_rate": 1.202822736309758e-07, "logits/chosen": 5462.9189453125, "logits/rejected": 4944.328125, "logps/chosen": -380.93890380859375, "logps/rejected": -490.4107360839844, "loss": 0.4935, "rewards/accuracies": 0.7166666984558105, "rewards/chosen": -122.89737701416016, "rewards/margins": 103.0918197631836, "rewards/rejected": -225.9892120361328, "step": 900 }, { "epoch": 0.7142857142857143, "grad_norm": 44.49383467719737, "learning_rate": 1.1447311009737299e-07, "logits/chosen": 5359.6708984375, "logits/rejected": 5030.1162109375, "logps/chosen": -391.5655822753906, "logps/rejected": -487.87115478515625, "loss": 0.5249, "rewards/accuracies": 0.75, "rewards/chosen": -133.9981689453125, "rewards/margins": 95.89762115478516, "rewards/rejected": -229.89578247070312, "step": 910 }, { "epoch": 0.7221350078492935, "grad_norm": 39.13787414975052, "learning_rate": 1.0876578900107053e-07, "logits/chosen": 5915.595703125, "logits/rejected": 4905.36962890625, "logps/chosen": -409.89666748046875, "logps/rejected": -458.7411193847656, "loss": 0.4992, "rewards/accuracies": 0.8166667222976685, "rewards/chosen": -123.1749496459961, "rewards/margins": 90.61244201660156, "rewards/rejected": -213.78738403320312, "step": 920 }, { "epoch": 0.7299843014128728, "grad_norm": 28.45244015965264, "learning_rate": 1.0316459914033793e-07, "logits/chosen": 5881.65283203125, "logits/rejected": 4327.70654296875, "logps/chosen": -418.02374267578125, "logps/rejected": -453.83935546875, "loss": 0.4843, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -130.07017517089844, "rewards/margins": 90.82231903076172, "rewards/rejected": -220.89248657226562, "step": 930 }, { "epoch": 0.7378335949764521, "grad_norm": 46.849192838715155, "learning_rate": 9.767374956053584e-08, "logits/chosen": 5717.37744140625, "logits/rejected": 4960.6591796875, "logps/chosen": -393.51995849609375, "logps/rejected": -481.38726806640625, "loss": 0.496, "rewards/accuracies": 0.7500000596046448, "rewards/chosen": -128.65396118164062, "rewards/margins": 105.57264709472656, "rewards/rejected": -234.2265625, "step": 940 }, { "epoch": 0.7456828885400314, "grad_norm": 35.63937558116584, "learning_rate": 9.229736639120561e-08, "logits/chosen": 5887.68359375, "logits/rejected": 5351.68896484375, "logps/chosen": -399.4278869628906, "logps/rejected": -471.6551208496094, "loss": 0.5235, "rewards/accuracies": 0.7333333492279053, "rewards/chosen": -135.86773681640625, "rewards/margins": 72.99427795410156, "rewards/rejected": -208.8620147705078, "step": 950 }, { "epoch": 0.7535321821036107, "grad_norm": 47.32741994534299, "learning_rate": 8.70394897454659e-08, "logits/chosen": 5760.6650390625, "logits/rejected": 5091.91943359375, "logps/chosen": -379.76678466796875, "logps/rejected": -456.99853515625, "loss": 0.484, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -115.75325012207031, "rewards/margins": 94.92268371582031, "rewards/rejected": -210.67593383789062, "step": 960 }, { "epoch": 0.7613814756671899, "grad_norm": 46.48423774841541, "learning_rate": 8.19040706840472e-08, "logits/chosen": 5845.96728515625, "logits/rejected": 4831.09814453125, "logps/chosen": -416.8921813964844, "logps/rejected": -476.6322326660156, "loss": 0.4821, "rewards/accuracies": 0.7083333730697632, "rewards/chosen": -127.76072692871094, "rewards/margins": 102.53706359863281, "rewards/rejected": -230.2977752685547, "step": 970 }, { "epoch": 0.7692307692307693, "grad_norm": 37.56654693333076, "learning_rate": 7.689496824624525e-08, "logits/chosen": 5535.7060546875, "logits/rejected": 4366.2099609375, "logps/chosen": -409.868408203125, "logps/rejected": -479.47723388671875, "loss": 0.4729, "rewards/accuracies": 0.7833333015441895, "rewards/chosen": -133.22642517089844, "rewards/margins": 107.56368255615234, "rewards/rejected": -240.7901153564453, "step": 980 }, { "epoch": 0.7770800627943485, "grad_norm": 48.82961102031894, "learning_rate": 7.201594655002458e-08, "logits/chosen": 5900.552734375, "logits/rejected": 4841.35400390625, "logps/chosen": -401.5061950683594, "logps/rejected": -461.838623046875, "loss": 0.484, "rewards/accuracies": 0.75, "rewards/chosen": -133.3988037109375, "rewards/margins": 92.15519714355469, "rewards/rejected": -225.5540008544922, "step": 990 }, { "epoch": 0.7849293563579278, "grad_norm": 42.04075049035118, "learning_rate": 6.727067196345099e-08, "logits/chosen": 5537.10205078125, "logits/rejected": 4615.935546875, "logps/chosen": -386.819091796875, "logps/rejected": -433.0892639160156, "loss": 0.4939, "rewards/accuracies": 0.7333333492279053, "rewards/chosen": -126.4975357055664, "rewards/margins": 80.91593170166016, "rewards/rejected": -207.41348266601562, "step": 1000 }, { "epoch": 0.792778649921507, "grad_norm": 39.432930289735246, "learning_rate": 6.26627103495786e-08, "logits/chosen": 5749.5244140625, "logits/rejected": 4712.25537109375, "logps/chosen": -375.5830078125, "logps/rejected": -432.10296630859375, "loss": 0.5081, "rewards/accuracies": 0.7166666388511658, "rewards/chosen": -116.16764068603516, "rewards/margins": 80.42024993896484, "rewards/rejected": -196.58787536621094, "step": 1010 }, { "epoch": 0.8006279434850864, "grad_norm": 42.33512249774745, "learning_rate": 5.8195524386862374e-08, "logits/chosen": 5819.30517578125, "logits/rejected": 5078.79150390625, "logps/chosen": -402.74078369140625, "logps/rejected": -492.830810546875, "loss": 0.4861, "rewards/accuracies": 0.7916666865348816, "rewards/chosen": -104.84419250488281, "rewards/margins": 114.38609313964844, "rewards/rejected": -219.2303009033203, "step": 1020 }, { "epoch": 0.8084772370486656, "grad_norm": 37.608058989225746, "learning_rate": 5.38724709671092e-08, "logits/chosen": 6264.0732421875, "logits/rejected": 5862.0419921875, "logps/chosen": -397.46209716796875, "logps/rejected": -495.5181579589844, "loss": 0.4768, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -119.66192626953125, "rewards/margins": 100.7459945678711, "rewards/rejected": -220.4079132080078, "step": 1030 }, { "epoch": 0.8163265306122449, "grad_norm": 40.42806275113912, "learning_rate": 4.969679867292276e-08, "logits/chosen": 5519.99755859375, "logits/rejected": 4945.90625, "logps/chosen": -395.88787841796875, "logps/rejected": -486.75103759765625, "loss": 0.4898, "rewards/accuracies": 0.7500000596046448, "rewards/chosen": -130.64459228515625, "rewards/margins": 102.53266906738281, "rewards/rejected": -233.17724609375, "step": 1040 }, { "epoch": 0.8241758241758241, "grad_norm": 57.792951342438336, "learning_rate": 4.5671645336537416e-08, "logits/chosen": 5589.9228515625, "logits/rejected": 4998.0732421875, "logps/chosen": -410.640625, "logps/rejected": -494.5079040527344, "loss": 0.4813, "rewards/accuracies": 0.7333332896232605, "rewards/chosen": -129.10745239257812, "rewards/margins": 99.67637634277344, "rewards/rejected": -228.7838134765625, "step": 1050 }, { "epoch": 0.8320251177394035, "grad_norm": 65.93775771111166, "learning_rate": 4.180003568187776e-08, "logits/chosen": 6950.94384765625, "logits/rejected": 5396.08203125, "logps/chosen": -445.6783142089844, "logps/rejected": -476.76348876953125, "loss": 0.5107, "rewards/accuracies": 0.7166666984558105, "rewards/chosen": -137.7556610107422, "rewards/margins": 74.47281646728516, "rewards/rejected": -212.22848510742188, "step": 1060 }, { "epoch": 0.8398744113029827, "grad_norm": 38.70868851954098, "learning_rate": 3.8084879051612144e-08, "logits/chosen": 5740.13427734375, "logits/rejected": 5175.44384765625, "logps/chosen": -401.1766357421875, "logps/rejected": -458.88983154296875, "loss": 0.4926, "rewards/accuracies": 0.7416666746139526, "rewards/chosen": -128.2843017578125, "rewards/margins": 93.37026977539062, "rewards/rejected": -221.65457153320312, "step": 1070 }, { "epoch": 0.847723704866562, "grad_norm": 32.75162608252022, "learning_rate": 3.452896722091128e-08, "logits/chosen": 6315.4775390625, "logits/rejected": 4809.1796875, "logps/chosen": -437.409423828125, "logps/rejected": -468.0005798339844, "loss": 0.4586, "rewards/accuracies": 0.8000000715255737, "rewards/chosen": -124.92161560058594, "rewards/margins": 101.8378677368164, "rewards/rejected": -226.7594757080078, "step": 1080 }, { "epoch": 0.8555729984301413, "grad_norm": 33.97476418315894, "learning_rate": 3.11349722995527e-08, "logits/chosen": 6288.51708984375, "logits/rejected": 4695.376953125, "logps/chosen": -410.7005920410156, "logps/rejected": -468.73748779296875, "loss": 0.4817, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -136.9029083251953, "rewards/margins": 82.16574096679688, "rewards/rejected": -219.0686492919922, "step": 1090 }, { "epoch": 0.8634222919937206, "grad_norm": 41.247061444792465, "learning_rate": 2.7905444723949762e-08, "logits/chosen": 6125.61474609375, "logits/rejected": 5000.64013671875, "logps/chosen": -429.1371154785156, "logps/rejected": -488.61846923828125, "loss": 0.4933, "rewards/accuracies": 0.7666667699813843, "rewards/chosen": -145.91433715820312, "rewards/margins": 110.5950698852539, "rewards/rejected": -256.5094299316406, "step": 1100 }, { "epoch": 0.8712715855572999, "grad_norm": 47.62834662420522, "learning_rate": 2.484281134061142e-08, "logits/chosen": 6494.0771484375, "logits/rejected": 5178.26513671875, "logps/chosen": -447.42950439453125, "logps/rejected": -501.88555908203125, "loss": 0.4894, "rewards/accuracies": 0.7666666507720947, "rewards/chosen": -138.15509033203125, "rewards/margins": 101.00972747802734, "rewards/rejected": -239.1648406982422, "step": 1110 }, { "epoch": 0.8791208791208791, "grad_norm": 51.3299256992955, "learning_rate": 2.194937358247506e-08, "logits/chosen": 6401.00244140625, "logits/rejected": 5139.669921875, "logps/chosen": -431.32470703125, "logps/rejected": -499.81622314453125, "loss": 0.4705, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -139.98660278320312, "rewards/margins": 100.34617614746094, "rewards/rejected": -240.33279418945312, "step": 1120 }, { "epoch": 0.8869701726844584, "grad_norm": 43.96318864451059, "learning_rate": 1.9227305739481612e-08, "logits/chosen": 5825.98388671875, "logits/rejected": 4487.8232421875, "logps/chosen": -395.4400634765625, "logps/rejected": -447.91632080078125, "loss": 0.4715, "rewards/accuracies": 0.7666666507720947, "rewards/chosen": -116.87007141113281, "rewards/margins": 106.44403076171875, "rewards/rejected": -223.31411743164062, "step": 1130 }, { "epoch": 0.8948194662480377, "grad_norm": 42.23031030576211, "learning_rate": 1.6678653324693787e-08, "logits/chosen": 6448.05615234375, "logits/rejected": 5131.9482421875, "logps/chosen": -437.029296875, "logps/rejected": -490.83782958984375, "loss": 0.4488, "rewards/accuracies": 0.7666667103767395, "rewards/chosen": -131.5898895263672, "rewards/margins": 97.64103698730469, "rewards/rejected": -229.23092651367188, "step": 1140 }, { "epoch": 0.902668759811617, "grad_norm": 35.04452076773732, "learning_rate": 1.4305331537183384e-08, "logits/chosen": 5712.69140625, "logits/rejected": 5130.47265625, "logps/chosen": -399.71881103515625, "logps/rejected": -485.8260192871094, "loss": 0.4769, "rewards/accuracies": 0.7166666984558105, "rewards/chosen": -132.52346801757812, "rewards/margins": 89.83201599121094, "rewards/rejected": -222.35549926757812, "step": 1150 }, { "epoch": 0.9105180533751962, "grad_norm": 38.74863585720943, "learning_rate": 1.2109123822844653e-08, "logits/chosen": 5841.828125, "logits/rejected": 4531.7880859375, "logps/chosen": -410.2086486816406, "logps/rejected": -458.80633544921875, "loss": 0.4863, "rewards/accuracies": 0.7333333492279053, "rewards/chosen": -139.53079223632812, "rewards/margins": 85.2570571899414, "rewards/rejected": -224.787841796875, "step": 1160 }, { "epoch": 0.9183673469387755, "grad_norm": 41.11649731671983, "learning_rate": 1.0091680534213387e-08, "logits/chosen": 6367.6767578125, "logits/rejected": 6075.7880859375, "logps/chosen": -418.6693420410156, "logps/rejected": -533.9754638671875, "loss": 0.4898, "rewards/accuracies": 0.7333333492279053, "rewards/chosen": -126.54231262207031, "rewards/margins": 103.62506103515625, "rewards/rejected": -230.16738891601562, "step": 1170 }, { "epoch": 0.9262166405023547, "grad_norm": 45.99374365039285, "learning_rate": 8.254517690300944e-09, "logits/chosen": 5604.158203125, "logits/rejected": 4977.4833984375, "logps/chosen": -398.29681396484375, "logps/rejected": -477.4459533691406, "loss": 0.4865, "rewards/accuracies": 0.7416666746139526, "rewards/chosen": -119.34525299072266, "rewards/margins": 99.28761291503906, "rewards/rejected": -218.6328582763672, "step": 1180 }, { "epoch": 0.9340659340659341, "grad_norm": 31.427571461447283, "learning_rate": 6.599015837372907e-09, "logits/chosen": 6136.6103515625, "logits/rejected": 5255.3720703125, "logps/chosen": -446.5904235839844, "logps/rejected": -511.08197021484375, "loss": 0.494, "rewards/accuracies": 0.6833333969116211, "rewards/chosen": -154.27175903320312, "rewards/margins": 92.97373962402344, "rewards/rejected": -247.24551391601562, "step": 1190 }, { "epoch": 0.9419152276295133, "grad_norm": 44.49841411933075, "learning_rate": 5.126419011529992e-09, "logits/chosen": 6378.0693359375, "logits/rejected": 5315.1552734375, "logps/chosen": -425.67828369140625, "logps/rejected": -504.85064697265625, "loss": 0.4639, "rewards/accuracies": 0.8333333730697632, "rewards/chosen": -120.93017578125, "rewards/margins": 125.771484375, "rewards/rejected": -246.70166015625, "step": 1200 }, { "epoch": 0.9497645211930926, "grad_norm": 46.301946932033985, "learning_rate": 3.837833803870177e-09, "logits/chosen": 5976.0302734375, "logits/rejected": 5170.73095703125, "logps/chosen": -413.837158203125, "logps/rejected": -494.04986572265625, "loss": 0.4959, "rewards/accuracies": 0.7416666746139526, "rewards/chosen": -134.03421020507812, "rewards/margins": 107.6988296508789, "rewards/rejected": -241.73306274414062, "step": 1210 }, { "epoch": 0.957613814756672, "grad_norm": 60.40747566974286, "learning_rate": 2.734228528934679e-09, "logits/chosen": 7431.4658203125, "logits/rejected": 5369.63525390625, "logps/chosen": -483.7549743652344, "logps/rejected": -520.4745483398438, "loss": 0.4907, "rewards/accuracies": 0.7333332896232605, "rewards/chosen": -137.33944702148438, "rewards/margins": 97.83613586425781, "rewards/rejected": -235.1755828857422, "step": 1220 }, { "epoch": 0.9654631083202512, "grad_norm": 41.529280528162516, "learning_rate": 1.8164324970625645e-09, "logits/chosen": 6534.43408203125, "logits/rejected": 5027.76416015625, "logps/chosen": -434.41461181640625, "logps/rejected": -487.7671813964844, "loss": 0.5081, "rewards/accuracies": 0.7583333253860474, "rewards/chosen": -126.0592041015625, "rewards/margins": 96.64753723144531, "rewards/rejected": -222.7067413330078, "step": 1230 }, { "epoch": 0.9733124018838305, "grad_norm": 42.77933523830055, "learning_rate": 1.0851353912008642e-09, "logits/chosen": 5607.98681640625, "logits/rejected": 5125.9951171875, "logps/chosen": -416.8941345214844, "logps/rejected": -513.7186279296875, "loss": 0.4823, "rewards/accuracies": 0.7416666746139526, "rewards/chosen": -143.94058227539062, "rewards/margins": 93.5618667602539, "rewards/rejected": -237.50244140625, "step": 1240 }, { "epoch": 0.9811616954474097, "grad_norm": 37.72195768092037, "learning_rate": 5.408867486384471e-10, "logits/chosen": 5727.11474609375, "logits/rejected": 4746.82763671875, "logps/chosen": -387.53179931640625, "logps/rejected": -445.4607849121094, "loss": 0.4804, "rewards/accuracies": 0.800000011920929, "rewards/chosen": -117.09318542480469, "rewards/margins": 105.014892578125, "rewards/rejected": -222.1080780029297, "step": 1250 }, { "epoch": 0.989010989010989, "grad_norm": 46.396982025433076, "learning_rate": 1.840955480532924e-10, "logits/chosen": 5508.7177734375, "logits/rejected": 5095.494140625, "logps/chosen": -396.3477478027344, "logps/rejected": -483.79315185546875, "loss": 0.4672, "rewards/accuracies": 0.7666667103767395, "rewards/chosen": -129.69497680664062, "rewards/margins": 96.39173126220703, "rewards/rejected": -226.08670043945312, "step": 1260 }, { "epoch": 0.9968602825745683, "grad_norm": 44.77134550316371, "learning_rate": 1.502990218302247e-11, "logits/chosen": 5705.7900390625, "logits/rejected": 4541.423828125, "logps/chosen": -400.3255920410156, "logps/rejected": -467.81878662109375, "loss": 0.4832, "rewards/accuracies": 0.8083332777023315, "rewards/chosen": -138.85206604003906, "rewards/margins": 100.80841827392578, "rewards/rejected": -239.66049194335938, "step": 1270 }, { "epoch": 1.0, "step": 1274, "total_flos": 0.0, "train_loss": 0.5271364345655336, "train_runtime": 13001.8761, "train_samples_per_second": 4.702, "train_steps_per_second": 0.098 } ], "logging_steps": 10, "max_steps": 1274, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 2000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": false, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 6, "trial_name": null, "trial_params": null }