{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 491, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1e-08, "logits/chosen": -2.6244959831237793, "logits/rejected": -2.5476584434509277, "logps/chosen": -148.28712463378906, "logps/rejected": -97.07695007324219, "loss": 0.6931, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.02, "learning_rate": 1e-07, "logits/chosen": -2.5370163917541504, "logits/rejected": -2.4581496715545654, "logps/chosen": -149.06556701660156, "logps/rejected": -90.43597412109375, "loss": 0.681, "rewards/accuracies": 0.5833333134651184, "rewards/chosen": 0.03589768707752228, "rewards/margins": 0.027119608595967293, "rewards/rejected": 0.008778076618909836, "step": 10 }, { "epoch": 0.04, "learning_rate": 2e-07, "logits/chosen": -2.5408926010131836, "logits/rejected": -2.4895992279052734, "logps/chosen": -144.8271484375, "logps/rejected": -86.21273040771484, "loss": 0.544, "rewards/accuracies": 0.875, "rewards/chosen": 0.26239949464797974, "rewards/margins": 0.43117108941078186, "rewards/rejected": -0.1687716245651245, "step": 20 }, { "epoch": 0.06, "learning_rate": 3e-07, "logits/chosen": -2.6375694274902344, "logits/rejected": -2.5197739601135254, "logps/chosen": -139.70712280273438, "logps/rejected": -95.44498443603516, "loss": 0.3616, "rewards/accuracies": 0.800000011920929, "rewards/chosen": 0.05464474484324455, "rewards/margins": 1.1275455951690674, "rewards/rejected": -1.0729008913040161, "step": 30 }, { "epoch": 0.08, "learning_rate": 4e-07, "logits/chosen": -2.462580442428589, "logits/rejected": -2.3860220909118652, "logps/chosen": -139.60357666015625, "logps/rejected": -104.787841796875, "loss": 0.2921, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": 0.09050186723470688, "rewards/margins": 1.936102271080017, "rewards/rejected": -1.8456003665924072, "step": 40 }, { "epoch": 0.1, "learning_rate": 5e-07, "logits/chosen": -2.497800588607788, "logits/rejected": -2.331794023513794, "logps/chosen": -176.67684936523438, "logps/rejected": -133.3009033203125, "loss": 0.2627, "rewards/accuracies": 0.862500011920929, "rewards/chosen": -0.7371834516525269, "rewards/margins": 2.163860321044922, "rewards/rejected": -2.9010438919067383, "step": 50 }, { "epoch": 0.12, "learning_rate": 4.886621315192743e-07, "logits/chosen": -2.5514307022094727, "logits/rejected": -2.447479486465454, "logps/chosen": -160.9045867919922, "logps/rejected": -135.33114624023438, "loss": 0.2211, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -0.9471993446350098, "rewards/margins": 2.4198856353759766, "rewards/rejected": -3.3670849800109863, "step": 60 }, { "epoch": 0.14, "learning_rate": 4.773242630385487e-07, "logits/chosen": -2.5035057067871094, "logits/rejected": -2.3295655250549316, "logps/chosen": -162.2573699951172, "logps/rejected": -130.7346954345703, "loss": 0.1829, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -0.5212399363517761, "rewards/margins": 3.431070327758789, "rewards/rejected": -3.9523098468780518, "step": 70 }, { "epoch": 0.16, "learning_rate": 4.6598639455782313e-07, "logits/chosen": -2.522507905960083, "logits/rejected": -2.3813233375549316, "logps/chosen": -178.6873779296875, "logps/rejected": -141.0321502685547, "loss": 0.1822, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -0.6599360704421997, "rewards/margins": 3.214824676513672, "rewards/rejected": -3.8747611045837402, "step": 80 }, { "epoch": 0.18, "learning_rate": 4.546485260770975e-07, "logits/chosen": -2.5989696979522705, "logits/rejected": -2.4465830326080322, "logps/chosen": -167.22085571289062, "logps/rejected": -131.60641479492188, "loss": 0.1684, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -0.6895953416824341, "rewards/margins": 3.479625701904297, "rewards/rejected": -4.169220924377441, "step": 90 }, { "epoch": 0.2, "learning_rate": 4.433106575963719e-07, "logits/chosen": -2.3363890647888184, "logits/rejected": -2.208930730819702, "logps/chosen": -226.7657470703125, "logps/rejected": -155.46226501464844, "loss": 1.378, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -6.209970951080322, "rewards/margins": -1.1396101713180542, "rewards/rejected": -5.070359230041504, "step": 100 }, { "epoch": 0.22, "learning_rate": 4.3197278911564623e-07, "logits/chosen": -2.2915709018707275, "logits/rejected": -2.2031328678131104, "logps/chosen": -154.38711547851562, "logps/rejected": -138.17401123046875, "loss": 0.1965, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -1.3441089391708374, "rewards/margins": 3.5139331817626953, "rewards/rejected": -4.858041286468506, "step": 110 }, { "epoch": 0.24, "learning_rate": 4.2063492063492063e-07, "logits/chosen": -2.2738468647003174, "logits/rejected": -2.1098406314849854, "logps/chosen": -156.9755401611328, "logps/rejected": -132.60494995117188, "loss": 0.176, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -1.06258225440979, "rewards/margins": 3.5116379261016846, "rewards/rejected": -4.574219703674316, "step": 120 }, { "epoch": 0.26, "learning_rate": 4.09297052154195e-07, "logits/chosen": -2.212961196899414, "logits/rejected": -2.104160785675049, "logps/chosen": -155.2751007080078, "logps/rejected": -145.32325744628906, "loss": 0.1783, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -1.956811547279358, "rewards/margins": 3.8180172443389893, "rewards/rejected": -5.774828910827637, "step": 130 }, { "epoch": 0.29, "learning_rate": 3.979591836734694e-07, "logits/chosen": -2.24934983253479, "logits/rejected": -2.0988452434539795, "logps/chosen": -164.42202758789062, "logps/rejected": -147.06228637695312, "loss": 0.182, "rewards/accuracies": 0.9375, "rewards/chosen": -1.2399647235870361, "rewards/margins": 4.234730243682861, "rewards/rejected": -5.474694728851318, "step": 140 }, { "epoch": 0.31, "learning_rate": 3.866213151927438e-07, "logits/chosen": -2.159578800201416, "logits/rejected": -2.0544795989990234, "logps/chosen": -145.58526611328125, "logps/rejected": -140.1253662109375, "loss": 0.1431, "rewards/accuracies": 0.9375, "rewards/chosen": -1.122462511062622, "rewards/margins": 4.34097146987915, "rewards/rejected": -5.463434219360352, "step": 150 }, { "epoch": 0.33, "learning_rate": 3.7528344671201813e-07, "logits/chosen": -2.339341640472412, "logits/rejected": -2.1767287254333496, "logps/chosen": -162.39453125, "logps/rejected": -140.1338348388672, "loss": 0.219, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -1.2674634456634521, "rewards/margins": 4.108874320983887, "rewards/rejected": -5.376338005065918, "step": 160 }, { "epoch": 0.35, "learning_rate": 3.6394557823129254e-07, "logits/chosen": -2.256037712097168, "logits/rejected": -2.0874805450439453, "logps/chosen": -171.171142578125, "logps/rejected": -147.9696044921875, "loss": 0.1341, "rewards/accuracies": 0.9375, "rewards/chosen": -1.6500320434570312, "rewards/margins": 4.346782684326172, "rewards/rejected": -5.996814250946045, "step": 170 }, { "epoch": 0.37, "learning_rate": 3.526077097505669e-07, "logits/chosen": -2.244986057281494, "logits/rejected": -2.1053173542022705, "logps/chosen": -162.7403564453125, "logps/rejected": -139.0584259033203, "loss": 0.1669, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -1.3248670101165771, "rewards/margins": 4.133967399597168, "rewards/rejected": -5.458834171295166, "step": 180 }, { "epoch": 0.39, "learning_rate": 3.412698412698413e-07, "logits/chosen": -2.294400453567505, "logits/rejected": -2.063782215118408, "logps/chosen": -177.40701293945312, "logps/rejected": -188.0184326171875, "loss": 0.1111, "rewards/accuracies": 1.0, "rewards/chosen": -1.5336018800735474, "rewards/margins": 7.072530269622803, "rewards/rejected": -8.606132507324219, "step": 190 }, { "epoch": 0.41, "learning_rate": 3.299319727891156e-07, "logits/chosen": -2.2452657222747803, "logits/rejected": -2.1023104190826416, "logps/chosen": -164.32254028320312, "logps/rejected": -154.12310791015625, "loss": 0.1856, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -2.1261470317840576, "rewards/margins": 4.135893821716309, "rewards/rejected": -6.2620415687561035, "step": 200 }, { "epoch": 0.43, "learning_rate": 3.1859410430839e-07, "logits/chosen": -2.3352227210998535, "logits/rejected": -2.1405560970306396, "logps/chosen": -154.53152465820312, "logps/rejected": -134.69961547851562, "loss": 0.1332, "rewards/accuracies": 0.9375, "rewards/chosen": -0.9065567255020142, "rewards/margins": 4.494719505310059, "rewards/rejected": -5.401275634765625, "step": 210 }, { "epoch": 0.45, "learning_rate": 3.072562358276644e-07, "logits/chosen": -2.3480963706970215, "logits/rejected": -2.1483328342437744, "logps/chosen": -159.91603088378906, "logps/rejected": -152.38380432128906, "loss": 0.1641, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -2.00056791305542, "rewards/margins": 4.964018821716309, "rewards/rejected": -6.9645867347717285, "step": 220 }, { "epoch": 0.47, "learning_rate": 2.9591836734693874e-07, "logits/chosen": -2.320943832397461, "logits/rejected": -2.178292751312256, "logps/chosen": -164.52239990234375, "logps/rejected": -156.4569854736328, "loss": 0.1432, "rewards/accuracies": 0.9375, "rewards/chosen": -1.483763337135315, "rewards/margins": 4.69757080078125, "rewards/rejected": -6.181333541870117, "step": 230 }, { "epoch": 0.49, "learning_rate": 2.8458049886621314e-07, "logits/chosen": -2.2857508659362793, "logits/rejected": -2.122419834136963, "logps/chosen": -148.7211456298828, "logps/rejected": -134.5128631591797, "loss": 0.1565, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -1.2494453191757202, "rewards/margins": 4.770277500152588, "rewards/rejected": -6.0197224617004395, "step": 240 }, { "epoch": 0.51, "learning_rate": 2.732426303854875e-07, "logits/chosen": -2.3791332244873047, "logits/rejected": -2.23042893409729, "logps/chosen": -174.9046173095703, "logps/rejected": -174.52374267578125, "loss": 0.1483, "rewards/accuracies": 0.949999988079071, "rewards/chosen": -1.7448902130126953, "rewards/margins": 4.780519008636475, "rewards/rejected": -6.525409698486328, "step": 250 }, { "epoch": 0.53, "learning_rate": 2.619047619047619e-07, "logits/chosen": -2.3673877716064453, "logits/rejected": -2.20123291015625, "logps/chosen": -151.64413452148438, "logps/rejected": -145.02633666992188, "loss": 0.15, "rewards/accuracies": 0.875, "rewards/chosen": -2.0061161518096924, "rewards/margins": 4.6594319343566895, "rewards/rejected": -6.665547847747803, "step": 260 }, { "epoch": 0.55, "learning_rate": 2.5056689342403624e-07, "logits/chosen": -2.236260414123535, "logits/rejected": -2.155104398727417, "logps/chosen": -166.03179931640625, "logps/rejected": -147.90829467773438, "loss": 0.1281, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -1.2687690258026123, "rewards/margins": 4.856295585632324, "rewards/rejected": -6.125064373016357, "step": 270 }, { "epoch": 0.57, "learning_rate": 2.3922902494331064e-07, "logits/chosen": -2.307720184326172, "logits/rejected": -2.1552658081054688, "logps/chosen": -177.2270965576172, "logps/rejected": -161.21737670898438, "loss": 0.1423, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -1.5429637432098389, "rewards/margins": 4.64047384262085, "rewards/rejected": -6.183438301086426, "step": 280 }, { "epoch": 0.59, "learning_rate": 2.2789115646258502e-07, "logits/chosen": -2.2872962951660156, "logits/rejected": -2.111124277114868, "logps/chosen": -151.91629028320312, "logps/rejected": -139.26318359375, "loss": 0.1764, "rewards/accuracies": 0.875, "rewards/chosen": -1.7618621587753296, "rewards/margins": 4.707967758178711, "rewards/rejected": -6.469829559326172, "step": 290 }, { "epoch": 0.61, "learning_rate": 2.165532879818594e-07, "logits/chosen": -2.2950973510742188, "logits/rejected": -2.1110997200012207, "logps/chosen": -151.4035186767578, "logps/rejected": -144.46658325195312, "loss": 0.1984, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -1.0890811681747437, "rewards/margins": 4.241808891296387, "rewards/rejected": -5.330889701843262, "step": 300 }, { "epoch": 0.63, "learning_rate": 2.0521541950113377e-07, "logits/chosen": -2.371819019317627, "logits/rejected": -2.220020294189453, "logps/chosen": -132.87916564941406, "logps/rejected": -135.39962768554688, "loss": 0.1227, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -1.924085259437561, "rewards/margins": 5.181375980377197, "rewards/rejected": -7.105461120605469, "step": 310 }, { "epoch": 0.65, "learning_rate": 1.9387755102040814e-07, "logits/chosen": -2.405116081237793, "logits/rejected": -2.2629637718200684, "logps/chosen": -160.4700164794922, "logps/rejected": -150.7443084716797, "loss": 0.137, "rewards/accuracies": 0.8999999761581421, "rewards/chosen": -2.134606122970581, "rewards/margins": 4.591971397399902, "rewards/rejected": -6.7265777587890625, "step": 320 }, { "epoch": 0.67, "learning_rate": 1.8253968253968252e-07, "logits/chosen": -2.368983745574951, "logits/rejected": -2.2834761142730713, "logps/chosen": -160.4642791748047, "logps/rejected": -160.0625762939453, "loss": 0.1244, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -1.8051602840423584, "rewards/margins": 4.877270698547363, "rewards/rejected": -6.682431221008301, "step": 330 }, { "epoch": 0.69, "learning_rate": 1.712018140589569e-07, "logits/chosen": -2.3076577186584473, "logits/rejected": -2.1015944480895996, "logps/chosen": -174.94973754882812, "logps/rejected": -178.3339080810547, "loss": 0.1237, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -2.0153822898864746, "rewards/margins": 6.191023349761963, "rewards/rejected": -8.206406593322754, "step": 340 }, { "epoch": 0.71, "learning_rate": 1.598639455782313e-07, "logits/chosen": -2.5221567153930664, "logits/rejected": -2.3901963233947754, "logps/chosen": -159.98475646972656, "logps/rejected": -157.09149169921875, "loss": 0.0978, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -1.7989521026611328, "rewards/margins": 4.653882026672363, "rewards/rejected": -6.452833652496338, "step": 350 }, { "epoch": 0.73, "learning_rate": 1.4852607709750567e-07, "logits/chosen": -2.4620490074157715, "logits/rejected": -2.253587245941162, "logps/chosen": -153.6101531982422, "logps/rejected": -151.8799285888672, "loss": 0.143, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -1.7074544429779053, "rewards/margins": 5.599391460418701, "rewards/rejected": -7.306845664978027, "step": 360 }, { "epoch": 0.75, "learning_rate": 1.3718820861678005e-07, "logits/chosen": -2.3479745388031006, "logits/rejected": -2.232555627822876, "logps/chosen": -166.34698486328125, "logps/rejected": -162.51980590820312, "loss": 0.1679, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -1.8680267333984375, "rewards/margins": 4.728573799133301, "rewards/rejected": -6.596600532531738, "step": 370 }, { "epoch": 0.77, "learning_rate": 1.2585034013605442e-07, "logits/chosen": -2.452205181121826, "logits/rejected": -2.236445903778076, "logps/chosen": -160.61105346679688, "logps/rejected": -151.0522918701172, "loss": 0.1506, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -1.6947262287139893, "rewards/margins": 4.786589622497559, "rewards/rejected": -6.481315612792969, "step": 380 }, { "epoch": 0.79, "learning_rate": 1.1451247165532879e-07, "logits/chosen": -2.442394971847534, "logits/rejected": -2.2332167625427246, "logps/chosen": -170.10598754882812, "logps/rejected": -160.43565368652344, "loss": 0.101, "rewards/accuracies": 0.9375, "rewards/chosen": -1.8190062046051025, "rewards/margins": 5.077141761779785, "rewards/rejected": -6.896147727966309, "step": 390 }, { "epoch": 0.81, "learning_rate": 1.0317460317460316e-07, "logits/chosen": -2.4227921962738037, "logits/rejected": -2.2378575801849365, "logps/chosen": -182.52713012695312, "logps/rejected": -169.83901977539062, "loss": 0.086, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -1.6897770166397095, "rewards/margins": 5.666071891784668, "rewards/rejected": -7.3558478355407715, "step": 400 }, { "epoch": 0.84, "learning_rate": 9.183673469387755e-08, "logits/chosen": -2.3523762226104736, "logits/rejected": -2.1948838233947754, "logps/chosen": -169.2480926513672, "logps/rejected": -179.06275939941406, "loss": 0.1139, "rewards/accuracies": 0.9624999761581421, "rewards/chosen": -1.6853822469711304, "rewards/margins": 7.397521018981934, "rewards/rejected": -9.082902908325195, "step": 410 }, { "epoch": 0.86, "learning_rate": 8.049886621315193e-08, "logits/chosen": -2.3608529567718506, "logits/rejected": -2.246788501739502, "logps/chosen": -165.41151428222656, "logps/rejected": -169.1031494140625, "loss": 0.1225, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -1.6766393184661865, "rewards/margins": 5.58035945892334, "rewards/rejected": -7.2569990158081055, "step": 420 }, { "epoch": 0.88, "learning_rate": 6.91609977324263e-08, "logits/chosen": -2.404496908187866, "logits/rejected": -2.184279441833496, "logps/chosen": -183.42774963378906, "logps/rejected": -190.3188934326172, "loss": 0.1032, "rewards/accuracies": 0.9375, "rewards/chosen": -1.3860682249069214, "rewards/margins": 7.569392204284668, "rewards/rejected": -8.955459594726562, "step": 430 }, { "epoch": 0.9, "learning_rate": 5.782312925170068e-08, "logits/chosen": -2.285003662109375, "logits/rejected": -2.115624189376831, "logps/chosen": -130.6668243408203, "logps/rejected": -122.2939453125, "loss": 0.1088, "rewards/accuracies": 0.9125000238418579, "rewards/chosen": -1.462835669517517, "rewards/margins": 4.940153121948242, "rewards/rejected": -6.402989387512207, "step": 440 }, { "epoch": 0.92, "learning_rate": 4.648526077097505e-08, "logits/chosen": -2.3510639667510986, "logits/rejected": -2.217038631439209, "logps/chosen": -164.84243774414062, "logps/rejected": -159.15789794921875, "loss": 0.0928, "rewards/accuracies": 0.887499988079071, "rewards/chosen": -1.6966283321380615, "rewards/margins": 5.2207136154174805, "rewards/rejected": -6.917341709136963, "step": 450 }, { "epoch": 0.94, "learning_rate": 3.5147392290249434e-08, "logits/chosen": -2.392794132232666, "logits/rejected": -2.2205255031585693, "logps/chosen": -174.13099670410156, "logps/rejected": -163.31275939941406, "loss": 0.1361, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -1.1660579442977905, "rewards/margins": 5.76992130279541, "rewards/rejected": -6.93597936630249, "step": 460 }, { "epoch": 0.96, "learning_rate": 2.3809523809523807e-08, "logits/chosen": -2.3290793895721436, "logits/rejected": -2.172487735748291, "logps/chosen": -172.22470092773438, "logps/rejected": -166.8828125, "loss": 0.1397, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -2.161699056625366, "rewards/margins": 5.623523712158203, "rewards/rejected": -7.78522253036499, "step": 470 }, { "epoch": 0.98, "learning_rate": 1.2471655328798186e-08, "logits/chosen": -2.4058690071105957, "logits/rejected": -2.255786418914795, "logps/chosen": -177.1675262451172, "logps/rejected": -168.27767944335938, "loss": 0.1171, "rewards/accuracies": 0.925000011920929, "rewards/chosen": -1.5147063732147217, "rewards/margins": 5.647317886352539, "rewards/rejected": -7.16202449798584, "step": 480 }, { "epoch": 1.0, "learning_rate": 1.1337868480725624e-09, "logits/chosen": -2.3738300800323486, "logits/rejected": -2.2069907188415527, "logps/chosen": -154.16586303710938, "logps/rejected": -149.5966033935547, "loss": 0.1296, "rewards/accuracies": 0.9750000238418579, "rewards/chosen": -1.6616309881210327, "rewards/margins": 5.4048004150390625, "rewards/rejected": -7.066430568695068, "step": 490 }, { "epoch": 1.0, "step": 491, "total_flos": 0.0, "train_loss": 0.20032393871645335, "train_runtime": 8133.5998, "train_samples_per_second": 1.928, "train_steps_per_second": 0.06 } ], "logging_steps": 10, "max_steps": 491, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 500, "total_flos": 0.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }