| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "eval_steps": 1000, | |
| "global_step": 1145, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004366812227074236, | |
| "grad_norm": 0.47458936036527805, | |
| "learning_rate": 4.347826086956522e-08, | |
| "logits/chosen": -1.077685832977295, | |
| "logits/rejected": -0.9262686967849731, | |
| "logps/chosen": -270.19927978515625, | |
| "logps/rejected": -284.0829772949219, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.043668122270742356, | |
| "grad_norm": 0.523539374370341, | |
| "learning_rate": 4.347826086956522e-07, | |
| "logits/chosen": -0.9952558875083923, | |
| "logits/rejected": -1.0596376657485962, | |
| "logps/chosen": -300.6241760253906, | |
| "logps/rejected": -255.2275390625, | |
| "loss": 0.6929, | |
| "rewards/accuracies": 0.4583333432674408, | |
| "rewards/chosen": 0.0006611866410821676, | |
| "rewards/margins": 0.0006012354861013591, | |
| "rewards/rejected": 5.995116225676611e-05, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.08733624454148471, | |
| "grad_norm": 0.5365126766074364, | |
| "learning_rate": 8.695652173913044e-07, | |
| "logits/chosen": -1.1060526371002197, | |
| "logits/rejected": -1.0136970281600952, | |
| "logps/chosen": -282.16522216796875, | |
| "logps/rejected": -316.65264892578125, | |
| "loss": 0.6925, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.0014794444432482123, | |
| "rewards/margins": 0.0008522253483533859, | |
| "rewards/rejected": 0.0006272192695178092, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.13100436681222707, | |
| "grad_norm": 0.580247902784444, | |
| "learning_rate": 1.3043478260869566e-06, | |
| "logits/chosen": -1.0645825862884521, | |
| "logits/rejected": -1.1114238500595093, | |
| "logps/chosen": -287.91937255859375, | |
| "logps/rejected": -264.2998352050781, | |
| "loss": 0.69, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 0.00400827219709754, | |
| "rewards/margins": 0.006094515323638916, | |
| "rewards/rejected": -0.002086243825033307, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.17467248908296942, | |
| "grad_norm": 0.45391259957771296, | |
| "learning_rate": 1.7391304347826088e-06, | |
| "logits/chosen": -1.0716311931610107, | |
| "logits/rejected": -1.1120277643203735, | |
| "logps/chosen": -285.3992919921875, | |
| "logps/rejected": -272.4431457519531, | |
| "loss": 0.6833, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": 0.008399474434554577, | |
| "rewards/margins": 0.01975000463426113, | |
| "rewards/rejected": -0.011350530199706554, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.2183406113537118, | |
| "grad_norm": 0.46277317667378487, | |
| "learning_rate": 2.173913043478261e-06, | |
| "logits/chosen": -1.111750841140747, | |
| "logits/rejected": -1.0212160348892212, | |
| "logps/chosen": -264.515380859375, | |
| "logps/rejected": -304.44281005859375, | |
| "loss": 0.6728, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.018858566880226135, | |
| "rewards/margins": 0.04448834806680679, | |
| "rewards/rejected": -0.02562977746129036, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.26200873362445415, | |
| "grad_norm": 0.538292399167113, | |
| "learning_rate": 2.6086956521739132e-06, | |
| "logits/chosen": -1.070939302444458, | |
| "logits/rejected": -1.1072248220443726, | |
| "logps/chosen": -285.0442810058594, | |
| "logps/rejected": -262.81597900390625, | |
| "loss": 0.654, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": 0.03614560887217522, | |
| "rewards/margins": 0.09129359573125839, | |
| "rewards/rejected": -0.05514797568321228, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.3056768558951965, | |
| "grad_norm": 0.8400838184943603, | |
| "learning_rate": 3.043478260869566e-06, | |
| "logits/chosen": -1.1011688709259033, | |
| "logits/rejected": -1.1285476684570312, | |
| "logps/chosen": -279.41290283203125, | |
| "logps/rejected": -276.6390686035156, | |
| "loss": 0.6152, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": 0.04447701945900917, | |
| "rewards/margins": 0.16786250472068787, | |
| "rewards/rejected": -0.123385488986969, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.34934497816593885, | |
| "grad_norm": 1.363956535309491, | |
| "learning_rate": 3.4782608695652175e-06, | |
| "logits/chosen": -1.1735438108444214, | |
| "logits/rejected": -1.0770442485809326, | |
| "logps/chosen": -256.97564697265625, | |
| "logps/rejected": -330.7611083984375, | |
| "loss": 0.514, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.10425807535648346, | |
| "rewards/margins": 0.4447970986366272, | |
| "rewards/rejected": -0.34053897857666016, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.3930131004366812, | |
| "grad_norm": 1.0652559890564237, | |
| "learning_rate": 3.91304347826087e-06, | |
| "logits/chosen": -1.0946518182754517, | |
| "logits/rejected": -1.0877329111099243, | |
| "logps/chosen": -266.5739440917969, | |
| "logps/rejected": -361.3409729003906, | |
| "loss": 0.278, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.2085898369550705, | |
| "rewards/margins": 1.1870415210723877, | |
| "rewards/rejected": -0.9784518480300903, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.4366812227074236, | |
| "grad_norm": 0.3433521660739356, | |
| "learning_rate": 4.347826086956522e-06, | |
| "logits/chosen": -1.055832862854004, | |
| "logits/rejected": -0.9789533615112305, | |
| "logps/chosen": -219.6240692138672, | |
| "logps/rejected": -508.79132080078125, | |
| "loss": 0.0735, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.38825297355651855, | |
| "rewards/margins": 2.7440783977508545, | |
| "rewards/rejected": -2.355825662612915, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.48034934497816595, | |
| "grad_norm": 0.11711005718135792, | |
| "learning_rate": 4.782608695652174e-06, | |
| "logits/chosen": -0.9485265016555786, | |
| "logits/rejected": -0.9038203358650208, | |
| "logps/chosen": -269.6318664550781, | |
| "logps/rejected": -716.6309814453125, | |
| "loss": 0.018, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.2308163344860077, | |
| "rewards/margins": 4.760537147521973, | |
| "rewards/rejected": -4.529721260070801, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.5240174672489083, | |
| "grad_norm": 0.056145921913412625, | |
| "learning_rate": 4.999709285361594e-06, | |
| "logits/chosen": -0.9236040115356445, | |
| "logits/rejected": -0.8355329632759094, | |
| "logps/chosen": -271.9773254394531, | |
| "logps/rejected": -864.2019653320312, | |
| "loss": 0.0057, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.2535458505153656, | |
| "rewards/margins": 6.195461273193359, | |
| "rewards/rejected": -5.941915035247803, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.5676855895196506, | |
| "grad_norm": 0.03098316075816059, | |
| "learning_rate": 4.997383973910631e-06, | |
| "logits/chosen": -0.8370451927185059, | |
| "logits/rejected": -0.6872082948684692, | |
| "logps/chosen": -263.49090576171875, | |
| "logps/rejected": -959.0076293945312, | |
| "loss": 0.0029, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.2465527057647705, | |
| "rewards/margins": 7.134703159332275, | |
| "rewards/rejected": -6.888150215148926, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.611353711790393, | |
| "grad_norm": 0.01781004140188894, | |
| "learning_rate": 4.9927355140895775e-06, | |
| "logits/chosen": -0.8542898297309875, | |
| "logits/rejected": -0.6182370185852051, | |
| "logps/chosen": -256.8946533203125, | |
| "logps/rejected": -1113.857666015625, | |
| "loss": 0.002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.1456744223833084, | |
| "rewards/margins": 8.21844482421875, | |
| "rewards/rejected": -8.072771072387695, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6550218340611353, | |
| "grad_norm": 0.0626793578117912, | |
| "learning_rate": 4.985768230048011e-06, | |
| "logits/chosen": -0.8041971325874329, | |
| "logits/rejected": -0.645085334777832, | |
| "logps/chosen": -274.4373779296875, | |
| "logps/rejected": -1066.285400390625, | |
| "loss": 0.0019, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.16038499772548676, | |
| "rewards/margins": 8.415651321411133, | |
| "rewards/rejected": -8.255266189575195, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6986899563318777, | |
| "grad_norm": 0.016666200992927045, | |
| "learning_rate": 4.976488602981748e-06, | |
| "logits/chosen": -0.8367046117782593, | |
| "logits/rejected": -0.6429373025894165, | |
| "logps/chosen": -258.5212707519531, | |
| "logps/rejected": -1201.1590576171875, | |
| "loss": 0.001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.2543124556541443, | |
| "rewards/margins": 9.140569686889648, | |
| "rewards/rejected": -8.88625717163086, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.74235807860262, | |
| "grad_norm": 0.07994746225935201, | |
| "learning_rate": 4.9649052651038255e-06, | |
| "logits/chosen": -0.7291176915168762, | |
| "logits/rejected": -0.5780690908432007, | |
| "logps/chosen": -279.2346496582031, | |
| "logps/rejected": -1126.931396484375, | |
| "loss": 0.0009, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.16093984246253967, | |
| "rewards/margins": 8.925050735473633, | |
| "rewards/rejected": -8.764110565185547, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.7860262008733624, | |
| "grad_norm": 0.01171334168304518, | |
| "learning_rate": 4.9510289916145295e-06, | |
| "logits/chosen": -0.7453992962837219, | |
| "logits/rejected": -0.5254510641098022, | |
| "logps/chosen": -256.3668518066406, | |
| "logps/rejected": -1209.734619140625, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.1778395026922226, | |
| "rewards/margins": 9.462778091430664, | |
| "rewards/rejected": -9.28493881225586, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.8296943231441049, | |
| "grad_norm": 0.017372174016347157, | |
| "learning_rate": 4.934872690677953e-06, | |
| "logits/chosen": -0.7389127612113953, | |
| "logits/rejected": -0.5030776858329773, | |
| "logps/chosen": -271.6781921386719, | |
| "logps/rejected": -1233.863037109375, | |
| "loss": 0.0007, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.13182204961776733, | |
| "rewards/margins": 9.499404907226562, | |
| "rewards/rejected": -9.367582321166992, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.8733624454148472, | |
| "grad_norm": 0.005099638792861995, | |
| "learning_rate": 4.9164513914144005e-06, | |
| "logits/chosen": -0.6438565254211426, | |
| "logits/rejected": -0.3725091516971588, | |
| "logps/chosen": -258.71893310546875, | |
| "logps/rejected": -1232.365478515625, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.11995329707860947, | |
| "rewards/margins": 9.731719017028809, | |
| "rewards/rejected": -9.611766815185547, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.9170305676855895, | |
| "grad_norm": 0.007449286013455516, | |
| "learning_rate": 4.8957822299198045e-06, | |
| "logits/chosen": -0.7947207689285278, | |
| "logits/rejected": -0.5452377200126648, | |
| "logps/chosen": -277.94647216796875, | |
| "logps/rejected": -1344.936767578125, | |
| "loss": 0.0006, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.07659594714641571, | |
| "rewards/margins": 10.560900688171387, | |
| "rewards/rejected": -10.484306335449219, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.9606986899563319, | |
| "grad_norm": 0.025567537238137088, | |
| "learning_rate": 4.872884433325169e-06, | |
| "logits/chosen": -0.7050657272338867, | |
| "logits/rejected": -0.49923476576805115, | |
| "logps/chosen": -286.5049743652344, | |
| "logps/rejected": -1234.7254638671875, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.08888520300388336, | |
| "rewards/margins": 9.898573875427246, | |
| "rewards/rejected": -9.809688568115234, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 1.0043668122270741, | |
| "grad_norm": 0.03935313476723805, | |
| "learning_rate": 4.847779301910868e-06, | |
| "logits/chosen": -0.7885675430297852, | |
| "logits/rejected": -0.48259955644607544, | |
| "logps/chosen": -235.9558563232422, | |
| "logps/rejected": -1333.6160888671875, | |
| "loss": 0.0005, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.1478557288646698, | |
| "rewards/margins": 10.591659545898438, | |
| "rewards/rejected": -10.443804740905762, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 1.0480349344978166, | |
| "grad_norm": 0.019331140450817564, | |
| "learning_rate": 4.820490189292415e-06, | |
| "logits/chosen": -0.6384273767471313, | |
| "logits/rejected": -0.32141706347465515, | |
| "logps/chosen": -256.14813232421875, | |
| "logps/rejected": -1328.8856201171875, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.03197110444307327, | |
| "rewards/margins": 10.722203254699707, | |
| "rewards/rejected": -10.690231323242188, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.091703056768559, | |
| "grad_norm": 0.001917220346998511, | |
| "learning_rate": 4.791042480696179e-06, | |
| "logits/chosen": -0.8021818399429321, | |
| "logits/rejected": -0.5436561107635498, | |
| "logps/chosen": -285.2548828125, | |
| "logps/rejected": -1359.247802734375, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.09929420799016953, | |
| "rewards/margins": 10.846672058105469, | |
| "rewards/rejected": -10.747377395629883, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.1353711790393013, | |
| "grad_norm": 0.009816737966176867, | |
| "learning_rate": 4.759463569345205e-06, | |
| "logits/chosen": -0.7113679051399231, | |
| "logits/rejected": -0.45295238494873047, | |
| "logps/chosen": -268.17083740234375, | |
| "logps/rejected": -1343.6666259765625, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.10034344345331192, | |
| "rewards/margins": 10.877279281616211, | |
| "rewards/rejected": -10.776936531066895, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.1790393013100438, | |
| "grad_norm": 0.003536564037828518, | |
| "learning_rate": 4.725782830977145e-06, | |
| "logits/chosen": -0.7203564047813416, | |
| "logits/rejected": -0.4096035957336426, | |
| "logps/chosen": -257.78900146484375, | |
| "logps/rejected": -1433.7529296875, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.05666704103350639, | |
| "rewards/margins": 11.476617813110352, | |
| "rewards/rejected": -11.419949531555176, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.222707423580786, | |
| "grad_norm": 0.0031083495470266283, | |
| "learning_rate": 4.690031596517984e-06, | |
| "logits/chosen": -0.6830414533615112, | |
| "logits/rejected": -0.3940260410308838, | |
| "logps/chosen": -252.91641235351562, | |
| "logps/rejected": -1419.6212158203125, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.10215254127979279, | |
| "rewards/margins": 11.425537109375, | |
| "rewards/rejected": -11.323384284973145, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.2663755458515285, | |
| "grad_norm": 0.0031010184167246683, | |
| "learning_rate": 4.652243122936987e-06, | |
| "logits/chosen": -0.7523640394210815, | |
| "logits/rejected": -0.4601876735687256, | |
| "logps/chosen": -267.94482421875, | |
| "logps/rejected": -1480.731201171875, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.141164630651474, | |
| "rewards/margins": 12.00882339477539, | |
| "rewards/rejected": -11.867658615112305, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.3100436681222707, | |
| "grad_norm": 0.03893958758819322, | |
| "learning_rate": 4.612452562309975e-06, | |
| "logits/chosen": -0.6891266703605652, | |
| "logits/rejected": -0.3029942512512207, | |
| "logps/chosen": -241.3924560546875, | |
| "logps/rejected": -1440.799072265625, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.11804157495498657, | |
| "rewards/margins": 11.638472557067871, | |
| "rewards/rejected": -11.520431518554688, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.3537117903930131, | |
| "grad_norm": 0.0020487177308325163, | |
| "learning_rate": 4.570696929119717e-06, | |
| "logits/chosen": -0.6615210771560669, | |
| "logits/rejected": -0.40224489569664, | |
| "logps/chosen": -264.6934814453125, | |
| "logps/rejected": -1409.147216796875, | |
| "loss": 0.0003, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.0794425830245018, | |
| "rewards/margins": 11.551889419555664, | |
| "rewards/rejected": -11.47244644165039, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.3973799126637554, | |
| "grad_norm": 0.0017833767780479168, | |
| "learning_rate": 4.527015065823841e-06, | |
| "logits/chosen": -0.6528028249740601, | |
| "logits/rejected": -0.4198477864265442, | |
| "logps/chosen": -277.9957580566406, | |
| "logps/rejected": -1393.375244140625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.09100461006164551, | |
| "rewards/margins": 11.455917358398438, | |
| "rewards/rejected": -11.364913940429688, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.4410480349344978, | |
| "grad_norm": 0.007670342609009089, | |
| "learning_rate": 4.481447606722309e-06, | |
| "logits/chosen": -0.6851311326026917, | |
| "logits/rejected": -0.44172611832618713, | |
| "logps/chosen": -280.6171875, | |
| "logps/rejected": -1444.6007080078125, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.1056002750992775, | |
| "rewards/margins": 11.923059463500977, | |
| "rewards/rejected": -11.817459106445312, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.48471615720524, | |
| "grad_norm": 0.004356336133031361, | |
| "learning_rate": 4.434036940158062e-06, | |
| "logits/chosen": -0.7159875631332397, | |
| "logits/rejected": -0.26825979351997375, | |
| "logps/chosen": -226.5546875, | |
| "logps/rejected": -1620.3592529296875, | |
| "loss": 0.0004, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.12559999525547028, | |
| "rewards/margins": 13.081733703613281, | |
| "rewards/rejected": -12.956133842468262, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.5283842794759825, | |
| "grad_norm": 0.006924018476345942, | |
| "learning_rate": 4.384827169085993e-06, | |
| "logits/chosen": -0.722930908203125, | |
| "logits/rejected": -0.54079270362854, | |
| "logps/chosen": -305.4542541503906, | |
| "logps/rejected": -1445.2802734375, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.036442067474126816, | |
| "rewards/margins": 11.9239501953125, | |
| "rewards/rejected": -11.887507438659668, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.572052401746725, | |
| "grad_norm": 0.00824202881851081, | |
| "learning_rate": 4.333864070046938e-06, | |
| "logits/chosen": -0.6452621221542358, | |
| "logits/rejected": -0.2813284993171692, | |
| "logps/chosen": -266.8562927246094, | |
| "logps/rejected": -1508.1966552734375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.023952916264533997, | |
| "rewards/margins": 12.330470085144043, | |
| "rewards/rejected": -12.306516647338867, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.6157205240174672, | |
| "grad_norm": 0.0028510754034206916, | |
| "learning_rate": 4.28119505058483e-06, | |
| "logits/chosen": -0.5809464454650879, | |
| "logits/rejected": -0.29732316732406616, | |
| "logps/chosen": -292.64154052734375, | |
| "logps/rejected": -1430.4862060546875, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.08253034949302673, | |
| "rewards/margins": 11.704535484313965, | |
| "rewards/rejected": -11.787064552307129, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.6593886462882095, | |
| "grad_norm": 0.004071842447232605, | |
| "learning_rate": 4.226869105146658e-06, | |
| "logits/chosen": -0.6976742148399353, | |
| "logits/rejected": -0.374077707529068, | |
| "logps/chosen": -267.6322326660156, | |
| "logps/rejected": -1627.9765625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.027708619832992554, | |
| "rewards/margins": 13.307157516479492, | |
| "rewards/rejected": -13.279451370239258, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.703056768558952, | |
| "grad_norm": 0.0030757362239471637, | |
| "learning_rate": 4.170936769506222e-06, | |
| "logits/chosen": -0.6652247309684753, | |
| "logits/rejected": -0.3628092110157013, | |
| "logps/chosen": -288.57257080078125, | |
| "logps/rejected": -1520.739990234375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.052382566034793854, | |
| "rewards/margins": 12.51613998413086, | |
| "rewards/rejected": -12.568521499633789, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.7467248908296944, | |
| "grad_norm": 0.003598764645985237, | |
| "learning_rate": 4.1134500737541026e-06, | |
| "logits/chosen": -0.6975280046463013, | |
| "logits/rejected": -0.3748314380645752, | |
| "logps/chosen": -277.32843017578125, | |
| "logps/rejected": -1550.5653076171875, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.06987209618091583, | |
| "rewards/margins": 12.744722366333008, | |
| "rewards/rejected": -12.674848556518555, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.7903930131004366, | |
| "grad_norm": 0.007268762647014337, | |
| "learning_rate": 4.054462493897569e-06, | |
| "logits/chosen": -0.7424625754356384, | |
| "logits/rejected": -0.30923086404800415, | |
| "logps/chosen": -234.93551635742188, | |
| "logps/rejected": -1707.387451171875, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.10221578180789948, | |
| "rewards/margins": 13.916183471679688, | |
| "rewards/rejected": -13.81396770477295, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.8340611353711789, | |
| "grad_norm": 0.0008727719544739926, | |
| "learning_rate": 3.994028902115439e-06, | |
| "logits/chosen": -0.6766982674598694, | |
| "logits/rejected": -0.22497108578681946, | |
| "logps/chosen": -246.74490356445312, | |
| "logps/rejected": -1593.43115234375, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.03030253015458584, | |
| "rewards/margins": 12.960153579711914, | |
| "rewards/rejected": -12.990457534790039, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.8777292576419216, | |
| "grad_norm": 0.0015384551125673842, | |
| "learning_rate": 3.932205515714189e-06, | |
| "logits/chosen": -0.7006328701972961, | |
| "logits/rejected": -0.38360628485679626, | |
| "logps/chosen": -298.765380859375, | |
| "logps/rejected": -1612.6402587890625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.06020375341176987, | |
| "rewards/margins": 13.307812690734863, | |
| "rewards/rejected": -13.368016242980957, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.9213973799126638, | |
| "grad_norm": 0.008179209300348753, | |
| "learning_rate": 3.86904984483277e-06, | |
| "logits/chosen": -0.7087584137916565, | |
| "logits/rejected": -0.4246200919151306, | |
| "logps/chosen": -304.3985900878906, | |
| "logps/rejected": -1534.143798828125, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.08279000222682953, | |
| "rewards/margins": 12.747483253479004, | |
| "rewards/rejected": -12.830273628234863, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.965065502183406, | |
| "grad_norm": 0.0014160258405897827, | |
| "learning_rate": 3.8046206389447916e-06, | |
| "logits/chosen": -0.6982067823410034, | |
| "logits/rejected": -0.3726228177547455, | |
| "logps/chosen": -276.84039306640625, | |
| "logps/rejected": -1593.4991455078125, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.04725092276930809, | |
| "rewards/margins": 13.1595458984375, | |
| "rewards/rejected": -13.20679759979248, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.0087336244541483, | |
| "grad_norm": 0.0017767298206001502, | |
| "learning_rate": 3.738977832207839e-06, | |
| "logits/chosen": -0.6970980763435364, | |
| "logits/rejected": -0.3680294156074524, | |
| "logps/chosen": -261.3289794921875, | |
| "logps/rejected": -1555.7623291015625, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.0009517133003100753, | |
| "rewards/margins": 12.788881301879883, | |
| "rewards/rejected": -12.787931442260742, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 2.052401746724891, | |
| "grad_norm": 0.001535205426650719, | |
| "learning_rate": 3.6721824877107588e-06, | |
| "logits/chosen": -0.7333104014396667, | |
| "logits/rejected": -0.45936185121536255, | |
| "logps/chosen": -308.5245666503906, | |
| "logps/rejected": -1585.037841796875, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.0660587027668953, | |
| "rewards/margins": 13.10444450378418, | |
| "rewards/rejected": -13.170503616333008, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 2.096069868995633, | |
| "grad_norm": 0.005158531675148358, | |
| "learning_rate": 3.604296740670768e-06, | |
| "logits/chosen": -0.7633829116821289, | |
| "logits/rejected": -0.33872467279434204, | |
| "logps/chosen": -260.2208557128906, | |
| "logps/rejected": -1724.6669921875, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.010257990099489689, | |
| "rewards/margins": 14.2053804397583, | |
| "rewards/rejected": -14.195123672485352, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.1397379912663754, | |
| "grad_norm": 0.006727466994391036, | |
| "learning_rate": 3.5353837406332464e-06, | |
| "logits/chosen": -0.7862290143966675, | |
| "logits/rejected": -0.45894408226013184, | |
| "logps/chosen": -251.74417114257812, | |
| "logps/rejected": -1686.0699462890625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.0987875908613205, | |
| "rewards/margins": 13.866643905639648, | |
| "rewards/rejected": -13.767855644226074, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.183406113537118, | |
| "grad_norm": 0.005878784292188106, | |
| "learning_rate": 3.4655075927279576e-06, | |
| "logits/chosen": -0.7339245080947876, | |
| "logits/rejected": -0.42568492889404297, | |
| "logps/chosen": -266.9426574707031, | |
| "logps/rejected": -1574.525634765625, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.030187349766492844, | |
| "rewards/margins": 13.049545288085938, | |
| "rewards/rejected": -13.079732894897461, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.2270742358078603, | |
| "grad_norm": 0.0017376152792194543, | |
| "learning_rate": 3.3947332980363552e-06, | |
| "logits/chosen": -0.5695234537124634, | |
| "logits/rejected": -0.19781272113323212, | |
| "logps/chosen": -270.3812561035156, | |
| "logps/rejected": -1615.6109619140625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.023705072700977325, | |
| "rewards/margins": 13.492143630981445, | |
| "rewards/rejected": -13.515848159790039, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.2707423580786026, | |
| "grad_norm": 0.0008884061380883483, | |
| "learning_rate": 3.3231266931254546e-06, | |
| "logits/chosen": -0.7268373370170593, | |
| "logits/rejected": -0.4064566493034363, | |
| "logps/chosen": -288.7190246582031, | |
| "logps/rejected": -1659.5142822265625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.004935880657285452, | |
| "rewards/margins": 13.8540620803833, | |
| "rewards/rejected": -13.858996391296387, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.314410480349345, | |
| "grad_norm": 0.0012752069272254601, | |
| "learning_rate": 3.250754388804495e-06, | |
| "logits/chosen": -0.7004576921463013, | |
| "logits/rejected": -0.3160724639892578, | |
| "logps/chosen": -260.39801025390625, | |
| "logps/rejected": -1695.2933349609375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.0040903836488723755, | |
| "rewards/margins": 14.074589729309082, | |
| "rewards/rejected": -14.078680038452148, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.3580786026200875, | |
| "grad_norm": 0.0010551393838541235, | |
| "learning_rate": 3.1776837081613893e-06, | |
| "logits/chosen": -0.6237174868583679, | |
| "logits/rejected": -0.2775794267654419, | |
| "logps/chosen": -275.13775634765625, | |
| "logps/rejected": -1654.786376953125, | |
| "loss": 0.0002, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.028180230408906937, | |
| "rewards/margins": 13.729413032531738, | |
| "rewards/rejected": -13.757593154907227, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.4017467248908297, | |
| "grad_norm": 0.010736168566735408, | |
| "learning_rate": 3.1039826239365754e-06, | |
| "logits/chosen": -0.6697626113891602, | |
| "logits/rejected": -0.3747546076774597, | |
| "logps/chosen": -265.6369323730469, | |
| "logps/rejected": -1629.97998046875, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.04310118779540062, | |
| "rewards/margins": 13.556437492370605, | |
| "rewards/rejected": -13.513336181640625, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.445414847161572, | |
| "grad_norm": 0.001439648164211988, | |
| "learning_rate": 3.0297196952925533e-06, | |
| "logits/chosen": -0.6776083111763, | |
| "logits/rejected": -0.4190545678138733, | |
| "logps/chosen": -304.7749328613281, | |
| "logps/rejected": -1601.60791015625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.04750397801399231, | |
| "rewards/margins": 13.419174194335938, | |
| "rewards/rejected": -13.466680526733398, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.489082969432314, | |
| "grad_norm": 0.0013072815191190771, | |
| "learning_rate": 2.9549640040379043e-06, | |
| "logits/chosen": -0.7228536605834961, | |
| "logits/rejected": -0.4134379029273987, | |
| "logps/chosen": -293.1953430175781, | |
| "logps/rejected": -1718.9398193359375, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.06104196235537529, | |
| "rewards/margins": 14.203579902648926, | |
| "rewards/rejected": -14.264620780944824, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.532751091703057, | |
| "grad_norm": 0.0008461963585448364, | |
| "learning_rate": 2.8797850903651274e-06, | |
| "logits/chosen": -0.6659802794456482, | |
| "logits/rejected": -0.360958993434906, | |
| "logps/chosen": -270.3370056152344, | |
| "logps/rejected": -1665.4697265625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.0008804827812127769, | |
| "rewards/margins": 13.956453323364258, | |
| "rewards/rejected": -13.9573335647583, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.576419213973799, | |
| "grad_norm": 0.0023387701751764562, | |
| "learning_rate": 2.804252888162079e-06, | |
| "logits/chosen": -0.8161109685897827, | |
| "logits/rejected": -0.4084259569644928, | |
| "logps/chosen": -272.53216552734375, | |
| "logps/rejected": -1786.5172119140625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.03314363956451416, | |
| "rewards/margins": 14.708500862121582, | |
| "rewards/rejected": -14.741642951965332, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.6200873362445414, | |
| "grad_norm": 0.0018232538424911746, | |
| "learning_rate": 2.7284376599571776e-06, | |
| "logits/chosen": -0.7031590342521667, | |
| "logits/rejected": -0.37507158517837524, | |
| "logps/chosen": -274.2117919921875, | |
| "logps/rejected": -1641.595458984375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.04686348885297775, | |
| "rewards/margins": 13.67021656036377, | |
| "rewards/rejected": -13.717081069946289, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.6637554585152836, | |
| "grad_norm": 0.000990528476960238, | |
| "learning_rate": 2.652409931558898e-06, | |
| "logits/chosen": -0.7374690175056458, | |
| "logits/rejected": -0.43666791915893555, | |
| "logps/chosen": -283.0863342285156, | |
| "logps/rejected": -1705.676513671875, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.040116313844919205, | |
| "rewards/margins": 14.263372421264648, | |
| "rewards/rejected": -14.303491592407227, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.7074235807860263, | |
| "grad_norm": 0.0009319562615286963, | |
| "learning_rate": 2.5762404264503538e-06, | |
| "logits/chosen": -0.6617239713668823, | |
| "logits/rejected": -0.28903594613075256, | |
| "logps/chosen": -267.43609619140625, | |
| "logps/rejected": -1748.5576171875, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.03135783225297928, | |
| "rewards/margins": 14.624079704284668, | |
| "rewards/rejected": -14.592721939086914, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.7510917030567685, | |
| "grad_norm": 0.013526422059541143, | |
| "learning_rate": 2.5e-06, | |
| "logits/chosen": -0.7105010151863098, | |
| "logits/rejected": -0.4328455328941345, | |
| "logps/chosen": -281.528564453125, | |
| "logps/rejected": -1675.9830322265625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.0196510162204504, | |
| "rewards/margins": 13.972375869750977, | |
| "rewards/rejected": -13.992025375366211, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.7947598253275108, | |
| "grad_norm": 0.0003686574185361761, | |
| "learning_rate": 2.423759573549647e-06, | |
| "logits/chosen": -0.6781324148178101, | |
| "logits/rejected": -0.2962037920951843, | |
| "logps/chosen": -263.18975830078125, | |
| "logps/rejected": -1710.5943603515625, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.022569647058844566, | |
| "rewards/margins": 14.243792533874512, | |
| "rewards/rejected": -14.266362190246582, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.8384279475982535, | |
| "grad_norm": 0.0009413555719320386, | |
| "learning_rate": 2.3475900684411027e-06, | |
| "logits/chosen": -0.6239091157913208, | |
| "logits/rejected": -0.22716188430786133, | |
| "logps/chosen": -257.69842529296875, | |
| "logps/rejected": -1695.5068359375, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.07734155654907227, | |
| "rewards/margins": 14.082303047180176, | |
| "rewards/rejected": -14.159643173217773, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.8820960698689957, | |
| "grad_norm": 0.000759592485033725, | |
| "learning_rate": 2.2715623400428228e-06, | |
| "logits/chosen": -0.694677472114563, | |
| "logits/rejected": -0.3776795268058777, | |
| "logps/chosen": -282.89202880859375, | |
| "logps/rejected": -1710.0386962890625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.0032756985165178776, | |
| "rewards/margins": 14.365133285522461, | |
| "rewards/rejected": -14.361857414245605, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.925764192139738, | |
| "grad_norm": 0.019784253772424126, | |
| "learning_rate": 2.1957471118379213e-06, | |
| "logits/chosen": -0.7745624780654907, | |
| "logits/rejected": -0.4288128912448883, | |
| "logps/chosen": -304.70574951171875, | |
| "logps/rejected": -1791.9283447265625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.05531958490610123, | |
| "rewards/margins": 14.97740364074707, | |
| "rewards/rejected": -15.032722473144531, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.96943231441048, | |
| "grad_norm": 0.004147931478685181, | |
| "learning_rate": 2.120214909634873e-06, | |
| "logits/chosen": -0.7153472304344177, | |
| "logits/rejected": -0.43083199858665466, | |
| "logps/chosen": -297.9048767089844, | |
| "logps/rejected": -1724.3980712890625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.03991527482867241, | |
| "rewards/margins": 14.473901748657227, | |
| "rewards/rejected": -14.513816833496094, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 3.013100436681223, | |
| "grad_norm": 0.0013738234416668116, | |
| "learning_rate": 2.045035995962097e-06, | |
| "logits/chosen": -0.7798169255256653, | |
| "logits/rejected": -0.43747028708457947, | |
| "logps/chosen": -279.0494079589844, | |
| "logps/rejected": -1814.386962890625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.06170396879315376, | |
| "rewards/margins": 15.159332275390625, | |
| "rewards/rejected": -15.221035957336426, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 3.056768558951965, | |
| "grad_norm": 0.0031209059954659443, | |
| "learning_rate": 1.970280304707447e-06, | |
| "logits/chosen": -0.714453399181366, | |
| "logits/rejected": -0.36034974455833435, | |
| "logps/chosen": -269.9892272949219, | |
| "logps/rejected": -1740.939208984375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": 0.002551588462665677, | |
| "rewards/margins": 14.63127326965332, | |
| "rewards/rejected": -14.62872314453125, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.1004366812227073, | |
| "grad_norm": 0.00034210696214912053, | |
| "learning_rate": 1.8960173760634257e-06, | |
| "logits/chosen": -0.7932115793228149, | |
| "logits/rejected": -0.5189129114151001, | |
| "logps/chosen": -322.1393737792969, | |
| "logps/rejected": -1711.671875, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.08139355480670929, | |
| "rewards/margins": 14.367878913879395, | |
| "rewards/rejected": -14.449272155761719, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 3.14410480349345, | |
| "grad_norm": 0.0013002362609120892, | |
| "learning_rate": 1.8223162918386122e-06, | |
| "logits/chosen": -0.7495840787887573, | |
| "logits/rejected": -0.3989501893520355, | |
| "logps/chosen": -285.207275390625, | |
| "logps/rejected": -1773.8642578125, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.09717712551355362, | |
| "rewards/margins": 14.81560230255127, | |
| "rewards/rejected": -14.9127779006958, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 3.1877729257641922, | |
| "grad_norm": 0.008009764157245521, | |
| "learning_rate": 1.7492456111955052e-06, | |
| "logits/chosen": -0.6571623086929321, | |
| "logits/rejected": -0.30920231342315674, | |
| "logps/chosen": -278.0492248535156, | |
| "logps/rejected": -1723.248046875, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.0555524006485939, | |
| "rewards/margins": 14.493664741516113, | |
| "rewards/rejected": -14.549217224121094, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 3.2314410480349345, | |
| "grad_norm": 0.002618631891345944, | |
| "learning_rate": 1.6768733068745468e-06, | |
| "logits/chosen": -0.7488084435462952, | |
| "logits/rejected": -0.4106837809085846, | |
| "logps/chosen": -282.20538330078125, | |
| "logps/rejected": -1826.658203125, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.08377159386873245, | |
| "rewards/margins": 15.273724555969238, | |
| "rewards/rejected": -15.35749626159668, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 3.2751091703056767, | |
| "grad_norm": 0.00023217650976632204, | |
| "learning_rate": 1.6052667019636462e-06, | |
| "logits/chosen": -0.7862998247146606, | |
| "logits/rejected": -0.44631749391555786, | |
| "logps/chosen": -276.70867919921875, | |
| "logps/rejected": -1797.5888671875, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.024743851274251938, | |
| "rewards/margins": 15.001612663269043, | |
| "rewards/rejected": -15.026357650756836, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.3187772925764194, | |
| "grad_norm": 0.01865811559978352, | |
| "learning_rate": 1.5344924072720434e-06, | |
| "logits/chosen": -0.7188819050788879, | |
| "logits/rejected": -0.4172704815864563, | |
| "logps/chosen": -285.13140869140625, | |
| "logps/rejected": -1677.7828369140625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.061395276337862015, | |
| "rewards/margins": 14.092081069946289, | |
| "rewards/rejected": -14.153475761413574, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 3.3624454148471616, | |
| "grad_norm": 0.00039610874442709104, | |
| "learning_rate": 1.4646162593667535e-06, | |
| "logits/chosen": -0.7479511499404907, | |
| "logits/rejected": -0.4791211187839508, | |
| "logps/chosen": -314.30169677734375, | |
| "logps/rejected": -1737.6015625, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.053110092878341675, | |
| "rewards/margins": 14.654457092285156, | |
| "rewards/rejected": -14.70756721496582, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 3.406113537117904, | |
| "grad_norm": 0.0020264522877018994, | |
| "learning_rate": 1.3957032593292319e-06, | |
| "logits/chosen": -0.6951030492782593, | |
| "logits/rejected": -0.34219890832901, | |
| "logps/chosen": -285.89373779296875, | |
| "logps/rejected": -1782.4566650390625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.08678509294986725, | |
| "rewards/margins": 14.822759628295898, | |
| "rewards/rejected": -14.9095458984375, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 3.449781659388646, | |
| "grad_norm": 0.0018935186827857544, | |
| "learning_rate": 1.3278175122892416e-06, | |
| "logits/chosen": -0.6852272748947144, | |
| "logits/rejected": -0.33605360984802246, | |
| "logps/chosen": -259.1459045410156, | |
| "logps/rejected": -1725.716796875, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.036944665014743805, | |
| "rewards/margins": 14.497613906860352, | |
| "rewards/rejected": -14.534558296203613, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 3.493449781659389, | |
| "grad_norm": 0.015223829550391095, | |
| "learning_rate": 1.261022167792161e-06, | |
| "logits/chosen": -0.7397433519363403, | |
| "logits/rejected": -0.36608046293258667, | |
| "logps/chosen": -281.9190673828125, | |
| "logps/rejected": -1852.1806640625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.03298696130514145, | |
| "rewards/margins": 15.54675006866455, | |
| "rewards/rejected": -15.579736709594727, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.537117903930131, | |
| "grad_norm": 0.010378807533355724, | |
| "learning_rate": 1.195379361055209e-06, | |
| "logits/chosen": -0.6590791940689087, | |
| "logits/rejected": -0.28567662835121155, | |
| "logps/chosen": -273.66009521484375, | |
| "logps/rejected": -1854.084716796875, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.010397704318165779, | |
| "rewards/margins": 15.640172004699707, | |
| "rewards/rejected": -15.650568962097168, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 3.5807860262008733, | |
| "grad_norm": 0.016204955970958463, | |
| "learning_rate": 1.1309501551672303e-06, | |
| "logits/chosen": -0.5972493886947632, | |
| "logits/rejected": -0.2114742249250412, | |
| "logps/chosen": -269.32476806640625, | |
| "logps/rejected": -1736.829833984375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.052963800728321075, | |
| "rewards/margins": 14.625813484191895, | |
| "rewards/rejected": -14.678777694702148, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 3.6244541484716155, | |
| "grad_norm": 0.0008894376854101312, | |
| "learning_rate": 1.0677944842858112e-06, | |
| "logits/chosen": -0.6675132513046265, | |
| "logits/rejected": -0.26750123500823975, | |
| "logps/chosen": -279.586181640625, | |
| "logps/rejected": -1758.112060546875, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.050345610827207565, | |
| "rewards/margins": 14.880023002624512, | |
| "rewards/rejected": -14.930368423461914, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 3.668122270742358, | |
| "grad_norm": 0.0004087774037557016, | |
| "learning_rate": 1.005971097884561e-06, | |
| "logits/chosen": -0.705937922000885, | |
| "logits/rejected": -0.3553905785083771, | |
| "logps/chosen": -300.83929443359375, | |
| "logps/rejected": -1823.3248291015625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.14001181721687317, | |
| "rewards/margins": 15.270605087280273, | |
| "rewards/rejected": -15.410616874694824, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 3.7117903930131004, | |
| "grad_norm": 0.0006949799978363079, | |
| "learning_rate": 9.455375061024319e-07, | |
| "logits/chosen": -0.7351874709129333, | |
| "logits/rejected": -0.3912436366081238, | |
| "logps/chosen": -282.2884826660156, | |
| "logps/rejected": -1754.314453125, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.06751924753189087, | |
| "rewards/margins": 14.702392578125, | |
| "rewards/rejected": -14.769912719726562, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.7554585152838427, | |
| "grad_norm": 0.006463044951812143, | |
| "learning_rate": 8.86549926245898e-07, | |
| "logits/chosen": -0.6956661939620972, | |
| "logits/rejected": -0.3181700110435486, | |
| "logps/chosen": -287.1238708496094, | |
| "logps/rejected": -1798.494384765625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.10201337188482285, | |
| "rewards/margins": 14.971240043640137, | |
| "rewards/rejected": -15.073251724243164, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 3.7991266375545854, | |
| "grad_norm": 0.0007296140394687955, | |
| "learning_rate": 8.29063230493779e-07, | |
| "logits/chosen": -0.7505240440368652, | |
| "logits/rejected": -0.3770305812358856, | |
| "logps/chosen": -265.15679931640625, | |
| "logps/rejected": -1858.8929443359375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.02774529531598091, | |
| "rewards/margins": 15.584371566772461, | |
| "rewards/rejected": -15.6121187210083, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 3.8427947598253276, | |
| "grad_norm": 0.008816147256943409, | |
| "learning_rate": 7.731308948533431e-07, | |
| "logits/chosen": -0.7113201022148132, | |
| "logits/rejected": -0.4147794246673584, | |
| "logps/chosen": -302.5737609863281, | |
| "logps/rejected": -1753.0826416015625, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.04450942575931549, | |
| "rewards/margins": 14.832493782043457, | |
| "rewards/rejected": -14.87700366973877, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 3.88646288209607, | |
| "grad_norm": 0.005372270249774767, | |
| "learning_rate": 7.188049494151703e-07, | |
| "logits/chosen": -0.6462150812149048, | |
| "logits/rejected": -0.2772805094718933, | |
| "logps/chosen": -277.2967834472656, | |
| "logps/rejected": -1723.9173583984375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.0981786921620369, | |
| "rewards/margins": 14.48997688293457, | |
| "rewards/rejected": -14.588155746459961, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 3.930131004366812, | |
| "grad_norm": 0.0007509813802026183, | |
| "learning_rate": 6.661359299530626e-07, | |
| "logits/chosen": -0.6561316847801208, | |
| "logits/rejected": -0.47288408875465393, | |
| "logps/chosen": -322.491455078125, | |
| "logps/rejected": -1582.517333984375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.14367668330669403, | |
| "rewards/margins": 13.482531547546387, | |
| "rewards/rejected": -13.626208305358887, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.9737991266375547, | |
| "grad_norm": 0.002905923093028766, | |
| "learning_rate": 6.151728309140071e-07, | |
| "logits/chosen": -0.7203381061553955, | |
| "logits/rejected": -0.382064551115036, | |
| "logps/chosen": -285.7363586425781, | |
| "logps/rejected": -1801.4898681640625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.07807151228189468, | |
| "rewards/margins": 15.15675163269043, | |
| "rewards/rejected": -15.234823226928711, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 4.0174672489082965, | |
| "grad_norm": 0.0020680720730432436, | |
| "learning_rate": 5.659630598419391e-07, | |
| "logits/chosen": -0.724728524684906, | |
| "logits/rejected": -0.3629612326622009, | |
| "logps/chosen": -295.44073486328125, | |
| "logps/rejected": -1742.753173828125, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.13527199625968933, | |
| "rewards/margins": 14.602720260620117, | |
| "rewards/rejected": -14.737993240356445, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 4.06113537117904, | |
| "grad_norm": 0.012272686016895323, | |
| "learning_rate": 5.185523932776923e-07, | |
| "logits/chosen": -0.6944893598556519, | |
| "logits/rejected": -0.3898335099220276, | |
| "logps/chosen": -287.7585144042969, | |
| "logps/rejected": -1794.3408203125, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.05893585830926895, | |
| "rewards/margins": 15.143773078918457, | |
| "rewards/rejected": -15.202710151672363, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 4.104803493449782, | |
| "grad_norm": 0.0053926273432672945, | |
| "learning_rate": 4.7298493417616024e-07, | |
| "logits/chosen": -0.6781089305877686, | |
| "logits/rejected": -0.40882062911987305, | |
| "logps/chosen": -303.1540222167969, | |
| "logps/rejected": -1783.308837890625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.1440161019563675, | |
| "rewards/margins": 14.9929780960083, | |
| "rewards/rejected": -15.136993408203125, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 4.148471615720524, | |
| "grad_norm": 0.0010027396216351911, | |
| "learning_rate": 4.293030708802834e-07, | |
| "logits/chosen": -0.7281903028488159, | |
| "logits/rejected": -0.4005254805088043, | |
| "logps/chosen": -281.6249694824219, | |
| "logps/rejected": -1874.48828125, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.086191326379776, | |
| "rewards/margins": 15.799471855163574, | |
| "rewards/rejected": -15.885663032531738, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.192139737991266, | |
| "grad_norm": 0.0027350722737438755, | |
| "learning_rate": 3.875474376900254e-07, | |
| "logits/chosen": -0.6054420471191406, | |
| "logits/rejected": -0.2656559944152832, | |
| "logps/chosen": -283.4683532714844, | |
| "logps/rejected": -1737.546630859375, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.1006595641374588, | |
| "rewards/margins": 14.711895942687988, | |
| "rewards/rejected": -14.812556266784668, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 4.235807860262009, | |
| "grad_norm": 0.006760599239543763, | |
| "learning_rate": 3.4775687706301437e-07, | |
| "logits/chosen": -0.7967843413352966, | |
| "logits/rejected": -0.42173296213150024, | |
| "logps/chosen": -270.96148681640625, | |
| "logps/rejected": -1896.1468505859375, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.02015690691769123, | |
| "rewards/margins": 15.979083061218262, | |
| "rewards/rejected": -15.999239921569824, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 4.279475982532751, | |
| "grad_norm": 0.0008520732124624599, | |
| "learning_rate": 3.0996840348201717e-07, | |
| "logits/chosen": -0.7633962631225586, | |
| "logits/rejected": -0.4977475106716156, | |
| "logps/chosen": -313.98822021484375, | |
| "logps/rejected": -1697.9468994140625, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.18659470975399017, | |
| "rewards/margins": 14.239301681518555, | |
| "rewards/rejected": -14.425895690917969, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 4.323144104803493, | |
| "grad_norm": 0.0013051411886959893, | |
| "learning_rate": 2.742171690228562e-07, | |
| "logits/chosen": -0.7120814323425293, | |
| "logits/rejected": -0.3730967342853546, | |
| "logps/chosen": -298.5450134277344, | |
| "logps/rejected": -1744.666015625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.15538612008094788, | |
| "rewards/margins": 14.598345756530762, | |
| "rewards/rejected": -14.75373363494873, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 4.366812227074236, | |
| "grad_norm": 0.002698793985143946, | |
| "learning_rate": 2.405364306547955e-07, | |
| "logits/chosen": -0.7172932028770447, | |
| "logits/rejected": -0.47384780645370483, | |
| "logps/chosen": -322.2046813964844, | |
| "logps/rejected": -1694.6380615234375, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.10717038065195084, | |
| "rewards/margins": 14.218195915222168, | |
| "rewards/rejected": -14.325366020202637, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.4104803493449785, | |
| "grad_norm": 0.00023879898445148078, | |
| "learning_rate": 2.0895751930382125e-07, | |
| "logits/chosen": -0.6613215804100037, | |
| "logits/rejected": -0.33443623781204224, | |
| "logps/chosen": -295.16278076171875, | |
| "logps/rejected": -1768.1917724609375, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.03670508787035942, | |
| "rewards/margins": 14.837631225585938, | |
| "rewards/rejected": -14.874334335327148, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 4.454148471615721, | |
| "grad_norm": 0.00023568210161956636, | |
| "learning_rate": 1.7950981070758488e-07, | |
| "logits/chosen": -0.6994235515594482, | |
| "logits/rejected": -0.41796016693115234, | |
| "logps/chosen": -287.3175048828125, | |
| "logps/rejected": -1685.6611328125, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.08301335573196411, | |
| "rewards/margins": 14.271142959594727, | |
| "rewards/rejected": -14.354156494140625, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 4.497816593886463, | |
| "grad_norm": 0.0073133867021571005, | |
| "learning_rate": 1.5222069808913303e-07, | |
| "logits/chosen": -0.6575132012367249, | |
| "logits/rejected": -0.3197494447231293, | |
| "logps/chosen": -300.7660217285156, | |
| "logps/rejected": -1779.563232421875, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.06065019965171814, | |
| "rewards/margins": 15.048309326171875, | |
| "rewards/rejected": -15.10895824432373, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 4.541484716157205, | |
| "grad_norm": 0.00034831682457422486, | |
| "learning_rate": 1.271155666748311e-07, | |
| "logits/chosen": -0.7639699578285217, | |
| "logits/rejected": -0.43667006492614746, | |
| "logps/chosen": -308.33502197265625, | |
| "logps/rejected": -1791.947998046875, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.1204371303319931, | |
| "rewards/margins": 15.054000854492188, | |
| "rewards/rejected": -15.1744384765625, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 4.585152838427947, | |
| "grad_norm": 0.0003771930949580811, | |
| "learning_rate": 1.0421777008019663e-07, | |
| "logits/chosen": -0.7766512632369995, | |
| "logits/rejected": -0.3912786841392517, | |
| "logps/chosen": -293.3248596191406, | |
| "logps/rejected": -1899.7158203125, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.07415080815553665, | |
| "rewards/margins": 15.951011657714844, | |
| "rewards/rejected": -16.02515983581543, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 4.62882096069869, | |
| "grad_norm": 0.0005156543022691973, | |
| "learning_rate": 8.354860858560021e-08, | |
| "logits/chosen": -0.7269026637077332, | |
| "logits/rejected": -0.3382445275783539, | |
| "logps/chosen": -295.33038330078125, | |
| "logps/rejected": -1880.168212890625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.10000383853912354, | |
| "rewards/margins": 15.843725204467773, | |
| "rewards/rejected": -15.94372844696045, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 4.672489082969433, | |
| "grad_norm": 0.0012206160223788651, | |
| "learning_rate": 6.512730932204698e-08, | |
| "logits/chosen": -0.653832197189331, | |
| "logits/rejected": -0.3418962359428406, | |
| "logps/chosen": -320.64703369140625, | |
| "logps/rejected": -1710.6214599609375, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.06000875309109688, | |
| "rewards/margins": 14.490182876586914, | |
| "rewards/rejected": -14.550191879272461, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 4.716157205240175, | |
| "grad_norm": 0.002712454997690386, | |
| "learning_rate": 4.897100838547081e-08, | |
| "logits/chosen": -0.6822378039360046, | |
| "logits/rejected": -0.3716619312763214, | |
| "logps/chosen": -306.8667907714844, | |
| "logps/rejected": -1840.0302734375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.12169905006885529, | |
| "rewards/margins": 15.402539253234863, | |
| "rewards/rejected": -15.524238586425781, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 4.759825327510917, | |
| "grad_norm": 0.00027359144331405783, | |
| "learning_rate": 3.5094734896174985e-08, | |
| "logits/chosen": -0.7298842668533325, | |
| "logits/rejected": -0.33024948835372925, | |
| "logps/chosen": -271.23095703125, | |
| "logps/rejected": -1889.5804443359375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.09704066812992096, | |
| "rewards/margins": 15.853654861450195, | |
| "rewards/rejected": -15.950695991516113, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 4.8034934497816595, | |
| "grad_norm": 0.0027199074701782094, | |
| "learning_rate": 2.351139701825267e-08, | |
| "logits/chosen": -0.6037005186080933, | |
| "logits/rejected": -0.26832693815231323, | |
| "logps/chosen": -272.974853515625, | |
| "logps/rejected": -1676.4384765625, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.06281115859746933, | |
| "rewards/margins": 14.172335624694824, | |
| "rewards/rejected": -14.235147476196289, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.847161572052402, | |
| "grad_norm": 0.0007236973858890073, | |
| "learning_rate": 1.4231769951990326e-08, | |
| "logits/chosen": -0.6609109044075012, | |
| "logits/rejected": -0.33717867732048035, | |
| "logps/chosen": -295.76898193359375, | |
| "logps/rejected": -1793.028076171875, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.19633208215236664, | |
| "rewards/margins": 15.07104206085205, | |
| "rewards/rejected": -15.267372131347656, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 4.890829694323144, | |
| "grad_norm": 0.0002222821524569821, | |
| "learning_rate": 7.264485910423447e-09, | |
| "logits/chosen": -0.6304144263267517, | |
| "logits/rejected": -0.2548811733722687, | |
| "logps/chosen": -274.11383056640625, | |
| "logps/rejected": -1818.213134765625, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.09864435344934464, | |
| "rewards/margins": 15.282160758972168, | |
| "rewards/rejected": -15.380805969238281, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 4.934497816593886, | |
| "grad_norm": 0.0096819850101334, | |
| "learning_rate": 2.6160260893692833e-09, | |
| "logits/chosen": -0.6611170768737793, | |
| "logits/rejected": -0.354715496301651, | |
| "logps/chosen": -308.61688232421875, | |
| "logps/rejected": -1704.0927734375, | |
| "loss": 0.0001, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.12653221189975739, | |
| "rewards/margins": 14.360208511352539, | |
| "rewards/rejected": -14.486740112304688, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 4.978165938864628, | |
| "grad_norm": 0.0002604769137792249, | |
| "learning_rate": 2.9071463840540936e-10, | |
| "logits/chosen": -0.7605892419815063, | |
| "logits/rejected": -0.5090384483337402, | |
| "logps/chosen": -336.0806579589844, | |
| "logps/rejected": -1813.3853759765625, | |
| "loss": 0.0, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -0.10771320015192032, | |
| "rewards/margins": 15.362016677856445, | |
| "rewards/rejected": -15.469728469848633, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 1145, | |
| "total_flos": 0.0, | |
| "train_loss": 0.04901049308554833, | |
| "train_runtime": 10890.5088, | |
| "train_samples_per_second": 6.722, | |
| "train_steps_per_second": 0.105 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1145, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 5, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |