diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,10756 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 1.0, + "eval_steps": 7000, + "global_step": 7642, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 6.535947712418301e-09, + "logits/chosen": -2.6548311710357666, + "logits/rejected": -2.625453472137451, + "logps/chosen": -241.47927856445312, + "logps/rejected": -255.76844787597656, + "loss": 0.6931, + "rewards/accuracies": 0.0, + "rewards/chosen": 0.0, + "rewards/margins": 0.0, + "rewards/rejected": 0.0, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 6.535947712418302e-08, + "logits/chosen": -2.6198513507843018, + "logits/rejected": -2.599431037902832, + "logps/chosen": -295.6170959472656, + "logps/rejected": -264.95758056640625, + "loss": 0.6933, + "rewards/accuracies": 0.3888888955116272, + "rewards/chosen": -0.00039648558595217764, + "rewards/margins": -0.00022628647275269032, + "rewards/rejected": -0.0001701990404399112, + "step": 10 + }, + { + "epoch": 0.0, + "learning_rate": 1.3071895424836603e-07, + "logits/chosen": -2.641977310180664, + "logits/rejected": -2.625978946685791, + "logps/chosen": -285.5495300292969, + "logps/rejected": -288.0072937011719, + "loss": 0.6933, + "rewards/accuracies": 0.48750001192092896, + "rewards/chosen": 0.0002985417377203703, + "rewards/margins": -0.00038930849405005574, + "rewards/rejected": 0.000687850231770426, + "step": 20 + }, + { + "epoch": 0.0, + "learning_rate": 1.9607843137254904e-07, + "logits/chosen": -2.634305238723755, + "logits/rejected": -2.5763115882873535, + "logps/chosen": -252.57080078125, + "logps/rejected": -230.24319458007812, + "loss": 0.6934, + "rewards/accuracies": 0.4749999940395355, + "rewards/chosen": -0.00017283143824897707, + "rewards/margins": -0.0005285438382998109, + "rewards/rejected": 0.0003557124291546643, + "step": 30 + }, + { + "epoch": 0.01, + "learning_rate": 2.6143790849673207e-07, + "logits/chosen": -2.6947145462036133, + "logits/rejected": -2.6161351203918457, + "logps/chosen": -283.332275390625, + "logps/rejected": -263.6659851074219, + "loss": 0.6934, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.00043302043923176825, + "rewards/margins": -0.00046260812086984515, + "rewards/rejected": 2.9587605240521953e-05, + "step": 40 + }, + { + "epoch": 0.01, + "learning_rate": 3.267973856209151e-07, + "logits/chosen": -2.6638424396514893, + "logits/rejected": -2.6378073692321777, + "logps/chosen": -243.68295288085938, + "logps/rejected": -243.103271484375, + "loss": 0.693, + "rewards/accuracies": 0.5375000238418579, + "rewards/chosen": 0.0005886313738301396, + "rewards/margins": 0.00029269742663018405, + "rewards/rejected": 0.0002959339471999556, + "step": 50 + }, + { + "epoch": 0.01, + "learning_rate": 3.921568627450981e-07, + "logits/chosen": -2.6192870140075684, + "logits/rejected": -2.6178603172302246, + "logps/chosen": -300.85760498046875, + "logps/rejected": -263.17523193359375, + "loss": 0.694, + "rewards/accuracies": 0.4625000059604645, + "rewards/chosen": -0.0010113344760611653, + "rewards/margins": -0.0017379462951794267, + "rewards/rejected": 0.0007266116444952786, + "step": 60 + }, + { + "epoch": 0.01, + "learning_rate": 4.5751633986928105e-07, + "logits/chosen": -2.663377285003662, + "logits/rejected": -2.6458141803741455, + "logps/chosen": -264.3880615234375, + "logps/rejected": -221.2679443359375, + "loss": 0.6927, + "rewards/accuracies": 0.550000011920929, + "rewards/chosen": 0.0004358245641924441, + "rewards/margins": 0.00086634122999385, + "rewards/rejected": -0.0004305167240090668, + "step": 70 + }, + { + "epoch": 0.01, + "learning_rate": 5.228758169934641e-07, + "logits/chosen": -2.6054091453552246, + "logits/rejected": -2.5989038944244385, + "logps/chosen": -311.854248046875, + "logps/rejected": -308.6831970214844, + "loss": 0.6929, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.0007162230904214084, + "rewards/margins": 0.0005422987742349505, + "rewards/rejected": 0.00017392440349794924, + "step": 80 + }, + { + "epoch": 0.01, + "learning_rate": 5.882352941176471e-07, + "logits/chosen": -2.702660322189331, + "logits/rejected": -2.663552761077881, + "logps/chosen": -275.44598388671875, + "logps/rejected": -243.2969970703125, + "loss": 0.6936, + "rewards/accuracies": 0.4124999940395355, + "rewards/chosen": -0.00016839444288052619, + "rewards/margins": -0.0009770167525857687, + "rewards/rejected": 0.000808622338809073, + "step": 90 + }, + { + "epoch": 0.01, + "learning_rate": 6.535947712418302e-07, + "logits/chosen": -2.6264312267303467, + "logits/rejected": -2.6374175548553467, + "logps/chosen": -247.1935577392578, + "logps/rejected": -254.58740234375, + "loss": 0.6928, + "rewards/accuracies": 0.5249999761581421, + "rewards/chosen": 0.0008388921851292253, + "rewards/margins": 0.0007487462717108428, + "rewards/rejected": 9.014595707412809e-05, + "step": 100 + }, + { + "epoch": 0.01, + "learning_rate": 7.189542483660131e-07, + "logits/chosen": -2.6622517108917236, + "logits/rejected": -2.608503580093384, + "logps/chosen": -290.3687744140625, + "logps/rejected": -271.18170166015625, + "loss": 0.6928, + "rewards/accuracies": 0.5375000238418579, + "rewards/chosen": 0.0006900489097461104, + "rewards/margins": 0.0007947798003442585, + "rewards/rejected": -0.0001047308323904872, + "step": 110 + }, + { + "epoch": 0.02, + "learning_rate": 7.843137254901962e-07, + "logits/chosen": -2.660090446472168, + "logits/rejected": -2.6123783588409424, + "logps/chosen": -308.0357360839844, + "logps/rejected": -249.29544067382812, + "loss": 0.6934, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.00019220352987758815, + "rewards/margins": -0.0005792457377538085, + "rewards/rejected": 0.0003870422951877117, + "step": 120 + }, + { + "epoch": 0.02, + "learning_rate": 8.496732026143792e-07, + "logits/chosen": -2.6670804023742676, + "logits/rejected": -2.6804003715515137, + "logps/chosen": -271.88018798828125, + "logps/rejected": -242.92874145507812, + "loss": 0.6929, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": 0.0007403042982332408, + "rewards/margins": 0.0005177179118618369, + "rewards/rejected": 0.00022258632816374302, + "step": 130 + }, + { + "epoch": 0.02, + "learning_rate": 9.150326797385621e-07, + "logits/chosen": -2.6343283653259277, + "logits/rejected": -2.606320381164551, + "logps/chosen": -281.6643371582031, + "logps/rejected": -250.2583465576172, + "loss": 0.693, + "rewards/accuracies": 0.5249999761581421, + "rewards/chosen": 7.358348375419155e-05, + "rewards/margins": 0.000287846167339012, + "rewards/rejected": -0.00021426270541269332, + "step": 140 + }, + { + "epoch": 0.02, + "learning_rate": 9.80392156862745e-07, + "logits/chosen": -2.7104547023773193, + "logits/rejected": -2.631685495376587, + "logps/chosen": -340.4215393066406, + "logps/rejected": -304.089599609375, + "loss": 0.6928, + "rewards/accuracies": 0.5249999761581421, + "rewards/chosen": 0.0008256893488578498, + "rewards/margins": 0.0006710177985951304, + "rewards/rejected": 0.0001546714483993128, + "step": 150 + }, + { + "epoch": 0.02, + "learning_rate": 1.0457516339869283e-06, + "logits/chosen": -2.6912295818328857, + "logits/rejected": -2.6192638874053955, + "logps/chosen": -263.2707214355469, + "logps/rejected": -241.91943359375, + "loss": 0.6933, + "rewards/accuracies": 0.4749999940395355, + "rewards/chosen": -0.0006294053164310753, + "rewards/margins": -0.000346137210726738, + "rewards/rejected": -0.00028326810570433736, + "step": 160 + }, + { + "epoch": 0.02, + "learning_rate": 1.111111111111111e-06, + "logits/chosen": -2.6141347885131836, + "logits/rejected": -2.589357852935791, + "logps/chosen": -247.6897430419922, + "logps/rejected": -230.26284790039062, + "loss": 0.6932, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.00013610209862235934, + "rewards/margins": -0.00014482846017926931, + "rewards/rejected": 0.00028093045693822205, + "step": 170 + }, + { + "epoch": 0.02, + "learning_rate": 1.1764705882352942e-06, + "logits/chosen": -2.7081546783447266, + "logits/rejected": -2.591892957687378, + "logps/chosen": -356.08349609375, + "logps/rejected": -272.71075439453125, + "loss": 0.6931, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.0010515575995668769, + "rewards/margins": 0.00011522386921569705, + "rewards/rejected": 0.0009363336721435189, + "step": 180 + }, + { + "epoch": 0.02, + "learning_rate": 1.2418300653594772e-06, + "logits/chosen": -2.612846851348877, + "logits/rejected": -2.5740773677825928, + "logps/chosen": -279.548583984375, + "logps/rejected": -228.6970672607422, + "loss": 0.6932, + "rewards/accuracies": 0.44999998807907104, + "rewards/chosen": 0.0006955896387808025, + "rewards/margins": -0.00014482141705229878, + "rewards/rejected": 0.0008404110558331013, + "step": 190 + }, + { + "epoch": 0.03, + "learning_rate": 1.3071895424836604e-06, + "logits/chosen": -2.6737027168273926, + "logits/rejected": -2.5836992263793945, + "logps/chosen": -264.43951416015625, + "logps/rejected": -266.1790466308594, + "loss": 0.6926, + "rewards/accuracies": 0.5874999761581421, + "rewards/chosen": 0.0010568450670689344, + "rewards/margins": 0.0010566504206508398, + "rewards/rejected": 1.9475992019124533e-07, + "step": 200 + }, + { + "epoch": 0.03, + "learning_rate": 1.3725490196078434e-06, + "logits/chosen": -2.732459545135498, + "logits/rejected": -2.702866792678833, + "logps/chosen": -288.2290954589844, + "logps/rejected": -264.72613525390625, + "loss": 0.693, + "rewards/accuracies": 0.6000000238418579, + "rewards/chosen": 0.0013281672727316618, + "rewards/margins": 0.0002143843739759177, + "rewards/rejected": 0.0011137829860672355, + "step": 210 + }, + { + "epoch": 0.03, + "learning_rate": 1.4379084967320261e-06, + "logits/chosen": -2.617934465408325, + "logits/rejected": -2.5537524223327637, + "logps/chosen": -289.5554504394531, + "logps/rejected": -263.4027404785156, + "loss": 0.6929, + "rewards/accuracies": 0.512499988079071, + "rewards/chosen": 0.001351059996522963, + "rewards/margins": 0.000558468047529459, + "rewards/rejected": 0.000792592007201165, + "step": 220 + }, + { + "epoch": 0.03, + "learning_rate": 1.5032679738562091e-06, + "logits/chosen": -2.569349765777588, + "logits/rejected": -2.5612125396728516, + "logps/chosen": -237.69454956054688, + "logps/rejected": -280.23577880859375, + "loss": 0.6931, + "rewards/accuracies": 0.5249999761581421, + "rewards/chosen": 0.0010062854271382093, + "rewards/margins": 3.270909655839205e-05, + "rewards/rejected": 0.0009735762141644955, + "step": 230 + }, + { + "epoch": 0.03, + "learning_rate": 1.5686274509803923e-06, + "logits/chosen": -2.637214183807373, + "logits/rejected": -2.617504119873047, + "logps/chosen": -255.1501007080078, + "logps/rejected": -268.76898193359375, + "loss": 0.6926, + "rewards/accuracies": 0.550000011920929, + "rewards/chosen": 0.0017043532570824027, + "rewards/margins": 0.001065735355950892, + "rewards/rejected": 0.0006386177847161889, + "step": 240 + }, + { + "epoch": 0.03, + "learning_rate": 1.6339869281045753e-06, + "logits/chosen": -2.6625113487243652, + "logits/rejected": -2.5959858894348145, + "logps/chosen": -323.64447021484375, + "logps/rejected": -278.6389465332031, + "loss": 0.6923, + "rewards/accuracies": 0.574999988079071, + "rewards/chosen": 0.0023534377105534077, + "rewards/margins": 0.0016259655822068453, + "rewards/rejected": 0.000727472361177206, + "step": 250 + }, + { + "epoch": 0.03, + "learning_rate": 1.6993464052287585e-06, + "logits/chosen": -2.689392566680908, + "logits/rejected": -2.63670015335083, + "logps/chosen": -301.657470703125, + "logps/rejected": -264.7417297363281, + "loss": 0.6922, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": 0.0035001803189516068, + "rewards/margins": 0.0018760201055556536, + "rewards/rejected": 0.0016241597477346659, + "step": 260 + }, + { + "epoch": 0.04, + "learning_rate": 1.7647058823529414e-06, + "logits/chosen": -2.639979839324951, + "logits/rejected": -2.599606990814209, + "logps/chosen": -255.54641723632812, + "logps/rejected": -233.1532440185547, + "loss": 0.6924, + "rewards/accuracies": 0.5874999761581421, + "rewards/chosen": 0.0029199840500950813, + "rewards/margins": 0.001584300072863698, + "rewards/rejected": 0.0013356839772313833, + "step": 270 + }, + { + "epoch": 0.04, + "learning_rate": 1.8300653594771242e-06, + "logits/chosen": -2.7038564682006836, + "logits/rejected": -2.6199021339416504, + "logps/chosen": -325.8532409667969, + "logps/rejected": -290.75335693359375, + "loss": 0.6924, + "rewards/accuracies": 0.574999988079071, + "rewards/chosen": 0.0041234977543354034, + "rewards/margins": 0.0016002748161554337, + "rewards/rejected": 0.0025232231710106134, + "step": 280 + }, + { + "epoch": 0.04, + "learning_rate": 1.8954248366013072e-06, + "logits/chosen": -2.647611379623413, + "logits/rejected": -2.6253955364227295, + "logps/chosen": -296.51324462890625, + "logps/rejected": -272.4481506347656, + "loss": 0.6922, + "rewards/accuracies": 0.574999988079071, + "rewards/chosen": 0.004413033369928598, + "rewards/margins": 0.001867209910415113, + "rewards/rejected": 0.0025458228774368763, + "step": 290 + }, + { + "epoch": 0.04, + "learning_rate": 1.96078431372549e-06, + "logits/chosen": -2.742915391921997, + "logits/rejected": -2.6598916053771973, + "logps/chosen": -306.30755615234375, + "logps/rejected": -253.3156280517578, + "loss": 0.6917, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": 0.005059394054114819, + "rewards/margins": 0.0029503516852855682, + "rewards/rejected": 0.002109042601659894, + "step": 300 + }, + { + "epoch": 0.04, + "learning_rate": 2.0261437908496734e-06, + "logits/chosen": -2.6307213306427, + "logits/rejected": -2.6080451011657715, + "logps/chosen": -303.6937561035156, + "logps/rejected": -290.9927062988281, + "loss": 0.6923, + "rewards/accuracies": 0.574999988079071, + "rewards/chosen": 0.004248838871717453, + "rewards/margins": 0.001757316873408854, + "rewards/rejected": 0.0024915223475545645, + "step": 310 + }, + { + "epoch": 0.04, + "learning_rate": 2.0915032679738565e-06, + "logits/chosen": -2.6685638427734375, + "logits/rejected": -2.5951087474823, + "logps/chosen": -280.6381530761719, + "logps/rejected": -276.507568359375, + "loss": 0.6917, + "rewards/accuracies": 0.574999988079071, + "rewards/chosen": 0.006356862839311361, + "rewards/margins": 0.0029213307425379753, + "rewards/rejected": 0.0034355330280959606, + "step": 320 + }, + { + "epoch": 0.04, + "learning_rate": 2.1568627450980393e-06, + "logits/chosen": -2.6920762062072754, + "logits/rejected": -2.6318111419677734, + "logps/chosen": -307.9992370605469, + "logps/rejected": -256.708984375, + "loss": 0.6918, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": 0.005926724523305893, + "rewards/margins": 0.0028058711905032396, + "rewards/rejected": 0.0031208537984639406, + "step": 330 + }, + { + "epoch": 0.04, + "learning_rate": 2.222222222222222e-06, + "logits/chosen": -2.6575253009796143, + "logits/rejected": -2.5525221824645996, + "logps/chosen": -278.5721740722656, + "logps/rejected": -232.0401611328125, + "loss": 0.6916, + "rewards/accuracies": 0.5874999761581421, + "rewards/chosen": 0.007162533700466156, + "rewards/margins": 0.0030813287012279034, + "rewards/rejected": 0.004081204999238253, + "step": 340 + }, + { + "epoch": 0.05, + "learning_rate": 2.2875816993464053e-06, + "logits/chosen": -2.7125134468078613, + "logits/rejected": -2.624390125274658, + "logps/chosen": -332.53369140625, + "logps/rejected": -271.7375793457031, + "loss": 0.6913, + "rewards/accuracies": 0.5874999761581421, + "rewards/chosen": 0.008111939765512943, + "rewards/margins": 0.003727226285263896, + "rewards/rejected": 0.004384713713079691, + "step": 350 + }, + { + "epoch": 0.05, + "learning_rate": 2.3529411764705885e-06, + "logits/chosen": -2.694873094558716, + "logits/rejected": -2.6326351165771484, + "logps/chosen": -266.7177734375, + "logps/rejected": -263.2748107910156, + "loss": 0.6916, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": 0.008503630757331848, + "rewards/margins": 0.0031798859126865864, + "rewards/rejected": 0.005323744844645262, + "step": 360 + }, + { + "epoch": 0.05, + "learning_rate": 2.4183006535947716e-06, + "logits/chosen": -2.6473262310028076, + "logits/rejected": -2.641078472137451, + "logps/chosen": -270.0342712402344, + "logps/rejected": -247.8629913330078, + "loss": 0.6915, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": 0.00798349641263485, + "rewards/margins": 0.0034046161454170942, + "rewards/rejected": 0.004578880965709686, + "step": 370 + }, + { + "epoch": 0.05, + "learning_rate": 2.4836601307189544e-06, + "logits/chosen": -2.6707968711853027, + "logits/rejected": -2.631885290145874, + "logps/chosen": -295.2742614746094, + "logps/rejected": -257.59832763671875, + "loss": 0.6914, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": 0.010473817586898804, + "rewards/margins": 0.0036211558617651463, + "rewards/rejected": 0.00685266125947237, + "step": 380 + }, + { + "epoch": 0.05, + "learning_rate": 2.549019607843137e-06, + "logits/chosen": -2.6127049922943115, + "logits/rejected": -2.5472323894500732, + "logps/chosen": -311.64892578125, + "logps/rejected": -251.50717163085938, + "loss": 0.6901, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": 0.011281657963991165, + "rewards/margins": 0.0060841976664960384, + "rewards/rejected": 0.005197459366172552, + "step": 390 + }, + { + "epoch": 0.05, + "learning_rate": 2.6143790849673208e-06, + "logits/chosen": -2.7016243934631348, + "logits/rejected": -2.702817440032959, + "logps/chosen": -295.6642150878906, + "logps/rejected": -304.6461486816406, + "loss": 0.6912, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": 0.010914499871432781, + "rewards/margins": 0.0038754392880946398, + "rewards/rejected": 0.00703906174749136, + "step": 400 + }, + { + "epoch": 0.05, + "learning_rate": 2.6797385620915036e-06, + "logits/chosen": -2.6141860485076904, + "logits/rejected": -2.561201572418213, + "logps/chosen": -268.1377258300781, + "logps/rejected": -220.4593505859375, + "loss": 0.6906, + "rewards/accuracies": 0.574999988079071, + "rewards/chosen": 0.011765848845243454, + "rewards/margins": 0.005084649659693241, + "rewards/rejected": 0.0066812001168727875, + "step": 410 + }, + { + "epoch": 0.05, + "learning_rate": 2.7450980392156867e-06, + "logits/chosen": -2.6364364624023438, + "logits/rejected": -2.5971553325653076, + "logps/chosen": -289.3436584472656, + "logps/rejected": -284.9223327636719, + "loss": 0.6894, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": 0.013207967393100262, + "rewards/margins": 0.007489602081477642, + "rewards/rejected": 0.005718364380300045, + "step": 420 + }, + { + "epoch": 0.06, + "learning_rate": 2.8104575163398695e-06, + "logits/chosen": -2.6831023693084717, + "logits/rejected": -2.6227285861968994, + "logps/chosen": -290.2114562988281, + "logps/rejected": -271.54486083984375, + "loss": 0.6899, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": 0.01368993241339922, + "rewards/margins": 0.006646145135164261, + "rewards/rejected": 0.007043786346912384, + "step": 430 + }, + { + "epoch": 0.06, + "learning_rate": 2.8758169934640523e-06, + "logits/chosen": -2.7199249267578125, + "logits/rejected": -2.7110390663146973, + "logps/chosen": -279.10302734375, + "logps/rejected": -250.34030151367188, + "loss": 0.6902, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": 0.014295792207121849, + "rewards/margins": 0.005877715535461903, + "rewards/rejected": 0.008418076671659946, + "step": 440 + }, + { + "epoch": 0.06, + "learning_rate": 2.9411764705882355e-06, + "logits/chosen": -2.684593439102173, + "logits/rejected": -2.6899209022521973, + "logps/chosen": -297.78277587890625, + "logps/rejected": -310.8353576660156, + "loss": 0.6901, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": 0.014938647858798504, + "rewards/margins": 0.006166353821754456, + "rewards/rejected": 0.008772294968366623, + "step": 450 + }, + { + "epoch": 0.06, + "learning_rate": 3.0065359477124182e-06, + "logits/chosen": -2.615818977355957, + "logits/rejected": -2.5617904663085938, + "logps/chosen": -265.1422424316406, + "logps/rejected": -251.06689453125, + "loss": 0.69, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": 0.017867058515548706, + "rewards/margins": 0.006379131227731705, + "rewards/rejected": 0.01148792915046215, + "step": 460 + }, + { + "epoch": 0.06, + "learning_rate": 3.071895424836602e-06, + "logits/chosen": -2.648271083831787, + "logits/rejected": -2.64194917678833, + "logps/chosen": -285.4848937988281, + "logps/rejected": -263.63787841796875, + "loss": 0.6891, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": 0.02088136412203312, + "rewards/margins": 0.00808627251535654, + "rewards/rejected": 0.012795092537999153, + "step": 470 + }, + { + "epoch": 0.06, + "learning_rate": 3.1372549019607846e-06, + "logits/chosen": -2.61929988861084, + "logits/rejected": -2.6027560234069824, + "logps/chosen": -292.32537841796875, + "logps/rejected": -243.3027801513672, + "loss": 0.6883, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": 0.022852329537272453, + "rewards/margins": 0.009881490841507912, + "rewards/rejected": 0.012970837764441967, + "step": 480 + }, + { + "epoch": 0.06, + "learning_rate": 3.2026143790849674e-06, + "logits/chosen": -2.693856716156006, + "logits/rejected": -2.605499267578125, + "logps/chosen": -280.8228759765625, + "logps/rejected": -227.39205932617188, + "loss": 0.6875, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": 0.01996048167347908, + "rewards/margins": 0.011412231251597404, + "rewards/rejected": 0.00854825135320425, + "step": 490 + }, + { + "epoch": 0.07, + "learning_rate": 3.2679738562091506e-06, + "logits/chosen": -2.656646728515625, + "logits/rejected": -2.5914053916931152, + "logps/chosen": -259.90399169921875, + "logps/rejected": -237.15554809570312, + "loss": 0.6897, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": 0.017155682668089867, + "rewards/margins": 0.007095979992300272, + "rewards/rejected": 0.010059705004096031, + "step": 500 + }, + { + "epoch": 0.07, + "learning_rate": 3.3333333333333333e-06, + "logits/chosen": -2.609172821044922, + "logits/rejected": -2.599658727645874, + "logps/chosen": -270.3688659667969, + "logps/rejected": -264.7857360839844, + "loss": 0.6881, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": 0.023332804441452026, + "rewards/margins": 0.0103679858148098, + "rewards/rejected": 0.012964817695319653, + "step": 510 + }, + { + "epoch": 0.07, + "learning_rate": 3.398692810457517e-06, + "logits/chosen": -2.6225383281707764, + "logits/rejected": -2.5244414806365967, + "logps/chosen": -259.9064025878906, + "logps/rejected": -264.4110107421875, + "loss": 0.688, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": 0.018774518743157387, + "rewards/margins": 0.010412798263132572, + "rewards/rejected": 0.00836171954870224, + "step": 520 + }, + { + "epoch": 0.07, + "learning_rate": 3.4640522875816997e-06, + "logits/chosen": -2.6073412895202637, + "logits/rejected": -2.557931900024414, + "logps/chosen": -267.40045166015625, + "logps/rejected": -240.80545043945312, + "loss": 0.6859, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": 0.02611006423830986, + "rewards/margins": 0.014706981368362904, + "rewards/rejected": 0.011403081007301807, + "step": 530 + }, + { + "epoch": 0.07, + "learning_rate": 3.529411764705883e-06, + "logits/chosen": -2.656104803085327, + "logits/rejected": -2.5741233825683594, + "logps/chosen": -297.51910400390625, + "logps/rejected": -266.53363037109375, + "loss": 0.6859, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": 0.02494840696454048, + "rewards/margins": 0.01465518493205309, + "rewards/rejected": 0.010293224826455116, + "step": 540 + }, + { + "epoch": 0.07, + "learning_rate": 3.5947712418300657e-06, + "logits/chosen": -2.6402623653411865, + "logits/rejected": -2.60183048248291, + "logps/chosen": -259.5213317871094, + "logps/rejected": -248.22152709960938, + "loss": 0.686, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": 0.02932048961520195, + "rewards/margins": 0.014470371417701244, + "rewards/rejected": 0.014850117266178131, + "step": 550 + }, + { + "epoch": 0.07, + "learning_rate": 3.6601307189542484e-06, + "logits/chosen": -2.669956684112549, + "logits/rejected": -2.617684841156006, + "logps/chosen": -293.4594421386719, + "logps/rejected": -301.4708251953125, + "loss": 0.6858, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": 0.027803082019090652, + "rewards/margins": 0.014931738376617432, + "rewards/rejected": 0.012871342711150646, + "step": 560 + }, + { + "epoch": 0.07, + "learning_rate": 3.7254901960784316e-06, + "logits/chosen": -2.7117550373077393, + "logits/rejected": -2.6599695682525635, + "logps/chosen": -272.2671203613281, + "logps/rejected": -257.97760009765625, + "loss": 0.6863, + "rewards/accuracies": 0.75, + "rewards/chosen": 0.027036011219024658, + "rewards/margins": 0.014011162333190441, + "rewards/rejected": 0.013024847023189068, + "step": 570 + }, + { + "epoch": 0.08, + "learning_rate": 3.7908496732026144e-06, + "logits/chosen": -2.68557071685791, + "logits/rejected": -2.588757276535034, + "logps/chosen": -318.75567626953125, + "logps/rejected": -289.80572509765625, + "loss": 0.6833, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": 0.03800877183675766, + "rewards/margins": 0.020139571279287338, + "rewards/rejected": 0.017869198694825172, + "step": 580 + }, + { + "epoch": 0.08, + "learning_rate": 3.856209150326798e-06, + "logits/chosen": -2.68744158744812, + "logits/rejected": -2.597557544708252, + "logps/chosen": -297.6092529296875, + "logps/rejected": -259.01068115234375, + "loss": 0.683, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": 0.03749191015958786, + "rewards/margins": 0.020818348973989487, + "rewards/rejected": 0.016673561185598373, + "step": 590 + }, + { + "epoch": 0.08, + "learning_rate": 3.92156862745098e-06, + "logits/chosen": -2.6166014671325684, + "logits/rejected": -2.5153515338897705, + "logps/chosen": -265.6646728515625, + "logps/rejected": -248.9541473388672, + "loss": 0.686, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": 0.03428620472550392, + "rewards/margins": 0.014840200543403625, + "rewards/rejected": 0.019446004182100296, + "step": 600 + }, + { + "epoch": 0.08, + "learning_rate": 3.986928104575164e-06, + "logits/chosen": -2.6381916999816895, + "logits/rejected": -2.559558868408203, + "logps/chosen": -240.59716796875, + "logps/rejected": -205.5615692138672, + "loss": 0.6846, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": 0.03674664348363876, + "rewards/margins": 0.017332863062620163, + "rewards/rejected": 0.0194137804210186, + "step": 610 + }, + { + "epoch": 0.08, + "learning_rate": 4.052287581699347e-06, + "logits/chosen": -2.628159999847412, + "logits/rejected": -2.564661741256714, + "logps/chosen": -300.2850646972656, + "logps/rejected": -268.270263671875, + "loss": 0.6853, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": 0.03930061683058739, + "rewards/margins": 0.01672186143696308, + "rewards/rejected": 0.022578757256269455, + "step": 620 + }, + { + "epoch": 0.08, + "learning_rate": 4.11764705882353e-06, + "logits/chosen": -2.6141462326049805, + "logits/rejected": -2.5952517986297607, + "logps/chosen": -281.1327819824219, + "logps/rejected": -259.13372802734375, + "loss": 0.6856, + "rewards/accuracies": 0.625, + "rewards/chosen": 0.03937515243887901, + "rewards/margins": 0.015657316893339157, + "rewards/rejected": 0.023717842996120453, + "step": 630 + }, + { + "epoch": 0.08, + "learning_rate": 4.183006535947713e-06, + "logits/chosen": -2.684302806854248, + "logits/rejected": -2.613997220993042, + "logps/chosen": -273.54119873046875, + "logps/rejected": -260.19879150390625, + "loss": 0.683, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": 0.03751297667622566, + "rewards/margins": 0.02131773717701435, + "rewards/rejected": 0.01619523949921131, + "step": 640 + }, + { + "epoch": 0.09, + "learning_rate": 4.2483660130718954e-06, + "logits/chosen": -2.638551950454712, + "logits/rejected": -2.6129698753356934, + "logps/chosen": -271.42877197265625, + "logps/rejected": -257.5237731933594, + "loss": 0.6855, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": 0.02996763028204441, + "rewards/margins": 0.01616152748465538, + "rewards/rejected": 0.01380610466003418, + "step": 650 + }, + { + "epoch": 0.09, + "learning_rate": 4.313725490196079e-06, + "logits/chosen": -2.657040596008301, + "logits/rejected": -2.6005265712738037, + "logps/chosen": -317.86322021484375, + "logps/rejected": -262.32012939453125, + "loss": 0.6827, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": 0.03117840550839901, + "rewards/margins": 0.021571574732661247, + "rewards/rejected": 0.009606831707060337, + "step": 660 + }, + { + "epoch": 0.09, + "learning_rate": 4.379084967320262e-06, + "logits/chosen": -2.7391180992126465, + "logits/rejected": -2.6445021629333496, + "logps/chosen": -301.35858154296875, + "logps/rejected": -317.3902282714844, + "loss": 0.68, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": 0.03976508975028992, + "rewards/margins": 0.02696283720433712, + "rewards/rejected": 0.012802252545952797, + "step": 670 + }, + { + "epoch": 0.09, + "learning_rate": 4.444444444444444e-06, + "logits/chosen": -2.729236602783203, + "logits/rejected": -2.6563024520874023, + "logps/chosen": -289.63568115234375, + "logps/rejected": -266.2176513671875, + "loss": 0.6824, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": 0.04282195121049881, + "rewards/margins": 0.022346172481775284, + "rewards/rejected": 0.020475778728723526, + "step": 680 + }, + { + "epoch": 0.09, + "learning_rate": 4.509803921568628e-06, + "logits/chosen": -2.685117244720459, + "logits/rejected": -2.63557505607605, + "logps/chosen": -311.70892333984375, + "logps/rejected": -271.84857177734375, + "loss": 0.6824, + "rewards/accuracies": 0.75, + "rewards/chosen": 0.03460826724767685, + "rewards/margins": 0.02228071726858616, + "rewards/rejected": 0.01232755184173584, + "step": 690 + }, + { + "epoch": 0.09, + "learning_rate": 4.5751633986928105e-06, + "logits/chosen": -2.667707920074463, + "logits/rejected": -2.654254913330078, + "logps/chosen": -286.30767822265625, + "logps/rejected": -274.2806701660156, + "loss": 0.6803, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": 0.02861057221889496, + "rewards/margins": 0.026673460379242897, + "rewards/rejected": 0.0019371159141883254, + "step": 700 + }, + { + "epoch": 0.09, + "learning_rate": 4.640522875816994e-06, + "logits/chosen": -2.6690964698791504, + "logits/rejected": -2.618622303009033, + "logps/chosen": -293.69586181640625, + "logps/rejected": -277.69219970703125, + "loss": 0.677, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": 0.034188129007816315, + "rewards/margins": 0.03403637558221817, + "rewards/rejected": 0.00015175435692071915, + "step": 710 + }, + { + "epoch": 0.09, + "learning_rate": 4.705882352941177e-06, + "logits/chosen": -2.6516525745391846, + "logits/rejected": -2.6103947162628174, + "logps/chosen": -333.8087158203125, + "logps/rejected": -309.9309387207031, + "loss": 0.6782, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": 0.03538156673312187, + "rewards/margins": 0.03180857375264168, + "rewards/rejected": 0.003572994377464056, + "step": 720 + }, + { + "epoch": 0.1, + "learning_rate": 4.77124183006536e-06, + "logits/chosen": -2.6342673301696777, + "logits/rejected": -2.615257978439331, + "logps/chosen": -294.0147399902344, + "logps/rejected": -282.20562744140625, + "loss": 0.6768, + "rewards/accuracies": 0.6875, + "rewards/chosen": 0.03988462686538696, + "rewards/margins": 0.03406631201505661, + "rewards/rejected": 0.005818313919007778, + "step": 730 + }, + { + "epoch": 0.1, + "learning_rate": 4.836601307189543e-06, + "logits/chosen": -2.627166271209717, + "logits/rejected": -2.5512778759002686, + "logps/chosen": -312.60919189453125, + "logps/rejected": -265.62127685546875, + "loss": 0.6762, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": 0.048726730048656464, + "rewards/margins": 0.03582502156496048, + "rewards/rejected": 0.01290170568972826, + "step": 740 + }, + { + "epoch": 0.1, + "learning_rate": 4.901960784313726e-06, + "logits/chosen": -2.6120457649230957, + "logits/rejected": -2.5613958835601807, + "logps/chosen": -252.6622772216797, + "logps/rejected": -243.0688934326172, + "loss": 0.6721, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": 0.03125419467687607, + "rewards/margins": 0.04380708932876587, + "rewards/rejected": -0.01255289651453495, + "step": 750 + }, + { + "epoch": 0.1, + "learning_rate": 4.967320261437909e-06, + "logits/chosen": -2.653477668762207, + "logits/rejected": -2.5420901775360107, + "logps/chosen": -284.51300048828125, + "logps/rejected": -243.9912109375, + "loss": 0.6685, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": 0.033526089042425156, + "rewards/margins": 0.051363784819841385, + "rewards/rejected": -0.01783769577741623, + "step": 760 + }, + { + "epoch": 0.1, + "learning_rate": 4.99999347843947e-06, + "logits/chosen": -2.6220593452453613, + "logits/rejected": -2.6276183128356934, + "logps/chosen": -297.4268493652344, + "logps/rejected": -277.671875, + "loss": 0.6758, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": 0.026646548882126808, + "rewards/margins": 0.03750267252326012, + "rewards/rejected": -0.010856127366423607, + "step": 770 + }, + { + "epoch": 0.1, + "learning_rate": 4.999941306159375e-06, + "logits/chosen": -2.6617603302001953, + "logits/rejected": -2.5644752979278564, + "logps/chosen": -310.10235595703125, + "logps/rejected": -277.78509521484375, + "loss": 0.661, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": 0.016902543604373932, + "rewards/margins": 0.0686565712094307, + "rewards/rejected": -0.05175402760505676, + "step": 780 + }, + { + "epoch": 0.1, + "learning_rate": 4.999836962687967e-06, + "logits/chosen": -2.5809264183044434, + "logits/rejected": -2.5831658840179443, + "logps/chosen": -296.1529541015625, + "logps/rejected": -299.0150146484375, + "loss": 0.6699, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.0022638884838670492, + "rewards/margins": 0.05030262470245361, + "rewards/rejected": -0.052566517144441605, + "step": 790 + }, + { + "epoch": 0.1, + "learning_rate": 4.999680450202786e-06, + "logits/chosen": -2.4812557697296143, + "logits/rejected": -2.4974255561828613, + "logps/chosen": -275.08343505859375, + "logps/rejected": -295.24969482421875, + "loss": 0.6736, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.010703706182539463, + "rewards/margins": 0.04286975413560867, + "rewards/rejected": -0.05357345938682556, + "step": 800 + }, + { + "epoch": 0.11, + "learning_rate": 4.999471771970087e-06, + "logits/chosen": -2.5986766815185547, + "logits/rejected": -2.5497448444366455, + "logps/chosen": -304.4241027832031, + "logps/rejected": -274.26422119140625, + "loss": 0.6694, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.009451848454773426, + "rewards/margins": 0.0511636957526207, + "rewards/rejected": -0.06061554700136185, + "step": 810 + }, + { + "epoch": 0.11, + "learning_rate": 4.999210932344767e-06, + "logits/chosen": -2.683938503265381, + "logits/rejected": -2.6293458938598633, + "logps/chosen": -324.17730712890625, + "logps/rejected": -284.7932434082031, + "loss": 0.6633, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.0026789132971316576, + "rewards/margins": 0.06414036452770233, + "rewards/rejected": -0.06681927293539047, + "step": 820 + }, + { + "epoch": 0.11, + "learning_rate": 4.998897936770281e-06, + "logits/chosen": -2.5737874507904053, + "logits/rejected": -2.6123242378234863, + "logps/chosen": -250.7256317138672, + "logps/rejected": -282.9770202636719, + "loss": 0.6741, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.033450886607170105, + "rewards/margins": 0.042477719485759735, + "rewards/rejected": -0.07592860609292984, + "step": 830 + }, + { + "epoch": 0.11, + "learning_rate": 4.998532791778521e-06, + "logits/chosen": -2.5229029655456543, + "logits/rejected": -2.441638469696045, + "logps/chosen": -313.8226318359375, + "logps/rejected": -248.78286743164062, + "loss": 0.6595, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.03411540389060974, + "rewards/margins": 0.07376129925251007, + "rewards/rejected": -0.10787669569253922, + "step": 840 + }, + { + "epoch": 0.11, + "learning_rate": 4.9981155049896885e-06, + "logits/chosen": -2.5809288024902344, + "logits/rejected": -2.529169797897339, + "logps/chosen": -299.9226379394531, + "logps/rejected": -322.46075439453125, + "loss": 0.6682, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.05599908158183098, + "rewards/margins": 0.05655177682638168, + "rewards/rejected": -0.11255085468292236, + "step": 850 + }, + { + "epoch": 0.11, + "learning_rate": 4.997646085112126e-06, + "logits/chosen": -2.623964309692383, + "logits/rejected": -2.556356906890869, + "logps/chosen": -265.13909912109375, + "logps/rejected": -266.3385009765625, + "loss": 0.6642, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.05545296519994736, + "rewards/margins": 0.06523346155881882, + "rewards/rejected": -0.12068643420934677, + "step": 860 + }, + { + "epoch": 0.11, + "learning_rate": 4.997124541942141e-06, + "logits/chosen": -2.5906269550323486, + "logits/rejected": -2.548358201980591, + "logps/chosen": -300.20611572265625, + "logps/rejected": -296.0473327636719, + "loss": 0.666, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.01703239232301712, + "rewards/margins": 0.060006629675626755, + "rewards/rejected": -0.07703902572393417, + "step": 870 + }, + { + "epoch": 0.12, + "learning_rate": 4.996550886363801e-06, + "logits/chosen": -2.5821096897125244, + "logits/rejected": -2.533576726913452, + "logps/chosen": -270.1414794921875, + "logps/rejected": -289.24896240234375, + "loss": 0.6758, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.015453094616532326, + "rewards/margins": 0.040653444826602936, + "rewards/rejected": -0.05610654503107071, + "step": 880 + }, + { + "epoch": 0.12, + "learning_rate": 4.995925130348706e-06, + "logits/chosen": -2.612670421600342, + "logits/rejected": -2.5754215717315674, + "logps/chosen": -270.81439208984375, + "logps/rejected": -278.97735595703125, + "loss": 0.6735, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": 0.009032963775098324, + "rewards/margins": 0.0453304797410965, + "rewards/rejected": -0.0362975150346756, + "step": 890 + }, + { + "epoch": 0.12, + "learning_rate": 4.995247286955734e-06, + "logits/chosen": -2.596879482269287, + "logits/rejected": -2.5179409980773926, + "logps/chosen": -263.36505126953125, + "logps/rejected": -231.7235565185547, + "loss": 0.666, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.026518285274505615, + "rewards/margins": 0.06129720062017441, + "rewards/rejected": -0.08781548589468002, + "step": 900 + }, + { + "epoch": 0.12, + "learning_rate": 4.994517370330779e-06, + "logits/chosen": -2.5293946266174316, + "logits/rejected": -2.5589756965637207, + "logps/chosen": -290.03570556640625, + "logps/rejected": -323.95782470703125, + "loss": 0.6561, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.01835092343389988, + "rewards/margins": 0.0834575966000557, + "rewards/rejected": -0.10180851072072983, + "step": 910 + }, + { + "epoch": 0.12, + "learning_rate": 4.993735395706446e-06, + "logits/chosen": -2.588839054107666, + "logits/rejected": -2.508923053741455, + "logps/chosen": -245.86978149414062, + "logps/rejected": -251.2750701904297, + "loss": 0.6497, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.008198278024792671, + "rewards/margins": 0.098419688642025, + "rewards/rejected": -0.10661796480417252, + "step": 920 + }, + { + "epoch": 0.12, + "learning_rate": 4.992901379401737e-06, + "logits/chosen": -2.655158519744873, + "logits/rejected": -2.6311938762664795, + "logps/chosen": -325.3824157714844, + "logps/rejected": -301.60150146484375, + "loss": 0.657, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": 0.017573771998286247, + "rewards/margins": 0.08263130486011505, + "rewards/rejected": -0.06505753099918365, + "step": 930 + }, + { + "epoch": 0.12, + "learning_rate": 4.992015338821711e-06, + "logits/chosen": -2.5817184448242188, + "logits/rejected": -2.5037972927093506, + "logps/chosen": -297.3199768066406, + "logps/rejected": -289.2392578125, + "loss": 0.6555, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.008880408480763435, + "rewards/margins": 0.08480361104011536, + "rewards/rejected": -0.09368401765823364, + "step": 940 + }, + { + "epoch": 0.12, + "learning_rate": 4.991077292457117e-06, + "logits/chosen": -2.609684705734253, + "logits/rejected": -2.5322821140289307, + "logps/chosen": -291.2019958496094, + "logps/rejected": -267.49969482421875, + "loss": 0.6496, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.008642762899398804, + "rewards/margins": 0.09803656488656998, + "rewards/rejected": -0.10667933523654938, + "step": 950 + }, + { + "epoch": 0.13, + "learning_rate": 4.990087259884016e-06, + "logits/chosen": -2.6076533794403076, + "logits/rejected": -2.554643154144287, + "logps/chosen": -267.25732421875, + "logps/rejected": -248.2394561767578, + "loss": 0.6481, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.03241444751620293, + "rewards/margins": 0.10384710133075714, + "rewards/rejected": -0.13626155257225037, + "step": 960 + }, + { + "epoch": 0.13, + "learning_rate": 4.989045261763362e-06, + "logits/chosen": -2.595069646835327, + "logits/rejected": -2.5618221759796143, + "logps/chosen": -297.8432922363281, + "logps/rejected": -288.5405578613281, + "loss": 0.6413, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.01923452876508236, + "rewards/margins": 0.11537112295627594, + "rewards/rejected": -0.13460564613342285, + "step": 970 + }, + { + "epoch": 0.13, + "learning_rate": 4.98795131984058e-06, + "logits/chosen": -2.508185625076294, + "logits/rejected": -2.502739667892456, + "logps/chosen": -274.4787902832031, + "logps/rejected": -250.0324249267578, + "loss": 0.6503, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.030040761455893517, + "rewards/margins": 0.09917756170034409, + "rewards/rejected": -0.12921833992004395, + "step": 980 + }, + { + "epoch": 0.13, + "learning_rate": 4.986805456945107e-06, + "logits/chosen": -2.5590319633483887, + "logits/rejected": -2.514545440673828, + "logps/chosen": -238.1454620361328, + "logps/rejected": -229.8460235595703, + "loss": 0.6393, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": 0.031098058447241783, + "rewards/margins": 0.12257441133260727, + "rewards/rejected": -0.09147634357213974, + "step": 990 + }, + { + "epoch": 0.13, + "learning_rate": 4.985607696989919e-06, + "logits/chosen": -2.6065621376037598, + "logits/rejected": -2.605027675628662, + "logps/chosen": -287.16986083984375, + "logps/rejected": -285.27789306640625, + "loss": 0.6501, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.012579170055687428, + "rewards/margins": 0.10108743607997894, + "rewards/rejected": -0.1136665940284729, + "step": 1000 + }, + { + "epoch": 0.13, + "learning_rate": 4.984358064971026e-06, + "logits/chosen": -2.560281753540039, + "logits/rejected": -2.490305185317993, + "logps/chosen": -288.9957580566406, + "logps/rejected": -264.52166748046875, + "loss": 0.6591, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.08325138688087463, + "rewards/margins": 0.08632998168468475, + "rewards/rejected": -0.1695813685655594, + "step": 1010 + }, + { + "epoch": 0.13, + "learning_rate": 4.983056586966958e-06, + "logits/chosen": -2.5651133060455322, + "logits/rejected": -2.4909870624542236, + "logps/chosen": -334.9750061035156, + "logps/rejected": -276.69317626953125, + "loss": 0.6293, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.04752253741025925, + "rewards/margins": 0.1456046998500824, + "rewards/rejected": -0.19312722980976105, + "step": 1020 + }, + { + "epoch": 0.13, + "learning_rate": 4.981703290138215e-06, + "logits/chosen": -2.631713390350342, + "logits/rejected": -2.581732988357544, + "logps/chosen": -326.59320068359375, + "logps/rejected": -311.9397888183594, + "loss": 0.6382, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.05011386424303055, + "rewards/margins": 0.1374647617340088, + "rewards/rejected": -0.18757864832878113, + "step": 1030 + }, + { + "epoch": 0.14, + "learning_rate": 4.980298202726706e-06, + "logits/chosen": -2.691251516342163, + "logits/rejected": -2.624222993850708, + "logps/chosen": -328.1868591308594, + "logps/rejected": -298.66473388671875, + "loss": 0.6088, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.007155093364417553, + "rewards/margins": 0.19580598175525665, + "rewards/rejected": -0.20296108722686768, + "step": 1040 + }, + { + "epoch": 0.14, + "learning_rate": 4.978841354055148e-06, + "logits/chosen": -2.548856019973755, + "logits/rejected": -2.522235631942749, + "logps/chosen": -281.12835693359375, + "logps/rejected": -274.5442199707031, + "loss": 0.6485, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.0306741651147604, + "rewards/margins": 0.1061779260635376, + "rewards/rejected": -0.13685208559036255, + "step": 1050 + }, + { + "epoch": 0.14, + "learning_rate": 4.977332774526471e-06, + "logits/chosen": -2.5694425106048584, + "logits/rejected": -2.4372596740722656, + "logps/chosen": -288.7796325683594, + "logps/rejected": -246.0713653564453, + "loss": 0.6226, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": 0.016771093010902405, + "rewards/margins": 0.16713151335716248, + "rewards/rejected": -0.15036042034626007, + "step": 1060 + }, + { + "epoch": 0.14, + "learning_rate": 4.97577249562317e-06, + "logits/chosen": -2.4907641410827637, + "logits/rejected": -2.469911813735962, + "logps/chosen": -273.491455078125, + "logps/rejected": -273.80584716796875, + "loss": 0.6279, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": 0.045688532292842865, + "rewards/margins": 0.15459074079990387, + "rewards/rejected": -0.10890217870473862, + "step": 1070 + }, + { + "epoch": 0.14, + "learning_rate": 4.974160549906652e-06, + "logits/chosen": -2.4853806495666504, + "logits/rejected": -2.428475856781006, + "logps/chosen": -265.21478271484375, + "logps/rejected": -256.92156982421875, + "loss": 0.6438, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.01522951852530241, + "rewards/margins": 0.12189844995737076, + "rewards/rejected": -0.13712796568870544, + "step": 1080 + }, + { + "epoch": 0.14, + "learning_rate": 4.972496971016559e-06, + "logits/chosen": -2.506234884262085, + "logits/rejected": -2.49314546585083, + "logps/chosen": -270.284912109375, + "logps/rejected": -250.1904754638672, + "loss": 0.6173, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": 0.011768155731260777, + "rewards/margins": 0.1889091283082962, + "rewards/rejected": -0.1771409809589386, + "step": 1090 + }, + { + "epoch": 0.14, + "learning_rate": 4.9707817936700635e-06, + "logits/chosen": -2.5315027236938477, + "logits/rejected": -2.4613234996795654, + "logps/chosen": -289.7018737792969, + "logps/rejected": -279.943115234375, + "loss": 0.6455, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.04312612861394882, + "rewards/margins": 0.1247950941324234, + "rewards/rejected": -0.16792121529579163, + "step": 1100 + }, + { + "epoch": 0.15, + "learning_rate": 4.969015053661142e-06, + "logits/chosen": -2.532196521759033, + "logits/rejected": -2.4918360710144043, + "logps/chosen": -304.68524169921875, + "logps/rejected": -285.6064453125, + "loss": 0.657, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.10881493240594864, + "rewards/margins": 0.09789443016052246, + "rewards/rejected": -0.2067093551158905, + "step": 1110 + }, + { + "epoch": 0.15, + "learning_rate": 4.967196787859835e-06, + "logits/chosen": -2.560375928878784, + "logits/rejected": -2.490992784500122, + "logps/chosen": -307.4469299316406, + "logps/rejected": -310.51800537109375, + "loss": 0.644, + "rewards/accuracies": 0.6000000238418579, + "rewards/chosen": -0.09717439115047455, + "rewards/margins": 0.12193155288696289, + "rewards/rejected": -0.21910595893859863, + "step": 1120 + }, + { + "epoch": 0.15, + "learning_rate": 4.965327034211469e-06, + "logits/chosen": -2.5642170906066895, + "logits/rejected": -2.4959988594055176, + "logps/chosen": -319.49249267578125, + "logps/rejected": -290.97015380859375, + "loss": 0.6031, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.07515348494052887, + "rewards/margins": 0.2159455269575119, + "rewards/rejected": -0.29109901189804077, + "step": 1130 + }, + { + "epoch": 0.15, + "learning_rate": 4.96340583173587e-06, + "logits/chosen": -2.554924249649048, + "logits/rejected": -2.4227776527404785, + "logps/chosen": -267.262451171875, + "logps/rejected": -251.24148559570312, + "loss": 0.6181, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.12634123861789703, + "rewards/margins": 0.18214096128940582, + "rewards/rejected": -0.30848219990730286, + "step": 1140 + }, + { + "epoch": 0.15, + "learning_rate": 4.96143322052655e-06, + "logits/chosen": -2.5284454822540283, + "logits/rejected": -2.573359251022339, + "logps/chosen": -309.14788818359375, + "logps/rejected": -300.38909912109375, + "loss": 0.6614, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.07307339459657669, + "rewards/margins": 0.09202548116445541, + "rewards/rejected": -0.1650988608598709, + "step": 1150 + }, + { + "epoch": 0.15, + "learning_rate": 4.959409241749864e-06, + "logits/chosen": -2.555816650390625, + "logits/rejected": -2.461007595062256, + "logps/chosen": -291.8132629394531, + "logps/rejected": -305.55926513671875, + "loss": 0.6302, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.036139845848083496, + "rewards/margins": 0.15513625741004944, + "rewards/rejected": -0.19127610325813293, + "step": 1160 + }, + { + "epoch": 0.15, + "learning_rate": 4.957333937644159e-06, + "logits/chosen": -2.5880494117736816, + "logits/rejected": -2.537034511566162, + "logps/chosen": -310.62689208984375, + "logps/rejected": -296.87628173828125, + "loss": 0.6264, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.07262597978115082, + "rewards/margins": 0.16844379901885986, + "rewards/rejected": -0.24106979370117188, + "step": 1170 + }, + { + "epoch": 0.15, + "learning_rate": 4.955207351518885e-06, + "logits/chosen": -2.5874555110931396, + "logits/rejected": -2.574850559234619, + "logps/chosen": -288.90997314453125, + "logps/rejected": -264.8959045410156, + "loss": 0.641, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.11980124562978745, + "rewards/margins": 0.13190236687660217, + "rewards/rejected": -0.2517036199569702, + "step": 1180 + }, + { + "epoch": 0.16, + "learning_rate": 4.953029527753699e-06, + "logits/chosen": -2.5432021617889404, + "logits/rejected": -2.5398898124694824, + "logps/chosen": -270.48175048828125, + "logps/rejected": -346.1780090332031, + "loss": 0.6798, + "rewards/accuracies": 0.6000000238418579, + "rewards/chosen": -0.2177748680114746, + "rewards/margins": 0.06020503118634224, + "rewards/rejected": -0.27797991037368774, + "step": 1190 + }, + { + "epoch": 0.16, + "learning_rate": 4.95080051179753e-06, + "logits/chosen": -2.4899439811706543, + "logits/rejected": -2.48333740234375, + "logps/chosen": -287.114013671875, + "logps/rejected": -292.2990417480469, + "loss": 0.6561, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.24787192046642303, + "rewards/margins": 0.09571520984172821, + "rewards/rejected": -0.34358716011047363, + "step": 1200 + }, + { + "epoch": 0.16, + "learning_rate": 4.948520350167637e-06, + "logits/chosen": -2.502089738845825, + "logits/rejected": -2.456737995147705, + "logps/chosen": -306.0370788574219, + "logps/rejected": -303.1842346191406, + "loss": 0.6316, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.2280178964138031, + "rewards/margins": 0.15330472588539124, + "rewards/rejected": -0.38132262229919434, + "step": 1210 + }, + { + "epoch": 0.16, + "learning_rate": 4.946189090448639e-06, + "logits/chosen": -2.4915668964385986, + "logits/rejected": -2.4508776664733887, + "logps/chosen": -318.48333740234375, + "logps/rejected": -276.1875, + "loss": 0.6827, + "rewards/accuracies": 0.5874999761581421, + "rewards/chosen": -0.28770923614501953, + "rewards/margins": 0.04758835583925247, + "rewards/rejected": -0.3352976143360138, + "step": 1220 + }, + { + "epoch": 0.16, + "learning_rate": 4.943806781291515e-06, + "logits/chosen": -2.53739070892334, + "logits/rejected": -2.490952491760254, + "logps/chosen": -289.3685607910156, + "logps/rejected": -271.5808410644531, + "loss": 0.5964, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.12012603133916855, + "rewards/margins": 0.23385348916053772, + "rewards/rejected": -0.35397952795028687, + "step": 1230 + }, + { + "epoch": 0.16, + "learning_rate": 4.941373472412595e-06, + "logits/chosen": -2.5729668140411377, + "logits/rejected": -2.527350902557373, + "logps/chosen": -339.57049560546875, + "logps/rejected": -328.00103759765625, + "loss": 0.638, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.07913851737976074, + "rewards/margins": 0.15087315440177917, + "rewards/rejected": -0.23001165688037872, + "step": 1240 + }, + { + "epoch": 0.16, + "learning_rate": 4.938889214592521e-06, + "logits/chosen": -2.572094678878784, + "logits/rejected": -2.5283236503601074, + "logps/chosen": -339.66009521484375, + "logps/rejected": -312.3247375488281, + "loss": 0.6597, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.002257655607536435, + "rewards/margins": 0.09324416518211365, + "rewards/rejected": -0.09550182521343231, + "step": 1250 + }, + { + "epoch": 0.16, + "learning_rate": 4.936354059675186e-06, + "logits/chosen": -2.5886847972869873, + "logits/rejected": -2.5199790000915527, + "logps/chosen": -288.98883056640625, + "logps/rejected": -260.17230224609375, + "loss": 0.6458, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": 0.006508971564471722, + "rewards/margins": 0.12008893489837646, + "rewards/rejected": -0.11357997357845306, + "step": 1260 + }, + { + "epoch": 0.17, + "learning_rate": 4.933768060566654e-06, + "logits/chosen": -2.577394485473633, + "logits/rejected": -2.5192856788635254, + "logps/chosen": -280.8554992675781, + "logps/rejected": -268.8490905761719, + "loss": 0.6352, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": 0.03389470651745796, + "rewards/margins": 0.13913270831108093, + "rewards/rejected": -0.10523799806833267, + "step": 1270 + }, + { + "epoch": 0.17, + "learning_rate": 4.931131271234052e-06, + "logits/chosen": -2.534980535507202, + "logits/rejected": -2.440115451812744, + "logps/chosen": -279.3022155761719, + "logps/rejected": -248.8036651611328, + "loss": 0.6346, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.05754392594099045, + "rewards/margins": 0.14556458592414856, + "rewards/rejected": -0.2031085044145584, + "step": 1280 + }, + { + "epoch": 0.17, + "learning_rate": 4.928443746704448e-06, + "logits/chosen": -2.5330698490142822, + "logits/rejected": -2.457183361053467, + "logps/chosen": -332.3924560546875, + "logps/rejected": -298.8351745605469, + "loss": 0.6147, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.10134587436914444, + "rewards/margins": 0.19787076115608215, + "rewards/rejected": -0.299216628074646, + "step": 1290 + }, + { + "epoch": 0.17, + "learning_rate": 4.925705543063703e-06, + "logits/chosen": -2.470041513442993, + "logits/rejected": -2.4728071689605713, + "logps/chosen": -304.2667541503906, + "logps/rejected": -353.01348876953125, + "loss": 0.6143, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.10572239011526108, + "rewards/margins": 0.19867298007011414, + "rewards/rejected": -0.3043953776359558, + "step": 1300 + }, + { + "epoch": 0.17, + "learning_rate": 4.922916717455297e-06, + "logits/chosen": -2.537431478500366, + "logits/rejected": -2.429407835006714, + "logps/chosen": -330.64190673828125, + "logps/rejected": -278.8482971191406, + "loss": 0.6321, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.17234036326408386, + "rewards/margins": 0.16186991333961487, + "rewards/rejected": -0.33421024680137634, + "step": 1310 + }, + { + "epoch": 0.17, + "learning_rate": 4.920077328079136e-06, + "logits/chosen": -2.5637619495391846, + "logits/rejected": -2.483635425567627, + "logps/chosen": -323.4110412597656, + "logps/rejected": -288.7492370605469, + "loss": 0.6395, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.1858586072921753, + "rewards/margins": 0.14614257216453552, + "rewards/rejected": -0.3320012092590332, + "step": 1320 + }, + { + "epoch": 0.17, + "learning_rate": 4.9171874341903445e-06, + "logits/chosen": -2.5309529304504395, + "logits/rejected": -2.465804100036621, + "logps/chosen": -304.7217102050781, + "logps/rejected": -324.9732360839844, + "loss": 0.6059, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.1965772658586502, + "rewards/margins": 0.21101978421211243, + "rewards/rejected": -0.40759706497192383, + "step": 1330 + }, + { + "epoch": 0.18, + "learning_rate": 4.914247096098019e-06, + "logits/chosen": -2.5039734840393066, + "logits/rejected": -2.4910902976989746, + "logps/chosen": -294.86187744140625, + "logps/rejected": -305.4557189941406, + "loss": 0.6585, + "rewards/accuracies": 0.5874999761581421, + "rewards/chosen": -0.13758406043052673, + "rewards/margins": 0.10470825433731079, + "rewards/rejected": -0.24229232966899872, + "step": 1340 + }, + { + "epoch": 0.18, + "learning_rate": 4.911256375163977e-06, + "logits/chosen": -2.415588855743408, + "logits/rejected": -2.4556922912597656, + "logps/chosen": -239.3975372314453, + "logps/rejected": -279.1402893066406, + "loss": 0.6237, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.030233344063162804, + "rewards/margins": 0.19135653972625732, + "rewards/rejected": -0.22158987820148468, + "step": 1350 + }, + { + "epoch": 0.18, + "learning_rate": 4.908215333801474e-06, + "logits/chosen": -2.484656810760498, + "logits/rejected": -2.446765422821045, + "logps/chosen": -269.4073181152344, + "logps/rejected": -249.4739227294922, + "loss": 0.628, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.050433941185474396, + "rewards/margins": 0.16900701820850372, + "rewards/rejected": -0.21944093704223633, + "step": 1360 + }, + { + "epoch": 0.18, + "learning_rate": 4.9051240354739004e-06, + "logits/chosen": -2.4662208557128906, + "logits/rejected": -2.3930366039276123, + "logps/chosen": -333.33831787109375, + "logps/rejected": -317.85369873046875, + "loss": 0.6583, + "rewards/accuracies": 0.5874999761581421, + "rewards/chosen": -0.16577494144439697, + "rewards/margins": 0.1024014800786972, + "rewards/rejected": -0.268176406621933, + "step": 1370 + }, + { + "epoch": 0.18, + "learning_rate": 4.901982544693457e-06, + "logits/chosen": -2.519670248031616, + "logits/rejected": -2.4658124446868896, + "logps/chosen": -322.218017578125, + "logps/rejected": -315.02655029296875, + "loss": 0.5962, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.07793796062469482, + "rewards/margins": 0.24866263568401337, + "rewards/rejected": -0.3266006112098694, + "step": 1380 + }, + { + "epoch": 0.18, + "learning_rate": 4.898790927019809e-06, + "logits/chosen": -2.404968500137329, + "logits/rejected": -2.3655800819396973, + "logps/chosen": -303.9695739746094, + "logps/rejected": -311.675048828125, + "loss": 0.643, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.19383570551872253, + "rewards/margins": 0.14913451671600342, + "rewards/rejected": -0.34297025203704834, + "step": 1390 + }, + { + "epoch": 0.18, + "learning_rate": 4.895549249058718e-06, + "logits/chosen": -2.41062593460083, + "logits/rejected": -2.2885286808013916, + "logps/chosen": -290.38861083984375, + "logps/rejected": -284.9938659667969, + "loss": 0.6138, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.11517498642206192, + "rewards/margins": 0.2158602476119995, + "rewards/rejected": -0.33103522658348083, + "step": 1400 + }, + { + "epoch": 0.18, + "learning_rate": 4.892257578460656e-06, + "logits/chosen": -2.48399019241333, + "logits/rejected": -2.440911054611206, + "logps/chosen": -288.1855163574219, + "logps/rejected": -292.6690979003906, + "loss": 0.6476, + "rewards/accuracies": 0.550000011920929, + "rewards/chosen": -0.10677523910999298, + "rewards/margins": 0.14951439201831818, + "rewards/rejected": -0.25628960132598877, + "step": 1410 + }, + { + "epoch": 0.19, + "learning_rate": 4.888915983919383e-06, + "logits/chosen": -2.519620656967163, + "logits/rejected": -2.4930050373077393, + "logps/chosen": -333.12139892578125, + "logps/rejected": -337.2662048339844, + "loss": 0.6077, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.0033730086870491505, + "rewards/margins": 0.21503393352031708, + "rewards/rejected": -0.21840694546699524, + "step": 1420 + }, + { + "epoch": 0.19, + "learning_rate": 4.885524535170525e-06, + "logits/chosen": -2.4327306747436523, + "logits/rejected": -2.4226138591766357, + "logps/chosen": -300.2519836425781, + "logps/rejected": -272.81097412109375, + "loss": 0.6407, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.0004309698997531086, + "rewards/margins": 0.138811856508255, + "rewards/rejected": -0.13924282789230347, + "step": 1430 + }, + { + "epoch": 0.19, + "learning_rate": 4.882083302990113e-06, + "logits/chosen": -2.5328216552734375, + "logits/rejected": -2.519892692565918, + "logps/chosen": -327.641845703125, + "logps/rejected": -308.94189453125, + "loss": 0.6132, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": 0.06218453496694565, + "rewards/margins": 0.19737879931926727, + "rewards/rejected": -0.13519425690174103, + "step": 1440 + }, + { + "epoch": 0.19, + "learning_rate": 4.878592359193104e-06, + "logits/chosen": -2.459606647491455, + "logits/rejected": -2.4827868938446045, + "logps/chosen": -277.9227294921875, + "logps/rejected": -268.23663330078125, + "loss": 0.6152, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.019242966547608376, + "rewards/margins": 0.19757792353630066, + "rewards/rejected": -0.2168208807706833, + "step": 1450 + }, + { + "epoch": 0.19, + "learning_rate": 4.875051776631888e-06, + "logits/chosen": -2.4710533618927, + "logits/rejected": -2.410731554031372, + "logps/chosen": -368.6264343261719, + "logps/rejected": -335.9675598144531, + "loss": 0.6227, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.08454436808824539, + "rewards/margins": 0.19244490563869476, + "rewards/rejected": -0.27698928117752075, + "step": 1460 + }, + { + "epoch": 0.19, + "learning_rate": 4.871461629194764e-06, + "logits/chosen": -2.4960079193115234, + "logits/rejected": -2.3759419918060303, + "logps/chosen": -280.95648193359375, + "logps/rejected": -256.5384216308594, + "loss": 0.6421, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.14747636020183563, + "rewards/margins": 0.1381666362285614, + "rewards/rejected": -0.28564298152923584, + "step": 1470 + }, + { + "epoch": 0.19, + "learning_rate": 4.8678219918043984e-06, + "logits/chosen": -2.4740777015686035, + "logits/rejected": -2.385272979736328, + "logps/chosen": -293.96856689453125, + "logps/rejected": -260.2104797363281, + "loss": 0.634, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.10929224640130997, + "rewards/margins": 0.16633400321006775, + "rewards/rejected": -0.2756262421607971, + "step": 1480 + }, + { + "epoch": 0.19, + "learning_rate": 4.864132940416262e-06, + "logits/chosen": -2.496289014816284, + "logits/rejected": -2.3923065662384033, + "logps/chosen": -326.35113525390625, + "logps/rejected": -289.10894775390625, + "loss": 0.5945, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.05443965643644333, + "rewards/margins": 0.24763624370098114, + "rewards/rejected": -0.30207592248916626, + "step": 1490 + }, + { + "epoch": 0.2, + "learning_rate": 4.860394552017044e-06, + "logits/chosen": -2.4541478157043457, + "logits/rejected": -2.436887741088867, + "logps/chosen": -308.11859130859375, + "logps/rejected": -337.92633056640625, + "loss": 0.5989, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.14253541827201843, + "rewards/margins": 0.24313051998615265, + "rewards/rejected": -0.3856659531593323, + "step": 1500 + }, + { + "epoch": 0.2, + "learning_rate": 4.856606904623047e-06, + "logits/chosen": -2.4310736656188965, + "logits/rejected": -2.4193150997161865, + "logps/chosen": -250.90640258789062, + "logps/rejected": -278.09735107421875, + "loss": 0.6456, + "rewards/accuracies": 0.6000000238418579, + "rewards/chosen": -0.1735648810863495, + "rewards/margins": 0.14540037512779236, + "rewards/rejected": -0.31896525621414185, + "step": 1510 + }, + { + "epoch": 0.2, + "learning_rate": 4.852770077278557e-06, + "logits/chosen": -2.414815902709961, + "logits/rejected": -2.370915174484253, + "logps/chosen": -294.3868713378906, + "logps/rejected": -282.88751220703125, + "loss": 0.5838, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.11077413707971573, + "rewards/margins": 0.2743500769138336, + "rewards/rejected": -0.3851242661476135, + "step": 1520 + }, + { + "epoch": 0.2, + "learning_rate": 4.848884150054196e-06, + "logits/chosen": -2.4590086936950684, + "logits/rejected": -2.4183783531188965, + "logps/chosen": -283.24932861328125, + "logps/rejected": -260.66949462890625, + "loss": 0.6576, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.1314813196659088, + "rewards/margins": 0.12110692262649536, + "rewards/rejected": -0.2525882422924042, + "step": 1530 + }, + { + "epoch": 0.2, + "learning_rate": 4.8449492040452495e-06, + "logits/chosen": -2.4285058975219727, + "logits/rejected": -2.392300844192505, + "logps/chosen": -307.29901123046875, + "logps/rejected": -276.5957946777344, + "loss": 0.6042, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.09597836434841156, + "rewards/margins": 0.2392992526292801, + "rewards/rejected": -0.33527761697769165, + "step": 1540 + }, + { + "epoch": 0.2, + "learning_rate": 4.840965321369973e-06, + "logits/chosen": -2.4373369216918945, + "logits/rejected": -2.3086016178131104, + "logps/chosen": -297.9757080078125, + "logps/rejected": -282.6000061035156, + "loss": 0.6129, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.10073225200176239, + "rewards/margins": 0.20788928866386414, + "rewards/rejected": -0.3086215853691101, + "step": 1550 + }, + { + "epoch": 0.2, + "learning_rate": 4.8369325851678795e-06, + "logits/chosen": -2.4190502166748047, + "logits/rejected": -2.352062702178955, + "logps/chosen": -311.42169189453125, + "logps/rejected": -303.9734191894531, + "loss": 0.604, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.15880802273750305, + "rewards/margins": 0.22597630321979523, + "rewards/rejected": -0.3847842812538147, + "step": 1560 + }, + { + "epoch": 0.21, + "learning_rate": 4.832851079598007e-06, + "logits/chosen": -2.4244778156280518, + "logits/rejected": -2.3742895126342773, + "logps/chosen": -269.66949462890625, + "logps/rejected": -303.1509704589844, + "loss": 0.6242, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.141389861702919, + "rewards/margins": 0.1891460120677948, + "rewards/rejected": -0.3305358290672302, + "step": 1570 + }, + { + "epoch": 0.21, + "learning_rate": 4.828720889837158e-06, + "logits/chosen": -2.3640172481536865, + "logits/rejected": -2.316593647003174, + "logps/chosen": -301.21771240234375, + "logps/rejected": -269.7645568847656, + "loss": 0.6379, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.14845463633537292, + "rewards/margins": 0.16235341131687164, + "rewards/rejected": -0.3108080327510834, + "step": 1580 + }, + { + "epoch": 0.21, + "learning_rate": 4.824542102078125e-06, + "logits/chosen": -2.3453826904296875, + "logits/rejected": -2.334388494491577, + "logps/chosen": -284.1679992675781, + "logps/rejected": -272.05841064453125, + "loss": 0.6336, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.1061963215470314, + "rewards/margins": 0.1646105945110321, + "rewards/rejected": -0.2708069086074829, + "step": 1590 + }, + { + "epoch": 0.21, + "learning_rate": 4.820314803527888e-06, + "logits/chosen": -2.325477361679077, + "logits/rejected": -2.23325777053833, + "logps/chosen": -278.75384521484375, + "logps/rejected": -280.7860412597656, + "loss": 0.6183, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.20397250354290009, + "rewards/margins": 0.19710715115070343, + "rewards/rejected": -0.4010796546936035, + "step": 1600 + }, + { + "epoch": 0.21, + "learning_rate": 4.816039082405799e-06, + "logits/chosen": -2.3431990146636963, + "logits/rejected": -2.311972141265869, + "logps/chosen": -300.4079895019531, + "logps/rejected": -313.13934326171875, + "loss": 0.6239, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.19499923288822174, + "rewards/margins": 0.19155170023441315, + "rewards/rejected": -0.3865508735179901, + "step": 1610 + }, + { + "epoch": 0.21, + "learning_rate": 4.81171502794174e-06, + "logits/chosen": -2.419560194015503, + "logits/rejected": -2.391439914703369, + "logps/chosen": -334.3507995605469, + "logps/rejected": -320.0035095214844, + "loss": 0.6251, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.09380961209535599, + "rewards/margins": 0.1967724859714508, + "rewards/rejected": -0.2905820608139038, + "step": 1620 + }, + { + "epoch": 0.21, + "learning_rate": 4.8073427303742584e-06, + "logits/chosen": -2.4514782428741455, + "logits/rejected": -2.3883414268493652, + "logps/chosen": -251.7395477294922, + "logps/rejected": -262.00921630859375, + "loss": 0.5929, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.031386177986860275, + "rewards/margins": 0.2520635724067688, + "rewards/rejected": -0.2834497392177582, + "step": 1630 + }, + { + "epoch": 0.21, + "learning_rate": 4.802922280948685e-06, + "logits/chosen": -2.3628644943237305, + "logits/rejected": -2.358330011367798, + "logps/chosen": -255.8154296875, + "logps/rejected": -309.6222229003906, + "loss": 0.6305, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.105247363448143, + "rewards/margins": 0.1670643836259842, + "rewards/rejected": -0.2723117470741272, + "step": 1640 + }, + { + "epoch": 0.22, + "learning_rate": 4.798453771915231e-06, + "logits/chosen": -2.4843735694885254, + "logits/rejected": -2.3918941020965576, + "logps/chosen": -300.1679992675781, + "logps/rejected": -294.6748962402344, + "loss": 0.6062, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.10460405051708221, + "rewards/margins": 0.22866663336753845, + "rewards/rejected": -0.33327072858810425, + "step": 1650 + }, + { + "epoch": 0.22, + "learning_rate": 4.793937296527062e-06, + "logits/chosen": -2.4866888523101807, + "logits/rejected": -2.395613193511963, + "logps/chosen": -351.20257568359375, + "logps/rejected": -353.2774963378906, + "loss": 0.639, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.1539994478225708, + "rewards/margins": 0.17863675951957703, + "rewards/rejected": -0.33263617753982544, + "step": 1660 + }, + { + "epoch": 0.22, + "learning_rate": 4.78937294903835e-06, + "logits/chosen": -2.437178134918213, + "logits/rejected": -2.33906888961792, + "logps/chosen": -280.81884765625, + "logps/rejected": -280.7358093261719, + "loss": 0.5956, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.07330270111560822, + "rewards/margins": 0.2512477934360504, + "rewards/rejected": -0.32455044984817505, + "step": 1670 + }, + { + "epoch": 0.22, + "learning_rate": 4.78476082470231e-06, + "logits/chosen": -2.445133686065674, + "logits/rejected": -2.3581621646881104, + "logps/chosen": -282.26776123046875, + "logps/rejected": -274.84368896484375, + "loss": 0.6238, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.15945813059806824, + "rewards/margins": 0.19288024306297302, + "rewards/rejected": -0.35233837366104126, + "step": 1680 + }, + { + "epoch": 0.22, + "learning_rate": 4.780101019769212e-06, + "logits/chosen": -2.411268949508667, + "logits/rejected": -2.3567655086517334, + "logps/chosen": -329.9833984375, + "logps/rejected": -304.2483825683594, + "loss": 0.6298, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.16943173110485077, + "rewards/margins": 0.19025549292564392, + "rewards/rejected": -0.3596872091293335, + "step": 1690 + }, + { + "epoch": 0.22, + "learning_rate": 4.775393631484368e-06, + "logits/chosen": -2.3584446907043457, + "logits/rejected": -2.2917559146881104, + "logps/chosen": -315.8598937988281, + "logps/rejected": -333.73553466796875, + "loss": 0.5791, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.11365552991628647, + "rewards/margins": 0.29867830872535706, + "rewards/rejected": -0.4123338758945465, + "step": 1700 + }, + { + "epoch": 0.22, + "learning_rate": 4.770638758086105e-06, + "logits/chosen": -2.3838050365448, + "logits/rejected": -2.3302836418151855, + "logps/chosen": -302.39459228515625, + "logps/rejected": -289.2746887207031, + "loss": 0.632, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.17028506100177765, + "rewards/margins": 0.17936842143535614, + "rewards/rejected": -0.3496534526348114, + "step": 1710 + }, + { + "epoch": 0.23, + "learning_rate": 4.7658364988037184e-06, + "logits/chosen": -2.439105987548828, + "logits/rejected": -2.3976969718933105, + "logps/chosen": -337.58843994140625, + "logps/rejected": -325.3658142089844, + "loss": 0.6464, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.26347967982292175, + "rewards/margins": 0.14311228692531586, + "rewards/rejected": -0.4065919816493988, + "step": 1720 + }, + { + "epoch": 0.23, + "learning_rate": 4.760986953855395e-06, + "logits/chosen": -2.4409725666046143, + "logits/rejected": -2.438054323196411, + "logps/chosen": -294.93121337890625, + "logps/rejected": -326.00799560546875, + "loss": 0.6005, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2294178009033203, + "rewards/margins": 0.24227671325206757, + "rewards/rejected": -0.4716945290565491, + "step": 1730 + }, + { + "epoch": 0.23, + "learning_rate": 4.756090224446127e-06, + "logits/chosen": -2.4490959644317627, + "logits/rejected": -2.367088794708252, + "logps/chosen": -332.90606689453125, + "logps/rejected": -326.7022705078125, + "loss": 0.6089, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.08716649562120438, + "rewards/margins": 0.22671692073345184, + "rewards/rejected": -0.3138834536075592, + "step": 1740 + }, + { + "epoch": 0.23, + "learning_rate": 4.7511464127655945e-06, + "logits/chosen": -2.4698281288146973, + "logits/rejected": -2.319673538208008, + "logps/chosen": -297.69561767578125, + "logps/rejected": -289.238525390625, + "loss": 0.5916, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.1407977044582367, + "rewards/margins": 0.27011579275131226, + "rewards/rejected": -0.41091352701187134, + "step": 1750 + }, + { + "epoch": 0.23, + "learning_rate": 4.74615562198604e-06, + "logits/chosen": -2.367678165435791, + "logits/rejected": -2.2320752143859863, + "logps/chosen": -279.7342529296875, + "logps/rejected": -272.38128662109375, + "loss": 0.5787, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.07772514969110489, + "rewards/margins": 0.30095797777175903, + "rewards/rejected": -0.3786831200122833, + "step": 1760 + }, + { + "epoch": 0.23, + "learning_rate": 4.741117956260107e-06, + "logits/chosen": -2.333146333694458, + "logits/rejected": -2.2740590572357178, + "logps/chosen": -286.67266845703125, + "logps/rejected": -306.8995361328125, + "loss": 0.6015, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.05592029169201851, + "rewards/margins": 0.24995534121990204, + "rewards/rejected": -0.30587562918663025, + "step": 1770 + }, + { + "epoch": 0.23, + "learning_rate": 4.736033520718672e-06, + "logits/chosen": -2.433627128601074, + "logits/rejected": -2.277806043624878, + "logps/chosen": -322.6488952636719, + "logps/rejected": -315.2213134765625, + "loss": 0.5985, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.0807700827717781, + "rewards/margins": 0.254264771938324, + "rewards/rejected": -0.33503487706184387, + "step": 1780 + }, + { + "epoch": 0.23, + "learning_rate": 4.730902421468652e-06, + "logits/chosen": -2.4281609058380127, + "logits/rejected": -2.336761474609375, + "logps/chosen": -283.50677490234375, + "logps/rejected": -310.46502685546875, + "loss": 0.5985, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.17131593823432922, + "rewards/margins": 0.2610486149787903, + "rewards/rejected": -0.4323646128177643, + "step": 1790 + }, + { + "epoch": 0.24, + "learning_rate": 4.7257247655907854e-06, + "logits/chosen": -2.3688364028930664, + "logits/rejected": -2.3433785438537598, + "logps/chosen": -262.29986572265625, + "logps/rejected": -294.90924072265625, + "loss": 0.6086, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.15793509781360626, + "rewards/margins": 0.23940256237983704, + "rewards/rejected": -0.3973376154899597, + "step": 1800 + }, + { + "epoch": 0.24, + "learning_rate": 4.720500661137397e-06, + "logits/chosen": -2.392461061477661, + "logits/rejected": -2.333200454711914, + "logps/chosen": -284.260498046875, + "logps/rejected": -301.0771484375, + "loss": 0.6467, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.16365382075309753, + "rewards/margins": 0.15784037113189697, + "rewards/rejected": -0.3214941620826721, + "step": 1810 + }, + { + "epoch": 0.24, + "learning_rate": 4.71523021713015e-06, + "logits/chosen": -2.4759411811828613, + "logits/rejected": -2.4524388313293457, + "logps/chosen": -392.81787109375, + "logps/rejected": -414.74249267578125, + "loss": 0.6457, + "rewards/accuracies": 0.6000000238418579, + "rewards/chosen": -0.17052128911018372, + "rewards/margins": 0.17021805047988892, + "rewards/rejected": -0.34073930978775024, + "step": 1820 + }, + { + "epoch": 0.24, + "learning_rate": 4.709913543557761e-06, + "logits/chosen": -2.418494701385498, + "logits/rejected": -2.372173547744751, + "logps/chosen": -339.1895751953125, + "logps/rejected": -339.2401123046875, + "loss": 0.6153, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.1340826153755188, + "rewards/margins": 0.2253703773021698, + "rewards/rejected": -0.359453022480011, + "step": 1830 + }, + { + "epoch": 0.24, + "learning_rate": 4.704550751373715e-06, + "logits/chosen": -2.375180244445801, + "logits/rejected": -2.283752679824829, + "logps/chosen": -379.5498046875, + "logps/rejected": -350.1661071777344, + "loss": 0.5828, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.08494368940591812, + "rewards/margins": 0.30550602078437805, + "rewards/rejected": -0.3904497027397156, + "step": 1840 + }, + { + "epoch": 0.24, + "learning_rate": 4.699141952493941e-06, + "logits/chosen": -2.389798641204834, + "logits/rejected": -2.350865602493286, + "logps/chosen": -288.4355773925781, + "logps/rejected": -306.73883056640625, + "loss": 0.635, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.1179732233285904, + "rewards/margins": 0.1771775782108307, + "rewards/rejected": -0.2951507866382599, + "step": 1850 + }, + { + "epoch": 0.24, + "learning_rate": 4.6936872597944814e-06, + "logits/chosen": -2.4748127460479736, + "logits/rejected": -2.3655686378479004, + "logps/chosen": -321.76165771484375, + "logps/rejected": -273.697509765625, + "loss": 0.5639, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.04060537368059158, + "rewards/margins": 0.3436310291290283, + "rewards/rejected": -0.3842363953590393, + "step": 1860 + }, + { + "epoch": 0.24, + "learning_rate": 4.688186787109136e-06, + "logits/chosen": -2.412353515625, + "logits/rejected": -2.365421772003174, + "logps/chosen": -324.1878662109375, + "logps/rejected": -338.0693359375, + "loss": 0.5989, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.10340994596481323, + "rewards/margins": 0.2609453797340393, + "rewards/rejected": -0.3643553555011749, + "step": 1870 + }, + { + "epoch": 0.25, + "learning_rate": 4.682640649227085e-06, + "logits/chosen": -2.4154398441314697, + "logits/rejected": -2.353245973587036, + "logps/chosen": -287.70440673828125, + "logps/rejected": -274.77886962890625, + "loss": 0.5911, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.026734206825494766, + "rewards/margins": 0.2650344967842102, + "rewards/rejected": -0.2917686700820923, + "step": 1880 + }, + { + "epoch": 0.25, + "learning_rate": 4.677048961890492e-06, + "logits/chosen": -2.302427291870117, + "logits/rejected": -2.2373576164245605, + "logps/chosen": -300.24896240234375, + "logps/rejected": -291.5721130371094, + "loss": 0.6059, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.10195889323949814, + "rewards/margins": 0.23982873558998108, + "rewards/rejected": -0.3417876362800598, + "step": 1890 + }, + { + "epoch": 0.25, + "learning_rate": 4.671411841792096e-06, + "logits/chosen": -2.3422927856445312, + "logits/rejected": -2.354828357696533, + "logps/chosen": -332.606201171875, + "logps/rejected": -353.5418395996094, + "loss": 0.6258, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.1744678020477295, + "rewards/margins": 0.19613692164421082, + "rewards/rejected": -0.3706046938896179, + "step": 1900 + }, + { + "epoch": 0.25, + "learning_rate": 4.665729406572764e-06, + "logits/chosen": -2.350491523742676, + "logits/rejected": -2.3193857669830322, + "logps/chosen": -270.14398193359375, + "logps/rejected": -286.48931884765625, + "loss": 0.6241, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2020377218723297, + "rewards/margins": 0.2181723564863205, + "rewards/rejected": -0.42021018266677856, + "step": 1910 + }, + { + "epoch": 0.25, + "learning_rate": 4.660001774819048e-06, + "logits/chosen": -2.4870200157165527, + "logits/rejected": -2.4069786071777344, + "logps/chosen": -337.8989562988281, + "logps/rejected": -291.6131591796875, + "loss": 0.611, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.15691247582435608, + "rewards/margins": 0.23032662272453308, + "rewards/rejected": -0.38723909854888916, + "step": 1920 + }, + { + "epoch": 0.25, + "learning_rate": 4.654229066060702e-06, + "logits/chosen": -2.3907480239868164, + "logits/rejected": -2.2987189292907715, + "logps/chosen": -294.33331298828125, + "logps/rejected": -267.0952453613281, + "loss": 0.6379, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.13681192696094513, + "rewards/margins": 0.15825247764587402, + "rewards/rejected": -0.29506438970565796, + "step": 1930 + }, + { + "epoch": 0.25, + "learning_rate": 4.648411400768193e-06, + "logits/chosen": -2.471407890319824, + "logits/rejected": -2.3333897590637207, + "logps/chosen": -313.52508544921875, + "logps/rejected": -257.5264892578125, + "loss": 0.5945, + "rewards/accuracies": 0.8125, + "rewards/chosen": -0.13640491664409637, + "rewards/margins": 0.24762806296348572, + "rewards/rejected": -0.3840329647064209, + "step": 1940 + }, + { + "epoch": 0.26, + "learning_rate": 4.642548900350182e-06, + "logits/chosen": -2.3417105674743652, + "logits/rejected": -2.3160438537597656, + "logps/chosen": -264.9716491699219, + "logps/rejected": -276.3714599609375, + "loss": 0.6235, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.16017001867294312, + "rewards/margins": 0.20394861698150635, + "rewards/rejected": -0.36411863565444946, + "step": 1950 + }, + { + "epoch": 0.26, + "learning_rate": 4.636641687150994e-06, + "logits/chosen": -2.4112541675567627, + "logits/rejected": -2.403289794921875, + "logps/chosen": -330.2745056152344, + "logps/rejected": -314.89324951171875, + "loss": 0.6105, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.09830652922391891, + "rewards/margins": 0.237725168466568, + "rewards/rejected": -0.3360317051410675, + "step": 1960 + }, + { + "epoch": 0.26, + "learning_rate": 4.6306898844480615e-06, + "logits/chosen": -2.3712351322174072, + "logits/rejected": -2.327747344970703, + "logps/chosen": -335.15362548828125, + "logps/rejected": -307.4205017089844, + "loss": 0.6308, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.07365657389163971, + "rewards/margins": 0.17441605031490326, + "rewards/rejected": -0.24807262420654297, + "step": 1970 + }, + { + "epoch": 0.26, + "learning_rate": 4.624693616449358e-06, + "logits/chosen": -2.3876092433929443, + "logits/rejected": -2.3262836933135986, + "logps/chosen": -246.8582000732422, + "logps/rejected": -252.2131805419922, + "loss": 0.6357, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.0633595809340477, + "rewards/margins": 0.17477674782276154, + "rewards/rejected": -0.23813633620738983, + "step": 1980 + }, + { + "epoch": 0.26, + "learning_rate": 4.6186530082908e-06, + "logits/chosen": -2.3876285552978516, + "logits/rejected": -2.316649913787842, + "logps/chosen": -334.5726318359375, + "logps/rejected": -336.7871398925781, + "loss": 0.6257, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.11085058748722076, + "rewards/margins": 0.2251889705657959, + "rewards/rejected": -0.33603957295417786, + "step": 1990 + }, + { + "epoch": 0.26, + "learning_rate": 4.612568186033633e-06, + "logits/chosen": -2.4370970726013184, + "logits/rejected": -2.4089818000793457, + "logps/chosen": -277.29840087890625, + "logps/rejected": -299.14544677734375, + "loss": 0.6438, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.16993489861488342, + "rewards/margins": 0.1625920981168747, + "rewards/rejected": -0.3325270116329193, + "step": 2000 + }, + { + "epoch": 0.26, + "learning_rate": 4.6064392766618125e-06, + "logits/chosen": -2.2922611236572266, + "logits/rejected": -2.2452197074890137, + "logps/chosen": -318.24725341796875, + "logps/rejected": -313.4471740722656, + "loss": 0.6207, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.13722370564937592, + "rewards/margins": 0.21375449001789093, + "rewards/rejected": -0.35097816586494446, + "step": 2010 + }, + { + "epoch": 0.26, + "learning_rate": 4.60026640807934e-06, + "logits/chosen": -2.378413438796997, + "logits/rejected": -2.32692289352417, + "logps/chosen": -306.7994079589844, + "logps/rejected": -286.31939697265625, + "loss": 0.6027, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.07194646447896957, + "rewards/margins": 0.23152203857898712, + "rewards/rejected": -0.3034684956073761, + "step": 2020 + }, + { + "epoch": 0.27, + "learning_rate": 4.594049709107604e-06, + "logits/chosen": -2.423443078994751, + "logits/rejected": -2.2880520820617676, + "logps/chosen": -334.21600341796875, + "logps/rejected": -334.7265625, + "loss": 0.6122, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.14375853538513184, + "rewards/margins": 0.24089816212654114, + "rewards/rejected": -0.38465672731399536, + "step": 2030 + }, + { + "epoch": 0.27, + "learning_rate": 4.587789309482687e-06, + "logits/chosen": -2.3012590408325195, + "logits/rejected": -2.27238130569458, + "logps/chosen": -275.12115478515625, + "logps/rejected": -334.7784118652344, + "loss": 0.5961, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.1960037648677826, + "rewards/margins": 0.2675577998161316, + "rewards/rejected": -0.4635615944862366, + "step": 2040 + }, + { + "epoch": 0.27, + "learning_rate": 4.581485339852659e-06, + "logits/chosen": -2.305781364440918, + "logits/rejected": -2.244325637817383, + "logps/chosen": -337.34027099609375, + "logps/rejected": -359.4820861816406, + "loss": 0.6366, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.18363210558891296, + "rewards/margins": 0.20024652779102325, + "rewards/rejected": -0.383878618478775, + "step": 2050 + }, + { + "epoch": 0.27, + "learning_rate": 4.5751379317748514e-06, + "logits/chosen": -2.362804889678955, + "logits/rejected": -2.304513692855835, + "logps/chosen": -295.06793212890625, + "logps/rejected": -303.84881591796875, + "loss": 0.6507, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.22672036290168762, + "rewards/margins": 0.13963665068149567, + "rewards/rejected": -0.3663569986820221, + "step": 2060 + }, + { + "epoch": 0.27, + "learning_rate": 4.56874721771311e-06, + "logits/chosen": -2.4059696197509766, + "logits/rejected": -2.3265323638916016, + "logps/chosen": -321.29754638671875, + "logps/rejected": -305.83929443359375, + "loss": 0.6215, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.25780874490737915, + "rewards/margins": 0.2031063586473465, + "rewards/rejected": -0.46091508865356445, + "step": 2070 + }, + { + "epoch": 0.27, + "learning_rate": 4.562313331035032e-06, + "logits/chosen": -2.3071093559265137, + "logits/rejected": -2.2454941272735596, + "logps/chosen": -297.61773681640625, + "logps/rejected": -331.1474304199219, + "loss": 0.6275, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.35569310188293457, + "rewards/margins": 0.18869885802268982, + "rewards/rejected": -0.5443919897079468, + "step": 2080 + }, + { + "epoch": 0.27, + "learning_rate": 4.555836406009183e-06, + "logits/chosen": -2.3374013900756836, + "logits/rejected": -2.3034305572509766, + "logps/chosen": -285.54107666015625, + "logps/rejected": -282.3943786621094, + "loss": 0.6324, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.3528543710708618, + "rewards/margins": 0.17680703103542328, + "rewards/rejected": -0.5296614170074463, + "step": 2090 + }, + { + "epoch": 0.27, + "learning_rate": 4.5493165778022945e-06, + "logits/chosen": -2.3471615314483643, + "logits/rejected": -2.2503392696380615, + "logps/chosen": -281.76812744140625, + "logps/rejected": -268.25872802734375, + "loss": 0.6143, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.2267017811536789, + "rewards/margins": 0.2128499299287796, + "rewards/rejected": -0.4395516812801361, + "step": 2100 + }, + { + "epoch": 0.28, + "learning_rate": 4.542753982476443e-06, + "logits/chosen": -2.360699415206909, + "logits/rejected": -2.2485642433166504, + "logps/chosen": -327.87591552734375, + "logps/rejected": -306.8153991699219, + "loss": 0.5964, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.08724816143512726, + "rewards/margins": 0.26559630036354065, + "rewards/rejected": -0.3528444468975067, + "step": 2110 + }, + { + "epoch": 0.28, + "learning_rate": 4.53614875698621e-06, + "logits/chosen": -2.4491329193115234, + "logits/rejected": -2.346787214279175, + "logps/chosen": -337.62335205078125, + "logps/rejected": -348.6912536621094, + "loss": 0.6047, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.12890265882015228, + "rewards/margins": 0.24755427241325378, + "rewards/rejected": -0.3764569163322449, + "step": 2120 + }, + { + "epoch": 0.28, + "learning_rate": 4.529501039175824e-06, + "logits/chosen": -2.3807685375213623, + "logits/rejected": -2.352710247039795, + "logps/chosen": -259.227294921875, + "logps/rejected": -269.9205322265625, + "loss": 0.6442, + "rewards/accuracies": 0.550000011920929, + "rewards/chosen": -0.16931673884391785, + "rewards/margins": 0.17628124356269836, + "rewards/rejected": -0.3455979824066162, + "step": 2130 + }, + { + "epoch": 0.28, + "learning_rate": 4.522810967776287e-06, + "logits/chosen": -2.3062472343444824, + "logits/rejected": -2.1800289154052734, + "logps/chosen": -315.9835510253906, + "logps/rejected": -264.95050048828125, + "loss": 0.5979, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.18467222154140472, + "rewards/margins": 0.2567806541919708, + "rewards/rejected": -0.44145289063453674, + "step": 2140 + }, + { + "epoch": 0.28, + "learning_rate": 4.516078682402473e-06, + "logits/chosen": -2.3751935958862305, + "logits/rejected": -2.2700018882751465, + "logps/chosen": -330.53070068359375, + "logps/rejected": -322.2341003417969, + "loss": 0.595, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.24924437701702118, + "rewards/margins": 0.28462594747543335, + "rewards/rejected": -0.5338703393936157, + "step": 2150 + }, + { + "epoch": 0.28, + "learning_rate": 4.509304323550221e-06, + "logits/chosen": -2.3613998889923096, + "logits/rejected": -2.3095054626464844, + "logps/chosen": -354.7084655761719, + "logps/rejected": -343.9185485839844, + "loss": 0.6226, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.20070073008537292, + "rewards/margins": 0.2098993957042694, + "rewards/rejected": -0.41060012578964233, + "step": 2160 + }, + { + "epoch": 0.28, + "learning_rate": 4.502488032593398e-06, + "logits/chosen": -2.422234058380127, + "logits/rejected": -2.3208489418029785, + "logps/chosen": -316.1452941894531, + "logps/rejected": -304.63330078125, + "loss": 0.6185, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.18992137908935547, + "rewards/margins": 0.2037736475467682, + "rewards/rejected": -0.39369502663612366, + "step": 2170 + }, + { + "epoch": 0.29, + "learning_rate": 4.495629951780951e-06, + "logits/chosen": -2.3926033973693848, + "logits/rejected": -2.2609989643096924, + "logps/chosen": -369.9411315917969, + "logps/rejected": -305.66607666015625, + "loss": 0.5797, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.07531694322824478, + "rewards/margins": 0.2898343801498413, + "rewards/rejected": -0.3651513159275055, + "step": 2180 + }, + { + "epoch": 0.29, + "learning_rate": 4.488730224233941e-06, + "logits/chosen": -2.316122531890869, + "logits/rejected": -2.250302314758301, + "logps/chosen": -248.08547973632812, + "logps/rejected": -300.73223876953125, + "loss": 0.5419, + "rewards/accuracies": 0.8500000238418579, + "rewards/chosen": -0.08749908953905106, + "rewards/margins": 0.3842558264732361, + "rewards/rejected": -0.47175493836402893, + "step": 2190 + }, + { + "epoch": 0.29, + "learning_rate": 4.481788993942547e-06, + "logits/chosen": -2.3757452964782715, + "logits/rejected": -2.2821907997131348, + "logps/chosen": -303.1983947753906, + "logps/rejected": -289.0472106933594, + "loss": 0.5995, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.12067332118749619, + "rewards/margins": 0.26477959752082825, + "rewards/rejected": -0.38545292615890503, + "step": 2200 + }, + { + "epoch": 0.29, + "learning_rate": 4.474806405763076e-06, + "logits/chosen": -2.3512368202209473, + "logits/rejected": -2.3097450733184814, + "logps/chosen": -301.57452392578125, + "logps/rejected": -299.31671142578125, + "loss": 0.6031, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.11587338149547577, + "rewards/margins": 0.25784656405448914, + "rewards/rejected": -0.3737199306488037, + "step": 2210 + }, + { + "epoch": 0.29, + "learning_rate": 4.4677826054149235e-06, + "logits/chosen": -2.338404417037964, + "logits/rejected": -2.253673553466797, + "logps/chosen": -340.2875671386719, + "logps/rejected": -330.8304748535156, + "loss": 0.6391, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.10927800834178925, + "rewards/margins": 0.17066845297813416, + "rewards/rejected": -0.2799464762210846, + "step": 2220 + }, + { + "epoch": 0.29, + "learning_rate": 4.460717739477543e-06, + "logits/chosen": -2.355332612991333, + "logits/rejected": -2.3418946266174316, + "logps/chosen": -287.7314758300781, + "logps/rejected": -341.2978515625, + "loss": 0.6354, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.12638050317764282, + "rewards/margins": 0.18259502947330475, + "rewards/rejected": -0.30897554755210876, + "step": 2230 + }, + { + "epoch": 0.29, + "learning_rate": 4.4536119553873866e-06, + "logits/chosen": -2.311370372772217, + "logits/rejected": -2.35019588470459, + "logps/chosen": -278.7158203125, + "logps/rejected": -300.1242370605469, + "loss": 0.5766, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.039584096521139145, + "rewards/margins": 0.3132474422454834, + "rewards/rejected": -0.35283154249191284, + "step": 2240 + }, + { + "epoch": 0.29, + "learning_rate": 4.446465401434824e-06, + "logits/chosen": -2.3198676109313965, + "logits/rejected": -2.2937235832214355, + "logps/chosen": -269.0504455566406, + "logps/rejected": -327.19183349609375, + "loss": 0.6296, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.15721771121025085, + "rewards/margins": 0.2044394314289093, + "rewards/rejected": -0.36165714263916016, + "step": 2250 + }, + { + "epoch": 0.3, + "learning_rate": 4.43927822676105e-06, + "logits/chosen": -2.3916709423065186, + "logits/rejected": -2.2888083457946777, + "logps/chosen": -322.9589538574219, + "logps/rejected": -301.18609619140625, + "loss": 0.6161, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.116670623421669, + "rewards/margins": 0.22785842418670654, + "rewards/rejected": -0.34452906250953674, + "step": 2260 + }, + { + "epoch": 0.3, + "learning_rate": 4.432050581354972e-06, + "logits/chosen": -2.3398289680480957, + "logits/rejected": -2.2326717376708984, + "logps/chosen": -295.10894775390625, + "logps/rejected": -317.585693359375, + "loss": 0.6182, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.139591783285141, + "rewards/margins": 0.2277776300907135, + "rewards/rejected": -0.36736947298049927, + "step": 2270 + }, + { + "epoch": 0.3, + "learning_rate": 4.424782616050078e-06, + "logits/chosen": -2.2997286319732666, + "logits/rejected": -2.237948417663574, + "logps/chosen": -307.8941650390625, + "logps/rejected": -298.88006591796875, + "loss": 0.614, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.11191853135824203, + "rewards/margins": 0.23675408959388733, + "rewards/rejected": -0.34867262840270996, + "step": 2280 + }, + { + "epoch": 0.3, + "learning_rate": 4.4174744825212954e-06, + "logits/chosen": -2.381026268005371, + "logits/rejected": -2.3788371086120605, + "logps/chosen": -274.66656494140625, + "logps/rejected": -307.1249084472656, + "loss": 0.6717, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.15815982222557068, + "rewards/margins": 0.12555231153964996, + "rewards/rejected": -0.28371211886405945, + "step": 2290 + }, + { + "epoch": 0.3, + "learning_rate": 4.410126333281815e-06, + "logits/chosen": -2.3236286640167236, + "logits/rejected": -2.2463207244873047, + "logps/chosen": -351.5771484375, + "logps/rejected": -385.1914978027344, + "loss": 0.6002, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.16838675737380981, + "rewards/margins": 0.2537475526332855, + "rewards/rejected": -0.42213425040245056, + "step": 2300 + }, + { + "epoch": 0.3, + "learning_rate": 4.402738321679918e-06, + "logits/chosen": -2.3222060203552246, + "logits/rejected": -2.245678186416626, + "logps/chosen": -319.8814392089844, + "logps/rejected": -279.41656494140625, + "loss": 0.5771, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.15020090341567993, + "rewards/margins": 0.31123560667037964, + "rewards/rejected": -0.4614364504814148, + "step": 2310 + }, + { + "epoch": 0.3, + "learning_rate": 4.395310601895772e-06, + "logits/chosen": -2.3784894943237305, + "logits/rejected": -2.3047304153442383, + "logps/chosen": -309.1102294921875, + "logps/rejected": -321.6300354003906, + "loss": 0.5812, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.11590411514043808, + "rewards/margins": 0.30960220098495483, + "rewards/rejected": -0.42550626397132874, + "step": 2320 + }, + { + "epoch": 0.3, + "learning_rate": 4.38784332893821e-06, + "logits/chosen": -2.3468894958496094, + "logits/rejected": -2.311635971069336, + "logps/chosen": -226.0956268310547, + "logps/rejected": -260.84552001953125, + "loss": 0.5672, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.12173782289028168, + "rewards/margins": 0.33285272121429443, + "rewards/rejected": -0.4545905590057373, + "step": 2330 + }, + { + "epoch": 0.31, + "learning_rate": 4.380336658641503e-06, + "logits/chosen": -2.4134764671325684, + "logits/rejected": -2.252642869949341, + "logps/chosen": -314.12115478515625, + "logps/rejected": -291.8180236816406, + "loss": 0.6069, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.1728154420852661, + "rewards/margins": 0.2426457703113556, + "rewards/rejected": -0.4154612123966217, + "step": 2340 + }, + { + "epoch": 0.31, + "learning_rate": 4.372790747662101e-06, + "logits/chosen": -2.280484199523926, + "logits/rejected": -2.2489004135131836, + "logps/chosen": -306.0282287597656, + "logps/rejected": -303.01141357421875, + "loss": 0.6283, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.2237296998500824, + "rewards/margins": 0.19994047284126282, + "rewards/rejected": -0.4236702024936676, + "step": 2350 + }, + { + "epoch": 0.31, + "learning_rate": 4.365205753475367e-06, + "logits/chosen": -2.366082191467285, + "logits/rejected": -2.2484962940216064, + "logps/chosen": -368.59075927734375, + "logps/rejected": -318.4290466308594, + "loss": 0.5859, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.11698893457651138, + "rewards/margins": 0.29912567138671875, + "rewards/rejected": -0.41611456871032715, + "step": 2360 + }, + { + "epoch": 0.31, + "learning_rate": 4.35758183437229e-06, + "logits/chosen": -2.410271167755127, + "logits/rejected": -2.3593297004699707, + "logps/chosen": -354.11431884765625, + "logps/rejected": -354.12945556640625, + "loss": 0.6349, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.022561874240636826, + "rewards/margins": 0.17640873789787292, + "rewards/rejected": -0.19897060096263885, + "step": 2370 + }, + { + "epoch": 0.31, + "learning_rate": 4.3499191494561835e-06, + "logits/chosen": -2.2106945514678955, + "logits/rejected": -2.1773438453674316, + "logps/chosen": -296.61639404296875, + "logps/rejected": -269.124755859375, + "loss": 0.6266, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.06345901638269424, + "rewards/margins": 0.19572456181049347, + "rewards/rejected": -0.2591835856437683, + "step": 2380 + }, + { + "epoch": 0.31, + "learning_rate": 4.3422178586393615e-06, + "logits/chosen": -2.4216809272766113, + "logits/rejected": -2.2997002601623535, + "logps/chosen": -341.6378173828125, + "logps/rejected": -307.9349670410156, + "loss": 0.5915, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.06652353703975677, + "rewards/margins": 0.2856021821498871, + "rewards/rejected": -0.35212570428848267, + "step": 2390 + }, + { + "epoch": 0.31, + "learning_rate": 4.334478122639804e-06, + "logits/chosen": -2.429971933364868, + "logits/rejected": -2.2178616523742676, + "logps/chosen": -286.44097900390625, + "logps/rejected": -248.7102813720703, + "loss": 0.5218, + "rewards/accuracies": 0.7875000238418579, + "rewards/chosen": -0.10870865732431412, + "rewards/margins": 0.4377427101135254, + "rewards/rejected": -0.5464513897895813, + "step": 2400 + }, + { + "epoch": 0.32, + "learning_rate": 4.3267001029778015e-06, + "logits/chosen": -2.367393732070923, + "logits/rejected": -2.246732711791992, + "logps/chosen": -330.3119201660156, + "logps/rejected": -277.81982421875, + "loss": 0.6332, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.17828808724880219, + "rewards/margins": 0.18477049469947815, + "rewards/rejected": -0.36305850744247437, + "step": 2410 + }, + { + "epoch": 0.32, + "learning_rate": 4.318883961972585e-06, + "logits/chosen": -2.2722887992858887, + "logits/rejected": -2.2594714164733887, + "logps/chosen": -294.3458251953125, + "logps/rejected": -307.031982421875, + "loss": 0.5982, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.17498214542865753, + "rewards/margins": 0.2718316614627838, + "rewards/rejected": -0.44681382179260254, + "step": 2420 + }, + { + "epoch": 0.32, + "learning_rate": 4.311029862738942e-06, + "logits/chosen": -2.3691306114196777, + "logits/rejected": -2.286330223083496, + "logps/chosen": -321.17352294921875, + "logps/rejected": -314.225830078125, + "loss": 0.6092, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.2745895981788635, + "rewards/margins": 0.22876712679862976, + "rewards/rejected": -0.5033567547798157, + "step": 2430 + }, + { + "epoch": 0.32, + "learning_rate": 4.303137969183804e-06, + "logits/chosen": -2.261221408843994, + "logits/rejected": -2.218248128890991, + "logps/chosen": -304.7114562988281, + "logps/rejected": -303.0655212402344, + "loss": 0.5703, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.22714805603027344, + "rewards/margins": 0.32687151432037354, + "rewards/rejected": -0.554019570350647, + "step": 2440 + }, + { + "epoch": 0.32, + "learning_rate": 4.295208446002832e-06, + "logits/chosen": -2.2834646701812744, + "logits/rejected": -2.1622109413146973, + "logps/chosen": -264.4088439941406, + "logps/rejected": -263.5638427734375, + "loss": 0.6082, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2190243899822235, + "rewards/margins": 0.25770148634910583, + "rewards/rejected": -0.47672590613365173, + "step": 2450 + }, + { + "epoch": 0.32, + "learning_rate": 4.287241458676981e-06, + "logits/chosen": -2.2780747413635254, + "logits/rejected": -2.2258801460266113, + "logps/chosen": -355.08612060546875, + "logps/rejected": -348.1125793457031, + "loss": 0.5692, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.19551631808280945, + "rewards/margins": 0.35570698976516724, + "rewards/rejected": -0.5512233376502991, + "step": 2460 + }, + { + "epoch": 0.32, + "learning_rate": 4.279237173469043e-06, + "logits/chosen": -2.3363661766052246, + "logits/rejected": -2.1859169006347656, + "logps/chosen": -296.65423583984375, + "logps/rejected": -286.93377685546875, + "loss": 0.5738, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.12626975774765015, + "rewards/margins": 0.3321135342121124, + "rewards/rejected": -0.4583832621574402, + "step": 2470 + }, + { + "epoch": 0.32, + "learning_rate": 4.271195757420177e-06, + "logits/chosen": -2.2018187046051025, + "logits/rejected": -2.1672024726867676, + "logps/chosen": -309.4765319824219, + "logps/rejected": -344.6891174316406, + "loss": 0.6229, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.24267332255840302, + "rewards/margins": 0.23746450245380402, + "rewards/rejected": -0.48013782501220703, + "step": 2480 + }, + { + "epoch": 0.33, + "learning_rate": 4.263117378346425e-06, + "logits/chosen": -2.28141450881958, + "logits/rejected": -2.1525015830993652, + "logps/chosen": -332.60894775390625, + "logps/rejected": -308.6076354980469, + "loss": 0.5955, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.1525574028491974, + "rewards/margins": 0.30104440450668335, + "rewards/rejected": -0.4536018371582031, + "step": 2490 + }, + { + "epoch": 0.33, + "learning_rate": 4.255002204835208e-06, + "logits/chosen": -2.2987420558929443, + "logits/rejected": -2.2409260272979736, + "logps/chosen": -328.59521484375, + "logps/rejected": -330.57098388671875, + "loss": 0.578, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.12822741270065308, + "rewards/margins": 0.3138895630836487, + "rewards/rejected": -0.44211697578430176, + "step": 2500 + }, + { + "epoch": 0.33, + "learning_rate": 4.246850406241812e-06, + "logits/chosen": -2.283529758453369, + "logits/rejected": -2.196633815765381, + "logps/chosen": -248.55825805664062, + "logps/rejected": -302.5537109375, + "loss": 0.5715, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.17185014486312866, + "rewards/margins": 0.3294577896595001, + "rewards/rejected": -0.5013079047203064, + "step": 2510 + }, + { + "epoch": 0.33, + "learning_rate": 4.2386621526858465e-06, + "logits/chosen": -2.2872681617736816, + "logits/rejected": -2.2363412380218506, + "logps/chosen": -260.451171875, + "logps/rejected": -248.4807891845703, + "loss": 0.5905, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.1857258677482605, + "rewards/margins": 0.2745230197906494, + "rewards/rejected": -0.4602488577365875, + "step": 2520 + }, + { + "epoch": 0.33, + "learning_rate": 4.2304376150477015e-06, + "logits/chosen": -2.3132100105285645, + "logits/rejected": -2.2969963550567627, + "logps/chosen": -313.3431701660156, + "logps/rejected": -330.83404541015625, + "loss": 0.5747, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.1469540297985077, + "rewards/margins": 0.33261358737945557, + "rewards/rejected": -0.47956761717796326, + "step": 2530 + }, + { + "epoch": 0.33, + "learning_rate": 4.222176964964977e-06, + "logits/chosen": -2.196845531463623, + "logits/rejected": -2.145677089691162, + "logps/chosen": -312.68353271484375, + "logps/rejected": -317.6732482910156, + "loss": 0.6325, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.2049153745174408, + "rewards/margins": 0.18959246575832367, + "rewards/rejected": -0.39450788497924805, + "step": 2540 + }, + { + "epoch": 0.33, + "learning_rate": 4.213880374828903e-06, + "logits/chosen": -2.3408188819885254, + "logits/rejected": -2.2571680545806885, + "logps/chosen": -268.51995849609375, + "logps/rejected": -301.3713073730469, + "loss": 0.5891, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.17139479517936707, + "rewards/margins": 0.2834908962249756, + "rewards/rejected": -0.45488566160202026, + "step": 2550 + }, + { + "epoch": 0.33, + "learning_rate": 4.2055480177807406e-06, + "logits/chosen": -2.170337200164795, + "logits/rejected": -2.16973876953125, + "logps/chosen": -252.72805786132812, + "logps/rejected": -257.45928955078125, + "loss": 0.6601, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.18463487923145294, + "rewards/margins": 0.1279645711183548, + "rewards/rejected": -0.3125994801521301, + "step": 2560 + }, + { + "epoch": 0.34, + "learning_rate": 4.1971800677081696e-06, + "logits/chosen": -2.2201130390167236, + "logits/rejected": -2.2000250816345215, + "logps/chosen": -306.4814147949219, + "logps/rejected": -351.8539123535156, + "loss": 0.6612, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.1903466135263443, + "rewards/margins": 0.14342021942138672, + "rewards/rejected": -0.3337668478488922, + "step": 2570 + }, + { + "epoch": 0.34, + "learning_rate": 4.188776699241661e-06, + "logits/chosen": -2.365649700164795, + "logits/rejected": -2.2430615425109863, + "logps/chosen": -313.85162353515625, + "logps/rejected": -308.8529357910156, + "loss": 0.6126, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.11727418005466461, + "rewards/margins": 0.23288528621196747, + "rewards/rejected": -0.3501594662666321, + "step": 2580 + }, + { + "epoch": 0.34, + "learning_rate": 4.180338087750827e-06, + "logits/chosen": -2.283839702606201, + "logits/rejected": -2.2854695320129395, + "logps/chosen": -288.41754150390625, + "logps/rejected": -333.4906311035156, + "loss": 0.5952, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.1470477283000946, + "rewards/margins": 0.2957322895526886, + "rewards/rejected": -0.4427799582481384, + "step": 2590 + }, + { + "epoch": 0.34, + "learning_rate": 4.1718644093407704e-06, + "logits/chosen": -2.215942859649658, + "logits/rejected": -2.168015241622925, + "logps/chosen": -300.6781311035156, + "logps/rejected": -312.72198486328125, + "loss": 0.5495, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.11697547137737274, + "rewards/margins": 0.38940751552581787, + "rewards/rejected": -0.5063830614089966, + "step": 2600 + }, + { + "epoch": 0.34, + "learning_rate": 4.163355840848401e-06, + "logits/chosen": -2.3301100730895996, + "logits/rejected": -2.21736478805542, + "logps/chosen": -333.5172119140625, + "logps/rejected": -344.1437072753906, + "loss": 0.6089, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.2591416835784912, + "rewards/margins": 0.27335575222969055, + "rewards/rejected": -0.5324974656105042, + "step": 2610 + }, + { + "epoch": 0.34, + "learning_rate": 4.154812559838748e-06, + "logits/chosen": -2.299487590789795, + "logits/rejected": -2.2091782093048096, + "logps/chosen": -278.13702392578125, + "logps/rejected": -286.05548095703125, + "loss": 0.5407, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.0895254835486412, + "rewards/margins": 0.4064970016479492, + "rewards/rejected": -0.49602246284484863, + "step": 2620 + }, + { + "epoch": 0.34, + "learning_rate": 4.146234744601259e-06, + "logits/chosen": -2.2448086738586426, + "logits/rejected": -2.1447150707244873, + "logps/chosen": -272.74371337890625, + "logps/rejected": -309.2991638183594, + "loss": 0.5772, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.15012232959270477, + "rewards/margins": 0.323434978723526, + "rewards/rejected": -0.47355732321739197, + "step": 2630 + }, + { + "epoch": 0.35, + "learning_rate": 4.137622574146071e-06, + "logits/chosen": -2.321953296661377, + "logits/rejected": -2.2377140522003174, + "logps/chosen": -271.127197265625, + "logps/rejected": -257.08197021484375, + "loss": 0.5904, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.04710864648222923, + "rewards/margins": 0.2709490954875946, + "rewards/rejected": -0.3180577754974365, + "step": 2640 + }, + { + "epoch": 0.35, + "learning_rate": 4.12897622820028e-06, + "logits/chosen": -2.323284387588501, + "logits/rejected": -2.216844081878662, + "logps/chosen": -327.4410400390625, + "logps/rejected": -322.1961975097656, + "loss": 0.5974, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.12447996437549591, + "rewards/margins": 0.26364922523498535, + "rewards/rejected": -0.38812920451164246, + "step": 2650 + }, + { + "epoch": 0.35, + "learning_rate": 4.120295887204191e-06, + "logits/chosen": -2.2313876152038574, + "logits/rejected": -2.1354150772094727, + "logps/chosen": -286.85870361328125, + "logps/rejected": -324.2660827636719, + "loss": 0.6137, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.21455173194408417, + "rewards/margins": 0.26458728313446045, + "rewards/rejected": -0.4791390299797058, + "step": 2660 + }, + { + "epoch": 0.35, + "learning_rate": 4.111581732307548e-06, + "logits/chosen": -2.239713668823242, + "logits/rejected": -2.2120895385742188, + "logps/chosen": -273.3136901855469, + "logps/rejected": -301.74273681640625, + "loss": 0.6098, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.2157611846923828, + "rewards/margins": 0.26702648401260376, + "rewards/rejected": -0.4827876687049866, + "step": 2670 + }, + { + "epoch": 0.35, + "learning_rate": 4.1028339453657595e-06, + "logits/chosen": -2.2836809158325195, + "logits/rejected": -2.0904288291931152, + "logps/chosen": -344.65362548828125, + "logps/rejected": -314.09149169921875, + "loss": 0.6117, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.12366046011447906, + "rewards/margins": 0.24424687027931213, + "rewards/rejected": -0.36790731549263, + "step": 2680 + }, + { + "epoch": 0.35, + "learning_rate": 4.094052708936096e-06, + "logits/chosen": -2.309826612472534, + "logits/rejected": -2.2577881813049316, + "logps/chosen": -316.89727783203125, + "logps/rejected": -366.1764221191406, + "loss": 0.5631, + "rewards/accuracies": 0.7875000238418579, + "rewards/chosen": -0.08284427970647812, + "rewards/margins": 0.36028558015823364, + "rewards/rejected": -0.44312983751296997, + "step": 2690 + }, + { + "epoch": 0.35, + "learning_rate": 4.0852382062738874e-06, + "logits/chosen": -2.305793285369873, + "logits/rejected": -2.2325210571289062, + "logps/chosen": -323.45245361328125, + "logps/rejected": -338.0271301269531, + "loss": 0.6095, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.22808530926704407, + "rewards/margins": 0.24091526865959167, + "rewards/rejected": -0.46900051832199097, + "step": 2700 + }, + { + "epoch": 0.35, + "learning_rate": 4.076390621328693e-06, + "logits/chosen": -2.338980197906494, + "logits/rejected": -2.2027366161346436, + "logps/chosen": -348.8518981933594, + "logps/rejected": -330.1253967285156, + "loss": 0.5792, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.20264069736003876, + "rewards/margins": 0.3276185095310211, + "rewards/rejected": -0.5302591323852539, + "step": 2710 + }, + { + "epoch": 0.36, + "learning_rate": 4.067510138740467e-06, + "logits/chosen": -2.2085351943969727, + "logits/rejected": -2.179727077484131, + "logps/chosen": -337.2236022949219, + "logps/rejected": -348.954833984375, + "loss": 0.5933, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.2019720822572708, + "rewards/margins": 0.29704728722572327, + "rewards/rejected": -0.4990193843841553, + "step": 2720 + }, + { + "epoch": 0.36, + "learning_rate": 4.058596943835703e-06, + "logits/chosen": -2.2849490642547607, + "logits/rejected": -2.255769729614258, + "logps/chosen": -313.56585693359375, + "logps/rejected": -291.57489013671875, + "loss": 0.627, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.16551968455314636, + "rewards/margins": 0.2221347838640213, + "rewards/rejected": -0.38765448331832886, + "step": 2730 + }, + { + "epoch": 0.36, + "learning_rate": 4.049651222623568e-06, + "logits/chosen": -2.3381128311157227, + "logits/rejected": -2.279048204421997, + "logps/chosen": -303.53460693359375, + "logps/rejected": -284.5528869628906, + "loss": 0.6096, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.17010393738746643, + "rewards/margins": 0.2440837174654007, + "rewards/rejected": -0.4141876697540283, + "step": 2740 + }, + { + "epoch": 0.36, + "learning_rate": 4.040673161792014e-06, + "logits/chosen": -2.297952651977539, + "logits/rejected": -2.2086384296417236, + "logps/chosen": -343.0641174316406, + "logps/rejected": -349.12786865234375, + "loss": 0.608, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.19907456636428833, + "rewards/margins": 0.2641879916191101, + "rewards/rejected": -0.46326255798339844, + "step": 2750 + }, + { + "epoch": 0.36, + "learning_rate": 4.031662948703896e-06, + "logits/chosen": -2.2400102615356445, + "logits/rejected": -2.1351914405822754, + "logps/chosen": -308.60504150390625, + "logps/rejected": -325.3629455566406, + "loss": 0.5625, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.1372842639684677, + "rewards/margins": 0.3827325701713562, + "rewards/rejected": -0.5200168490409851, + "step": 2760 + }, + { + "epoch": 0.36, + "learning_rate": 4.022620771393047e-06, + "logits/chosen": -2.3050320148468018, + "logits/rejected": -2.191411256790161, + "logps/chosen": -324.8857116699219, + "logps/rejected": -308.9805908203125, + "loss": 0.61, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.254262775182724, + "rewards/margins": 0.25628453493118286, + "rewards/rejected": -0.5105473399162292, + "step": 2770 + }, + { + "epoch": 0.36, + "learning_rate": 4.013546818560362e-06, + "logits/chosen": -2.2292933464050293, + "logits/rejected": -2.0870003700256348, + "logps/chosen": -292.7982177734375, + "logps/rejected": -288.9156188964844, + "loss": 0.6229, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.24121983349323273, + "rewards/margins": 0.21186479926109314, + "rewards/rejected": -0.45308464765548706, + "step": 2780 + }, + { + "epoch": 0.37, + "learning_rate": 4.00444127956986e-06, + "logits/chosen": -2.3071656227111816, + "logits/rejected": -2.1425764560699463, + "logps/chosen": -282.1644592285156, + "logps/rejected": -288.5877990722656, + "loss": 0.5743, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.1937316209077835, + "rewards/margins": 0.3192608952522278, + "rewards/rejected": -0.5129925012588501, + "step": 2790 + }, + { + "epoch": 0.37, + "learning_rate": 3.9953043444447255e-06, + "logits/chosen": -2.3431785106658936, + "logits/rejected": -2.2949059009552, + "logps/chosen": -337.19873046875, + "logps/rejected": -321.8359375, + "loss": 0.6295, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.14888326823711395, + "rewards/margins": 0.20381757616996765, + "rewards/rejected": -0.3527008593082428, + "step": 2800 + }, + { + "epoch": 0.37, + "learning_rate": 3.986136203863355e-06, + "logits/chosen": -2.269035816192627, + "logits/rejected": -2.258761167526245, + "logps/chosen": -334.3866271972656, + "logps/rejected": -335.5721435546875, + "loss": 0.5786, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.1394425928592682, + "rewards/margins": 0.3137834966182709, + "rewards/rejected": -0.45322608947753906, + "step": 2810 + }, + { + "epoch": 0.37, + "learning_rate": 3.976937049155365e-06, + "logits/chosen": -2.244459629058838, + "logits/rejected": -2.1737873554229736, + "logps/chosen": -295.0531005859375, + "logps/rejected": -328.18231201171875, + "loss": 0.5907, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.1519058644771576, + "rewards/margins": 0.28696325421333313, + "rewards/rejected": -0.43886908888816833, + "step": 2820 + }, + { + "epoch": 0.37, + "learning_rate": 3.967707072297608e-06, + "logits/chosen": -2.15556001663208, + "logits/rejected": -2.0936989784240723, + "logps/chosen": -322.49542236328125, + "logps/rejected": -345.54522705078125, + "loss": 0.5865, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.18620704114437103, + "rewards/margins": 0.29388242959976196, + "rewards/rejected": -0.4800894856452942, + "step": 2830 + }, + { + "epoch": 0.37, + "learning_rate": 3.958446465910159e-06, + "logits/chosen": -2.356562852859497, + "logits/rejected": -2.3032262325286865, + "logps/chosen": -305.7391662597656, + "logps/rejected": -330.58135986328125, + "loss": 0.5959, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.2515236735343933, + "rewards/margins": 0.27204760909080505, + "rewards/rejected": -0.5235711932182312, + "step": 2840 + }, + { + "epoch": 0.37, + "learning_rate": 3.9491554232523066e-06, + "logits/chosen": -2.154636859893799, + "logits/rejected": -2.174771785736084, + "logps/chosen": -259.4274597167969, + "logps/rejected": -301.49591064453125, + "loss": 0.6512, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.2738287150859833, + "rewards/margins": 0.16962867975234985, + "rewards/rejected": -0.4434574246406555, + "step": 2850 + }, + { + "epoch": 0.37, + "learning_rate": 3.939834138218505e-06, + "logits/chosen": -2.259753704071045, + "logits/rejected": -2.2131614685058594, + "logps/chosen": -316.70062255859375, + "logps/rejected": -321.30499267578125, + "loss": 0.5859, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.22981849312782288, + "rewards/margins": 0.32320156693458557, + "rewards/rejected": -0.5530200004577637, + "step": 2860 + }, + { + "epoch": 0.38, + "learning_rate": 3.930482805334339e-06, + "logits/chosen": -2.2335028648376465, + "logits/rejected": -2.1571171283721924, + "logps/chosen": -287.03802490234375, + "logps/rejected": -297.360107421875, + "loss": 0.6134, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.24314312636852264, + "rewards/margins": 0.2636451721191406, + "rewards/rejected": -0.5067883133888245, + "step": 2870 + }, + { + "epoch": 0.38, + "learning_rate": 3.921101619752464e-06, + "logits/chosen": -2.2365384101867676, + "logits/rejected": -2.118043899536133, + "logps/chosen": -315.28839111328125, + "logps/rejected": -292.41571044921875, + "loss": 0.6048, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.2544742524623871, + "rewards/margins": 0.25104787945747375, + "rewards/rejected": -0.5055221319198608, + "step": 2880 + }, + { + "epoch": 0.38, + "learning_rate": 3.911690777248525e-06, + "logits/chosen": -2.279517650604248, + "logits/rejected": -2.1619081497192383, + "logps/chosen": -331.37713623046875, + "logps/rejected": -343.8858337402344, + "loss": 0.6157, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.28338202834129333, + "rewards/margins": 0.22550642490386963, + "rewards/rejected": -0.5088884234428406, + "step": 2890 + }, + { + "epoch": 0.38, + "learning_rate": 3.902250474217079e-06, + "logits/chosen": -2.249533176422119, + "logits/rejected": -2.1728508472442627, + "logps/chosen": -290.2388000488281, + "logps/rejected": -288.2944030761719, + "loss": 0.5999, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.212762713432312, + "rewards/margins": 0.2684236168861389, + "rewards/rejected": -0.4811863303184509, + "step": 2900 + }, + { + "epoch": 0.38, + "learning_rate": 3.892780907667495e-06, + "logits/chosen": -2.254171848297119, + "logits/rejected": -2.1818957328796387, + "logps/chosen": -290.6761169433594, + "logps/rejected": -308.3992614746094, + "loss": 0.5687, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.10069400072097778, + "rewards/margins": 0.34916025400161743, + "rewards/rejected": -0.4498542845249176, + "step": 2910 + }, + { + "epoch": 0.38, + "learning_rate": 3.883282275219837e-06, + "logits/chosen": -2.3390860557556152, + "logits/rejected": -2.244075059890747, + "logps/chosen": -319.77874755859375, + "logps/rejected": -308.5218200683594, + "loss": 0.5906, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.23237352073192596, + "rewards/margins": 0.3077944219112396, + "rewards/rejected": -0.5401679873466492, + "step": 2920 + }, + { + "epoch": 0.38, + "learning_rate": 3.873754775100751e-06, + "logits/chosen": -2.332498073577881, + "logits/rejected": -2.244035482406616, + "logps/chosen": -316.33770751953125, + "logps/rejected": -314.38739013671875, + "loss": 0.5997, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.1779693067073822, + "rewards/margins": 0.3119404911994934, + "rewards/rejected": -0.48990970849990845, + "step": 2930 + }, + { + "epoch": 0.38, + "learning_rate": 3.8641986061393145e-06, + "logits/chosen": -2.3143229484558105, + "logits/rejected": -2.261376142501831, + "logps/chosen": -287.69525146484375, + "logps/rejected": -336.7581787109375, + "loss": 0.5772, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.08865178376436234, + "rewards/margins": 0.3206256926059723, + "rewards/rejected": -0.40927743911743164, + "step": 2940 + }, + { + "epoch": 0.39, + "learning_rate": 3.854613967762898e-06, + "logits/chosen": -2.2918875217437744, + "logits/rejected": -2.1621129512786865, + "logps/chosen": -347.6598815917969, + "logps/rejected": -332.0389709472656, + "loss": 0.5778, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.14384640753269196, + "rewards/margins": 0.33262744545936584, + "rewards/rejected": -0.476473867893219, + "step": 2950 + }, + { + "epoch": 0.39, + "learning_rate": 3.845001059992999e-06, + "logits/chosen": -2.2963719367980957, + "logits/rejected": -2.2243804931640625, + "logps/chosen": -321.4737854003906, + "logps/rejected": -306.0363464355469, + "loss": 0.5931, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.2161785364151001, + "rewards/margins": 0.285264253616333, + "rewards/rejected": -0.5014427900314331, + "step": 2960 + }, + { + "epoch": 0.39, + "learning_rate": 3.835360083441067e-06, + "logits/chosen": -2.1902008056640625, + "logits/rejected": -2.1949942111968994, + "logps/chosen": -284.7296447753906, + "logps/rejected": -296.80828857421875, + "loss": 0.559, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.12577447295188904, + "rewards/margins": 0.37559282779693604, + "rewards/rejected": -0.5013672113418579, + "step": 2970 + }, + { + "epoch": 0.39, + "learning_rate": 3.825691239304318e-06, + "logits/chosen": -2.2690746784210205, + "logits/rejected": -2.194706439971924, + "logps/chosen": -297.1219177246094, + "logps/rejected": -316.86395263671875, + "loss": 0.6104, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.19138354063034058, + "rewards/margins": 0.24811045825481415, + "rewards/rejected": -0.43949398398399353, + "step": 2980 + }, + { + "epoch": 0.39, + "learning_rate": 3.8159947293615385e-06, + "logits/chosen": -2.3479580879211426, + "logits/rejected": -2.2541697025299072, + "logps/chosen": -332.5345764160156, + "logps/rejected": -315.8721618652344, + "loss": 0.5872, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.10395509004592896, + "rewards/margins": 0.29379114508628845, + "rewards/rejected": -0.3977462351322174, + "step": 2990 + }, + { + "epoch": 0.39, + "learning_rate": 3.806270755968866e-06, + "logits/chosen": -2.34431791305542, + "logits/rejected": -2.254798412322998, + "logps/chosen": -364.5201721191406, + "logps/rejected": -317.5235290527344, + "loss": 0.5968, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.18700167536735535, + "rewards/margins": 0.317229688167572, + "rewards/rejected": -0.504231333732605, + "step": 3000 + }, + { + "epoch": 0.39, + "learning_rate": 3.7965195220555784e-06, + "logits/chosen": -2.1608643531799316, + "logits/rejected": -2.1240034103393555, + "logps/chosen": -289.5939025878906, + "logps/rejected": -296.3310241699219, + "loss": 0.6677, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.3151804506778717, + "rewards/margins": 0.11579956114292145, + "rewards/rejected": -0.43098002672195435, + "step": 3010 + }, + { + "epoch": 0.4, + "learning_rate": 3.786741231119847e-06, + "logits/chosen": -2.3387038707733154, + "logits/rejected": -2.179028272628784, + "logps/chosen": -307.25726318359375, + "logps/rejected": -297.0262756347656, + "loss": 0.6305, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.12563619017601013, + "rewards/margins": 0.220838263630867, + "rewards/rejected": -0.34647443890571594, + "step": 3020 + }, + { + "epoch": 0.4, + "learning_rate": 3.7769360872244992e-06, + "logits/chosen": -2.262904405593872, + "logits/rejected": -2.238494634628296, + "logps/chosen": -280.38232421875, + "logps/rejected": -277.2829284667969, + "loss": 0.6257, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.11235074698925018, + "rewards/margins": 0.2052387297153473, + "rewards/rejected": -0.3175894618034363, + "step": 3030 + }, + { + "epoch": 0.4, + "learning_rate": 3.767104294992754e-06, + "logits/chosen": -2.3239455223083496, + "logits/rejected": -2.2261970043182373, + "logps/chosen": -333.36273193359375, + "logps/rejected": -263.41082763671875, + "loss": 0.621, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.1215156763792038, + "rewards/margins": 0.2308710515499115, + "rewards/rejected": -0.3523866832256317, + "step": 3040 + }, + { + "epoch": 0.4, + "learning_rate": 3.7572460596039524e-06, + "logits/chosen": -2.2785532474517822, + "logits/rejected": -2.289318084716797, + "logps/chosen": -303.37408447265625, + "logps/rejected": -311.6347351074219, + "loss": 0.64, + "rewards/accuracies": 0.5874999761581421, + "rewards/chosen": -0.20798878371715546, + "rewards/margins": 0.19912748038768768, + "rewards/rejected": -0.40711626410484314, + "step": 3050 + }, + { + "epoch": 0.4, + "learning_rate": 3.74736158678928e-06, + "logits/chosen": -2.177031993865967, + "logits/rejected": -2.113590717315674, + "logps/chosen": -389.597412109375, + "logps/rejected": -324.69122314453125, + "loss": 0.5969, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.13935567438602448, + "rewards/margins": 0.30283206701278687, + "rewards/rejected": -0.4421877861022949, + "step": 3060 + }, + { + "epoch": 0.4, + "learning_rate": 3.7374510828274673e-06, + "logits/chosen": -2.3563344478607178, + "logits/rejected": -2.245603561401367, + "logps/chosen": -370.6936340332031, + "logps/rejected": -324.93255615234375, + "loss": 0.6248, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.1752164363861084, + "rewards/margins": 0.24540957808494568, + "rewards/rejected": -0.4206259846687317, + "step": 3070 + }, + { + "epoch": 0.4, + "learning_rate": 3.72751475454049e-06, + "logits/chosen": -2.223323345184326, + "logits/rejected": -2.1870429515838623, + "logps/chosen": -312.41802978515625, + "logps/rejected": -312.8993225097656, + "loss": 0.6326, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.13783670961856842, + "rewards/margins": 0.20678336918354034, + "rewards/rejected": -0.3446200489997864, + "step": 3080 + }, + { + "epoch": 0.4, + "learning_rate": 3.7175528092892503e-06, + "logits/chosen": -2.192528009414673, + "logits/rejected": -2.089958667755127, + "logps/chosen": -296.0037536621094, + "logps/rejected": -298.6204528808594, + "loss": 0.6075, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.1506446897983551, + "rewards/margins": 0.2501722574234009, + "rewards/rejected": -0.4008169174194336, + "step": 3090 + }, + { + "epoch": 0.41, + "learning_rate": 3.7075654549692498e-06, + "logits/chosen": -2.3563451766967773, + "logits/rejected": -2.253929853439331, + "logps/chosen": -373.33721923828125, + "logps/rejected": -308.59918212890625, + "loss": 0.5507, + "rewards/accuracies": 0.875, + "rewards/chosen": -0.03195354714989662, + "rewards/margins": 0.36858731508255005, + "rewards/rejected": -0.40054088830947876, + "step": 3100 + }, + { + "epoch": 0.41, + "learning_rate": 3.697552900006249e-06, + "logits/chosen": -2.2431931495666504, + "logits/rejected": -2.239027500152588, + "logps/chosen": -294.050048828125, + "logps/rejected": -288.5823669433594, + "loss": 0.6181, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.15926004946231842, + "rewards/margins": 0.2313537299633026, + "rewards/rejected": -0.3906137943267822, + "step": 3110 + }, + { + "epoch": 0.41, + "learning_rate": 3.6875153533519244e-06, + "logits/chosen": -2.153017520904541, + "logits/rejected": -2.1035304069519043, + "logps/chosen": -333.7864685058594, + "logps/rejected": -318.5226135253906, + "loss": 0.6047, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.09763199090957642, + "rewards/margins": 0.26820889115333557, + "rewards/rejected": -0.3658409118652344, + "step": 3120 + }, + { + "epoch": 0.41, + "learning_rate": 3.6774530244794992e-06, + "logits/chosen": -2.2834887504577637, + "logits/rejected": -2.151353597640991, + "logps/chosen": -315.78375244140625, + "logps/rejected": -276.2959289550781, + "loss": 0.6187, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.14533431828022003, + "rewards/margins": 0.24671044945716858, + "rewards/rejected": -0.3920448124408722, + "step": 3130 + }, + { + "epoch": 0.41, + "learning_rate": 3.667366123379378e-06, + "logits/chosen": -2.2238709926605225, + "logits/rejected": -2.1253607273101807, + "logps/chosen": -255.31527709960938, + "logps/rejected": -252.7228240966797, + "loss": 0.6183, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.22929397225379944, + "rewards/margins": 0.2054261714220047, + "rewards/rejected": -0.4347201883792877, + "step": 3140 + }, + { + "epoch": 0.41, + "learning_rate": 3.6572548605547607e-06, + "logits/chosen": -2.3579158782958984, + "logits/rejected": -2.274282217025757, + "logps/chosen": -282.13775634765625, + "logps/rejected": -323.9889221191406, + "loss": 0.5839, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.15901920199394226, + "rewards/margins": 0.33014026284217834, + "rewards/rejected": -0.4891594350337982, + "step": 3150 + }, + { + "epoch": 0.41, + "learning_rate": 3.6471194470172538e-06, + "logits/chosen": -2.29618501663208, + "logits/rejected": -2.2774035930633545, + "logps/chosen": -303.40814208984375, + "logps/rejected": -333.0376892089844, + "loss": 0.5978, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.17918583750724792, + "rewards/margins": 0.28591203689575195, + "rewards/rejected": -0.46509790420532227, + "step": 3160 + }, + { + "epoch": 0.41, + "learning_rate": 3.636960094282461e-06, + "logits/chosen": -2.196422576904297, + "logits/rejected": -2.139151096343994, + "logps/chosen": -320.49676513671875, + "logps/rejected": -305.7890319824219, + "loss": 0.5728, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.1928102970123291, + "rewards/margins": 0.34613439440727234, + "rewards/rejected": -0.538944661617279, + "step": 3170 + }, + { + "epoch": 0.42, + "learning_rate": 3.6267770143655743e-06, + "logits/chosen": -2.4047622680664062, + "logits/rejected": -2.283409595489502, + "logps/chosen": -320.8899841308594, + "logps/rejected": -308.2359924316406, + "loss": 0.6233, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.23850727081298828, + "rewards/margins": 0.23102286458015442, + "rewards/rejected": -0.4695301949977875, + "step": 3180 + }, + { + "epoch": 0.42, + "learning_rate": 3.6165704197769484e-06, + "logits/chosen": -2.2761237621307373, + "logits/rejected": -2.2387335300445557, + "logps/chosen": -295.4513854980469, + "logps/rejected": -303.45062255859375, + "loss": 0.6285, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.14914533495903015, + "rewards/margins": 0.2022182047367096, + "rewards/rejected": -0.35136356949806213, + "step": 3190 + }, + { + "epoch": 0.42, + "learning_rate": 3.606340523517663e-06, + "logits/chosen": -2.248460054397583, + "logits/rejected": -2.194171667098999, + "logps/chosen": -284.29449462890625, + "logps/rejected": -288.8624572753906, + "loss": 0.6104, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.07691571861505508, + "rewards/margins": 0.2391362488269806, + "rewards/rejected": -0.3160519599914551, + "step": 3200 + }, + { + "epoch": 0.42, + "learning_rate": 3.5960875390750793e-06, + "logits/chosen": -2.325347900390625, + "logits/rejected": -2.1235616207122803, + "logps/chosen": -360.14239501953125, + "logps/rejected": -321.9942932128906, + "loss": 0.5561, + "rewards/accuracies": 0.7875000238418579, + "rewards/chosen": -0.04620830714702606, + "rewards/margins": 0.38283371925354004, + "rewards/rejected": -0.4290420114994049, + "step": 3210 + }, + { + "epoch": 0.42, + "learning_rate": 3.585811680418386e-06, + "logits/chosen": -2.24568510055542, + "logits/rejected": -2.1574959754943848, + "logps/chosen": -310.50616455078125, + "logps/rejected": -308.9969177246094, + "loss": 0.5897, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.08727524429559708, + "rewards/margins": 0.29882222414016724, + "rewards/rejected": -0.3860974907875061, + "step": 3220 + }, + { + "epoch": 0.42, + "learning_rate": 3.5755131619941347e-06, + "logits/chosen": -2.201096296310425, + "logits/rejected": -2.157496452331543, + "logps/chosen": -288.51776123046875, + "logps/rejected": -278.84356689453125, + "loss": 0.611, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.13764691352844238, + "rewards/margins": 0.2472527027130127, + "rewards/rejected": -0.3848995864391327, + "step": 3230 + }, + { + "epoch": 0.42, + "learning_rate": 3.565192198721759e-06, + "logits/chosen": -2.207019567489624, + "logits/rejected": -2.1581149101257324, + "logps/chosen": -274.2110290527344, + "logps/rejected": -330.1119079589844, + "loss": 0.5841, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.17945356667041779, + "rewards/margins": 0.3046710789203644, + "rewards/rejected": -0.4841246008872986, + "step": 3240 + }, + { + "epoch": 0.43, + "learning_rate": 3.5548490059890965e-06, + "logits/chosen": -2.2094249725341797, + "logits/rejected": -2.1251931190490723, + "logps/chosen": -338.31134033203125, + "logps/rejected": -347.03143310546875, + "loss": 0.5808, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.14935877919197083, + "rewards/margins": 0.34563302993774414, + "rewards/rejected": -0.4949917793273926, + "step": 3250 + }, + { + "epoch": 0.43, + "learning_rate": 3.5444837996478903e-06, + "logits/chosen": -2.287841558456421, + "logits/rejected": -2.161818027496338, + "logps/chosen": -342.252685546875, + "logps/rejected": -380.78582763671875, + "loss": 0.6197, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.20302172005176544, + "rewards/margins": 0.259146511554718, + "rewards/rejected": -0.4621681571006775, + "step": 3260 + }, + { + "epoch": 0.43, + "learning_rate": 3.534096796009282e-06, + "logits/chosen": -2.1436102390289307, + "logits/rejected": -2.121990203857422, + "logps/chosen": -290.8398742675781, + "logps/rejected": -306.20050048828125, + "loss": 0.6164, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.12765438854694366, + "rewards/margins": 0.23902735114097595, + "rewards/rejected": -0.3666817545890808, + "step": 3270 + }, + { + "epoch": 0.43, + "learning_rate": 3.5236882118393046e-06, + "logits/chosen": -2.2807412147521973, + "logits/rejected": -2.20418119430542, + "logps/chosen": -330.96759033203125, + "logps/rejected": -313.71533203125, + "loss": 0.6203, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.08276918530464172, + "rewards/margins": 0.22776314616203308, + "rewards/rejected": -0.3105323016643524, + "step": 3280 + }, + { + "epoch": 0.43, + "learning_rate": 3.5132582643543513e-06, + "logits/chosen": -2.199484348297119, + "logits/rejected": -2.11887264251709, + "logps/chosen": -308.62176513671875, + "logps/rejected": -303.9732666015625, + "loss": 0.613, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.10497989505529404, + "rewards/margins": 0.22992253303527832, + "rewards/rejected": -0.33490240573883057, + "step": 3290 + }, + { + "epoch": 0.43, + "learning_rate": 3.5028071712166456e-06, + "logits/chosen": -2.2492740154266357, + "logits/rejected": -2.0946688652038574, + "logps/chosen": -314.8434143066406, + "logps/rejected": -283.57696533203125, + "loss": 0.5991, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.07174123823642731, + "rewards/margins": 0.2797015309333801, + "rewards/rejected": -0.35144275426864624, + "step": 3300 + }, + { + "epoch": 0.43, + "learning_rate": 3.4923351505297008e-06, + "logits/chosen": -2.3448548316955566, + "logits/rejected": -2.2648229598999023, + "logps/chosen": -326.4363098144531, + "logps/rejected": -330.6997375488281, + "loss": 0.5816, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.10261914879083633, + "rewards/margins": 0.299100786447525, + "rewards/rejected": -0.4017198979854584, + "step": 3310 + }, + { + "epoch": 0.43, + "learning_rate": 3.481842420833766e-06, + "logits/chosen": -2.3306705951690674, + "logits/rejected": -2.2149271965026855, + "logps/chosen": -368.263916015625, + "logps/rejected": -313.85577392578125, + "loss": 0.5753, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.1251881867647171, + "rewards/margins": 0.32386136054992676, + "rewards/rejected": -0.4490494728088379, + "step": 3320 + }, + { + "epoch": 0.44, + "learning_rate": 3.4713292011012645e-06, + "logits/chosen": -2.1722724437713623, + "logits/rejected": -2.05879282951355, + "logps/chosen": -327.58026123046875, + "logps/rejected": -323.82806396484375, + "loss": 0.6109, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.16994589567184448, + "rewards/margins": 0.2712971270084381, + "rewards/rejected": -0.4412430226802826, + "step": 3330 + }, + { + "epoch": 0.44, + "learning_rate": 3.4607957107322277e-06, + "logits/chosen": -2.2359910011291504, + "logits/rejected": -2.0647006034851074, + "logps/chosen": -275.21966552734375, + "logps/rejected": -279.6451110839844, + "loss": 0.5917, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.22481286525726318, + "rewards/margins": 0.28406020998954773, + "rewards/rejected": -0.5088731050491333, + "step": 3340 + }, + { + "epoch": 0.44, + "learning_rate": 3.4502421695497112e-06, + "logits/chosen": -2.1291165351867676, + "logits/rejected": -2.114325523376465, + "logps/chosen": -304.02459716796875, + "logps/rejected": -307.4547424316406, + "loss": 0.6274, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.1552276909351349, + "rewards/margins": 0.20688049495220184, + "rewards/rejected": -0.3621082007884979, + "step": 3350 + }, + { + "epoch": 0.44, + "learning_rate": 3.4396687977952137e-06, + "logits/chosen": -2.171285629272461, + "logits/rejected": -2.0632450580596924, + "logps/chosen": -295.75616455078125, + "logps/rejected": -300.7738342285156, + "loss": 0.5886, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.18257856369018555, + "rewards/margins": 0.3184903562068939, + "rewards/rejected": -0.5010689496994019, + "step": 3360 + }, + { + "epoch": 0.44, + "learning_rate": 3.429075816124075e-06, + "logits/chosen": -2.2236578464508057, + "logits/rejected": -2.1825549602508545, + "logps/chosen": -332.3364562988281, + "logps/rejected": -344.52398681640625, + "loss": 0.5825, + "rewards/accuracies": 0.8125, + "rewards/chosen": -0.23878760635852814, + "rewards/margins": 0.3120969235897064, + "rewards/rejected": -0.5508845448493958, + "step": 3370 + }, + { + "epoch": 0.44, + "learning_rate": 3.418463445600874e-06, + "logits/chosen": -2.254668712615967, + "logits/rejected": -2.173424243927002, + "logps/chosen": -275.3207092285156, + "logps/rejected": -298.79443359375, + "loss": 0.606, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.33824622631073, + "rewards/margins": 0.2654719650745392, + "rewards/rejected": -0.6037181615829468, + "step": 3380 + }, + { + "epoch": 0.44, + "learning_rate": 3.4078319076948173e-06, + "logits/chosen": -2.1863226890563965, + "logits/rejected": -2.1001029014587402, + "logps/chosen": -281.54888916015625, + "logps/rejected": -294.13311767578125, + "loss": 0.5669, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.18318873643875122, + "rewards/margins": 0.3577658236026764, + "rewards/rejected": -0.5409545302391052, + "step": 3390 + }, + { + "epoch": 0.44, + "learning_rate": 3.3971814242751123e-06, + "logits/chosen": -2.2467076778411865, + "logits/rejected": -2.1341750621795654, + "logps/chosen": -327.49407958984375, + "logps/rejected": -304.19384765625, + "loss": 0.6118, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.07793440669775009, + "rewards/margins": 0.24999889731407166, + "rewards/rejected": -0.32793331146240234, + "step": 3400 + }, + { + "epoch": 0.45, + "learning_rate": 3.386512217606339e-06, + "logits/chosen": -2.256131410598755, + "logits/rejected": -2.2085914611816406, + "logps/chosen": -301.0712890625, + "logps/rejected": -293.40728759765625, + "loss": 0.5734, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.15557502210140228, + "rewards/margins": 0.33370381593704224, + "rewards/rejected": -0.4892788529396057, + "step": 3410 + }, + { + "epoch": 0.45, + "learning_rate": 3.375824510343816e-06, + "logits/chosen": -2.2942771911621094, + "logits/rejected": -2.183180570602417, + "logps/chosen": -329.44525146484375, + "logps/rejected": -359.43402099609375, + "loss": 0.5946, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.1809997260570526, + "rewards/margins": 0.2896903157234192, + "rewards/rejected": -0.4706901013851166, + "step": 3420 + }, + { + "epoch": 0.45, + "learning_rate": 3.3651185255289466e-06, + "logits/chosen": -2.295954465866089, + "logits/rejected": -2.2029337882995605, + "logps/chosen": -350.3366394042969, + "logps/rejected": -360.73248291015625, + "loss": 0.5818, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.23058481514453888, + "rewards/margins": 0.31552547216415405, + "rewards/rejected": -0.5461102724075317, + "step": 3430 + }, + { + "epoch": 0.45, + "learning_rate": 3.354394486584568e-06, + "logits/chosen": -2.2945914268493652, + "logits/rejected": -2.179950714111328, + "logps/chosen": -354.335205078125, + "logps/rejected": -368.111328125, + "loss": 0.5819, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.21776917576789856, + "rewards/margins": 0.302031934261322, + "rewards/rejected": -0.519801139831543, + "step": 3440 + }, + { + "epoch": 0.45, + "learning_rate": 3.3436526173102913e-06, + "logits/chosen": -2.2314987182617188, + "logits/rejected": -2.1264474391937256, + "logps/chosen": -274.8760986328125, + "logps/rejected": -247.54135131835938, + "loss": 0.5977, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.21922855079174042, + "rewards/margins": 0.2775750160217285, + "rewards/rejected": -0.4968036115169525, + "step": 3450 + }, + { + "epoch": 0.45, + "learning_rate": 3.3328931418778254e-06, + "logits/chosen": -2.2393717765808105, + "logits/rejected": -2.203644275665283, + "logps/chosen": -310.0876770019531, + "logps/rejected": -333.08660888671875, + "loss": 0.5576, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.1818048506975174, + "rewards/margins": 0.3758290708065033, + "rewards/rejected": -0.5576339960098267, + "step": 3460 + }, + { + "epoch": 0.45, + "learning_rate": 3.3221162848263028e-06, + "logits/chosen": -2.255612850189209, + "logits/rejected": -2.183682680130005, + "logps/chosen": -335.07220458984375, + "logps/rejected": -319.34234619140625, + "loss": 0.5432, + "rewards/accuracies": 0.824999988079071, + "rewards/chosen": -0.18285012245178223, + "rewards/margins": 0.40342992544174194, + "rewards/rejected": -0.5862800478935242, + "step": 3470 + }, + { + "epoch": 0.46, + "learning_rate": 3.3113222710575914e-06, + "logits/chosen": -2.288682460784912, + "logits/rejected": -2.2233240604400635, + "logps/chosen": -342.3880310058594, + "logps/rejected": -330.94049072265625, + "loss": 0.6524, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.2884979248046875, + "rewards/margins": 0.15531443059444427, + "rewards/rejected": -0.44381237030029297, + "step": 3480 + }, + { + "epoch": 0.46, + "learning_rate": 3.300511325831603e-06, + "logits/chosen": -2.1967105865478516, + "logits/rejected": -2.1920440196990967, + "logps/chosen": -306.7802429199219, + "logps/rejected": -332.65289306640625, + "loss": 0.6126, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.17921099066734314, + "rewards/margins": 0.2644205093383789, + "rewards/rejected": -0.44363147020339966, + "step": 3490 + }, + { + "epoch": 0.46, + "learning_rate": 3.289683674761592e-06, + "logits/chosen": -2.21291184425354, + "logits/rejected": -2.155174493789673, + "logps/chosen": -310.1589660644531, + "logps/rejected": -302.49102783203125, + "loss": 0.5816, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.19234074652194977, + "rewards/margins": 0.31977444887161255, + "rewards/rejected": -0.5121151804924011, + "step": 3500 + }, + { + "epoch": 0.46, + "learning_rate": 3.2788395438094444e-06, + "logits/chosen": -2.2448463439941406, + "logits/rejected": -2.154503583908081, + "logps/chosen": -322.6603698730469, + "logps/rejected": -321.0233154296875, + "loss": 0.603, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.22834105789661407, + "rewards/margins": 0.29928749799728394, + "rewards/rejected": -0.5276285409927368, + "step": 3510 + }, + { + "epoch": 0.46, + "learning_rate": 3.2679791592809653e-06, + "logits/chosen": -2.2332160472869873, + "logits/rejected": -2.134450912475586, + "logps/chosen": -271.142822265625, + "logps/rejected": -309.9905090332031, + "loss": 0.5777, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.07708028703927994, + "rewards/margins": 0.30101966857910156, + "rewards/rejected": -0.3780999183654785, + "step": 3520 + }, + { + "epoch": 0.46, + "learning_rate": 3.257102747821157e-06, + "logits/chosen": -2.184156894683838, + "logits/rejected": -2.0749685764312744, + "logps/chosen": -334.580322265625, + "logps/rejected": -338.4678955078125, + "loss": 0.5468, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.09229175001382828, + "rewards/margins": 0.4070521295070648, + "rewards/rejected": -0.49934396147727966, + "step": 3530 + }, + { + "epoch": 0.46, + "learning_rate": 3.246210536409484e-06, + "logits/chosen": -2.1600894927978516, + "logits/rejected": -2.0873522758483887, + "logps/chosen": -318.50640869140625, + "logps/rejected": -305.25787353515625, + "loss": 0.6145, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.20227620005607605, + "rewards/margins": 0.24205967783927917, + "rewards/rejected": -0.44433584809303284, + "step": 3540 + }, + { + "epoch": 0.46, + "learning_rate": 3.235302752355142e-06, + "logits/chosen": -2.2977747917175293, + "logits/rejected": -2.1594138145446777, + "logps/chosen": -337.0153503417969, + "logps/rejected": -287.40789794921875, + "loss": 0.5581, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.0762878805398941, + "rewards/margins": 0.37493664026260376, + "rewards/rejected": -0.45122450590133667, + "step": 3550 + }, + { + "epoch": 0.47, + "learning_rate": 3.2243796232923097e-06, + "logits/chosen": -2.3173668384552, + "logits/rejected": -2.1900227069854736, + "logps/chosen": -362.87994384765625, + "logps/rejected": -369.14581298828125, + "loss": 0.6038, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.2711828351020813, + "rewards/margins": 0.27714723348617554, + "rewards/rejected": -0.5483301877975464, + "step": 3560 + }, + { + "epoch": 0.47, + "learning_rate": 3.2134413771754037e-06, + "logits/chosen": -2.3003287315368652, + "logits/rejected": -2.2316155433654785, + "logps/chosen": -307.1923828125, + "logps/rejected": -319.0071105957031, + "loss": 0.5999, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.2506977617740631, + "rewards/margins": 0.2686203718185425, + "rewards/rejected": -0.519318163394928, + "step": 3570 + }, + { + "epoch": 0.47, + "learning_rate": 3.2024882422743118e-06, + "logits/chosen": -2.206987142562866, + "logits/rejected": -2.154073476791382, + "logps/chosen": -273.85357666015625, + "logps/rejected": -308.89007568359375, + "loss": 0.5649, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.20727932453155518, + "rewards/margins": 0.34459999203681946, + "rewards/rejected": -0.5518792867660522, + "step": 3580 + }, + { + "epoch": 0.47, + "learning_rate": 3.1915204471696425e-06, + "logits/chosen": -2.298491954803467, + "logits/rejected": -2.126559019088745, + "logps/chosen": -301.7465515136719, + "logps/rejected": -271.80535888671875, + "loss": 0.5882, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.13729017972946167, + "rewards/margins": 0.3090158998966217, + "rewards/rejected": -0.4463060796260834, + "step": 3590 + }, + { + "epoch": 0.47, + "learning_rate": 3.180538220747943e-06, + "logits/chosen": -2.240255355834961, + "logits/rejected": -2.1766419410705566, + "logps/chosen": -330.14813232421875, + "logps/rejected": -320.18756103515625, + "loss": 0.6079, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.24947145581245422, + "rewards/margins": 0.28124094009399414, + "rewards/rejected": -0.530712366104126, + "step": 3600 + }, + { + "epoch": 0.47, + "learning_rate": 3.1695417921969287e-06, + "logits/chosen": -2.3270153999328613, + "logits/rejected": -2.2739806175231934, + "logps/chosen": -294.16253662109375, + "logps/rejected": -328.5826721191406, + "loss": 0.6349, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.14072641730308533, + "rewards/margins": 0.1998475342988968, + "rewards/rejected": -0.3405739665031433, + "step": 3610 + }, + { + "epoch": 0.47, + "learning_rate": 3.158531391000697e-06, + "logits/chosen": -2.2572436332702637, + "logits/rejected": -2.1108286380767822, + "logps/chosen": -315.2557373046875, + "logps/rejected": -303.2156982421875, + "loss": 0.582, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.14727096259593964, + "rewards/margins": 0.34550607204437256, + "rewards/rejected": -0.4927770495414734, + "step": 3620 + }, + { + "epoch": 0.48, + "learning_rate": 3.147507246934943e-06, + "logits/chosen": -2.273355722427368, + "logits/rejected": -2.232914447784424, + "logps/chosen": -340.02618408203125, + "logps/rejected": -342.54217529296875, + "loss": 0.5948, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.0910380631685257, + "rewards/margins": 0.2883290648460388, + "rewards/rejected": -0.3793671727180481, + "step": 3630 + }, + { + "epoch": 0.48, + "learning_rate": 3.136469590062158e-06, + "logits/chosen": -2.1601414680480957, + "logits/rejected": -2.0744059085845947, + "logps/chosen": -277.2970275878906, + "logps/rejected": -271.2074890136719, + "loss": 0.5807, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.1494699865579605, + "rewards/margins": 0.32256048917770386, + "rewards/rejected": -0.4720304608345032, + "step": 3640 + }, + { + "epoch": 0.48, + "learning_rate": 3.1254186507268354e-06, + "logits/chosen": -2.272613763809204, + "logits/rejected": -2.1787149906158447, + "logps/chosen": -267.9743347167969, + "logps/rejected": -291.6812438964844, + "loss": 0.6001, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.07630626112222672, + "rewards/margins": 0.2765544354915619, + "rewards/rejected": -0.3528607189655304, + "step": 3650 + }, + { + "epoch": 0.48, + "learning_rate": 3.114354659550656e-06, + "logits/chosen": -2.1525888442993164, + "logits/rejected": -2.1212618350982666, + "logps/chosen": -322.45074462890625, + "logps/rejected": -308.0700378417969, + "loss": 0.5766, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.24668438732624054, + "rewards/margins": 0.34969207644462585, + "rewards/rejected": -0.5963764190673828, + "step": 3660 + }, + { + "epoch": 0.48, + "learning_rate": 3.1032778474276816e-06, + "logits/chosen": -2.366992473602295, + "logits/rejected": -2.2086708545684814, + "logps/chosen": -319.9839782714844, + "logps/rejected": -298.745361328125, + "loss": 0.5549, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.2000807523727417, + "rewards/margins": 0.3787020444869995, + "rewards/rejected": -0.5787827968597412, + "step": 3670 + }, + { + "epoch": 0.48, + "learning_rate": 3.092188445519532e-06, + "logits/chosen": -2.26643967628479, + "logits/rejected": -2.128983974456787, + "logps/chosen": -326.4763488769531, + "logps/rejected": -297.0374755859375, + "loss": 0.6034, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.24593958258628845, + "rewards/margins": 0.30430930852890015, + "rewards/rejected": -0.550248920917511, + "step": 3680 + }, + { + "epoch": 0.48, + "learning_rate": 3.081086685250565e-06, + "logits/chosen": -2.2758965492248535, + "logits/rejected": -2.1524596214294434, + "logps/chosen": -292.1910400390625, + "logps/rejected": -272.1326904296875, + "loss": 0.5854, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.2058788239955902, + "rewards/margins": 0.31988587975502014, + "rewards/rejected": -0.5257646441459656, + "step": 3690 + }, + { + "epoch": 0.48, + "learning_rate": 3.0699727983030434e-06, + "logits/chosen": -2.3243601322174072, + "logits/rejected": -2.2180685997009277, + "logps/chosen": -332.9953308105469, + "logps/rejected": -295.5639953613281, + "loss": 0.6358, + "rewards/accuracies": 0.574999988079071, + "rewards/chosen": -0.23166723549365997, + "rewards/margins": 0.2067691832780838, + "rewards/rejected": -0.4384364187717438, + "step": 3700 + }, + { + "epoch": 0.49, + "learning_rate": 3.058847016612301e-06, + "logits/chosen": -2.23081636428833, + "logits/rejected": -2.207549571990967, + "logps/chosen": -353.3510437011719, + "logps/rejected": -366.8516540527344, + "loss": 0.6399, + "rewards/accuracies": 0.6000000238418579, + "rewards/chosen": -0.19398830831050873, + "rewards/margins": 0.19586136937141418, + "rewards/rejected": -0.3898496925830841, + "step": 3710 + }, + { + "epoch": 0.49, + "learning_rate": 3.0477095723619034e-06, + "logits/chosen": -2.3202261924743652, + "logits/rejected": -2.171468734741211, + "logps/chosen": -286.7763366699219, + "logps/rejected": -291.25006103515625, + "loss": 0.5881, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.18975184857845306, + "rewards/margins": 0.3185967803001404, + "rewards/rejected": -0.5083485841751099, + "step": 3720 + }, + { + "epoch": 0.49, + "learning_rate": 3.0365606979788003e-06, + "logits/chosen": -2.29597544670105, + "logits/rejected": -2.201775074005127, + "logps/chosen": -308.6613464355469, + "logps/rejected": -297.501220703125, + "loss": 0.589, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.16128845512866974, + "rewards/margins": 0.29569411277770996, + "rewards/rejected": -0.4569825530052185, + "step": 3730 + }, + { + "epoch": 0.49, + "learning_rate": 3.0254006261284786e-06, + "logits/chosen": -2.2917492389678955, + "logits/rejected": -2.20141339302063, + "logps/chosen": -305.4287109375, + "logps/rejected": -313.76934814453125, + "loss": 0.5959, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.22600774466991425, + "rewards/margins": 0.28799906373023987, + "rewards/rejected": -0.5140067934989929, + "step": 3740 + }, + { + "epoch": 0.49, + "learning_rate": 3.0142295897101032e-06, + "logits/chosen": -2.1403932571411133, + "logits/rejected": -1.9929752349853516, + "logps/chosen": -326.08441162109375, + "logps/rejected": -297.84808349609375, + "loss": 0.5601, + "rewards/accuracies": 0.8125, + "rewards/chosen": -0.1513725221157074, + "rewards/margins": 0.3802407681941986, + "rewards/rejected": -0.5316132307052612, + "step": 3750 + }, + { + "epoch": 0.49, + "learning_rate": 3.0030478218516578e-06, + "logits/chosen": -2.2112417221069336, + "logits/rejected": -2.1325557231903076, + "logps/chosen": -309.48089599609375, + "logps/rejected": -310.97509765625, + "loss": 0.588, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.14635725319385529, + "rewards/margins": 0.3053987920284271, + "rewards/rejected": -0.4517560601234436, + "step": 3760 + }, + { + "epoch": 0.49, + "learning_rate": 2.9918555559050826e-06, + "logits/chosen": -2.1812379360198975, + "logits/rejected": -2.1582610607147217, + "logps/chosen": -297.3588562011719, + "logps/rejected": -288.1904296875, + "loss": 0.5738, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.13978731632232666, + "rewards/margins": 0.3152059316635132, + "rewards/rejected": -0.45499324798583984, + "step": 3770 + }, + { + "epoch": 0.49, + "learning_rate": 2.980653025441399e-06, + "logits/chosen": -2.294768810272217, + "logits/rejected": -2.217179775238037, + "logps/chosen": -335.85235595703125, + "logps/rejected": -327.75543212890625, + "loss": 0.5307, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.053152166306972504, + "rewards/margins": 0.4419528841972351, + "rewards/rejected": -0.4951050877571106, + "step": 3780 + }, + { + "epoch": 0.5, + "learning_rate": 2.969440464245841e-06, + "logits/chosen": -2.312551975250244, + "logits/rejected": -2.216581344604492, + "logps/chosen": -335.02862548828125, + "logps/rejected": -301.836181640625, + "loss": 0.6369, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.15713906288146973, + "rewards/margins": 0.18496128916740417, + "rewards/rejected": -0.3421003222465515, + "step": 3790 + }, + { + "epoch": 0.5, + "learning_rate": 2.95821810631297e-06, + "logits/chosen": -2.155395030975342, + "logits/rejected": -2.148796796798706, + "logps/chosen": -260.44122314453125, + "logps/rejected": -276.7882080078125, + "loss": 0.5771, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.0877716988325119, + "rewards/margins": 0.322742760181427, + "rewards/rejected": -0.4105144441127777, + "step": 3800 + }, + { + "epoch": 0.5, + "learning_rate": 2.946986185841801e-06, + "logits/chosen": -2.1743924617767334, + "logits/rejected": -2.0896475315093994, + "logps/chosen": -299.45391845703125, + "logps/rejected": -304.2263488769531, + "loss": 0.6052, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.12790082395076752, + "rewards/margins": 0.24636109173297882, + "rewards/rejected": -0.37426191568374634, + "step": 3810 + }, + { + "epoch": 0.5, + "learning_rate": 2.935744937230903e-06, + "logits/chosen": -2.1896514892578125, + "logits/rejected": -2.161184310913086, + "logps/chosen": -324.8854675292969, + "logps/rejected": -304.50994873046875, + "loss": 0.6399, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.06740345805883408, + "rewards/margins": 0.1831163913011551, + "rewards/rejected": -0.25051984190940857, + "step": 3820 + }, + { + "epoch": 0.5, + "learning_rate": 2.924494595073517e-06, + "logits/chosen": -2.2635974884033203, + "logits/rejected": -2.2178092002868652, + "logps/chosen": -305.7368469238281, + "logps/rejected": -304.89007568359375, + "loss": 0.5615, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.032348960638046265, + "rewards/margins": 0.38179102540016174, + "rewards/rejected": -0.41413992643356323, + "step": 3830 + }, + { + "epoch": 0.5, + "learning_rate": 2.9132353941526575e-06, + "logits/chosen": -2.2265233993530273, + "logits/rejected": -2.157034158706665, + "logps/chosen": -296.8994445800781, + "logps/rejected": -320.8399963378906, + "loss": 0.6032, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.1254245936870575, + "rewards/margins": 0.27609962224960327, + "rewards/rejected": -0.40152424573898315, + "step": 3840 + }, + { + "epoch": 0.5, + "learning_rate": 2.901967569436209e-06, + "logits/chosen": -2.2328522205352783, + "logits/rejected": -2.1735124588012695, + "logps/chosen": -250.29086303710938, + "logps/rejected": -267.7702941894531, + "loss": 0.5602, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.062219761312007904, + "rewards/margins": 0.36623015999794006, + "rewards/rejected": -0.42844992876052856, + "step": 3850 + }, + { + "epoch": 0.51, + "learning_rate": 2.89069135607203e-06, + "logits/chosen": -2.1895928382873535, + "logits/rejected": -2.1317780017852783, + "logps/chosen": -311.4351806640625, + "logps/rejected": -313.31988525390625, + "loss": 0.6074, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.10760854184627533, + "rewards/margins": 0.24679633975028992, + "rewards/rejected": -0.35440492630004883, + "step": 3860 + }, + { + "epoch": 0.51, + "learning_rate": 2.8794069893830386e-06, + "logits/chosen": -2.198737382888794, + "logits/rejected": -2.1195740699768066, + "logps/chosen": -287.1357116699219, + "logps/rejected": -308.64666748046875, + "loss": 0.5917, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.1630997508764267, + "rewards/margins": 0.29942062497138977, + "rewards/rejected": -0.46252042055130005, + "step": 3870 + }, + { + "epoch": 0.51, + "learning_rate": 2.8681147048623038e-06, + "logits/chosen": -2.2363996505737305, + "logits/rejected": -2.2083020210266113, + "logps/chosen": -323.7630310058594, + "logps/rejected": -343.34991455078125, + "loss": 0.5567, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.06575950235128403, + "rewards/margins": 0.38548240065574646, + "rewards/rejected": -0.4512419104576111, + "step": 3880 + }, + { + "epoch": 0.51, + "learning_rate": 2.8568147381681333e-06, + "logits/chosen": -2.246504068374634, + "logits/rejected": -2.189868688583374, + "logps/chosen": -288.006591796875, + "logps/rejected": -303.1795959472656, + "loss": 0.6026, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.1896587759256363, + "rewards/margins": 0.28125303983688354, + "rewards/rejected": -0.47091180086135864, + "step": 3890 + }, + { + "epoch": 0.51, + "learning_rate": 2.8455073251191533e-06, + "logits/chosen": -2.279578447341919, + "logits/rejected": -2.235502243041992, + "logps/chosen": -310.79351806640625, + "logps/rejected": -327.28033447265625, + "loss": 0.6095, + "rewards/accuracies": 0.6000000238418579, + "rewards/chosen": -0.2548527121543884, + "rewards/margins": 0.2784276306629181, + "rewards/rejected": -0.5332802534103394, + "step": 3900 + }, + { + "epoch": 0.51, + "learning_rate": 2.8341927016893887e-06, + "logits/chosen": -2.243419647216797, + "logits/rejected": -2.213569164276123, + "logps/chosen": -282.7080993652344, + "logps/rejected": -317.5572509765625, + "loss": 0.591, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.28008633852005005, + "rewards/margins": 0.30951592326164246, + "rewards/rejected": -0.5896022915840149, + "step": 3910 + }, + { + "epoch": 0.51, + "learning_rate": 2.822871104003335e-06, + "logits/chosen": -2.1288013458251953, + "logits/rejected": -2.0994820594787598, + "logps/chosen": -268.18475341796875, + "logps/rejected": -329.8453063964844, + "loss": 0.5971, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.19896329939365387, + "rewards/margins": 0.28691792488098145, + "rewards/rejected": -0.48588117957115173, + "step": 3920 + }, + { + "epoch": 0.51, + "learning_rate": 2.8115427683310355e-06, + "logits/chosen": -2.309434413909912, + "logits/rejected": -2.18902850151062, + "logps/chosen": -343.51385498046875, + "logps/rejected": -341.13525390625, + "loss": 0.5761, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.19033750891685486, + "rewards/margins": 0.3401159346103668, + "rewards/rejected": -0.5304533839225769, + "step": 3930 + }, + { + "epoch": 0.52, + "learning_rate": 2.8002079310831477e-06, + "logits/chosen": -2.25565242767334, + "logits/rejected": -2.157276153564453, + "logps/chosen": -297.09967041015625, + "logps/rejected": -317.29083251953125, + "loss": 0.6199, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.21050819754600525, + "rewards/margins": 0.23793046176433563, + "rewards/rejected": -0.4484386444091797, + "step": 3940 + }, + { + "epoch": 0.52, + "learning_rate": 2.7888668288060095e-06, + "logits/chosen": -2.258208751678467, + "logits/rejected": -2.150857925415039, + "logps/chosen": -326.5715637207031, + "logps/rejected": -338.069580078125, + "loss": 0.5258, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.1631345897912979, + "rewards/margins": 0.44891101121902466, + "rewards/rejected": -0.6120456457138062, + "step": 3950 + }, + { + "epoch": 0.52, + "learning_rate": 2.7775196981767044e-06, + "logits/chosen": -2.1642158031463623, + "logits/rejected": -2.065582275390625, + "logps/chosen": -278.7409973144531, + "logps/rejected": -314.7583923339844, + "loss": 0.5873, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.19401176273822784, + "rewards/margins": 0.34037813544273376, + "rewards/rejected": -0.5343899726867676, + "step": 3960 + }, + { + "epoch": 0.52, + "learning_rate": 2.7661667759981213e-06, + "logits/chosen": -2.245973587036133, + "logits/rejected": -2.2456717491149902, + "logps/chosen": -271.9554138183594, + "logps/rejected": -294.5726623535156, + "loss": 0.6176, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.22680599987506866, + "rewards/margins": 0.2192671298980713, + "rewards/rejected": -0.44607311487197876, + "step": 3970 + }, + { + "epoch": 0.52, + "learning_rate": 2.7548082991940137e-06, + "logits/chosen": -2.23134183883667, + "logits/rejected": -2.1554768085479736, + "logps/chosen": -308.91314697265625, + "logps/rejected": -320.1424865722656, + "loss": 0.6458, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.20446991920471191, + "rewards/margins": 0.19974420964717865, + "rewards/rejected": -0.40421414375305176, + "step": 3980 + }, + { + "epoch": 0.52, + "learning_rate": 2.743444504804051e-06, + "logits/chosen": -2.1393723487854004, + "logits/rejected": -2.0969605445861816, + "logps/chosen": -254.2072296142578, + "logps/rejected": -263.64495849609375, + "loss": 0.6221, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.21768343448638916, + "rewards/margins": 0.21589119732379913, + "rewards/rejected": -0.4335746169090271, + "step": 3990 + }, + { + "epoch": 0.52, + "learning_rate": 2.7320756299788788e-06, + "logits/chosen": -2.199575901031494, + "logits/rejected": -2.108891010284424, + "logps/chosen": -316.3847961425781, + "logps/rejected": -306.77606201171875, + "loss": 0.6127, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.16566872596740723, + "rewards/margins": 0.25379401445388794, + "rewards/rejected": -0.41946274042129517, + "step": 4000 + }, + { + "epoch": 0.52, + "learning_rate": 2.7207019119751644e-06, + "logits/chosen": -2.2674145698547363, + "logits/rejected": -2.1696324348449707, + "logps/chosen": -276.8014221191406, + "logps/rejected": -265.08660888671875, + "loss": 0.5671, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.1784783899784088, + "rewards/margins": 0.3498672842979431, + "rewards/rejected": -0.5283456444740295, + "step": 4010 + }, + { + "epoch": 0.53, + "learning_rate": 2.7093235881506474e-06, + "logits/chosen": -2.3143670558929443, + "logits/rejected": -2.181687831878662, + "logps/chosen": -314.69378662109375, + "logps/rejected": -334.4461975097656, + "loss": 0.5555, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.1640726625919342, + "rewards/margins": 0.4091968536376953, + "rewards/rejected": -0.5732695460319519, + "step": 4020 + }, + { + "epoch": 0.53, + "learning_rate": 2.6979408959591863e-06, + "logits/chosen": -2.280393123626709, + "logits/rejected": -2.2724785804748535, + "logps/chosen": -272.15374755859375, + "logps/rejected": -299.1646728515625, + "loss": 0.6125, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.13850513100624084, + "rewards/margins": 0.2782619595527649, + "rewards/rejected": -0.41676706075668335, + "step": 4030 + }, + { + "epoch": 0.53, + "learning_rate": 2.6865540729458034e-06, + "logits/chosen": -2.2067208290100098, + "logits/rejected": -2.104708194732666, + "logps/chosen": -318.2445373535156, + "logps/rejected": -334.36676025390625, + "loss": 0.5605, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.09667529910802841, + "rewards/margins": 0.3725287914276123, + "rewards/rejected": -0.4692040979862213, + "step": 4040 + }, + { + "epoch": 0.53, + "learning_rate": 2.675163356741726e-06, + "logits/chosen": -2.2189674377441406, + "logits/rejected": -2.1255784034729004, + "logps/chosen": -267.58892822265625, + "logps/rejected": -300.7120666503906, + "loss": 0.5771, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.03780367970466614, + "rewards/margins": 0.3484765887260437, + "rewards/rejected": -0.38628023862838745, + "step": 4050 + }, + { + "epoch": 0.53, + "learning_rate": 2.6637689850594285e-06, + "logits/chosen": -2.2345590591430664, + "logits/rejected": -2.1449902057647705, + "logps/chosen": -273.7065124511719, + "logps/rejected": -324.08111572265625, + "loss": 0.6185, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.11849425733089447, + "rewards/margins": 0.24025535583496094, + "rewards/rejected": -0.3587496876716614, + "step": 4060 + }, + { + "epoch": 0.53, + "learning_rate": 2.652371195687671e-06, + "logits/chosen": -2.2655909061431885, + "logits/rejected": -2.1847379207611084, + "logps/chosen": -287.54351806640625, + "logps/rejected": -315.51531982421875, + "loss": 0.5583, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.022312898188829422, + "rewards/margins": 0.3931361138820648, + "rewards/rejected": -0.41544899344444275, + "step": 4070 + }, + { + "epoch": 0.53, + "learning_rate": 2.64097022648654e-06, + "logits/chosen": -2.220186710357666, + "logits/rejected": -2.1098077297210693, + "logps/chosen": -325.7334289550781, + "logps/rejected": -305.6304016113281, + "loss": 0.5655, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.08791736513376236, + "rewards/margins": 0.38575097918510437, + "rewards/rejected": -0.4736683964729309, + "step": 4080 + }, + { + "epoch": 0.54, + "learning_rate": 2.6295663153824774e-06, + "logits/chosen": -2.2417075634002686, + "logits/rejected": -2.1070821285247803, + "logps/chosen": -322.048828125, + "logps/rejected": -293.63092041015625, + "loss": 0.5736, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.11551324278116226, + "rewards/margins": 0.3490517735481262, + "rewards/rejected": -0.4645650386810303, + "step": 4090 + }, + { + "epoch": 0.54, + "learning_rate": 2.6181597003633218e-06, + "logits/chosen": -2.284554958343506, + "logits/rejected": -2.167095899581909, + "logps/chosen": -289.63763427734375, + "logps/rejected": -300.6242370605469, + "loss": 0.5748, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.05315490812063217, + "rewards/margins": 0.34837380051612854, + "rewards/rejected": -0.4015286862850189, + "step": 4100 + }, + { + "epoch": 0.54, + "learning_rate": 2.606750619473342e-06, + "logits/chosen": -2.1678881645202637, + "logits/rejected": -2.1289010047912598, + "logps/chosen": -276.3775634765625, + "logps/rejected": -320.9483337402344, + "loss": 0.5601, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.14479228854179382, + "rewards/margins": 0.36358359456062317, + "rewards/rejected": -0.508375883102417, + "step": 4110 + }, + { + "epoch": 0.54, + "learning_rate": 2.595339310808262e-06, + "logits/chosen": -2.2610087394714355, + "logits/rejected": -2.250720500946045, + "logps/chosen": -279.8154296875, + "logps/rejected": -314.9268493652344, + "loss": 0.5925, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.12273641675710678, + "rewards/margins": 0.29857784509658813, + "rewards/rejected": -0.4213142395019531, + "step": 4120 + }, + { + "epoch": 0.54, + "learning_rate": 2.5839260125103004e-06, + "logits/chosen": -2.2299766540527344, + "logits/rejected": -2.088897228240967, + "logps/chosen": -290.0578308105469, + "logps/rejected": -304.0356750488281, + "loss": 0.5816, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.1891263872385025, + "rewards/margins": 0.3399195075035095, + "rewards/rejected": -0.5290459394454956, + "step": 4130 + }, + { + "epoch": 0.54, + "learning_rate": 2.5725109627631984e-06, + "logits/chosen": -2.2458443641662598, + "logits/rejected": -2.2014641761779785, + "logps/chosen": -317.4651794433594, + "logps/rejected": -290.9559020996094, + "loss": 0.5728, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.15963537991046906, + "rewards/margins": 0.34964263439178467, + "rewards/rejected": -0.5092780590057373, + "step": 4140 + }, + { + "epoch": 0.54, + "learning_rate": 2.5610943997872443e-06, + "logits/chosen": -2.21856951713562, + "logits/rejected": -2.118227481842041, + "logps/chosen": -304.394775390625, + "logps/rejected": -321.1172790527344, + "loss": 0.6039, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.26354244351387024, + "rewards/margins": 0.2619742751121521, + "rewards/rejected": -0.5255167484283447, + "step": 4150 + }, + { + "epoch": 0.54, + "learning_rate": 2.5496765618343096e-06, + "logits/chosen": -2.2561728954315186, + "logits/rejected": -2.2816646099090576, + "logps/chosen": -235.952880859375, + "logps/rejected": -308.1616516113281, + "loss": 0.5897, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.049363039433956146, + "rewards/margins": 0.29978281259536743, + "rewards/rejected": -0.3491458594799042, + "step": 4160 + }, + { + "epoch": 0.55, + "learning_rate": 2.538257687182871e-06, + "logits/chosen": -2.2296934127807617, + "logits/rejected": -2.24413800239563, + "logps/chosen": -312.1509094238281, + "logps/rejected": -329.59002685546875, + "loss": 0.5784, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.21270079910755157, + "rewards/margins": 0.3697042167186737, + "rewards/rejected": -0.5824050903320312, + "step": 4170 + }, + { + "epoch": 0.55, + "learning_rate": 2.526838014133041e-06, + "logits/chosen": -2.300278663635254, + "logits/rejected": -2.1520462036132812, + "logps/chosen": -295.089111328125, + "logps/rejected": -299.8688049316406, + "loss": 0.6147, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.17557980120182037, + "rewards/margins": 0.2640655040740967, + "rewards/rejected": -0.43964534997940063, + "step": 4180 + }, + { + "epoch": 0.55, + "learning_rate": 2.515417781001594e-06, + "logits/chosen": -2.202958583831787, + "logits/rejected": -2.1037344932556152, + "logps/chosen": -315.9857482910156, + "logps/rejected": -307.81396484375, + "loss": 0.5757, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.24192318320274353, + "rewards/margins": 0.34862837195396423, + "rewards/rejected": -0.5905515551567078, + "step": 4190 + }, + { + "epoch": 0.55, + "learning_rate": 2.503997226116992e-06, + "logits/chosen": -2.2134172916412354, + "logits/rejected": -2.1476619243621826, + "logps/chosen": -292.0931091308594, + "logps/rejected": -327.3158874511719, + "loss": 0.589, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.20902717113494873, + "rewards/margins": 0.29169192910194397, + "rewards/rejected": -0.5007191300392151, + "step": 4200 + }, + { + "epoch": 0.55, + "learning_rate": 2.4925765878144115e-06, + "logits/chosen": -2.272026777267456, + "logits/rejected": -2.180445909500122, + "logps/chosen": -282.70440673828125, + "logps/rejected": -305.32659912109375, + "loss": 0.5539, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.1601114571094513, + "rewards/margins": 0.39398443698883057, + "rewards/rejected": -0.5540959239006042, + "step": 4210 + }, + { + "epoch": 0.55, + "learning_rate": 2.4811561044307727e-06, + "logits/chosen": -2.174419403076172, + "logits/rejected": -2.157695770263672, + "logps/chosen": -276.0654296875, + "logps/rejected": -327.058837890625, + "loss": 0.5467, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.1962500512599945, + "rewards/margins": 0.4340830445289612, + "rewards/rejected": -0.6303331255912781, + "step": 4220 + }, + { + "epoch": 0.55, + "learning_rate": 2.469736014299758e-06, + "logits/chosen": -2.3431780338287354, + "logits/rejected": -2.2919249534606934, + "logps/chosen": -320.96673583984375, + "logps/rejected": -338.6656494140625, + "loss": 0.6172, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.2039066106081009, + "rewards/margins": 0.24904990196228027, + "rewards/rejected": -0.45295652747154236, + "step": 4230 + }, + { + "epoch": 0.55, + "learning_rate": 2.458316555746846e-06, + "logits/chosen": -2.3326783180236816, + "logits/rejected": -2.273663282394409, + "logps/chosen": -365.83270263671875, + "logps/rejected": -353.1064758300781, + "loss": 0.5618, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.17347517609596252, + "rewards/margins": 0.3686824440956116, + "rewards/rejected": -0.5421576499938965, + "step": 4240 + }, + { + "epoch": 0.56, + "learning_rate": 2.446897967084334e-06, + "logits/chosen": -2.271059274673462, + "logits/rejected": -2.1506505012512207, + "logps/chosen": -344.97119140625, + "logps/rejected": -309.66107177734375, + "loss": 0.5966, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.25669288635253906, + "rewards/margins": 0.31914716958999634, + "rewards/rejected": -0.5758401155471802, + "step": 4250 + }, + { + "epoch": 0.56, + "learning_rate": 2.4354804866063684e-06, + "logits/chosen": -2.248983860015869, + "logits/rejected": -2.182692289352417, + "logps/chosen": -312.4232177734375, + "logps/rejected": -328.446533203125, + "loss": 0.5894, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.22014348208904266, + "rewards/margins": 0.3003430664539337, + "rewards/rejected": -0.5204865336418152, + "step": 4260 + }, + { + "epoch": 0.56, + "learning_rate": 2.424064352583964e-06, + "logits/chosen": -2.297269344329834, + "logits/rejected": -2.1917881965637207, + "logps/chosen": -320.69378662109375, + "logps/rejected": -361.626220703125, + "loss": 0.5803, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.2521662414073944, + "rewards/margins": 0.34581252932548523, + "rewards/rejected": -0.5979788303375244, + "step": 4270 + }, + { + "epoch": 0.56, + "learning_rate": 2.4126498032600403e-06, + "logits/chosen": -2.2667434215545654, + "logits/rejected": -2.2319869995117188, + "logps/chosen": -316.8856201171875, + "logps/rejected": -350.779052734375, + "loss": 0.5954, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.17394258081912994, + "rewards/margins": 0.3134852349758148, + "rewards/rejected": -0.48742786049842834, + "step": 4280 + }, + { + "epoch": 0.56, + "learning_rate": 2.401237076844445e-06, + "logits/chosen": -2.2061569690704346, + "logits/rejected": -2.1190149784088135, + "logps/chosen": -285.7115173339844, + "logps/rejected": -270.5001525878906, + "loss": 0.6478, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.25898075103759766, + "rewards/margins": 0.17670731246471405, + "rewards/rejected": -0.4356880187988281, + "step": 4290 + }, + { + "epoch": 0.56, + "learning_rate": 2.38982641150898e-06, + "logits/chosen": -2.128105878829956, + "logits/rejected": -2.07566499710083, + "logps/chosen": -292.49517822265625, + "logps/rejected": -288.88555908203125, + "loss": 0.5645, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.18100666999816895, + "rewards/margins": 0.35543978214263916, + "rewards/rejected": -0.5364464521408081, + "step": 4300 + }, + { + "epoch": 0.56, + "learning_rate": 2.3784180453824414e-06, + "logits/chosen": -2.196662425994873, + "logits/rejected": -2.1289172172546387, + "logps/chosen": -286.9634094238281, + "logps/rejected": -309.788330078125, + "loss": 0.5923, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.18550142645835876, + "rewards/margins": 0.31308630108833313, + "rewards/rejected": -0.4985876977443695, + "step": 4310 + }, + { + "epoch": 0.57, + "learning_rate": 2.367012216545638e-06, + "logits/chosen": -2.244236469268799, + "logits/rejected": -2.163097620010376, + "logps/chosen": -284.93682861328125, + "logps/rejected": -281.2056884765625, + "loss": 0.6113, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2157663106918335, + "rewards/margins": 0.2603696882724762, + "rewards/rejected": -0.4761359691619873, + "step": 4320 + }, + { + "epoch": 0.57, + "learning_rate": 2.3556091630264294e-06, + "logits/chosen": -2.242006778717041, + "logits/rejected": -2.1157474517822266, + "logps/chosen": -292.1373596191406, + "logps/rejected": -280.36322021484375, + "loss": 0.5863, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.1108798012137413, + "rewards/margins": 0.3140495717525482, + "rewards/rejected": -0.4249293804168701, + "step": 4330 + }, + { + "epoch": 0.57, + "learning_rate": 2.344209122794757e-06, + "logits/chosen": -2.278151512145996, + "logits/rejected": -2.2355151176452637, + "logps/chosen": -307.44207763671875, + "logps/rejected": -342.2320556640625, + "loss": 0.5696, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.1577954888343811, + "rewards/margins": 0.3458147644996643, + "rewards/rejected": -0.5036102533340454, + "step": 4340 + }, + { + "epoch": 0.57, + "learning_rate": 2.3328123337576787e-06, + "logits/chosen": -2.1961028575897217, + "logits/rejected": -2.231940507888794, + "logps/chosen": -270.5798645019531, + "logps/rejected": -328.88958740234375, + "loss": 0.5705, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.14266864955425262, + "rewards/margins": 0.35884302854537964, + "rewards/rejected": -0.5015116930007935, + "step": 4350 + }, + { + "epoch": 0.57, + "learning_rate": 2.3214190337544017e-06, + "logits/chosen": -2.3711869716644287, + "logits/rejected": -2.2435050010681152, + "logps/chosen": -292.3756103515625, + "logps/rejected": -303.8966369628906, + "loss": 0.5751, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.16369611024856567, + "rewards/margins": 0.3327244818210602, + "rewards/rejected": -0.49642062187194824, + "step": 4360 + }, + { + "epoch": 0.57, + "learning_rate": 2.310029460551323e-06, + "logits/chosen": -2.2210006713867188, + "logits/rejected": -2.092313289642334, + "logps/chosen": -336.9834899902344, + "logps/rejected": -296.81842041015625, + "loss": 0.6183, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.1385176181793213, + "rewards/margins": 0.26691168546676636, + "rewards/rejected": -0.40542930364608765, + "step": 4370 + }, + { + "epoch": 0.57, + "learning_rate": 2.2986438518370645e-06, + "logits/chosen": -2.3118929862976074, + "logits/rejected": -2.172315835952759, + "logps/chosen": -293.66888427734375, + "logps/rejected": -308.866455078125, + "loss": 0.5572, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.20712819695472717, + "rewards/margins": 0.38452520966529846, + "rewards/rejected": -0.5916533470153809, + "step": 4380 + }, + { + "epoch": 0.57, + "learning_rate": 2.2872624452175123e-06, + "logits/chosen": -2.1337263584136963, + "logits/rejected": -2.096621036529541, + "logps/chosen": -317.05853271484375, + "logps/rejected": -306.6132507324219, + "loss": 0.5822, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.17859305441379547, + "rewards/margins": 0.31152716279029846, + "rewards/rejected": -0.4901202321052551, + "step": 4390 + }, + { + "epoch": 0.58, + "learning_rate": 2.2758854782108584e-06, + "logits/chosen": -2.2928977012634277, + "logits/rejected": -2.213534116744995, + "logps/chosen": -344.09814453125, + "logps/rejected": -357.140380859375, + "loss": 0.576, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.2843039333820343, + "rewards/margins": 0.3630077540874481, + "rewards/rejected": -0.6473116278648376, + "step": 4400 + }, + { + "epoch": 0.58, + "learning_rate": 2.2645131882426458e-06, + "logits/chosen": -2.24771785736084, + "logits/rejected": -2.217566728591919, + "logps/chosen": -296.27117919921875, + "logps/rejected": -318.70465087890625, + "loss": 0.6186, + "rewards/accuracies": 0.6000000238418579, + "rewards/chosen": -0.13453179597854614, + "rewards/margins": 0.254404753446579, + "rewards/rejected": -0.3889365792274475, + "step": 4410 + }, + { + "epoch": 0.58, + "learning_rate": 2.2531458126408154e-06, + "logits/chosen": -2.211513042449951, + "logits/rejected": -2.1245293617248535, + "logps/chosen": -298.334716796875, + "logps/rejected": -299.1788635253906, + "loss": 0.5585, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.22218851745128632, + "rewards/margins": 0.36935141682624817, + "rewards/rejected": -0.5915398597717285, + "step": 4420 + }, + { + "epoch": 0.58, + "learning_rate": 2.2417835886307452e-06, + "logits/chosen": -2.2878329753875732, + "logits/rejected": -2.191943645477295, + "logps/chosen": -295.5746154785156, + "logps/rejected": -303.7256774902344, + "loss": 0.5662, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.23160219192504883, + "rewards/margins": 0.3590404689311981, + "rewards/rejected": -0.5906426906585693, + "step": 4430 + }, + { + "epoch": 0.58, + "learning_rate": 2.2304267533303075e-06, + "logits/chosen": -2.198486089706421, + "logits/rejected": -2.1466586589813232, + "logps/chosen": -330.70770263671875, + "logps/rejected": -326.08502197265625, + "loss": 0.5688, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.1812635213136673, + "rewards/margins": 0.3645995259284973, + "rewards/rejected": -0.5458630323410034, + "step": 4440 + }, + { + "epoch": 0.58, + "learning_rate": 2.219075543744918e-06, + "logits/chosen": -2.313459634780884, + "logits/rejected": -2.2000479698181152, + "logps/chosen": -310.2533874511719, + "logps/rejected": -302.20257568359375, + "loss": 0.6248, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.2181706428527832, + "rewards/margins": 0.23994271457195282, + "rewards/rejected": -0.45811334252357483, + "step": 4450 + }, + { + "epoch": 0.58, + "learning_rate": 2.207730196762589e-06, + "logits/chosen": -2.1994261741638184, + "logits/rejected": -2.086202621459961, + "logps/chosen": -330.8420715332031, + "logps/rejected": -322.92034912109375, + "loss": 0.5502, + "rewards/accuracies": 0.7875000238418579, + "rewards/chosen": -0.2075098752975464, + "rewards/margins": 0.392425000667572, + "rewards/rejected": -0.5999349355697632, + "step": 4460 + }, + { + "epoch": 0.58, + "learning_rate": 2.1963909491489846e-06, + "logits/chosen": -2.270793914794922, + "logits/rejected": -2.2250256538391113, + "logps/chosen": -316.1448059082031, + "logps/rejected": -365.9324035644531, + "loss": 0.6155, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.19968464970588684, + "rewards/margins": 0.2766803205013275, + "rewards/rejected": -0.47636494040489197, + "step": 4470 + }, + { + "epoch": 0.59, + "learning_rate": 2.185058037542486e-06, + "logits/chosen": -2.3094911575317383, + "logits/rejected": -2.185676336288452, + "logps/chosen": -322.5010070800781, + "logps/rejected": -314.72601318359375, + "loss": 0.5884, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.19121722877025604, + "rewards/margins": 0.31124967336654663, + "rewards/rejected": -0.5024669170379639, + "step": 4480 + }, + { + "epoch": 0.59, + "learning_rate": 2.173731698449244e-06, + "logits/chosen": -2.2886509895324707, + "logits/rejected": -2.221942186355591, + "logps/chosen": -298.2712707519531, + "logps/rejected": -320.44842529296875, + "loss": 0.5943, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.17132337391376495, + "rewards/margins": 0.3021024167537689, + "rewards/rejected": -0.4734257757663727, + "step": 4490 + }, + { + "epoch": 0.59, + "learning_rate": 2.1624121682382495e-06, + "logits/chosen": -2.312828302383423, + "logits/rejected": -2.1678576469421387, + "logps/chosen": -290.67218017578125, + "logps/rejected": -297.09759521484375, + "loss": 0.5425, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.18776679039001465, + "rewards/margins": 0.4041364789009094, + "rewards/rejected": -0.5919032692909241, + "step": 4500 + }, + { + "epoch": 0.59, + "learning_rate": 2.1510996831363993e-06, + "logits/chosen": -2.2624051570892334, + "logits/rejected": -2.1440157890319824, + "logps/chosen": -303.673583984375, + "logps/rejected": -276.68798828125, + "loss": 0.6176, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.3465026021003723, + "rewards/margins": 0.2668396532535553, + "rewards/rejected": -0.61334228515625, + "step": 4510 + }, + { + "epoch": 0.59, + "learning_rate": 2.139794479223565e-06, + "logits/chosen": -2.177112579345703, + "logits/rejected": -2.1620678901672363, + "logps/chosen": -357.629638671875, + "logps/rejected": -324.1454162597656, + "loss": 0.5798, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.206990048289299, + "rewards/margins": 0.34105998277664185, + "rewards/rejected": -0.548050045967102, + "step": 4520 + }, + { + "epoch": 0.59, + "learning_rate": 2.128496792427669e-06, + "logits/chosen": -2.2875874042510986, + "logits/rejected": -2.273015260696411, + "logps/chosen": -324.6013488769531, + "logps/rejected": -359.5809326171875, + "loss": 0.5796, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.1470945179462433, + "rewards/margins": 0.33035987615585327, + "rewards/rejected": -0.47745442390441895, + "step": 4530 + }, + { + "epoch": 0.59, + "learning_rate": 2.117206858519758e-06, + "logits/chosen": -2.3104758262634277, + "logits/rejected": -2.3068227767944336, + "logps/chosen": -288.22686767578125, + "logps/rejected": -338.008056640625, + "loss": 0.616, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.23997744917869568, + "rewards/margins": 0.23832137882709503, + "rewards/rejected": -0.47829875349998474, + "step": 4540 + }, + { + "epoch": 0.6, + "learning_rate": 2.1059249131090844e-06, + "logits/chosen": -2.1803038120269775, + "logits/rejected": -2.1118412017822266, + "logps/chosen": -271.5584411621094, + "logps/rejected": -260.3912658691406, + "loss": 0.634, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.32383498549461365, + "rewards/margins": 0.23475320637226105, + "rewards/rejected": -0.5585881471633911, + "step": 4550 + }, + { + "epoch": 0.6, + "learning_rate": 2.094651191638189e-06, + "logits/chosen": -2.2164502143859863, + "logits/rejected": -2.1599996089935303, + "logps/chosen": -299.9985046386719, + "logps/rejected": -292.871826171875, + "loss": 0.6067, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.25585514307022095, + "rewards/margins": 0.26472902297973633, + "rewards/rejected": -0.520584225654602, + "step": 4560 + }, + { + "epoch": 0.6, + "learning_rate": 2.0833859293779867e-06, + "logits/chosen": -2.2270455360412598, + "logits/rejected": -2.142152786254883, + "logps/chosen": -311.3333435058594, + "logps/rejected": -325.640869140625, + "loss": 0.6001, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.21837690472602844, + "rewards/margins": 0.28140920400619507, + "rewards/rejected": -0.4997861385345459, + "step": 4570 + }, + { + "epoch": 0.6, + "learning_rate": 2.0721293614228568e-06, + "logits/chosen": -2.239487409591675, + "logits/rejected": -2.190485715866089, + "logps/chosen": -271.17694091796875, + "logps/rejected": -347.658447265625, + "loss": 0.5628, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.18228408694267273, + "rewards/margins": 0.365995854139328, + "rewards/rejected": -0.548279881477356, + "step": 4580 + }, + { + "epoch": 0.6, + "learning_rate": 2.060881722685742e-06, + "logits/chosen": -2.3012619018554688, + "logits/rejected": -2.3104777336120605, + "logps/chosen": -295.18853759765625, + "logps/rejected": -333.760986328125, + "loss": 0.5775, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.2056879997253418, + "rewards/margins": 0.32939228415489197, + "rewards/rejected": -0.5350803136825562, + "step": 4590 + }, + { + "epoch": 0.6, + "learning_rate": 2.049643247893235e-06, + "logits/chosen": -2.270509958267212, + "logits/rejected": -2.230644464492798, + "logps/chosen": -306.8868103027344, + "logps/rejected": -314.2762756347656, + "loss": 0.5663, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.2096973955631256, + "rewards/margins": 0.36910349130630493, + "rewards/rejected": -0.5788008570671082, + "step": 4600 + }, + { + "epoch": 0.6, + "learning_rate": 2.0384141715806903e-06, + "logits/chosen": -2.3155465126037598, + "logits/rejected": -2.258345365524292, + "logps/chosen": -343.16064453125, + "logps/rejected": -335.7472229003906, + "loss": 0.6168, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.23217563331127167, + "rewards/margins": 0.23074308037757874, + "rewards/rejected": -0.462918758392334, + "step": 4610 + }, + { + "epoch": 0.6, + "learning_rate": 2.0271947280873255e-06, + "logits/chosen": -2.323875904083252, + "logits/rejected": -2.2460594177246094, + "logps/chosen": -294.6390075683594, + "logps/rejected": -310.1608581542969, + "loss": 0.5651, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.13543495535850525, + "rewards/margins": 0.37415453791618347, + "rewards/rejected": -0.5095894932746887, + "step": 4620 + }, + { + "epoch": 0.61, + "learning_rate": 2.0159851515513302e-06, + "logits/chosen": -2.1863014698028564, + "logits/rejected": -2.179434299468994, + "logps/chosen": -251.57015991210938, + "logps/rejected": -300.84246826171875, + "loss": 0.6143, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.1544230431318283, + "rewards/margins": 0.2654711604118347, + "rewards/rejected": -0.4198942184448242, + "step": 4630 + }, + { + "epoch": 0.61, + "learning_rate": 2.004785675904982e-06, + "logits/chosen": -2.2885677814483643, + "logits/rejected": -2.2793614864349365, + "logps/chosen": -320.08502197265625, + "logps/rejected": -323.58355712890625, + "loss": 0.5914, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.14112447202205658, + "rewards/margins": 0.3043881356716156, + "rewards/rejected": -0.445512592792511, + "step": 4640 + }, + { + "epoch": 0.61, + "learning_rate": 1.9935965348697624e-06, + "logits/chosen": -2.28725266456604, + "logits/rejected": -2.18363094329834, + "logps/chosen": -309.8697814941406, + "logps/rejected": -277.9781799316406, + "loss": 0.5836, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.09367014467716217, + "rewards/margins": 0.32573920488357544, + "rewards/rejected": -0.4194093644618988, + "step": 4650 + }, + { + "epoch": 0.61, + "learning_rate": 1.9824179619514807e-06, + "logits/chosen": -2.2646570205688477, + "logits/rejected": -2.215155839920044, + "logps/chosen": -325.0639953613281, + "logps/rejected": -358.20281982421875, + "loss": 0.6222, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.236891508102417, + "rewards/margins": 0.2578909397125244, + "rewards/rejected": -0.4947824478149414, + "step": 4660 + }, + { + "epoch": 0.61, + "learning_rate": 1.9712501904354004e-06, + "logits/chosen": -2.2292869091033936, + "logits/rejected": -2.1636569499969482, + "logps/chosen": -292.38629150390625, + "logps/rejected": -309.7817687988281, + "loss": 0.5976, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.20154981315135956, + "rewards/margins": 0.2922808825969696, + "rewards/rejected": -0.49383068084716797, + "step": 4670 + }, + { + "epoch": 0.61, + "learning_rate": 1.960093453381369e-06, + "logits/chosen": -2.2603418827056885, + "logits/rejected": -2.2402851581573486, + "logps/chosen": -299.7186584472656, + "logps/rejected": -327.2197570800781, + "loss": 0.6625, + "rewards/accuracies": 0.512499988079071, + "rewards/chosen": -0.2904617488384247, + "rewards/margins": 0.16700880229473114, + "rewards/rejected": -0.45747050642967224, + "step": 4680 + }, + { + "epoch": 0.61, + "learning_rate": 1.948947983618962e-06, + "logits/chosen": -2.2831361293792725, + "logits/rejected": -2.186596155166626, + "logps/chosen": -331.02001953125, + "logps/rejected": -323.8455810546875, + "loss": 0.5772, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.2344851940870285, + "rewards/margins": 0.37284547090530396, + "rewards/rejected": -0.6073306798934937, + "step": 4690 + }, + { + "epoch": 0.62, + "learning_rate": 1.937814013742611e-06, + "logits/chosen": -2.192678928375244, + "logits/rejected": -2.1581597328186035, + "logps/chosen": -288.61614990234375, + "logps/rejected": -313.3251647949219, + "loss": 0.6163, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.24112264811992645, + "rewards/margins": 0.270085871219635, + "rewards/rejected": -0.5112084746360779, + "step": 4700 + }, + { + "epoch": 0.62, + "learning_rate": 1.9266917761067617e-06, + "logits/chosen": -2.252155065536499, + "logits/rejected": -2.150637149810791, + "logps/chosen": -274.75201416015625, + "logps/rejected": -309.49407958984375, + "loss": 0.5682, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.1783822923898697, + "rewards/margins": 0.39890679717063904, + "rewards/rejected": -0.5772891044616699, + "step": 4710 + }, + { + "epoch": 0.62, + "learning_rate": 1.915581502821017e-06, + "logits/chosen": -2.2457118034362793, + "logits/rejected": -2.202882766723633, + "logps/chosen": -311.6221618652344, + "logps/rejected": -326.28167724609375, + "loss": 0.6551, + "rewards/accuracies": 0.5874999761581421, + "rewards/chosen": -0.29532963037490845, + "rewards/margins": 0.16430024802684784, + "rewards/rejected": -0.4596298336982727, + "step": 4720 + }, + { + "epoch": 0.62, + "learning_rate": 1.9044834257452997e-06, + "logits/chosen": -2.262570381164551, + "logits/rejected": -2.277740001678467, + "logps/chosen": -322.4426574707031, + "logps/rejected": -324.764892578125, + "loss": 0.5724, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.19570091366767883, + "rewards/margins": 0.3290325105190277, + "rewards/rejected": -0.5247333645820618, + "step": 4730 + }, + { + "epoch": 0.62, + "learning_rate": 1.893397776485006e-06, + "logits/chosen": -2.195594072341919, + "logits/rejected": -2.083705425262451, + "logps/chosen": -264.0104064941406, + "logps/rejected": -286.9299621582031, + "loss": 0.5795, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.24587281048297882, + "rewards/margins": 0.34859025478363037, + "rewards/rejected": -0.594463050365448, + "step": 4740 + }, + { + "epoch": 0.62, + "learning_rate": 1.8823247863861804e-06, + "logits/chosen": -2.1810431480407715, + "logits/rejected": -2.1117537021636963, + "logps/chosen": -292.3841857910156, + "logps/rejected": -286.7529296875, + "loss": 0.5537, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.17687204480171204, + "rewards/margins": 0.40910300612449646, + "rewards/rejected": -0.5859750509262085, + "step": 4750 + }, + { + "epoch": 0.62, + "learning_rate": 1.8712646865306822e-06, + "logits/chosen": -2.2291183471679688, + "logits/rejected": -2.2378058433532715, + "logps/chosen": -306.16046142578125, + "logps/rejected": -347.13543701171875, + "loss": 0.5795, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.21971885859966278, + "rewards/margins": 0.3429061472415924, + "rewards/rejected": -0.562624990940094, + "step": 4760 + }, + { + "epoch": 0.62, + "learning_rate": 1.8602177077313631e-06, + "logits/chosen": -2.245903253555298, + "logits/rejected": -2.2234671115875244, + "logps/chosen": -287.87542724609375, + "logps/rejected": -343.1726989746094, + "loss": 0.597, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.1752200871706009, + "rewards/margins": 0.31497079133987427, + "rewards/rejected": -0.4901908040046692, + "step": 4770 + }, + { + "epoch": 0.63, + "learning_rate": 1.8491840805272546e-06, + "logits/chosen": -2.2528433799743652, + "logits/rejected": -2.20115327835083, + "logps/chosen": -285.2230529785156, + "logps/rejected": -294.3558044433594, + "loss": 0.5773, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.2560577988624573, + "rewards/margins": 0.3361159861087799, + "rewards/rejected": -0.5921737551689148, + "step": 4780 + }, + { + "epoch": 0.63, + "learning_rate": 1.8381640351787516e-06, + "logits/chosen": -2.1432268619537354, + "logits/rejected": -2.0488507747650146, + "logps/chosen": -305.54290771484375, + "logps/rejected": -316.4336242675781, + "loss": 0.5747, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.24591200053691864, + "rewards/margins": 0.3427841067314148, + "rewards/rejected": -0.5886961221694946, + "step": 4790 + }, + { + "epoch": 0.63, + "learning_rate": 1.8271578016628122e-06, + "logits/chosen": -2.3278307914733887, + "logits/rejected": -2.226720094680786, + "logps/chosen": -320.807373046875, + "logps/rejected": -316.5043640136719, + "loss": 0.6094, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.24610686302185059, + "rewards/margins": 0.2771647572517395, + "rewards/rejected": -0.5232715606689453, + "step": 4800 + }, + { + "epoch": 0.63, + "learning_rate": 1.8161656096681546e-06, + "logits/chosen": -2.283215284347534, + "logits/rejected": -2.1816093921661377, + "logps/chosen": -301.87738037109375, + "logps/rejected": -300.78631591796875, + "loss": 0.5584, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.18695667386054993, + "rewards/margins": 0.4141852855682373, + "rewards/rejected": -0.6011419296264648, + "step": 4810 + }, + { + "epoch": 0.63, + "learning_rate": 1.8051876885904645e-06, + "logits/chosen": -2.160775661468506, + "logits/rejected": -2.090919017791748, + "logps/chosen": -263.7298889160156, + "logps/rejected": -277.83697509765625, + "loss": 0.6028, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2347753793001175, + "rewards/margins": 0.30403420329093933, + "rewards/rejected": -0.538809597492218, + "step": 4820 + }, + { + "epoch": 0.63, + "learning_rate": 1.7942242675276098e-06, + "logits/chosen": -2.222561836242676, + "logits/rejected": -2.1378979682922363, + "logps/chosen": -311.54071044921875, + "logps/rejected": -351.65911865234375, + "loss": 0.5569, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.2278716266155243, + "rewards/margins": 0.4089781641960144, + "rewards/rejected": -0.6368497610092163, + "step": 4830 + }, + { + "epoch": 0.63, + "learning_rate": 1.783275575274856e-06, + "logits/chosen": -2.077249050140381, + "logits/rejected": -2.02012300491333, + "logps/chosen": -311.983154296875, + "logps/rejected": -328.8287658691406, + "loss": 0.541, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.2819608747959137, + "rewards/margins": 0.4067571759223938, + "rewards/rejected": -0.6887180209159851, + "step": 4840 + }, + { + "epoch": 0.63, + "learning_rate": 1.7723418403200943e-06, + "logits/chosen": -2.2702884674072266, + "logits/rejected": -2.1757566928863525, + "logps/chosen": -306.5168762207031, + "logps/rejected": -339.5025939941406, + "loss": 0.5797, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.22507329285144806, + "rewards/margins": 0.38953521847724915, + "rewards/rejected": -0.6146085262298584, + "step": 4850 + }, + { + "epoch": 0.64, + "learning_rate": 1.7614232908390748e-06, + "logits/chosen": -2.199558973312378, + "logits/rejected": -2.0967612266540527, + "logps/chosen": -322.54193115234375, + "logps/rejected": -318.96319580078125, + "loss": 0.5785, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.31979161500930786, + "rewards/margins": 0.35457512736320496, + "rewards/rejected": -0.6743667721748352, + "step": 4860 + }, + { + "epoch": 0.64, + "learning_rate": 1.7505201546906398e-06, + "logits/chosen": -2.3104164600372314, + "logits/rejected": -2.1747078895568848, + "logps/chosen": -321.0475158691406, + "logps/rejected": -316.3406677246094, + "loss": 0.5938, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.24703386425971985, + "rewards/margins": 0.32056108117103577, + "rewards/rejected": -0.5675949454307556, + "step": 4870 + }, + { + "epoch": 0.64, + "learning_rate": 1.7396326594119717e-06, + "logits/chosen": -2.1867446899414062, + "logits/rejected": -2.2044179439544678, + "logps/chosen": -318.033447265625, + "logps/rejected": -376.8894958496094, + "loss": 0.549, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.21440622210502625, + "rewards/margins": 0.4122908115386963, + "rewards/rejected": -0.6266970038414001, + "step": 4880 + }, + { + "epoch": 0.64, + "learning_rate": 1.7287610322138449e-06, + "logits/chosen": -2.2811503410339355, + "logits/rejected": -2.2072041034698486, + "logps/chosen": -287.03448486328125, + "logps/rejected": -296.5589904785156, + "loss": 0.5992, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.19398990273475647, + "rewards/margins": 0.2939120829105377, + "rewards/rejected": -0.4879019856452942, + "step": 4890 + }, + { + "epoch": 0.64, + "learning_rate": 1.7179054999758817e-06, + "logits/chosen": -2.2527401447296143, + "logits/rejected": -2.2034683227539062, + "logps/chosen": -322.36529541015625, + "logps/rejected": -313.5167541503906, + "loss": 0.604, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.22369468212127686, + "rewards/margins": 0.2823079228401184, + "rewards/rejected": -0.50600266456604, + "step": 4900 + }, + { + "epoch": 0.64, + "learning_rate": 1.7070662892418225e-06, + "logits/chosen": -2.2799415588378906, + "logits/rejected": -2.2384002208709717, + "logps/chosen": -292.792236328125, + "logps/rejected": -323.26556396484375, + "loss": 0.5765, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.17639829218387604, + "rewards/margins": 0.34047171473503113, + "rewards/rejected": -0.5168699622154236, + "step": 4910 + }, + { + "epoch": 0.64, + "learning_rate": 1.6962436262147913e-06, + "logits/chosen": -2.3212456703186035, + "logits/rejected": -2.231661319732666, + "logps/chosen": -346.7903137207031, + "logps/rejected": -369.8257751464844, + "loss": 0.6044, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.3193613588809967, + "rewards/margins": 0.2776809334754944, + "rewards/rejected": -0.5970422625541687, + "step": 4920 + }, + { + "epoch": 0.65, + "learning_rate": 1.6854377367525814e-06, + "logits/chosen": -2.229032516479492, + "logits/rejected": -2.1569199562072754, + "logps/chosen": -288.3301086425781, + "logps/rejected": -323.98944091796875, + "loss": 0.5499, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.1581147015094757, + "rewards/margins": 0.41132253408432007, + "rewards/rejected": -0.5694373250007629, + "step": 4930 + }, + { + "epoch": 0.65, + "learning_rate": 1.6746488463629362e-06, + "logits/chosen": -2.251732349395752, + "logits/rejected": -2.1786773204803467, + "logps/chosen": -331.1020812988281, + "logps/rejected": -317.1491394042969, + "loss": 0.5869, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.12536010146141052, + "rewards/margins": 0.3522440493106842, + "rewards/rejected": -0.47760409116744995, + "step": 4940 + }, + { + "epoch": 0.65, + "learning_rate": 1.6638771801988483e-06, + "logits/chosen": -2.2855029106140137, + "logits/rejected": -2.182276964187622, + "logps/chosen": -339.62530517578125, + "logps/rejected": -338.089111328125, + "loss": 0.6111, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.18541690707206726, + "rewards/margins": 0.2663753926753998, + "rewards/rejected": -0.45179229974746704, + "step": 4950 + }, + { + "epoch": 0.65, + "learning_rate": 1.653122963053857e-06, + "logits/chosen": -2.2749080657958984, + "logits/rejected": -2.250270366668701, + "logps/chosen": -281.3585510253906, + "logps/rejected": -291.1257629394531, + "loss": 0.5941, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.13789398968219757, + "rewards/margins": 0.2909575402736664, + "rewards/rejected": -0.42885151505470276, + "step": 4960 + }, + { + "epoch": 0.65, + "learning_rate": 1.6423864193573606e-06, + "logits/chosen": -2.2902145385742188, + "logits/rejected": -2.168369770050049, + "logps/chosen": -302.56756591796875, + "logps/rejected": -324.24639892578125, + "loss": 0.6049, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.179901584982872, + "rewards/margins": 0.27662497758865356, + "rewards/rejected": -0.45652657747268677, + "step": 4970 + }, + { + "epoch": 0.65, + "learning_rate": 1.6316677731699286e-06, + "logits/chosen": -2.2854247093200684, + "logits/rejected": -2.1676154136657715, + "logps/chosen": -315.88287353515625, + "logps/rejected": -309.0321350097656, + "loss": 0.5703, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.16615870594978333, + "rewards/margins": 0.34920066595077515, + "rewards/rejected": -0.5153594017028809, + "step": 4980 + }, + { + "epoch": 0.65, + "learning_rate": 1.6209672481786302e-06, + "logits/chosen": -2.227534770965576, + "logits/rejected": -2.1882176399230957, + "logps/chosen": -306.0838317871094, + "logps/rejected": -336.73150634765625, + "loss": 0.6091, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.2127208709716797, + "rewards/margins": 0.2690984606742859, + "rewards/rejected": -0.4818193316459656, + "step": 4990 + }, + { + "epoch": 0.65, + "learning_rate": 1.6102850676923616e-06, + "logits/chosen": -2.1546339988708496, + "logits/rejected": -2.2065951824188232, + "logps/chosen": -233.6758575439453, + "logps/rejected": -282.8135986328125, + "loss": 0.5674, + "rewards/accuracies": 0.7875000238418579, + "rewards/chosen": -0.12460102885961533, + "rewards/margins": 0.33533772826194763, + "rewards/rejected": -0.4599387049674988, + "step": 5000 + }, + { + "epoch": 0.66, + "learning_rate": 1.5996214546371888e-06, + "logits/chosen": -2.2153162956237793, + "logits/rejected": -2.1397464275360107, + "logps/chosen": -300.18560791015625, + "logps/rejected": -345.79058837890625, + "loss": 0.5581, + "rewards/accuracies": 0.8374999761581421, + "rewards/chosen": -0.09821708500385284, + "rewards/margins": 0.3663147985935211, + "rewards/rejected": -0.46453189849853516, + "step": 5010 + }, + { + "epoch": 0.66, + "learning_rate": 1.588976631551697e-06, + "logits/chosen": -2.229888439178467, + "logits/rejected": -2.027146339416504, + "logps/chosen": -322.75372314453125, + "logps/rejected": -313.32061767578125, + "loss": 0.5698, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.17541874945163727, + "rewards/margins": 0.34946078062057495, + "rewards/rejected": -0.5248795747756958, + "step": 5020 + }, + { + "epoch": 0.66, + "learning_rate": 1.5783508205823412e-06, + "logits/chosen": -2.2798500061035156, + "logits/rejected": -2.1787333488464355, + "logps/chosen": -289.74658203125, + "logps/rejected": -309.55584716796875, + "loss": 0.5514, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.11370769888162613, + "rewards/margins": 0.42184892296791077, + "rewards/rejected": -0.5355566143989563, + "step": 5030 + }, + { + "epoch": 0.66, + "learning_rate": 1.5677442434788143e-06, + "logits/chosen": -2.241020679473877, + "logits/rejected": -2.1790409088134766, + "logps/chosen": -347.45867919921875, + "logps/rejected": -309.4788513183594, + "loss": 0.6261, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.2387397736310959, + "rewards/margins": 0.20377786457538605, + "rewards/rejected": -0.4425176680088043, + "step": 5040 + }, + { + "epoch": 0.66, + "learning_rate": 1.5571571215894181e-06, + "logits/chosen": -2.144336223602295, + "logits/rejected": -2.112370491027832, + "logps/chosen": -271.4165954589844, + "logps/rejected": -289.98236083984375, + "loss": 0.57, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.16046592593193054, + "rewards/margins": 0.3745737373828888, + "rewards/rejected": -0.5350396633148193, + "step": 5050 + }, + { + "epoch": 0.66, + "learning_rate": 1.5465896758564452e-06, + "logits/chosen": -2.269864559173584, + "logits/rejected": -2.20538592338562, + "logps/chosen": -308.1907958984375, + "logps/rejected": -300.5450439453125, + "loss": 0.6381, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.1570548266172409, + "rewards/margins": 0.22571484744548798, + "rewards/rejected": -0.3827696442604065, + "step": 5060 + }, + { + "epoch": 0.66, + "learning_rate": 1.5360421268115653e-06, + "logits/chosen": -2.3499755859375, + "logits/rejected": -2.2601470947265625, + "logps/chosen": -296.19677734375, + "logps/rejected": -309.92144775390625, + "loss": 0.6088, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.1885501593351364, + "rewards/margins": 0.2522444427013397, + "rewards/rejected": -0.4407946467399597, + "step": 5070 + }, + { + "epoch": 0.66, + "learning_rate": 1.5255146945712267e-06, + "logits/chosen": -2.318502902984619, + "logits/rejected": -2.317073345184326, + "logps/chosen": -264.9830017089844, + "logps/rejected": -285.3392639160156, + "loss": 0.5724, + "rewards/accuracies": 0.7875000238418579, + "rewards/chosen": -0.04835155978798866, + "rewards/margins": 0.33312511444091797, + "rewards/rejected": -0.38147664070129395, + "step": 5080 + }, + { + "epoch": 0.67, + "learning_rate": 1.5150075988320594e-06, + "logits/chosen": -2.3264520168304443, + "logits/rejected": -2.158167600631714, + "logps/chosen": -272.5290222167969, + "logps/rejected": -264.8362731933594, + "loss": 0.5598, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.13282960653305054, + "rewards/margins": 0.40845727920532227, + "rewards/rejected": -0.5412868857383728, + "step": 5090 + }, + { + "epoch": 0.67, + "learning_rate": 1.5045210588662929e-06, + "logits/chosen": -2.243901491165161, + "logits/rejected": -2.2242465019226074, + "logps/chosen": -333.20440673828125, + "logps/rejected": -331.9303894042969, + "loss": 0.6221, + "rewards/accuracies": 0.6000000238418579, + "rewards/chosen": -0.2723377048969269, + "rewards/margins": 0.2254316508769989, + "rewards/rejected": -0.497769296169281, + "step": 5100 + }, + { + "epoch": 0.67, + "learning_rate": 1.4940552935171781e-06, + "logits/chosen": -2.185523509979248, + "logits/rejected": -2.1053202152252197, + "logps/chosen": -306.14935302734375, + "logps/rejected": -293.40130615234375, + "loss": 0.6166, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.19349311292171478, + "rewards/margins": 0.28787046670913696, + "rewards/rejected": -0.48136359453201294, + "step": 5110 + }, + { + "epoch": 0.67, + "learning_rate": 1.483610521194419e-06, + "logits/chosen": -2.1743483543395996, + "logits/rejected": -2.123823642730713, + "logps/chosen": -253.65994262695312, + "logps/rejected": -274.9971618652344, + "loss": 0.5659, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.13903601467609406, + "rewards/margins": 0.3621757924556732, + "rewards/rejected": -0.5012118220329285, + "step": 5120 + }, + { + "epoch": 0.67, + "learning_rate": 1.4731869598696226e-06, + "logits/chosen": -2.3371212482452393, + "logits/rejected": -2.2577626705169678, + "logps/chosen": -348.0635681152344, + "logps/rejected": -335.85894775390625, + "loss": 0.5872, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.15545479953289032, + "rewards/margins": 0.3256588876247406, + "rewards/rejected": -0.4811136722564697, + "step": 5130 + }, + { + "epoch": 0.67, + "learning_rate": 1.4627848270717387e-06, + "logits/chosen": -2.359225034713745, + "logits/rejected": -2.271225929260254, + "logps/chosen": -300.64215087890625, + "logps/rejected": -313.2225341796875, + "loss": 0.5655, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.1491098403930664, + "rewards/margins": 0.3669964671134949, + "rewards/rejected": -0.5161063075065613, + "step": 5140 + }, + { + "epoch": 0.67, + "learning_rate": 1.4524043398825277e-06, + "logits/chosen": -2.3282556533813477, + "logits/rejected": -2.105178117752075, + "logps/chosen": -325.2800598144531, + "logps/rejected": -251.1908416748047, + "loss": 0.5857, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.13584375381469727, + "rewards/margins": 0.3080797791481018, + "rewards/rejected": -0.44392356276512146, + "step": 5150 + }, + { + "epoch": 0.68, + "learning_rate": 1.4420457149320299e-06, + "logits/chosen": -2.3180902004241943, + "logits/rejected": -2.1738173961639404, + "logps/chosen": -325.9745178222656, + "logps/rejected": -324.75347900390625, + "loss": 0.5682, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.14252132177352905, + "rewards/margins": 0.35515937209129333, + "rewards/rejected": -0.49768075346946716, + "step": 5160 + }, + { + "epoch": 0.68, + "learning_rate": 1.431709168394042e-06, + "logits/chosen": -2.2476959228515625, + "logits/rejected": -2.187673568725586, + "logps/chosen": -276.9595031738281, + "logps/rejected": -304.9107360839844, + "loss": 0.6158, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.18238958716392517, + "rewards/margins": 0.2684926986694336, + "rewards/rejected": -0.45088228583335876, + "step": 5170 + }, + { + "epoch": 0.68, + "learning_rate": 1.4213949159816059e-06, + "logits/chosen": -2.0993666648864746, + "logits/rejected": -2.034672498703003, + "logps/chosen": -291.9742736816406, + "logps/rejected": -333.3852233886719, + "loss": 0.6012, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.2016802579164505, + "rewards/margins": 0.3297204375267029, + "rewards/rejected": -0.531400740146637, + "step": 5180 + }, + { + "epoch": 0.68, + "learning_rate": 1.4111031729425103e-06, + "logits/chosen": -2.280778169631958, + "logits/rejected": -2.2158453464508057, + "logps/chosen": -312.0046081542969, + "logps/rejected": -329.3743591308594, + "loss": 0.5698, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.11512058973312378, + "rewards/margins": 0.3418727219104767, + "rewards/rejected": -0.45699334144592285, + "step": 5190 + }, + { + "epoch": 0.68, + "learning_rate": 1.4008341540547965e-06, + "logits/chosen": -2.2735283374786377, + "logits/rejected": -2.2086353302001953, + "logps/chosen": -275.8941650390625, + "logps/rejected": -257.28326416015625, + "loss": 0.6039, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.2713925838470459, + "rewards/margins": 0.27395159006118774, + "rewards/rejected": -0.5453441143035889, + "step": 5200 + }, + { + "epoch": 0.68, + "learning_rate": 1.3905880736222737e-06, + "logits/chosen": -2.2202446460723877, + "logits/rejected": -2.152207851409912, + "logps/chosen": -274.55718994140625, + "logps/rejected": -284.71246337890625, + "loss": 0.5776, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2355513572692871, + "rewards/margins": 0.34016019105911255, + "rewards/rejected": -0.5757116079330444, + "step": 5210 + }, + { + "epoch": 0.68, + "learning_rate": 1.3803651454700531e-06, + "logits/chosen": -2.291025400161743, + "logits/rejected": -2.220370292663574, + "logps/chosen": -323.9850769042969, + "logps/rejected": -318.9848327636719, + "loss": 0.5719, + "rewards/accuracies": 0.7875000238418579, + "rewards/chosen": -0.15749236941337585, + "rewards/margins": 0.3263644278049469, + "rewards/rejected": -0.48385676741600037, + "step": 5220 + }, + { + "epoch": 0.68, + "learning_rate": 1.3701655829400773e-06, + "logits/chosen": -2.2412068843841553, + "logits/rejected": -2.1757941246032715, + "logps/chosen": -282.99298095703125, + "logps/rejected": -296.9221496582031, + "loss": 0.614, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.17335820198059082, + "rewards/margins": 0.25958600640296936, + "rewards/rejected": -0.4329442083835602, + "step": 5230 + }, + { + "epoch": 0.69, + "learning_rate": 1.3599895988866756e-06, + "logits/chosen": -2.22521710395813, + "logits/rejected": -2.205357074737549, + "logps/chosen": -273.3722229003906, + "logps/rejected": -316.802001953125, + "loss": 0.5615, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.22788643836975098, + "rewards/margins": 0.37798845767974854, + "rewards/rejected": -0.6058748960494995, + "step": 5240 + }, + { + "epoch": 0.69, + "learning_rate": 1.3498374056721198e-06, + "logits/chosen": -2.236832857131958, + "logits/rejected": -2.1135387420654297, + "logps/chosen": -347.28631591796875, + "logps/rejected": -304.4608459472656, + "loss": 0.6088, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.18296435475349426, + "rewards/margins": 0.2971632182598114, + "rewards/rejected": -0.48012763261795044, + "step": 5250 + }, + { + "epoch": 0.69, + "learning_rate": 1.3397092151621883e-06, + "logits/chosen": -2.181856632232666, + "logits/rejected": -2.118957042694092, + "logps/chosen": -314.9497985839844, + "logps/rejected": -348.48602294921875, + "loss": 0.6031, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.3234806954860687, + "rewards/margins": 0.2929115891456604, + "rewards/rejected": -0.6163923740386963, + "step": 5260 + }, + { + "epoch": 0.69, + "learning_rate": 1.3296052387217484e-06, + "logits/chosen": -2.3140857219696045, + "logits/rejected": -2.269150495529175, + "logps/chosen": -335.5849609375, + "logps/rejected": -337.6166687011719, + "loss": 0.5836, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.21779537200927734, + "rewards/margins": 0.3222483992576599, + "rewards/rejected": -0.5400437116622925, + "step": 5270 + }, + { + "epoch": 0.69, + "learning_rate": 1.3195256872103476e-06, + "logits/chosen": -2.371598720550537, + "logits/rejected": -2.2018089294433594, + "logps/chosen": -370.02215576171875, + "logps/rejected": -298.6385498046875, + "loss": 0.5612, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.17238309979438782, + "rewards/margins": 0.3724508583545685, + "rewards/rejected": -0.5448338985443115, + "step": 5280 + }, + { + "epoch": 0.69, + "learning_rate": 1.3094707709778068e-06, + "logits/chosen": -2.275477886199951, + "logits/rejected": -2.2253832817077637, + "logps/chosen": -335.6798400878906, + "logps/rejected": -341.5209655761719, + "loss": 0.6059, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.2835708260536194, + "rewards/margins": 0.2779124677181244, + "rewards/rejected": -0.5614833235740662, + "step": 5290 + }, + { + "epoch": 0.69, + "learning_rate": 1.2994406998598364e-06, + "logits/chosen": -2.320047378540039, + "logits/rejected": -2.2582650184631348, + "logps/chosen": -311.9765930175781, + "logps/rejected": -303.0313415527344, + "loss": 0.5961, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.1751178801059723, + "rewards/margins": 0.2839520573616028, + "rewards/rejected": -0.45906996726989746, + "step": 5300 + }, + { + "epoch": 0.69, + "learning_rate": 1.2894356831736558e-06, + "logits/chosen": -2.330653429031372, + "logits/rejected": -2.2683331966400146, + "logps/chosen": -347.83404541015625, + "logps/rejected": -331.4981994628906, + "loss": 0.6062, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.17227540910243988, + "rewards/margins": 0.2967098355293274, + "rewards/rejected": -0.4689852297306061, + "step": 5310 + }, + { + "epoch": 0.7, + "learning_rate": 1.2794559297136203e-06, + "logits/chosen": -2.1872544288635254, + "logits/rejected": -2.121631622314453, + "logps/chosen": -332.4129333496094, + "logps/rejected": -350.60662841796875, + "loss": 0.6709, + "rewards/accuracies": 0.5249999761581421, + "rewards/chosen": -0.31739237904548645, + "rewards/margins": 0.20196351408958435, + "rewards/rejected": -0.5193558931350708, + "step": 5320 + }, + { + "epoch": 0.7, + "learning_rate": 1.2695016477468724e-06, + "logits/chosen": -2.210880994796753, + "logits/rejected": -2.2127017974853516, + "logps/chosen": -269.5990295410156, + "logps/rejected": -329.0955810546875, + "loss": 0.541, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.14754262566566467, + "rewards/margins": 0.46367454528808594, + "rewards/rejected": -0.6112171411514282, + "step": 5330 + }, + { + "epoch": 0.7, + "learning_rate": 1.2595730450089874e-06, + "logits/chosen": -2.3433351516723633, + "logits/rejected": -2.269415855407715, + "logps/chosen": -343.03265380859375, + "logps/rejected": -328.55035400390625, + "loss": 0.538, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.07410895079374313, + "rewards/margins": 0.4277653694152832, + "rewards/rejected": -0.5018743276596069, + "step": 5340 + }, + { + "epoch": 0.7, + "learning_rate": 1.2496703286996433e-06, + "logits/chosen": -2.379992961883545, + "logits/rejected": -2.249950408935547, + "logps/chosen": -323.057373046875, + "logps/rejected": -333.5340270996094, + "loss": 0.6141, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.17264093458652496, + "rewards/margins": 0.26445716619491577, + "rewards/rejected": -0.43709808588027954, + "step": 5350 + }, + { + "epoch": 0.7, + "learning_rate": 1.2397937054782961e-06, + "logits/chosen": -2.3674094676971436, + "logits/rejected": -2.261693000793457, + "logps/chosen": -296.53131103515625, + "logps/rejected": -300.84710693359375, + "loss": 0.6406, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.2015656679868698, + "rewards/margins": 0.22162330150604248, + "rewards/rejected": -0.4231889843940735, + "step": 5360 + }, + { + "epoch": 0.7, + "learning_rate": 1.2299433814598635e-06, + "logits/chosen": -2.311593770980835, + "logits/rejected": -2.267444610595703, + "logps/chosen": -339.8316955566406, + "logps/rejected": -333.5635681152344, + "loss": 0.6398, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2215697318315506, + "rewards/margins": 0.17241211235523224, + "rewards/rejected": -0.39398181438446045, + "step": 5370 + }, + { + "epoch": 0.7, + "learning_rate": 1.2201195622104265e-06, + "logits/chosen": -2.2464287281036377, + "logits/rejected": -2.13893723487854, + "logps/chosen": -310.2950744628906, + "logps/rejected": -321.4704895019531, + "loss": 0.5879, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.20835845172405243, + "rewards/margins": 0.3142760992050171, + "rewards/rejected": -0.5226346254348755, + "step": 5380 + }, + { + "epoch": 0.71, + "learning_rate": 1.2103224527429417e-06, + "logits/chosen": -2.287074565887451, + "logits/rejected": -2.287384510040283, + "logps/chosen": -328.0266418457031, + "logps/rejected": -336.390380859375, + "loss": 0.6092, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.18557676672935486, + "rewards/margins": 0.281202107667923, + "rewards/rejected": -0.46677881479263306, + "step": 5390 + }, + { + "epoch": 0.71, + "learning_rate": 1.2005522575129559e-06, + "logits/chosen": -2.3571369647979736, + "logits/rejected": -2.2702527046203613, + "logps/chosen": -328.0036926269531, + "logps/rejected": -348.81451416015625, + "loss": 0.5618, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.12667694687843323, + "rewards/margins": 0.3967793583869934, + "rewards/rejected": -0.5234562754631042, + "step": 5400 + }, + { + "epoch": 0.71, + "learning_rate": 1.1908091804143469e-06, + "logits/chosen": -2.169992208480835, + "logits/rejected": -2.131552219390869, + "logps/chosen": -289.8910827636719, + "logps/rejected": -305.1105651855469, + "loss": 0.5734, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.14728592336177826, + "rewards/margins": 0.3489544093608856, + "rewards/rejected": -0.4962403178215027, + "step": 5410 + }, + { + "epoch": 0.71, + "learning_rate": 1.1810934247750649e-06, + "logits/chosen": -2.2190515995025635, + "logits/rejected": -2.098783016204834, + "logps/chosen": -337.6979675292969, + "logps/rejected": -306.4912109375, + "loss": 0.6128, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.20069074630737305, + "rewards/margins": 0.27513587474823, + "rewards/rejected": -0.47582656145095825, + "step": 5420 + }, + { + "epoch": 0.71, + "learning_rate": 1.1714051933528881e-06, + "logits/chosen": -2.259046792984009, + "logits/rejected": -2.174419403076172, + "logps/chosen": -307.9107971191406, + "logps/rejected": -321.9892578125, + "loss": 0.5619, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.17964601516723633, + "rewards/margins": 0.44221797585487366, + "rewards/rejected": -0.6218639612197876, + "step": 5430 + }, + { + "epoch": 0.71, + "learning_rate": 1.161744688331192e-06, + "logits/chosen": -2.2144017219543457, + "logits/rejected": -2.104426383972168, + "logps/chosen": -315.6021423339844, + "logps/rejected": -317.8662109375, + "loss": 0.5714, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.1677999347448349, + "rewards/margins": 0.3393593430519104, + "rewards/rejected": -0.5071592926979065, + "step": 5440 + }, + { + "epoch": 0.71, + "learning_rate": 1.152112111314733e-06, + "logits/chosen": -2.2710747718811035, + "logits/rejected": -2.3048489093780518, + "logps/chosen": -310.8573913574219, + "logps/rejected": -316.53350830078125, + "loss": 0.6252, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.22738580405712128, + "rewards/margins": 0.23117342591285706, + "rewards/rejected": -0.45855921506881714, + "step": 5450 + }, + { + "epoch": 0.71, + "learning_rate": 1.142507663325439e-06, + "logits/chosen": -2.2273240089416504, + "logits/rejected": -2.180992603302002, + "logps/chosen": -311.2198791503906, + "logps/rejected": -304.9374084472656, + "loss": 0.5482, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.12583717703819275, + "rewards/margins": 0.4427891671657562, + "rewards/rejected": -0.5686262845993042, + "step": 5460 + }, + { + "epoch": 0.72, + "learning_rate": 1.132931544798211e-06, + "logits/chosen": -2.2862751483917236, + "logits/rejected": -2.169010639190674, + "logps/chosen": -320.9517517089844, + "logps/rejected": -279.1144104003906, + "loss": 0.5979, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2036980837583542, + "rewards/margins": 0.29568424820899963, + "rewards/rejected": -0.499382346868515, + "step": 5470 + }, + { + "epoch": 0.72, + "learning_rate": 1.1233839555767482e-06, + "logits/chosen": -2.2294743061065674, + "logits/rejected": -2.21972918510437, + "logps/chosen": -264.0962829589844, + "logps/rejected": -317.7840270996094, + "loss": 0.5615, + "rewards/accuracies": 0.7875000238418579, + "rewards/chosen": -0.1876416802406311, + "rewards/margins": 0.3944639563560486, + "rewards/rejected": -0.5821055769920349, + "step": 5480 + }, + { + "epoch": 0.72, + "learning_rate": 1.1138650949093668e-06, + "logits/chosen": -2.2792954444885254, + "logits/rejected": -2.237781047821045, + "logps/chosen": -279.8136291503906, + "logps/rejected": -312.3643798828125, + "loss": 0.6221, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.22569458186626434, + "rewards/margins": 0.24749894440174103, + "rewards/rejected": -0.47319355607032776, + "step": 5490 + }, + { + "epoch": 0.72, + "learning_rate": 1.1043751614448543e-06, + "logits/chosen": -2.191702365875244, + "logits/rejected": -2.1757311820983887, + "logps/chosen": -306.8429260253906, + "logps/rejected": -328.25262451171875, + "loss": 0.6302, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.2591175138950348, + "rewards/margins": 0.23401156067848206, + "rewards/rejected": -0.49312910437583923, + "step": 5500 + }, + { + "epoch": 0.72, + "learning_rate": 1.0949143532283107e-06, + "logits/chosen": -2.2734391689300537, + "logits/rejected": -2.1344447135925293, + "logps/chosen": -287.9913635253906, + "logps/rejected": -288.32305908203125, + "loss": 0.5483, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.16467280685901642, + "rewards/margins": 0.41381826996803284, + "rewards/rejected": -0.5784910917282104, + "step": 5510 + }, + { + "epoch": 0.72, + "learning_rate": 1.0854828676970275e-06, + "logits/chosen": -2.1989667415618896, + "logits/rejected": -2.140624523162842, + "logps/chosen": -338.8399658203125, + "logps/rejected": -291.35333251953125, + "loss": 0.6008, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.23939767479896545, + "rewards/margins": 0.2871435582637787, + "rewards/rejected": -0.5265411734580994, + "step": 5520 + }, + { + "epoch": 0.72, + "learning_rate": 1.076080901676361e-06, + "logits/chosen": -2.2054896354675293, + "logits/rejected": -2.294689655303955, + "logps/chosen": -301.1555480957031, + "logps/rejected": -356.1802978515625, + "loss": 0.6367, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.27497249841690063, + "rewards/margins": 0.19801412522792816, + "rewards/rejected": -0.4729866087436676, + "step": 5530 + }, + { + "epoch": 0.72, + "learning_rate": 1.0667086513756234e-06, + "logits/chosen": -2.163573980331421, + "logits/rejected": -2.1624486446380615, + "logps/chosen": -258.3441162109375, + "logps/rejected": -276.3139343261719, + "loss": 0.581, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.22899429500102997, + "rewards/margins": 0.32109665870666504, + "rewards/rejected": -0.5500909090042114, + "step": 5540 + }, + { + "epoch": 0.73, + "learning_rate": 1.0573663123839912e-06, + "logits/chosen": -2.2281603813171387, + "logits/rejected": -2.146318197250366, + "logps/chosen": -252.2932891845703, + "logps/rejected": -278.700439453125, + "loss": 0.5812, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.21522720158100128, + "rewards/margins": 0.31266722083091736, + "rewards/rejected": -0.5278943777084351, + "step": 5550 + }, + { + "epoch": 0.73, + "learning_rate": 1.0480540796664251e-06, + "logits/chosen": -2.295581817626953, + "logits/rejected": -2.2284538745880127, + "logps/chosen": -358.788818359375, + "logps/rejected": -363.5187072753906, + "loss": 0.6118, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.22466108202934265, + "rewards/margins": 0.2881178855895996, + "rewards/rejected": -0.5127789974212646, + "step": 5560 + }, + { + "epoch": 0.73, + "learning_rate": 1.0387721475595978e-06, + "logits/chosen": -2.3117308616638184, + "logits/rejected": -2.255199909210205, + "logps/chosen": -343.64739990234375, + "logps/rejected": -330.49798583984375, + "loss": 0.622, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.2619249224662781, + "rewards/margins": 0.2441365271806717, + "rewards/rejected": -0.5060614347457886, + "step": 5570 + }, + { + "epoch": 0.73, + "learning_rate": 1.0295207097678378e-06, + "logits/chosen": -2.321887969970703, + "logits/rejected": -2.1694681644439697, + "logps/chosen": -303.7414855957031, + "logps/rejected": -286.21697998046875, + "loss": 0.6026, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.19890199601650238, + "rewards/margins": 0.3013504445552826, + "rewards/rejected": -0.5002524852752686, + "step": 5580 + }, + { + "epoch": 0.73, + "learning_rate": 1.0202999593590924e-06, + "logits/chosen": -2.1839795112609863, + "logits/rejected": -2.1160731315612793, + "logps/chosen": -280.7303161621094, + "logps/rejected": -253.0699005126953, + "loss": 0.6294, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.24383839964866638, + "rewards/margins": 0.2163550853729248, + "rewards/rejected": -0.4601934850215912, + "step": 5590 + }, + { + "epoch": 0.73, + "learning_rate": 1.011110088760891e-06, + "logits/chosen": -2.3758280277252197, + "logits/rejected": -2.244551181793213, + "logps/chosen": -311.04730224609375, + "logps/rejected": -292.23748779296875, + "loss": 0.5873, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.18764840066432953, + "rewards/margins": 0.3283648192882538, + "rewards/rejected": -0.5160131454467773, + "step": 5600 + }, + { + "epoch": 0.73, + "learning_rate": 1.0019512897563347e-06, + "logits/chosen": -2.1539273262023926, + "logits/rejected": -2.155287504196167, + "logps/chosen": -266.65643310546875, + "logps/rejected": -317.4012145996094, + "loss": 0.5188, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.14052417874336243, + "rewards/margins": 0.497483491897583, + "rewards/rejected": -0.638007640838623, + "step": 5610 + }, + { + "epoch": 0.74, + "learning_rate": 9.928237534800935e-07, + "logits/chosen": -2.3350729942321777, + "logits/rejected": -2.17598295211792, + "logps/chosen": -313.20025634765625, + "logps/rejected": -303.25201416015625, + "loss": 0.5729, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.22488637268543243, + "rewards/margins": 0.347454696893692, + "rewards/rejected": -0.5723410844802856, + "step": 5620 + }, + { + "epoch": 0.74, + "learning_rate": 9.837276704144174e-07, + "logits/chosen": -2.2017126083374023, + "logits/rejected": -2.249338388442993, + "logps/chosen": -263.5722961425781, + "logps/rejected": -320.7875671386719, + "loss": 0.6455, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.29073411226272583, + "rewards/margins": 0.1826065182685852, + "rewards/rejected": -0.47334060072898865, + "step": 5630 + }, + { + "epoch": 0.74, + "learning_rate": 9.746632303851569e-07, + "logits/chosen": -2.246596574783325, + "logits/rejected": -2.208622455596924, + "logps/chosen": -319.1954040527344, + "logps/rejected": -301.9804382324219, + "loss": 0.6029, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.14036308228969574, + "rewards/margins": 0.2734617292881012, + "rewards/rejected": -0.4138248562812805, + "step": 5640 + }, + { + "epoch": 0.74, + "learning_rate": 9.65630622557809e-07, + "logits/chosen": -2.1633477210998535, + "logits/rejected": -2.1405282020568848, + "logps/chosen": -287.43878173828125, + "logps/rejected": -305.71551513671875, + "loss": 0.6005, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.22302007675170898, + "rewards/margins": 0.29547959566116333, + "rewards/rejected": -0.5184996724128723, + "step": 5650 + }, + { + "epoch": 0.74, + "learning_rate": 9.56630035433561e-07, + "logits/chosen": -2.3428664207458496, + "logits/rejected": -2.2034668922424316, + "logps/chosen": -312.30511474609375, + "logps/rejected": -310.9969177246094, + "loss": 0.5626, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.20427866280078888, + "rewards/margins": 0.35165926814079285, + "rewards/rejected": -0.5559378862380981, + "step": 5660 + }, + { + "epoch": 0.74, + "learning_rate": 9.476616568453659e-07, + "logits/chosen": -2.236656904220581, + "logits/rejected": -2.1467082500457764, + "logps/chosen": -307.81842041015625, + "logps/rejected": -295.35455322265625, + "loss": 0.619, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.2767907679080963, + "rewards/margins": 0.26458436250686646, + "rewards/rejected": -0.5413750410079956, + "step": 5670 + }, + { + "epoch": 0.74, + "learning_rate": 9.387256739540162e-07, + "logits/chosen": -2.2104694843292236, + "logits/rejected": -2.100861072540283, + "logps/chosen": -258.29876708984375, + "logps/rejected": -254.37704467773438, + "loss": 0.5961, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.1629006713628769, + "rewards/margins": 0.29869306087493896, + "rewards/rejected": -0.46159371733665466, + "step": 5680 + }, + { + "epoch": 0.74, + "learning_rate": 9.298222732442377e-07, + "logits/chosen": -2.2853808403015137, + "logits/rejected": -2.258769989013672, + "logps/chosen": -291.1142883300781, + "logps/rejected": -339.2998352050781, + "loss": 0.5793, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.22501727938652039, + "rewards/margins": 0.332058846950531, + "rewards/rejected": -0.5570761561393738, + "step": 5690 + }, + { + "epoch": 0.75, + "learning_rate": 9.20951640520803e-07, + "logits/chosen": -2.382042407989502, + "logits/rejected": -2.300607919692993, + "logps/chosen": -333.5317077636719, + "logps/rejected": -361.81109619140625, + "loss": 0.553, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.13352376222610474, + "rewards/margins": 0.39602312445640564, + "rewards/rejected": -0.529546856880188, + "step": 5700 + }, + { + "epoch": 0.75, + "learning_rate": 9.121139609046484e-07, + "logits/chosen": -2.3873419761657715, + "logits/rejected": -2.246689796447754, + "logps/chosen": -295.42633056640625, + "logps/rejected": -291.67498779296875, + "loss": 0.5954, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.21414339542388916, + "rewards/margins": 0.28452545404434204, + "rewards/rejected": -0.4986688494682312, + "step": 5710 + }, + { + "epoch": 0.75, + "learning_rate": 9.033094188290121e-07, + "logits/chosen": -2.345534563064575, + "logits/rejected": -2.2926785945892334, + "logps/chosen": -300.8721008300781, + "logps/rejected": -347.0675048828125, + "loss": 0.5706, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.18854479491710663, + "rewards/margins": 0.3862067759037018, + "rewards/rejected": -0.5747515559196472, + "step": 5720 + }, + { + "epoch": 0.75, + "learning_rate": 8.945381980355889e-07, + "logits/chosen": -2.328420877456665, + "logits/rejected": -2.2627224922180176, + "logps/chosen": -318.6661376953125, + "logps/rejected": -330.42950439453125, + "loss": 0.6339, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.2633594572544098, + "rewards/margins": 0.20379576086997986, + "rewards/rejected": -0.4671551585197449, + "step": 5730 + }, + { + "epoch": 0.75, + "learning_rate": 8.858004815706919e-07, + "logits/chosen": -2.222224712371826, + "logits/rejected": -2.224364757537842, + "logps/chosen": -259.622802734375, + "logps/rejected": -296.14923095703125, + "loss": 0.5829, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.11960119009017944, + "rewards/margins": 0.35548511147499084, + "rewards/rejected": -0.4750862717628479, + "step": 5740 + }, + { + "epoch": 0.75, + "learning_rate": 8.77096451781432e-07, + "logits/chosen": -2.1508302688598633, + "logits/rejected": -2.1575701236724854, + "logps/chosen": -295.2816162109375, + "logps/rejected": -298.4773864746094, + "loss": 0.5663, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.16687360405921936, + "rewards/margins": 0.37276214361190796, + "rewards/rejected": -0.5396358370780945, + "step": 5750 + }, + { + "epoch": 0.75, + "learning_rate": 8.684262903119165e-07, + "logits/chosen": -2.2715201377868652, + "logits/rejected": -2.1461901664733887, + "logps/chosen": -334.562744140625, + "logps/rejected": -293.47265625, + "loss": 0.5984, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.23921248316764832, + "rewards/margins": 0.2952956557273865, + "rewards/rejected": -0.5345081090927124, + "step": 5760 + }, + { + "epoch": 0.76, + "learning_rate": 8.597901780994525e-07, + "logits/chosen": -2.140129804611206, + "logits/rejected": -2.0952887535095215, + "logps/chosen": -320.7549133300781, + "logps/rejected": -348.1116943359375, + "loss": 0.5804, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.16557027399539948, + "rewards/margins": 0.34092384576797485, + "rewards/rejected": -0.5064941048622131, + "step": 5770 + }, + { + "epoch": 0.76, + "learning_rate": 8.511882953707773e-07, + "logits/chosen": -2.194216012954712, + "logits/rejected": -2.2444090843200684, + "logps/chosen": -309.2511901855469, + "logps/rejected": -334.2317199707031, + "loss": 0.6026, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.23426124453544617, + "rewards/margins": 0.3184870779514313, + "rewards/rejected": -0.5527483224868774, + "step": 5780 + }, + { + "epoch": 0.76, + "learning_rate": 8.426208216382944e-07, + "logits/chosen": -2.069161891937256, + "logits/rejected": -2.0820701122283936, + "logps/chosen": -249.37881469726562, + "logps/rejected": -305.1940002441406, + "loss": 0.5999, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.19658584892749786, + "rewards/margins": 0.3157762289047241, + "rewards/rejected": -0.5123620629310608, + "step": 5790 + }, + { + "epoch": 0.76, + "learning_rate": 8.340879356963245e-07, + "logits/chosen": -2.275988817214966, + "logits/rejected": -2.204143524169922, + "logps/chosen": -266.86895751953125, + "logps/rejected": -306.22491455078125, + "loss": 0.5774, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.20006637275218964, + "rewards/margins": 0.33580079674720764, + "rewards/rejected": -0.5358671545982361, + "step": 5800 + }, + { + "epoch": 0.76, + "learning_rate": 8.255898156173777e-07, + "logits/chosen": -2.194227695465088, + "logits/rejected": -2.169670820236206, + "logps/chosen": -362.6389465332031, + "logps/rejected": -331.9048767089844, + "loss": 0.6589, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.358928382396698, + "rewards/margins": 0.15927556157112122, + "rewards/rejected": -0.5182039141654968, + "step": 5810 + }, + { + "epoch": 0.76, + "learning_rate": 8.171266387484389e-07, + "logits/chosen": -2.19954252243042, + "logits/rejected": -2.181100368499756, + "logps/chosen": -310.19866943359375, + "logps/rejected": -348.6551208496094, + "loss": 0.6145, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.24491801857948303, + "rewards/margins": 0.2709752917289734, + "rewards/rejected": -0.5158933401107788, + "step": 5820 + }, + { + "epoch": 0.76, + "learning_rate": 8.086985817072604e-07, + "logits/chosen": -2.2362565994262695, + "logits/rejected": -2.1750736236572266, + "logps/chosen": -279.22943115234375, + "logps/rejected": -297.84405517578125, + "loss": 0.5367, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.15669438242912292, + "rewards/margins": 0.4486146569252014, + "rewards/rejected": -0.605309009552002, + "step": 5830 + }, + { + "epoch": 0.76, + "learning_rate": 8.003058203786835e-07, + "logits/chosen": -2.2802937030792236, + "logits/rejected": -2.206637144088745, + "logps/chosen": -301.95965576171875, + "logps/rejected": -281.61566162109375, + "loss": 0.6613, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.2820906341075897, + "rewards/margins": 0.1556374877691269, + "rewards/rejected": -0.4377281069755554, + "step": 5840 + }, + { + "epoch": 0.77, + "learning_rate": 7.91948529910963e-07, + "logits/chosen": -2.3338422775268555, + "logits/rejected": -2.2146682739257812, + "logps/chosen": -379.86395263671875, + "logps/rejected": -342.8544006347656, + "loss": 0.6284, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.2647390067577362, + "rewards/margins": 0.25852879881858826, + "rewards/rejected": -0.5232678055763245, + "step": 5850 + }, + { + "epoch": 0.77, + "learning_rate": 7.836268847121126e-07, + "logits/chosen": -2.2316813468933105, + "logits/rejected": -2.1332125663757324, + "logps/chosen": -321.942626953125, + "logps/rejected": -333.6553039550781, + "loss": 0.5478, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.1956327110528946, + "rewards/margins": 0.3980453312397003, + "rewards/rejected": -0.5936779975891113, + "step": 5860 + }, + { + "epoch": 0.77, + "learning_rate": 7.753410584462681e-07, + "logits/chosen": -2.2441978454589844, + "logits/rejected": -2.209455966949463, + "logps/chosen": -319.6392517089844, + "logps/rejected": -367.57415771484375, + "loss": 0.5529, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.23563961684703827, + "rewards/margins": 0.39463010430336, + "rewards/rejected": -0.6302697062492371, + "step": 5870 + }, + { + "epoch": 0.77, + "learning_rate": 7.670912240300596e-07, + "logits/chosen": -2.261707067489624, + "logits/rejected": -2.168673038482666, + "logps/chosen": -306.07293701171875, + "logps/rejected": -354.01513671875, + "loss": 0.6089, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.24367579817771912, + "rewards/margins": 0.2957931160926819, + "rewards/rejected": -0.5394688844680786, + "step": 5880 + }, + { + "epoch": 0.77, + "learning_rate": 7.588775536290035e-07, + "logits/chosen": -2.3636014461517334, + "logits/rejected": -2.2855453491210938, + "logps/chosen": -313.52703857421875, + "logps/rejected": -322.87322998046875, + "loss": 0.5993, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.18039445579051971, + "rewards/margins": 0.3048645853996277, + "rewards/rejected": -0.4852589964866638, + "step": 5890 + }, + { + "epoch": 0.77, + "learning_rate": 7.507002186539147e-07, + "logits/chosen": -2.3298370838165283, + "logits/rejected": -2.264897584915161, + "logps/chosen": -329.05426025390625, + "logps/rejected": -335.554443359375, + "loss": 0.6083, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.2601746916770935, + "rewards/margins": 0.29222458600997925, + "rewards/rejected": -0.5523992776870728, + "step": 5900 + }, + { + "epoch": 0.77, + "learning_rate": 7.425593897573216e-07, + "logits/chosen": -2.230464458465576, + "logits/rejected": -2.0954625606536865, + "logps/chosen": -293.1115417480469, + "logps/rejected": -318.01263427734375, + "loss": 0.586, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.20525069534778595, + "rewards/margins": 0.3247843086719513, + "rewards/rejected": -0.5300350189208984, + "step": 5910 + }, + { + "epoch": 0.77, + "learning_rate": 7.344552368299088e-07, + "logits/chosen": -2.1736879348754883, + "logits/rejected": -2.1197733879089355, + "logps/chosen": -285.8775634765625, + "logps/rejected": -314.8118896484375, + "loss": 0.6024, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.3129540681838989, + "rewards/margins": 0.27547574043273926, + "rewards/rejected": -0.5884298086166382, + "step": 5920 + }, + { + "epoch": 0.78, + "learning_rate": 7.26387928996973e-07, + "logits/chosen": -2.2027297019958496, + "logits/rejected": -2.091571092605591, + "logps/chosen": -306.52398681640625, + "logps/rejected": -293.25689697265625, + "loss": 0.6152, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.2530812919139862, + "rewards/margins": 0.25087738037109375, + "rewards/rejected": -0.5039586424827576, + "step": 5930 + }, + { + "epoch": 0.78, + "learning_rate": 7.183576346148899e-07, + "logits/chosen": -2.2879271507263184, + "logits/rejected": -2.2315640449523926, + "logps/chosen": -338.07135009765625, + "logps/rejected": -343.3145751953125, + "loss": 0.5948, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.2316274642944336, + "rewards/margins": 0.30058106780052185, + "rewards/rejected": -0.5322085022926331, + "step": 5940 + }, + { + "epoch": 0.78, + "learning_rate": 7.103645212676044e-07, + "logits/chosen": -2.2760863304138184, + "logits/rejected": -2.2098329067230225, + "logps/chosen": -335.37030029296875, + "logps/rejected": -331.3144836425781, + "loss": 0.5758, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.167093425989151, + "rewards/margins": 0.3470625579357147, + "rewards/rejected": -0.5141559839248657, + "step": 5950 + }, + { + "epoch": 0.78, + "learning_rate": 7.024087557631318e-07, + "logits/chosen": -2.3430886268615723, + "logits/rejected": -2.2137045860290527, + "logps/chosen": -315.588623046875, + "logps/rejected": -324.27435302734375, + "loss": 0.5913, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.26683956384658813, + "rewards/margins": 0.3276747763156891, + "rewards/rejected": -0.5945142507553101, + "step": 5960 + }, + { + "epoch": 0.78, + "learning_rate": 6.944905041300739e-07, + "logits/chosen": -2.354771375656128, + "logits/rejected": -2.235982894897461, + "logps/chosen": -329.1302185058594, + "logps/rejected": -307.24176025390625, + "loss": 0.5805, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.13809750974178314, + "rewards/margins": 0.343759685754776, + "rewards/rejected": -0.48185721039772034, + "step": 5970 + }, + { + "epoch": 0.78, + "learning_rate": 6.866099316141606e-07, + "logits/chosen": -2.2135233879089355, + "logits/rejected": -2.17932391166687, + "logps/chosen": -323.75799560546875, + "logps/rejected": -317.0902404785156, + "loss": 0.5982, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.18628045916557312, + "rewards/margins": 0.316842257976532, + "rewards/rejected": -0.5031227469444275, + "step": 5980 + }, + { + "epoch": 0.78, + "learning_rate": 6.787672026747946e-07, + "logits/chosen": -2.1737163066864014, + "logits/rejected": -2.0860915184020996, + "logps/chosen": -288.33111572265625, + "logps/rejected": -328.392578125, + "loss": 0.5433, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.11049487441778183, + "rewards/margins": 0.43303370475769043, + "rewards/rejected": -0.54352867603302, + "step": 5990 + }, + { + "epoch": 0.79, + "learning_rate": 6.709624809816223e-07, + "logits/chosen": -2.2856686115264893, + "logits/rejected": -2.143397092819214, + "logps/chosen": -293.72772216796875, + "logps/rejected": -281.6180419921875, + "loss": 0.6046, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.23028914630413055, + "rewards/margins": 0.2749117314815521, + "rewards/rejected": -0.5052008628845215, + "step": 6000 + }, + { + "epoch": 0.79, + "learning_rate": 6.6319592941112e-07, + "logits/chosen": -2.276402711868286, + "logits/rejected": -2.199864149093628, + "logps/chosen": -315.3931579589844, + "logps/rejected": -298.86737060546875, + "loss": 0.587, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.23889532685279846, + "rewards/margins": 0.31959256529808044, + "rewards/rejected": -0.5584878921508789, + "step": 6010 + }, + { + "epoch": 0.79, + "learning_rate": 6.554677100431927e-07, + "logits/chosen": -2.239607572555542, + "logits/rejected": -2.155582904815674, + "logps/chosen": -269.63873291015625, + "logps/rejected": -292.4139709472656, + "loss": 0.5285, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.15219341218471527, + "rewards/margins": 0.4786094129085541, + "rewards/rejected": -0.6308028101921082, + "step": 6020 + }, + { + "epoch": 0.79, + "learning_rate": 6.4777798415779e-07, + "logits/chosen": -2.416595935821533, + "logits/rejected": -2.2838592529296875, + "logps/chosen": -318.8462829589844, + "logps/rejected": -312.1090393066406, + "loss": 0.5785, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.12302907556295395, + "rewards/margins": 0.3575192987918854, + "rewards/rejected": -0.4805483818054199, + "step": 6030 + }, + { + "epoch": 0.79, + "learning_rate": 6.401269122315451e-07, + "logits/chosen": -2.225639820098877, + "logits/rejected": -2.2105183601379395, + "logps/chosen": -317.6982727050781, + "logps/rejected": -369.0693359375, + "loss": 0.6053, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2531086504459381, + "rewards/margins": 0.2957480847835541, + "rewards/rejected": -0.5488567352294922, + "step": 6040 + }, + { + "epoch": 0.79, + "learning_rate": 6.325146539344196e-07, + "logits/chosen": -2.209933280944824, + "logits/rejected": -2.0943663120269775, + "logps/chosen": -409.1436462402344, + "logps/rejected": -398.69842529296875, + "loss": 0.5813, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.2911561131477356, + "rewards/margins": 0.36323466897010803, + "rewards/rejected": -0.654390811920166, + "step": 6050 + }, + { + "epoch": 0.79, + "learning_rate": 6.249413681263782e-07, + "logits/chosen": -2.232851982116699, + "logits/rejected": -2.155423879623413, + "logps/chosen": -250.1781768798828, + "logps/rejected": -332.4837951660156, + "loss": 0.5884, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.24195115268230438, + "rewards/margins": 0.35050472617149353, + "rewards/rejected": -0.5924558639526367, + "step": 6060 + }, + { + "epoch": 0.79, + "learning_rate": 6.174072128540686e-07, + "logits/chosen": -2.3279008865356445, + "logits/rejected": -2.26503586769104, + "logps/chosen": -337.4497985839844, + "logps/rejected": -307.052734375, + "loss": 0.6443, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.21359160542488098, + "rewards/margins": 0.2172565460205078, + "rewards/rejected": -0.4308481812477112, + "step": 6070 + }, + { + "epoch": 0.8, + "learning_rate": 6.099123453475245e-07, + "logits/chosen": -2.245133399963379, + "logits/rejected": -2.1051242351531982, + "logps/chosen": -337.2713623046875, + "logps/rejected": -332.1095275878906, + "loss": 0.5822, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.23662367463111877, + "rewards/margins": 0.3294454514980316, + "rewards/rejected": -0.5660691261291504, + "step": 6080 + }, + { + "epoch": 0.8, + "learning_rate": 6.024569220168836e-07, + "logits/chosen": -2.2474308013916016, + "logits/rejected": -2.104712724685669, + "logps/chosen": -322.6059265136719, + "logps/rejected": -311.0595397949219, + "loss": 0.5344, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.11445900052785873, + "rewards/margins": 0.4638898968696594, + "rewards/rejected": -0.5783488750457764, + "step": 6090 + }, + { + "epoch": 0.8, + "learning_rate": 5.950410984491268e-07, + "logits/chosen": -2.231515645980835, + "logits/rejected": -2.23624849319458, + "logps/chosen": -334.6339111328125, + "logps/rejected": -355.5390319824219, + "loss": 0.58, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.2133444845676422, + "rewards/margins": 0.3178350329399109, + "rewards/rejected": -0.5311795473098755, + "step": 6100 + }, + { + "epoch": 0.8, + "learning_rate": 5.876650294048262e-07, + "logits/chosen": -2.2630536556243896, + "logits/rejected": -2.2580368518829346, + "logps/chosen": -270.99591064453125, + "logps/rejected": -303.1621398925781, + "loss": 0.6118, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.19207841157913208, + "rewards/margins": 0.2853357791900635, + "rewards/rejected": -0.47741419076919556, + "step": 6110 + }, + { + "epoch": 0.8, + "learning_rate": 5.8032886881492e-07, + "logits/chosen": -2.174412250518799, + "logits/rejected": -2.121708393096924, + "logps/chosen": -249.01132202148438, + "logps/rejected": -254.09683227539062, + "loss": 0.6319, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.22126099467277527, + "rewards/margins": 0.22934678196907043, + "rewards/rejected": -0.4506077170372009, + "step": 6120 + }, + { + "epoch": 0.8, + "learning_rate": 5.730327697774988e-07, + "logits/chosen": -2.2567834854125977, + "logits/rejected": -2.223865270614624, + "logps/chosen": -280.71051025390625, + "logps/rejected": -335.3533630371094, + "loss": 0.5546, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.1094474345445633, + "rewards/margins": 0.37788665294647217, + "rewards/rejected": -0.48733407258987427, + "step": 6130 + }, + { + "epoch": 0.8, + "learning_rate": 5.657768845546068e-07, + "logits/chosen": -2.209315061569214, + "logits/rejected": -2.1145834922790527, + "logps/chosen": -242.4571533203125, + "logps/rejected": -268.93524169921875, + "loss": 0.5958, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.22153480350971222, + "rewards/margins": 0.29737162590026855, + "rewards/rejected": -0.5189064741134644, + "step": 6140 + }, + { + "epoch": 0.8, + "learning_rate": 5.585613645690713e-07, + "logits/chosen": -2.23826003074646, + "logits/rejected": -2.1479594707489014, + "logps/chosen": -283.0222473144531, + "logps/rejected": -324.4845886230469, + "loss": 0.548, + "rewards/accuracies": 0.7875000238418579, + "rewards/chosen": -0.18606118857860565, + "rewards/margins": 0.42992424964904785, + "rewards/rejected": -0.6159854531288147, + "step": 6150 + }, + { + "epoch": 0.81, + "learning_rate": 5.513863604013355e-07, + "logits/chosen": -2.3211870193481445, + "logits/rejected": -2.1886870861053467, + "logps/chosen": -334.17950439453125, + "logps/rejected": -308.362548828125, + "loss": 0.6228, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.24311676621437073, + "rewards/margins": 0.24683444201946259, + "rewards/rejected": -0.4899512231349945, + "step": 6160 + }, + { + "epoch": 0.81, + "learning_rate": 5.442520217863215e-07, + "logits/chosen": -2.227276563644409, + "logits/rejected": -2.1782267093658447, + "logps/chosen": -287.66583251953125, + "logps/rejected": -291.90948486328125, + "loss": 0.6353, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.2558891177177429, + "rewards/margins": 0.2297544926404953, + "rewards/rejected": -0.48564356565475464, + "step": 6170 + }, + { + "epoch": 0.81, + "learning_rate": 5.371584976103034e-07, + "logits/chosen": -2.28275990486145, + "logits/rejected": -2.173872470855713, + "logps/chosen": -303.195068359375, + "logps/rejected": -327.93853759765625, + "loss": 0.5368, + "rewards/accuracies": 0.8125, + "rewards/chosen": -0.16327349841594696, + "rewards/margins": 0.44478267431259155, + "rewards/rejected": -0.6080561876296997, + "step": 6180 + }, + { + "epoch": 0.81, + "learning_rate": 5.301059359077987e-07, + "logits/chosen": -2.2108511924743652, + "logits/rejected": -2.1684250831604004, + "logps/chosen": -270.13397216796875, + "logps/rejected": -290.10211181640625, + "loss": 0.6143, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.24181067943572998, + "rewards/margins": 0.2641329765319824, + "rewards/rejected": -0.5059436559677124, + "step": 6190 + }, + { + "epoch": 0.81, + "learning_rate": 5.230944838584806e-07, + "logits/chosen": -2.293006181716919, + "logits/rejected": -2.262521266937256, + "logps/chosen": -325.39898681640625, + "logps/rejected": -333.6467590332031, + "loss": 0.6232, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.1768663227558136, + "rewards/margins": 0.2859518826007843, + "rewards/rejected": -0.4628182053565979, + "step": 6200 + }, + { + "epoch": 0.81, + "learning_rate": 5.161242877841083e-07, + "logits/chosen": -2.361013412475586, + "logits/rejected": -2.2488346099853516, + "logps/chosen": -310.595458984375, + "logps/rejected": -313.78564453125, + "loss": 0.5627, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.18735520541667938, + "rewards/margins": 0.36751627922058105, + "rewards/rejected": -0.5548714399337769, + "step": 6210 + }, + { + "epoch": 0.81, + "learning_rate": 5.091954931454682e-07, + "logits/chosen": -2.1743760108947754, + "logits/rejected": -2.0993740558624268, + "logps/chosen": -299.3758239746094, + "logps/rejected": -320.70550537109375, + "loss": 0.5427, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.09906675666570663, + "rewards/margins": 0.4197840690612793, + "rewards/rejected": -0.5188508629798889, + "step": 6220 + }, + { + "epoch": 0.82, + "learning_rate": 5.023082445393446e-07, + "logits/chosen": -2.222538948059082, + "logits/rejected": -2.1469168663024902, + "logps/chosen": -296.52783203125, + "logps/rejected": -293.89630126953125, + "loss": 0.531, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.14179904758930206, + "rewards/margins": 0.4603515565395355, + "rewards/rejected": -0.602150559425354, + "step": 6230 + }, + { + "epoch": 0.82, + "learning_rate": 4.95462685695498e-07, + "logits/chosen": -2.299201488494873, + "logits/rejected": -2.2109179496765137, + "logps/chosen": -308.81817626953125, + "logps/rejected": -290.07586669921875, + "loss": 0.616, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.21971330046653748, + "rewards/margins": 0.26939094066619873, + "rewards/rejected": -0.4891042113304138, + "step": 6240 + }, + { + "epoch": 0.82, + "learning_rate": 4.88658959473666e-07, + "logits/chosen": -2.2667698860168457, + "logits/rejected": -2.178948402404785, + "logps/chosen": -315.88116455078125, + "logps/rejected": -302.3547668457031, + "loss": 0.5547, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.15843792259693146, + "rewards/margins": 0.40256786346435547, + "rewards/rejected": -0.5610057711601257, + "step": 6250 + }, + { + "epoch": 0.82, + "learning_rate": 4.818972078605821e-07, + "logits/chosen": -2.2810981273651123, + "logits/rejected": -2.12772274017334, + "logps/chosen": -371.85943603515625, + "logps/rejected": -299.86651611328125, + "loss": 0.6056, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.22580282390117645, + "rewards/margins": 0.3250388205051422, + "rewards/rejected": -0.5508416891098022, + "step": 6260 + }, + { + "epoch": 0.82, + "learning_rate": 4.7517757196701514e-07, + "logits/chosen": -2.3292527198791504, + "logits/rejected": -2.2128889560699463, + "logps/chosen": -277.4884338378906, + "logps/rejected": -277.57928466796875, + "loss": 0.5827, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.23419201374053955, + "rewards/margins": 0.32151398062705994, + "rewards/rejected": -0.5557060241699219, + "step": 6270 + }, + { + "epoch": 0.82, + "learning_rate": 4.6850019202482193e-07, + "logits/chosen": -2.2545182704925537, + "logits/rejected": -2.1212382316589355, + "logps/chosen": -278.05474853515625, + "logps/rejected": -286.8658447265625, + "loss": 0.6044, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.25792375206947327, + "rewards/margins": 0.3005824685096741, + "rewards/rejected": -0.5585063099861145, + "step": 6280 + }, + { + "epoch": 0.82, + "learning_rate": 4.618652073840188e-07, + "logits/chosen": -2.354627847671509, + "logits/rejected": -2.2349581718444824, + "logps/chosen": -331.49822998046875, + "logps/rejected": -317.4371337890625, + "loss": 0.5817, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.17250117659568787, + "rewards/margins": 0.3626854121685028, + "rewards/rejected": -0.5351866483688354, + "step": 6290 + }, + { + "epoch": 0.82, + "learning_rate": 4.5527275650987965e-07, + "logits/chosen": -2.2923061847686768, + "logits/rejected": -2.209237575531006, + "logps/chosen": -310.1547546386719, + "logps/rejected": -298.08636474609375, + "loss": 0.6077, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.25109967589378357, + "rewards/margins": 0.26999661326408386, + "rewards/rejected": -0.5210963487625122, + "step": 6300 + }, + { + "epoch": 0.83, + "learning_rate": 4.487229769800394e-07, + "logits/chosen": -2.2456047534942627, + "logits/rejected": -2.1408562660217285, + "logps/chosen": -261.54779052734375, + "logps/rejected": -265.80511474609375, + "loss": 0.6225, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.24030759930610657, + "rewards/margins": 0.2347225695848465, + "rewards/rejected": -0.4750301241874695, + "step": 6310 + }, + { + "epoch": 0.83, + "learning_rate": 4.422160054816285e-07, + "logits/chosen": -2.1729722023010254, + "logits/rejected": -2.0807228088378906, + "logps/chosen": -344.05810546875, + "logps/rejected": -352.92376708984375, + "loss": 0.6316, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.24153423309326172, + "rewards/margins": 0.21908824145793915, + "rewards/rejected": -0.4606224596500397, + "step": 6320 + }, + { + "epoch": 0.83, + "learning_rate": 4.35751977808416e-07, + "logits/chosen": -2.284636974334717, + "logits/rejected": -2.201692581176758, + "logps/chosen": -297.3533935546875, + "logps/rejected": -308.90155029296875, + "loss": 0.5235, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.12625914812088013, + "rewards/margins": 0.4717814326286316, + "rewards/rejected": -0.5980406403541565, + "step": 6330 + }, + { + "epoch": 0.83, + "learning_rate": 4.293310288579794e-07, + "logits/chosen": -2.0978596210479736, + "logits/rejected": -2.1035284996032715, + "logps/chosen": -291.7745056152344, + "logps/rejected": -306.15411376953125, + "loss": 0.6196, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.24561206996440887, + "rewards/margins": 0.26010841131210327, + "rewards/rejected": -0.5057204961776733, + "step": 6340 + }, + { + "epoch": 0.83, + "learning_rate": 4.2295329262888733e-07, + "logits/chosen": -2.2204253673553467, + "logits/rejected": -2.11824893951416, + "logps/chosen": -297.88568115234375, + "logps/rejected": -290.4247131347656, + "loss": 0.6025, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.26537007093429565, + "rewards/margins": 0.2906087636947632, + "rewards/rejected": -0.5559788942337036, + "step": 6350 + }, + { + "epoch": 0.83, + "learning_rate": 4.1661890221790316e-07, + "logits/chosen": -2.2358736991882324, + "logits/rejected": -2.2458510398864746, + "logps/chosen": -281.9924621582031, + "logps/rejected": -350.58941650390625, + "loss": 0.5342, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.15237683057785034, + "rewards/margins": 0.4450441002845764, + "rewards/rejected": -0.5974208116531372, + "step": 6360 + }, + { + "epoch": 0.83, + "learning_rate": 4.103279898172072e-07, + "logits/chosen": -2.322620153427124, + "logits/rejected": -2.168433427810669, + "logps/chosen": -366.12786865234375, + "logps/rejected": -325.152587890625, + "loss": 0.6199, + "rewards/accuracies": 0.6000000238418579, + "rewards/chosen": -0.29785770177841187, + "rewards/margins": 0.26100578904151917, + "rewards/rejected": -0.5588634610176086, + "step": 6370 + }, + { + "epoch": 0.83, + "learning_rate": 4.040806867116401e-07, + "logits/chosen": -2.250920057296753, + "logits/rejected": -2.2237958908081055, + "logps/chosen": -321.1445007324219, + "logps/rejected": -321.01239013671875, + "loss": 0.6241, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.22857800126075745, + "rewards/margins": 0.23127146065235138, + "rewards/rejected": -0.4598495066165924, + "step": 6380 + }, + { + "epoch": 0.84, + "learning_rate": 3.978771232759615e-07, + "logits/chosen": -2.2364799976348877, + "logits/rejected": -2.212056875228882, + "logps/chosen": -269.04595947265625, + "logps/rejected": -298.14642333984375, + "loss": 0.5804, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.1761271059513092, + "rewards/margins": 0.32325392961502075, + "rewards/rejected": -0.49938106536865234, + "step": 6390 + }, + { + "epoch": 0.84, + "learning_rate": 3.917174289721276e-07, + "logits/chosen": -2.2611424922943115, + "logits/rejected": -2.2784390449523926, + "logps/chosen": -315.00732421875, + "logps/rejected": -305.6526184082031, + "loss": 0.6236, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.2230949103832245, + "rewards/margins": 0.21592876315116882, + "rewards/rejected": -0.4390237331390381, + "step": 6400 + }, + { + "epoch": 0.84, + "learning_rate": 3.856017323465938e-07, + "logits/chosen": -2.191107749938965, + "logits/rejected": -2.1177146434783936, + "logps/chosen": -274.64263916015625, + "logps/rejected": -298.37091064453125, + "loss": 0.5915, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.2880602478981018, + "rewards/margins": 0.30864188075065613, + "rewards/rejected": -0.5967021584510803, + "step": 6410 + }, + { + "epoch": 0.84, + "learning_rate": 3.7953016102762695e-07, + "logits/chosen": -2.2297253608703613, + "logits/rejected": -2.2038657665252686, + "logps/chosen": -305.0720520019531, + "logps/rejected": -304.5875549316406, + "loss": 0.6349, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.19702813029289246, + "rewards/margins": 0.25031501054763794, + "rewards/rejected": -0.4473431706428528, + "step": 6420 + }, + { + "epoch": 0.84, + "learning_rate": 3.7350284172264493e-07, + "logits/chosen": -2.242567300796509, + "logits/rejected": -2.2194294929504395, + "logps/chosen": -260.15185546875, + "logps/rejected": -306.4600524902344, + "loss": 0.6072, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.2087719887495041, + "rewards/margins": 0.27035030722618103, + "rewards/rejected": -0.4791223108768463, + "step": 6430 + }, + { + "epoch": 0.84, + "learning_rate": 3.67519900215573e-07, + "logits/chosen": -2.237037420272827, + "logits/rejected": -2.1576311588287354, + "logps/chosen": -295.35089111328125, + "logps/rejected": -292.58917236328125, + "loss": 0.5995, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.20834775269031525, + "rewards/margins": 0.3026237487792969, + "rewards/rejected": -0.5109714865684509, + "step": 6440 + }, + { + "epoch": 0.84, + "learning_rate": 3.615814613642174e-07, + "logits/chosen": -2.2904200553894043, + "logits/rejected": -2.1129608154296875, + "logps/chosen": -343.78662109375, + "logps/rejected": -301.4750671386719, + "loss": 0.6, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.22187814116477966, + "rewards/margins": 0.28146690130233765, + "rewards/rejected": -0.5033451318740845, + "step": 6450 + }, + { + "epoch": 0.85, + "learning_rate": 3.5568764909765795e-07, + "logits/chosen": -2.2265584468841553, + "logits/rejected": -2.1241002082824707, + "logps/chosen": -330.27655029296875, + "logps/rejected": -335.1883239746094, + "loss": 0.5565, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.16477003693580627, + "rewards/margins": 0.3972473740577698, + "rewards/rejected": -0.5620173811912537, + "step": 6460 + }, + { + "epoch": 0.85, + "learning_rate": 3.498385864136672e-07, + "logits/chosen": -2.2943968772888184, + "logits/rejected": -2.254265308380127, + "logps/chosen": -307.8959655761719, + "logps/rejected": -323.3202819824219, + "loss": 0.6178, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.2675125002861023, + "rewards/margins": 0.26192277669906616, + "rewards/rejected": -0.5294352769851685, + "step": 6470 + }, + { + "epoch": 0.85, + "learning_rate": 3.440343953761363e-07, + "logits/chosen": -2.2510826587677, + "logits/rejected": -2.151731014251709, + "logps/chosen": -297.1509094238281, + "logps/rejected": -293.01715087890625, + "loss": 0.6359, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.2728116512298584, + "rewards/margins": 0.22913222014904022, + "rewards/rejected": -0.5019438862800598, + "step": 6480 + }, + { + "epoch": 0.85, + "learning_rate": 3.382751971125345e-07, + "logits/chosen": -2.225860834121704, + "logits/rejected": -2.175529956817627, + "logps/chosen": -287.80706787109375, + "logps/rejected": -287.97802734375, + "loss": 0.6297, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.24230583012104034, + "rewards/margins": 0.23965951800346375, + "rewards/rejected": -0.4819653630256653, + "step": 6490 + }, + { + "epoch": 0.85, + "learning_rate": 3.3256111181137753e-07, + "logits/chosen": -2.3095192909240723, + "logits/rejected": -2.2175674438476562, + "logps/chosen": -286.6315612792969, + "logps/rejected": -287.605224609375, + "loss": 0.5935, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.18542449176311493, + "rewards/margins": 0.2997186779975891, + "rewards/rejected": -0.48514318466186523, + "step": 6500 + }, + { + "epoch": 0.85, + "learning_rate": 3.2689225871971905e-07, + "logits/chosen": -2.213831663131714, + "logits/rejected": -2.195160150527954, + "logps/chosen": -316.28851318359375, + "logps/rejected": -342.26446533203125, + "loss": 0.6135, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.23785743117332458, + "rewards/margins": 0.2976071238517761, + "rewards/rejected": -0.5354645848274231, + "step": 6510 + }, + { + "epoch": 0.85, + "learning_rate": 3.2126875614066523e-07, + "logits/chosen": -2.273104190826416, + "logits/rejected": -2.214599609375, + "logps/chosen": -287.70855712890625, + "logps/rejected": -285.56036376953125, + "loss": 0.6276, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.2149469405412674, + "rewards/margins": 0.21491274237632751, + "rewards/rejected": -0.4298597276210785, + "step": 6520 + }, + { + "epoch": 0.85, + "learning_rate": 3.156907214309024e-07, + "logits/chosen": -2.1992926597595215, + "logits/rejected": -2.141411066055298, + "logps/chosen": -274.2710876464844, + "logps/rejected": -306.6649169921875, + "loss": 0.5484, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.1834583729505539, + "rewards/margins": 0.38944798707962036, + "rewards/rejected": -0.5729063749313354, + "step": 6530 + }, + { + "epoch": 0.86, + "learning_rate": 3.1015827099824923e-07, + "logits/chosen": -2.3543479442596436, + "logits/rejected": -2.218076229095459, + "logps/chosen": -315.0438537597656, + "logps/rejected": -313.69757080078125, + "loss": 0.5693, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.19667963683605194, + "rewards/margins": 0.34850531816482544, + "rewards/rejected": -0.5451849102973938, + "step": 6540 + }, + { + "epoch": 0.86, + "learning_rate": 3.0467152029922926e-07, + "logits/chosen": -2.1841235160827637, + "logits/rejected": -2.1445069313049316, + "logps/chosen": -273.4917907714844, + "logps/rejected": -334.92669677734375, + "loss": 0.6072, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2173459827899933, + "rewards/margins": 0.28793925046920776, + "rewards/rejected": -0.5052852630615234, + "step": 6550 + }, + { + "epoch": 0.86, + "learning_rate": 2.992305838366591e-07, + "logits/chosen": -2.3001112937927246, + "logits/rejected": -2.259284496307373, + "logps/chosen": -362.2890930175781, + "logps/rejected": -344.4498596191406, + "loss": 0.5852, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.22456571459770203, + "rewards/margins": 0.3367841839790344, + "rewards/rejected": -0.5613499283790588, + "step": 6560 + }, + { + "epoch": 0.86, + "learning_rate": 2.938355751572583e-07, + "logits/chosen": -2.2215077877044678, + "logits/rejected": -2.122424364089966, + "logps/chosen": -305.6630859375, + "logps/rejected": -292.48687744140625, + "loss": 0.6135, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.2404574155807495, + "rewards/margins": 0.25619426369667053, + "rewards/rejected": -0.49665170907974243, + "step": 6570 + }, + { + "epoch": 0.86, + "learning_rate": 2.8848660684928307e-07, + "logits/chosen": -2.212378978729248, + "logits/rejected": -2.1459293365478516, + "logps/chosen": -328.45721435546875, + "logps/rejected": -319.005126953125, + "loss": 0.5897, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.24658489227294922, + "rewards/margins": 0.32633015513420105, + "rewards/rejected": -0.5729150772094727, + "step": 6580 + }, + { + "epoch": 0.86, + "learning_rate": 2.8318379054017383e-07, + "logits/chosen": -2.2369682788848877, + "logits/rejected": -2.1657543182373047, + "logps/chosen": -320.33563232421875, + "logps/rejected": -298.6103820800781, + "loss": 0.5521, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.17626380920410156, + "rewards/margins": 0.4276823103427887, + "rewards/rejected": -0.6039462089538574, + "step": 6590 + }, + { + "epoch": 0.86, + "learning_rate": 2.779272368942246e-07, + "logits/chosen": -2.3483896255493164, + "logits/rejected": -2.2070584297180176, + "logps/chosen": -327.89691162109375, + "logps/rejected": -339.7986755371094, + "loss": 0.599, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.2045535296201706, + "rewards/margins": 0.3108772337436676, + "rewards/rejected": -0.5154308080673218, + "step": 6600 + }, + { + "epoch": 0.86, + "learning_rate": 2.7271705561027986e-07, + "logits/chosen": -2.2898526191711426, + "logits/rejected": -2.179412603378296, + "logps/chosen": -350.47845458984375, + "logps/rejected": -316.5538330078125, + "loss": 0.5658, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.1713074892759323, + "rewards/margins": 0.39373883605003357, + "rewards/rejected": -0.5650463104248047, + "step": 6610 + }, + { + "epoch": 0.87, + "learning_rate": 2.6755335541943677e-07, + "logits/chosen": -2.2905704975128174, + "logits/rejected": -2.2754604816436768, + "logps/chosen": -309.1745300292969, + "logps/rejected": -333.67938232421875, + "loss": 0.6222, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.283588171005249, + "rewards/margins": 0.24496212601661682, + "rewards/rejected": -0.5285503268241882, + "step": 6620 + }, + { + "epoch": 0.87, + "learning_rate": 2.62436244082781e-07, + "logits/chosen": -2.350175142288208, + "logits/rejected": -2.3214735984802246, + "logps/chosen": -322.21685791015625, + "logps/rejected": -337.0072326660156, + "loss": 0.5259, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.15177544951438904, + "rewards/margins": 0.4634989798069, + "rewards/rejected": -0.6152744293212891, + "step": 6630 + }, + { + "epoch": 0.87, + "learning_rate": 2.5736582838913836e-07, + "logits/chosen": -2.2806878089904785, + "logits/rejected": -2.1992342472076416, + "logps/chosen": -307.2468566894531, + "logps/rejected": -306.5870666503906, + "loss": 0.5171, + "rewards/accuracies": 0.800000011920929, + "rewards/chosen": -0.12974882125854492, + "rewards/margins": 0.5071049928665161, + "rewards/rejected": -0.6368538737297058, + "step": 6640 + }, + { + "epoch": 0.87, + "learning_rate": 2.5234221415284363e-07, + "logits/chosen": -2.2767059803009033, + "logits/rejected": -2.210958957672119, + "logps/chosen": -314.1791076660156, + "logps/rejected": -352.72564697265625, + "loss": 0.5694, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.1622014045715332, + "rewards/margins": 0.35927143692970276, + "rewards/rejected": -0.5214728713035583, + "step": 6650 + }, + { + "epoch": 0.87, + "learning_rate": 2.4736550621153375e-07, + "logits/chosen": -2.2079639434814453, + "logits/rejected": -2.1122167110443115, + "logps/chosen": -277.1203308105469, + "logps/rejected": -299.25885009765625, + "loss": 0.5394, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.2556740939617157, + "rewards/margins": 0.44593340158462524, + "rewards/rejected": -0.7016074657440186, + "step": 6660 + }, + { + "epoch": 0.87, + "learning_rate": 2.424358084239609e-07, + "logits/chosen": -2.2848687171936035, + "logits/rejected": -2.2923455238342285, + "logps/chosen": -297.78179931640625, + "logps/rejected": -319.9772033691406, + "loss": 0.6469, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.23676368594169617, + "rewards/margins": 0.20353002846240997, + "rewards/rejected": -0.44029369950294495, + "step": 6670 + }, + { + "epoch": 0.87, + "learning_rate": 2.3755322366782158e-07, + "logits/chosen": -2.2076756954193115, + "logits/rejected": -2.1373071670532227, + "logps/chosen": -310.51495361328125, + "logps/rejected": -308.01654052734375, + "loss": 0.5738, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.16560694575309753, + "rewards/margins": 0.35412055253982544, + "rewards/rejected": -0.5197274684906006, + "step": 6680 + }, + { + "epoch": 0.88, + "learning_rate": 2.3271785383761431e-07, + "logits/chosen": -2.2977707386016846, + "logits/rejected": -2.253223419189453, + "logps/chosen": -306.7425231933594, + "logps/rejected": -299.0722351074219, + "loss": 0.6451, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.2154873162508011, + "rewards/margins": 0.17152908444404602, + "rewards/rejected": -0.3870163559913635, + "step": 6690 + }, + { + "epoch": 0.88, + "learning_rate": 2.2792979984250978e-07, + "logits/chosen": -2.304884672164917, + "logits/rejected": -2.1841373443603516, + "logps/chosen": -325.85693359375, + "logps/rejected": -301.67694091796875, + "loss": 0.6013, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.16651174426078796, + "rewards/margins": 0.27157658338546753, + "rewards/rejected": -0.4380883276462555, + "step": 6700 + }, + { + "epoch": 0.88, + "learning_rate": 2.231891616042453e-07, + "logits/chosen": -2.3625378608703613, + "logits/rejected": -2.2966275215148926, + "logps/chosen": -317.78643798828125, + "logps/rejected": -343.6407470703125, + "loss": 0.558, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.21526435017585754, + "rewards/margins": 0.3885418772697449, + "rewards/rejected": -0.60380619764328, + "step": 6710 + }, + { + "epoch": 0.88, + "learning_rate": 2.1849603805504328e-07, + "logits/chosen": -2.3497023582458496, + "logits/rejected": -2.25719952583313, + "logps/chosen": -319.1162109375, + "logps/rejected": -331.0961608886719, + "loss": 0.5468, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.13552118837833405, + "rewards/margins": 0.41164129972457886, + "rewards/rejected": -0.5471624732017517, + "step": 6720 + }, + { + "epoch": 0.88, + "learning_rate": 2.1385052713554066e-07, + "logits/chosen": -2.1155714988708496, + "logits/rejected": -2.0738768577575684, + "logps/chosen": -286.7259521484375, + "logps/rejected": -323.79388427734375, + "loss": 0.5892, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.22641244530677795, + "rewards/margins": 0.31793564558029175, + "rewards/rejected": -0.5443481206893921, + "step": 6730 + }, + { + "epoch": 0.88, + "learning_rate": 2.0925272579274873e-07, + "logits/chosen": -2.2623276710510254, + "logits/rejected": -2.278806686401367, + "logps/chosen": -297.44622802734375, + "logps/rejected": -334.06915283203125, + "loss": 0.5437, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.14218921959400177, + "rewards/margins": 0.4180937707424164, + "rewards/rejected": -0.5602830052375793, + "step": 6740 + }, + { + "epoch": 0.88, + "learning_rate": 2.047027299780302e-07, + "logits/chosen": -2.2847766876220703, + "logits/rejected": -2.237086534500122, + "logps/chosen": -337.4761657714844, + "logps/rejected": -318.3897399902344, + "loss": 0.6143, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.17803582549095154, + "rewards/margins": 0.24196699261665344, + "rewards/rejected": -0.42000284790992737, + "step": 6750 + }, + { + "epoch": 0.88, + "learning_rate": 2.0020063464509492e-07, + "logits/chosen": -2.1941092014312744, + "logits/rejected": -2.1769931316375732, + "logps/chosen": -299.5726013183594, + "logps/rejected": -310.5265197753906, + "loss": 0.5557, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.15381093323230743, + "rewards/margins": 0.3809550404548645, + "rewards/rejected": -0.534765899181366, + "step": 6760 + }, + { + "epoch": 0.89, + "learning_rate": 1.957465337480191e-07, + "logits/chosen": -2.1837964057922363, + "logits/rejected": -2.1292548179626465, + "logps/chosen": -345.9893798828125, + "logps/rejected": -344.40460205078125, + "loss": 0.59, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.2072792500257492, + "rewards/margins": 0.3229365646839142, + "rewards/rejected": -0.530215859413147, + "step": 6770 + }, + { + "epoch": 0.89, + "learning_rate": 1.9134052023928622e-07, + "logits/chosen": -2.2238945960998535, + "logits/rejected": -2.1773924827575684, + "logps/chosen": -284.38330078125, + "logps/rejected": -311.63232421875, + "loss": 0.6043, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.22153577208518982, + "rewards/margins": 0.26465216279029846, + "rewards/rejected": -0.4861879348754883, + "step": 6780 + }, + { + "epoch": 0.89, + "learning_rate": 1.8698268606784392e-07, + "logits/chosen": -2.234797239303589, + "logits/rejected": -2.236541986465454, + "logps/chosen": -274.5113220214844, + "logps/rejected": -309.7568359375, + "loss": 0.5625, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.20977351069450378, + "rewards/margins": 0.41913795471191406, + "rewards/rejected": -0.6289114952087402, + "step": 6790 + }, + { + "epoch": 0.89, + "learning_rate": 1.826731221771866e-07, + "logits/chosen": -2.2946181297302246, + "logits/rejected": -2.1501200199127197, + "logps/chosen": -328.7485046386719, + "logps/rejected": -303.0035095214844, + "loss": 0.5848, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.22563433647155762, + "rewards/margins": 0.35804134607315063, + "rewards/rejected": -0.583675742149353, + "step": 6800 + }, + { + "epoch": 0.89, + "learning_rate": 1.7841191850345967e-07, + "logits/chosen": -2.17826509475708, + "logits/rejected": -2.160841464996338, + "logps/chosen": -316.8827819824219, + "logps/rejected": -346.2579650878906, + "loss": 0.5618, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.17746450006961823, + "rewards/margins": 0.3708667457103729, + "rewards/rejected": -0.5483312606811523, + "step": 6810 + }, + { + "epoch": 0.89, + "learning_rate": 1.7419916397357905e-07, + "logits/chosen": -2.2968761920928955, + "logits/rejected": -2.1889562606811523, + "logps/chosen": -332.37384033203125, + "logps/rejected": -314.2474670410156, + "loss": 0.6169, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.252440482378006, + "rewards/margins": 0.2617506682872772, + "rewards/rejected": -0.5141911506652832, + "step": 6820 + }, + { + "epoch": 0.89, + "learning_rate": 1.700349465033782e-07, + "logits/chosen": -2.2410268783569336, + "logits/rejected": -2.1834325790405273, + "logps/chosen": -312.6735534667969, + "logps/rejected": -321.71875, + "loss": 0.5946, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.23761789500713348, + "rewards/margins": 0.30250900983810425, + "rewards/rejected": -0.5401269197463989, + "step": 6830 + }, + { + "epoch": 0.9, + "learning_rate": 1.6591935299577227e-07, + "logits/chosen": -2.190650224685669, + "logits/rejected": -2.1460845470428467, + "logps/chosen": -296.0951232910156, + "logps/rejected": -275.2724609375, + "loss": 0.5734, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.18607425689697266, + "rewards/margins": 0.3766199052333832, + "rewards/rejected": -0.5626941919326782, + "step": 6840 + }, + { + "epoch": 0.9, + "learning_rate": 1.6185246933894338e-07, + "logits/chosen": -2.3041493892669678, + "logits/rejected": -2.2566075325012207, + "logps/chosen": -303.43121337890625, + "logps/rejected": -318.13983154296875, + "loss": 0.6156, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.20772568881511688, + "rewards/margins": 0.28159254789352417, + "rewards/rejected": -0.48931822180747986, + "step": 6850 + }, + { + "epoch": 0.9, + "learning_rate": 1.5783438040455097e-07, + "logits/chosen": -2.304779529571533, + "logits/rejected": -2.19102144241333, + "logps/chosen": -327.00958251953125, + "logps/rejected": -284.0123596191406, + "loss": 0.5961, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.2807047367095947, + "rewards/margins": 0.28873246908187866, + "rewards/rejected": -0.5694371461868286, + "step": 6860 + }, + { + "epoch": 0.9, + "learning_rate": 1.538651700459576e-07, + "logits/chosen": -2.248351573944092, + "logits/rejected": -2.1588175296783447, + "logps/chosen": -336.6203918457031, + "logps/rejected": -411.03387451171875, + "loss": 0.5578, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.15604709088802338, + "rewards/margins": 0.36909565329551697, + "rewards/rejected": -0.5251427888870239, + "step": 6870 + }, + { + "epoch": 0.9, + "learning_rate": 1.4994492109648151e-07, + "logits/chosen": -2.207669973373413, + "logits/rejected": -2.1326980590820312, + "logps/chosen": -282.05950927734375, + "logps/rejected": -282.82220458984375, + "loss": 0.569, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2813078761100769, + "rewards/margins": 0.37337321043014526, + "rewards/rejected": -0.6546810865402222, + "step": 6880 + }, + { + "epoch": 0.9, + "learning_rate": 1.4607371536766695e-07, + "logits/chosen": -2.261211395263672, + "logits/rejected": -2.1927568912506104, + "logps/chosen": -261.5356750488281, + "logps/rejected": -281.796630859375, + "loss": 0.5787, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.2290767878293991, + "rewards/margins": 0.351195752620697, + "rewards/rejected": -0.5802726149559021, + "step": 6890 + }, + { + "epoch": 0.9, + "learning_rate": 1.4225163364757655e-07, + "logits/chosen": -2.213453769683838, + "logits/rejected": -2.0844531059265137, + "logps/chosen": -322.1564636230469, + "logps/rejected": -316.3904724121094, + "loss": 0.6334, + "rewards/accuracies": 0.550000011920929, + "rewards/chosen": -0.307486891746521, + "rewards/margins": 0.20499345660209656, + "rewards/rejected": -0.5124803781509399, + "step": 6900 + }, + { + "epoch": 0.9, + "learning_rate": 1.3847875569910462e-07, + "logits/chosen": -2.2584121227264404, + "logits/rejected": -2.168458938598633, + "logps/chosen": -301.62115478515625, + "logps/rejected": -313.1435852050781, + "loss": 0.585, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.16885143518447876, + "rewards/margins": 0.32763198018074036, + "rewards/rejected": -0.4964834749698639, + "step": 6910 + }, + { + "epoch": 0.91, + "learning_rate": 1.3475516025831552e-07, + "logits/chosen": -2.291088819503784, + "logits/rejected": -2.2780730724334717, + "logps/chosen": -296.73297119140625, + "logps/rejected": -327.2522888183594, + "loss": 0.5713, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.1908951699733734, + "rewards/margins": 0.3621283173561096, + "rewards/rejected": -0.5530234575271606, + "step": 6920 + }, + { + "epoch": 0.91, + "learning_rate": 1.310809250327974e-07, + "logits/chosen": -2.0948023796081543, + "logits/rejected": -2.055344581604004, + "logps/chosen": -280.61175537109375, + "logps/rejected": -291.4289245605469, + "loss": 0.5726, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.19594351947307587, + "rewards/margins": 0.3705883324146271, + "rewards/rejected": -0.5665318369865417, + "step": 6930 + }, + { + "epoch": 0.91, + "learning_rate": 1.2745612670004153e-07, + "logits/chosen": -2.2199532985687256, + "logits/rejected": -2.161428928375244, + "logps/chosen": -286.59930419921875, + "logps/rejected": -384.32720947265625, + "loss": 0.5807, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.18531376123428345, + "rewards/margins": 0.3441842794418335, + "rewards/rejected": -0.5294980406761169, + "step": 6940 + }, + { + "epoch": 0.91, + "learning_rate": 1.2388084090584395e-07, + "logits/chosen": -2.2351741790771484, + "logits/rejected": -2.1834254264831543, + "logps/chosen": -285.2957458496094, + "logps/rejected": -326.4273986816406, + "loss": 0.6217, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.3242362439632416, + "rewards/margins": 0.27158844470977783, + "rewards/rejected": -0.5958247780799866, + "step": 6950 + }, + { + "epoch": 0.91, + "learning_rate": 1.2035514226272305e-07, + "logits/chosen": -2.201180934906006, + "logits/rejected": -2.1012425422668457, + "logps/chosen": -273.3121032714844, + "logps/rejected": -305.42120361328125, + "loss": 0.579, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.20711271464824677, + "rewards/margins": 0.35867878794670105, + "rewards/rejected": -0.5657914876937866, + "step": 6960 + }, + { + "epoch": 0.91, + "learning_rate": 1.1687910434836607e-07, + "logits/chosen": -2.195547580718994, + "logits/rejected": -2.1885972023010254, + "logps/chosen": -290.4573669433594, + "logps/rejected": -315.23175048828125, + "loss": 0.5884, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.23133409023284912, + "rewards/margins": 0.3339126706123352, + "rewards/rejected": -0.5652468204498291, + "step": 6970 + }, + { + "epoch": 0.91, + "learning_rate": 1.1345279970409128e-07, + "logits/chosen": -2.1727585792541504, + "logits/rejected": -2.179666519165039, + "logps/chosen": -260.8642883300781, + "logps/rejected": -298.676513671875, + "loss": 0.652, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.23396344482898712, + "rewards/margins": 0.17201301455497742, + "rewards/rejected": -0.40597644448280334, + "step": 6980 + }, + { + "epoch": 0.91, + "learning_rate": 1.1007629983333629e-07, + "logits/chosen": -2.186631679534912, + "logits/rejected": -2.111750364303589, + "logps/chosen": -292.9834289550781, + "logps/rejected": -299.646240234375, + "loss": 0.5754, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.18006864190101624, + "rewards/margins": 0.340657502412796, + "rewards/rejected": -0.5207261443138123, + "step": 6990 + }, + { + "epoch": 0.92, + "learning_rate": 1.067496752001626e-07, + "logits/chosen": -2.21193790435791, + "logits/rejected": -2.147188663482666, + "logps/chosen": -309.00604248046875, + "logps/rejected": -298.1776123046875, + "loss": 0.5796, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.2621694505214691, + "rewards/margins": 0.3532871603965759, + "rewards/rejected": -0.6154566407203674, + "step": 7000 + }, + { + "epoch": 0.92, + "eval_logits/chosen": -2.2260730266571045, + "eval_logits/rejected": -2.1487460136413574, + "eval_logps/chosen": -313.4948425292969, + "eval_logps/rejected": -320.43585205078125, + "eval_loss": 0.5940207242965698, + "eval_rewards/accuracies": 0.6940000057220459, + "eval_rewards/chosen": -0.21513542532920837, + "eval_rewards/margins": 0.30914872884750366, + "eval_rewards/rejected": -0.5242841243743896, + "eval_runtime": 1172.4889, + "eval_samples_per_second": 1.706, + "eval_steps_per_second": 0.213, + "step": 7000 + }, + { + "epoch": 0.92, + "learning_rate": 1.0347299522778909e-07, + "logits/chosen": -2.280367374420166, + "logits/rejected": -2.2171788215637207, + "logps/chosen": -338.8251953125, + "logps/rejected": -333.30853271484375, + "loss": 0.557, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.21220973134040833, + "rewards/margins": 0.3927744925022125, + "rewards/rejected": -0.6049842238426208, + "step": 7010 + }, + { + "epoch": 0.92, + "learning_rate": 1.0024632829713971e-07, + "logits/chosen": -2.219359874725342, + "logits/rejected": -2.2551379203796387, + "logps/chosen": -278.66119384765625, + "logps/rejected": -319.22418212890625, + "loss": 0.6177, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.2532063126564026, + "rewards/margins": 0.2454725205898285, + "rewards/rejected": -0.49867886304855347, + "step": 7020 + }, + { + "epoch": 0.92, + "learning_rate": 9.706974174541889e-08, + "logits/chosen": -2.2002689838409424, + "logits/rejected": -2.1216237545013428, + "logps/chosen": -328.0563049316406, + "logps/rejected": -313.7506103515625, + "loss": 0.5491, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.1435789167881012, + "rewards/margins": 0.4156878888607025, + "rewards/rejected": -0.5592668652534485, + "step": 7030 + }, + { + "epoch": 0.92, + "learning_rate": 9.39433018647043e-08, + "logits/chosen": -2.3666234016418457, + "logits/rejected": -2.3080554008483887, + "logps/chosen": -304.62506103515625, + "logps/rejected": -350.1346435546875, + "loss": 0.5554, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.13446173071861267, + "rewards/margins": 0.3920600116252899, + "rewards/rejected": -0.5265216827392578, + "step": 7040 + }, + { + "epoch": 0.92, + "learning_rate": 9.086707390056543e-08, + "logits/chosen": -2.3087592124938965, + "logits/rejected": -2.227196216583252, + "logps/chosen": -320.02374267578125, + "logps/rejected": -315.66668701171875, + "loss": 0.6072, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.18638381361961365, + "rewards/margins": 0.2937110960483551, + "rewards/rejected": -0.48009490966796875, + "step": 7050 + }, + { + "epoch": 0.92, + "learning_rate": 8.784112205070083e-08, + "logits/chosen": -2.2482821941375732, + "logits/rejected": -2.247669219970703, + "logps/chosen": -281.7313537597656, + "logps/rejected": -324.09564208984375, + "loss": 0.6207, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.22902318835258484, + "rewards/margins": 0.2263072431087494, + "rewards/rejected": -0.45533037185668945, + "step": 7060 + }, + { + "epoch": 0.93, + "learning_rate": 8.486550946359779e-08, + "logits/chosen": -2.293630599975586, + "logits/rejected": -2.1961753368377686, + "logps/chosen": -329.4664611816406, + "logps/rejected": -332.4908447265625, + "loss": 0.6724, + "rewards/accuracies": 0.574999988079071, + "rewards/chosen": -0.2469293624162674, + "rewards/margins": 0.12247602641582489, + "rewards/rejected": -0.3694053888320923, + "step": 7070 + }, + { + "epoch": 0.93, + "learning_rate": 8.194029823721556e-08, + "logits/chosen": -2.2293028831481934, + "logits/rejected": -2.113828659057617, + "logps/chosen": -293.24163818359375, + "logps/rejected": -302.811279296875, + "loss": 0.5748, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.19101731479167938, + "rewards/margins": 0.3309341371059418, + "rewards/rejected": -0.5219514966011047, + "step": 7080 + }, + { + "epoch": 0.93, + "learning_rate": 7.906554941768896e-08, + "logits/chosen": -2.2355294227600098, + "logits/rejected": -2.204115390777588, + "logps/chosen": -277.62774658203125, + "logps/rejected": -302.0038146972656, + "loss": 0.6123, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.2697092890739441, + "rewards/margins": 0.25456854701042175, + "rewards/rejected": -0.5242778062820435, + "step": 7090 + }, + { + "epoch": 0.93, + "learning_rate": 7.624132299805575e-08, + "logits/chosen": -2.2374064922332764, + "logits/rejected": -2.2581026554107666, + "logps/chosen": -345.123046875, + "logps/rejected": -358.58941650390625, + "loss": 0.553, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.20718857645988464, + "rewards/margins": 0.4173831045627594, + "rewards/rejected": -0.624571681022644, + "step": 7100 + }, + { + "epoch": 0.93, + "learning_rate": 7.346767791700127e-08, + "logits/chosen": -2.214167833328247, + "logits/rejected": -2.1587700843811035, + "logps/chosen": -344.684814453125, + "logps/rejected": -337.58135986328125, + "loss": 0.5272, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.1494571715593338, + "rewards/margins": 0.465663343667984, + "rewards/rejected": -0.615120530128479, + "step": 7110 + }, + { + "epoch": 0.93, + "learning_rate": 7.07446720576327e-08, + "logits/chosen": -2.194397449493408, + "logits/rejected": -2.1149685382843018, + "logps/chosen": -282.08489990234375, + "logps/rejected": -291.74896240234375, + "loss": 0.6077, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.27410179376602173, + "rewards/margins": 0.2866399884223938, + "rewards/rejected": -0.5607417821884155, + "step": 7120 + }, + { + "epoch": 0.93, + "learning_rate": 6.807236224626701e-08, + "logits/chosen": -2.2754781246185303, + "logits/rejected": -2.2205283641815186, + "logps/chosen": -315.891845703125, + "logps/rejected": -316.5805969238281, + "loss": 0.6225, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.20273521542549133, + "rewards/margins": 0.2261861264705658, + "rewards/rejected": -0.42892131209373474, + "step": 7130 + }, + { + "epoch": 0.93, + "learning_rate": 6.545080425124888e-08, + "logits/chosen": -2.2018957138061523, + "logits/rejected": -2.1935436725616455, + "logps/chosen": -297.48065185546875, + "logps/rejected": -327.8349609375, + "loss": 0.6065, + "rewards/accuracies": 0.762499988079071, + "rewards/chosen": -0.18633002042770386, + "rewards/margins": 0.31622928380966187, + "rewards/rejected": -0.5025593042373657, + "step": 7140 + }, + { + "epoch": 0.94, + "learning_rate": 6.288005278178382e-08, + "logits/chosen": -2.3038198947906494, + "logits/rejected": -2.2294774055480957, + "logps/chosen": -314.5107421875, + "logps/rejected": -312.3115234375, + "loss": 0.6334, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.17556893825531006, + "rewards/margins": 0.23685109615325928, + "rewards/rejected": -0.41242003440856934, + "step": 7150 + }, + { + "epoch": 0.94, + "learning_rate": 6.036016148679825e-08, + "logits/chosen": -2.3269429206848145, + "logits/rejected": -2.221116065979004, + "logps/chosen": -311.7928161621094, + "logps/rejected": -280.10498046875, + "loss": 0.5777, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.2013367861509323, + "rewards/margins": 0.3149644136428833, + "rewards/rejected": -0.5163012742996216, + "step": 7160 + }, + { + "epoch": 0.94, + "learning_rate": 5.7891182953819235e-08, + "logits/chosen": -2.309523105621338, + "logits/rejected": -2.2230565547943115, + "logps/chosen": -264.33099365234375, + "logps/rejected": -320.1153869628906, + "loss": 0.5605, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.16378042101860046, + "rewards/margins": 0.37534573674201965, + "rewards/rejected": -0.5391262173652649, + "step": 7170 + }, + { + "epoch": 0.94, + "learning_rate": 5.547316870787689e-08, + "logits/chosen": -2.3665127754211426, + "logits/rejected": -2.2079923152923584, + "logps/chosen": -327.3066101074219, + "logps/rejected": -321.58447265625, + "loss": 0.5633, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.20733241736888885, + "rewards/margins": 0.3998747169971466, + "rewards/rejected": -0.607207179069519, + "step": 7180 + }, + { + "epoch": 0.94, + "learning_rate": 5.310616921042927e-08, + "logits/chosen": -2.3350491523742676, + "logits/rejected": -2.193040370941162, + "logps/chosen": -350.3499450683594, + "logps/rejected": -335.13385009765625, + "loss": 0.5699, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.2007473260164261, + "rewards/margins": 0.3822272717952728, + "rewards/rejected": -0.5829745531082153, + "step": 7190 + }, + { + "epoch": 0.94, + "learning_rate": 5.079023385830939e-08, + "logits/chosen": -2.1871137619018555, + "logits/rejected": -2.092442274093628, + "logps/chosen": -355.10369873046875, + "logps/rejected": -324.44525146484375, + "loss": 0.5941, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.27507293224334717, + "rewards/margins": 0.32030484080314636, + "rewards/rejected": -0.5953778028488159, + "step": 7200 + }, + { + "epoch": 0.94, + "learning_rate": 4.8525410982695476e-08, + "logits/chosen": -2.265181541442871, + "logits/rejected": -2.1195521354675293, + "logps/chosen": -343.2086486816406, + "logps/rejected": -350.94854736328125, + "loss": 0.5739, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.15476097166538239, + "rewards/margins": 0.37012749910354614, + "rewards/rejected": -0.5248885154724121, + "step": 7210 + }, + { + "epoch": 0.94, + "learning_rate": 4.6311747848099e-08, + "logits/chosen": -2.2479965686798096, + "logits/rejected": -2.113255023956299, + "logps/chosen": -281.4270324707031, + "logps/rejected": -274.7620544433594, + "loss": 0.6068, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.237498477101326, + "rewards/margins": 0.29350370168685913, + "rewards/rejected": -0.5310021638870239, + "step": 7220 + }, + { + "epoch": 0.95, + "learning_rate": 4.4149290651382405e-08, + "logits/chosen": -2.2208240032196045, + "logits/rejected": -2.2298367023468018, + "logps/chosen": -249.2593994140625, + "logps/rejected": -297.3638916015625, + "loss": 0.6534, + "rewards/accuracies": 0.574999988079071, + "rewards/chosen": -0.23134097456932068, + "rewards/margins": 0.19426584243774414, + "rewards/rejected": -0.4256068170070648, + "step": 7230 + }, + { + "epoch": 0.95, + "learning_rate": 4.203808452079211e-08, + "logits/chosen": -2.2425343990325928, + "logits/rejected": -2.193664073944092, + "logps/chosen": -315.44464111328125, + "logps/rejected": -329.19805908203125, + "loss": 0.5699, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.1318148672580719, + "rewards/margins": 0.3500699996948242, + "rewards/rejected": -0.48188480734825134, + "step": 7240 + }, + { + "epoch": 0.95, + "learning_rate": 3.9978173515018427e-08, + "logits/chosen": -2.3699262142181396, + "logits/rejected": -2.2451012134552, + "logps/chosen": -357.103271484375, + "logps/rejected": -330.6025695800781, + "loss": 0.5719, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.20406512916088104, + "rewards/margins": 0.3894132673740387, + "rewards/rejected": -0.5934784412384033, + "step": 7250 + }, + { + "epoch": 0.95, + "learning_rate": 3.7969600622274614e-08, + "logits/chosen": -2.3619580268859863, + "logits/rejected": -2.3217098712921143, + "logps/chosen": -355.99786376953125, + "logps/rejected": -307.1109924316406, + "loss": 0.5836, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.1927039921283722, + "rewards/margins": 0.3221895396709442, + "rewards/rejected": -0.5148935317993164, + "step": 7260 + }, + { + "epoch": 0.95, + "learning_rate": 3.601240775940151e-08, + "logits/chosen": -2.2095894813537598, + "logits/rejected": -2.203479051589966, + "logps/chosen": -285.00537109375, + "logps/rejected": -307.1541748046875, + "loss": 0.5907, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.22207343578338623, + "rewards/margins": 0.3408282399177551, + "rewards/rejected": -0.5629016160964966, + "step": 7270 + }, + { + "epoch": 0.95, + "learning_rate": 3.410663577099071e-08, + "logits/chosen": -2.352631092071533, + "logits/rejected": -2.2568843364715576, + "logps/chosen": -367.5334167480469, + "logps/rejected": -351.14337158203125, + "loss": 0.6094, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.2421077936887741, + "rewards/margins": 0.2984387278556824, + "rewards/rejected": -0.5405465364456177, + "step": 7280 + }, + { + "epoch": 0.95, + "learning_rate": 3.2252324428534986e-08, + "logits/chosen": -2.25034499168396, + "logits/rejected": -2.159763813018799, + "logps/chosen": -340.13385009765625, + "logps/rejected": -316.416259765625, + "loss": 0.5969, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.12953811883926392, + "rewards/margins": 0.31046968698501587, + "rewards/rejected": -0.4400077760219574, + "step": 7290 + }, + { + "epoch": 0.96, + "learning_rate": 3.0449512429594486e-08, + "logits/chosen": -2.182349681854248, + "logits/rejected": -2.095487356185913, + "logps/chosen": -312.82342529296875, + "logps/rejected": -324.8750915527344, + "loss": 0.5346, + "rewards/accuracies": 0.7875000238418579, + "rewards/chosen": -0.13655495643615723, + "rewards/margins": 0.45154494047164917, + "rewards/rejected": -0.5880998373031616, + "step": 7300 + }, + { + "epoch": 0.96, + "learning_rate": 2.8698237396992956e-08, + "logits/chosen": -2.3066210746765137, + "logits/rejected": -2.314364194869995, + "logps/chosen": -305.68328857421875, + "logps/rejected": -333.3356018066406, + "loss": 0.5677, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.22517363727092743, + "rewards/margins": 0.3729446232318878, + "rewards/rejected": -0.598118245601654, + "step": 7310 + }, + { + "epoch": 0.96, + "learning_rate": 2.6998535878030584e-08, + "logits/chosen": -2.2450592517852783, + "logits/rejected": -2.1315126419067383, + "logps/chosen": -239.87503051757812, + "logps/rejected": -227.39389038085938, + "loss": 0.5835, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.17774894833564758, + "rewards/margins": 0.33111661672592163, + "rewards/rejected": -0.5088655352592468, + "step": 7320 + }, + { + "epoch": 0.96, + "learning_rate": 2.535044334372072e-08, + "logits/chosen": -2.214465618133545, + "logits/rejected": -2.114131212234497, + "logps/chosen": -373.511962890625, + "logps/rejected": -330.72393798828125, + "loss": 0.5968, + "rewards/accuracies": 0.6499999761581421, + "rewards/chosen": -0.21916863322257996, + "rewards/margins": 0.3310352861881256, + "rewards/rejected": -0.5502039194107056, + "step": 7330 + }, + { + "epoch": 0.96, + "learning_rate": 2.3753994188051853e-08, + "logits/chosen": -2.2956297397613525, + "logits/rejected": -2.1181273460388184, + "logps/chosen": -329.48443603515625, + "logps/rejected": -315.2552795410156, + "loss": 0.6199, + "rewards/accuracies": 0.637499988079071, + "rewards/chosen": -0.2757745385169983, + "rewards/margins": 0.2727423906326294, + "rewards/rejected": -0.5485169291496277, + "step": 7340 + }, + { + "epoch": 0.96, + "learning_rate": 2.220922172726764e-08, + "logits/chosen": -2.2476680278778076, + "logits/rejected": -2.197944164276123, + "logps/chosen": -306.67291259765625, + "logps/rejected": -296.61871337890625, + "loss": 0.6074, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.19910797476768494, + "rewards/margins": 0.2877027094364166, + "rewards/rejected": -0.4868106245994568, + "step": 7350 + }, + { + "epoch": 0.96, + "learning_rate": 2.071615819917244e-08, + "logits/chosen": -2.3216476440429688, + "logits/rejected": -2.3249900341033936, + "logps/chosen": -312.84942626953125, + "logps/rejected": -339.9403076171875, + "loss": 0.5956, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.19821664690971375, + "rewards/margins": 0.271074503660202, + "rewards/rejected": -0.4692911207675934, + "step": 7360 + }, + { + "epoch": 0.96, + "learning_rate": 1.9274834762459393e-08, + "logits/chosen": -2.2404820919036865, + "logits/rejected": -2.132887363433838, + "logps/chosen": -333.1466064453125, + "logps/rejected": -303.07623291015625, + "loss": 0.5646, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.22520729899406433, + "rewards/margins": 0.3837920129299164, + "rewards/rejected": -0.6089993715286255, + "step": 7370 + }, + { + "epoch": 0.97, + "learning_rate": 1.7885281496058947e-08, + "logits/chosen": -2.348447322845459, + "logits/rejected": -2.2594656944274902, + "logps/chosen": -327.52490234375, + "logps/rejected": -389.01995849609375, + "loss": 0.5819, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.22132794559001923, + "rewards/margins": 0.32333776354789734, + "rewards/rejected": -0.5446656942367554, + "step": 7380 + }, + { + "epoch": 0.97, + "learning_rate": 1.654752739851134e-08, + "logits/chosen": -2.352313280105591, + "logits/rejected": -2.190316677093506, + "logps/chosen": -320.6366271972656, + "logps/rejected": -305.8868408203125, + "loss": 0.5708, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.18807712197303772, + "rewards/margins": 0.3853645324707031, + "rewards/rejected": -0.5734416842460632, + "step": 7390 + }, + { + "epoch": 0.97, + "learning_rate": 1.526160038736235e-08, + "logits/chosen": -2.3725545406341553, + "logits/rejected": -2.2139534950256348, + "logps/chosen": -362.9305419921875, + "logps/rejected": -325.1018371582031, + "loss": 0.5774, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.149748295545578, + "rewards/margins": 0.3587859570980072, + "rewards/rejected": -0.5085342526435852, + "step": 7400 + }, + { + "epoch": 0.97, + "learning_rate": 1.402752729857959e-08, + "logits/chosen": -2.2219464778900146, + "logits/rejected": -2.208585262298584, + "logps/chosen": -288.1156005859375, + "logps/rejected": -335.0963439941406, + "loss": 0.5739, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.15629379451274872, + "rewards/margins": 0.34737879037857056, + "rewards/rejected": -0.5036724805831909, + "step": 7410 + }, + { + "epoch": 0.97, + "learning_rate": 1.2845333885992683e-08, + "logits/chosen": -2.2705419063568115, + "logits/rejected": -2.1903064250946045, + "logps/chosen": -314.6650390625, + "logps/rejected": -300.46478271484375, + "loss": 0.5884, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.1653841882944107, + "rewards/margins": 0.3278418183326721, + "rewards/rejected": -0.49322596192359924, + "step": 7420 + }, + { + "epoch": 0.97, + "learning_rate": 1.171504482075675e-08, + "logits/chosen": -2.343583583831787, + "logits/rejected": -2.254533290863037, + "logps/chosen": -351.45062255859375, + "logps/rejected": -369.24066162109375, + "loss": 0.5625, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.1713610142469406, + "rewards/margins": 0.397853285074234, + "rewards/rejected": -0.5692143440246582, + "step": 7430 + }, + { + "epoch": 0.97, + "learning_rate": 1.0636683690836147e-08, + "logits/chosen": -2.3275885581970215, + "logits/rejected": -2.2382407188415527, + "logps/chosen": -296.9913024902344, + "logps/rejected": -306.6391906738281, + "loss": 0.5994, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.17596189677715302, + "rewards/margins": 0.29039308428764343, + "rewards/rejected": -0.46635502576828003, + "step": 7440 + }, + { + "epoch": 0.97, + "learning_rate": 9.610273000513203e-09, + "logits/chosen": -2.3288352489471436, + "logits/rejected": -2.228902578353882, + "logps/chosen": -351.21826171875, + "logps/rejected": -350.92034912109375, + "loss": 0.6418, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.2834011912345886, + "rewards/margins": 0.19712185859680176, + "rewards/rejected": -0.480523020029068, + "step": 7450 + }, + { + "epoch": 0.98, + "learning_rate": 8.635834169918312e-09, + "logits/chosen": -2.24247407913208, + "logits/rejected": -2.2252285480499268, + "logps/chosen": -343.86529541015625, + "logps/rejected": -359.72381591796875, + "loss": 0.6262, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.2396010458469391, + "rewards/margins": 0.2377980500459671, + "rewards/rejected": -0.4773990511894226, + "step": 7460 + }, + { + "epoch": 0.98, + "learning_rate": 7.713387534582506e-09, + "logits/chosen": -2.22129487991333, + "logits/rejected": -2.1426587104797363, + "logps/chosen": -288.47857666015625, + "logps/rejected": -334.2923889160156, + "loss": 0.5616, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.15719589591026306, + "rewards/margins": 0.3799470365047455, + "rewards/rejected": -0.5371429920196533, + "step": 7470 + }, + { + "epoch": 0.98, + "learning_rate": 6.84295234501392e-09, + "logits/chosen": -2.261748790740967, + "logits/rejected": -2.1660752296447754, + "logps/chosen": -312.7343444824219, + "logps/rejected": -301.38897705078125, + "loss": 0.6069, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.220415398478508, + "rewards/margins": 0.27050694823265076, + "rewards/rejected": -0.49092239141464233, + "step": 7480 + }, + { + "epoch": 0.98, + "learning_rate": 6.024546766295325e-09, + "logits/chosen": -2.2253947257995605, + "logits/rejected": -2.0965373516082764, + "logps/chosen": -341.40618896484375, + "logps/rejected": -308.0964050292969, + "loss": 0.5498, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.17060282826423645, + "rewards/margins": 0.4203481078147888, + "rewards/rejected": -0.5909509658813477, + "step": 7490 + }, + { + "epoch": 0.98, + "learning_rate": 5.2581878777049895e-09, + "logits/chosen": -2.297600030899048, + "logits/rejected": -2.2106003761291504, + "logps/chosen": -320.5754089355469, + "logps/rejected": -340.50433349609375, + "loss": 0.5427, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.16367055475711823, + "rewards/margins": 0.4177732467651367, + "rewards/rejected": -0.5814438462257385, + "step": 7500 + }, + { + "epoch": 0.98, + "learning_rate": 4.543891672361411e-09, + "logits/chosen": -2.223428726196289, + "logits/rejected": -2.1965279579162598, + "logps/chosen": -319.80853271484375, + "logps/rejected": -334.6192321777344, + "loss": 0.6028, + "rewards/accuracies": 0.6625000238418579, + "rewards/chosen": -0.21012039482593536, + "rewards/margins": 0.2811013162136078, + "rewards/rejected": -0.49122172594070435, + "step": 7510 + }, + { + "epoch": 0.98, + "learning_rate": 3.881673056887747e-09, + "logits/chosen": -2.3167202472686768, + "logits/rejected": -2.2619433403015137, + "logps/chosen": -325.6050720214844, + "logps/rejected": -329.0150451660156, + "loss": 0.5812, + "rewards/accuracies": 0.7250000238418579, + "rewards/chosen": -0.18179884552955627, + "rewards/margins": 0.3695027232170105, + "rewards/rejected": -0.5513015985488892, + "step": 7520 + }, + { + "epoch": 0.99, + "learning_rate": 3.2715458511023425e-09, + "logits/chosen": -2.24223256111145, + "logits/rejected": -2.1757986545562744, + "logps/chosen": -318.0364685058594, + "logps/rejected": -343.64984130859375, + "loss": 0.6106, + "rewards/accuracies": 0.7124999761581421, + "rewards/chosen": -0.2036343812942505, + "rewards/margins": 0.2790573239326477, + "rewards/rejected": -0.48269176483154297, + "step": 7530 + }, + { + "epoch": 0.99, + "learning_rate": 2.7135227877289617e-09, + "logits/chosen": -2.2901463508605957, + "logits/rejected": -2.2843945026397705, + "logps/chosen": -294.41400146484375, + "logps/rejected": -338.856689453125, + "loss": 0.5917, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.15333044528961182, + "rewards/margins": 0.30261334776878357, + "rewards/rejected": -0.4559438228607178, + "step": 7540 + }, + { + "epoch": 0.99, + "learning_rate": 2.2076155121328326e-09, + "logits/chosen": -2.155514717102051, + "logits/rejected": -2.0566611289978027, + "logps/chosen": -282.373046875, + "logps/rejected": -292.1627502441406, + "loss": 0.5583, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.16147629916667938, + "rewards/margins": 0.39047032594680786, + "rewards/rejected": -0.5519465804100037, + "step": 7550 + }, + { + "epoch": 0.99, + "learning_rate": 1.7538345820755641e-09, + "logits/chosen": -2.2953784465789795, + "logits/rejected": -2.1645030975341797, + "logps/chosen": -347.6931457519531, + "logps/rejected": -315.5379638671875, + "loss": 0.5725, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.19179323315620422, + "rewards/margins": 0.33790022134780884, + "rewards/rejected": -0.5296934843063354, + "step": 7560 + }, + { + "epoch": 0.99, + "learning_rate": 1.3521894674961567e-09, + "logits/chosen": -2.23071551322937, + "logits/rejected": -2.198906421661377, + "logps/chosen": -329.29095458984375, + "logps/rejected": -332.01507568359375, + "loss": 0.6182, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.2797185778617859, + "rewards/margins": 0.24416124820709229, + "rewards/rejected": -0.5238798260688782, + "step": 7570 + }, + { + "epoch": 0.99, + "learning_rate": 1.0026885503131023e-09, + "logits/chosen": -2.2943203449249268, + "logits/rejected": -2.2038662433624268, + "logps/chosen": -325.7265930175781, + "logps/rejected": -290.61712646484375, + "loss": 0.5624, + "rewards/accuracies": 0.675000011920929, + "rewards/chosen": -0.28593507409095764, + "rewards/margins": 0.3805255591869354, + "rewards/rejected": -0.6664606332778931, + "step": 7580 + }, + { + "epoch": 0.99, + "learning_rate": 7.053391242492491e-10, + "logits/chosen": -2.140082836151123, + "logits/rejected": -2.0679380893707275, + "logps/chosen": -318.735107421875, + "logps/rejected": -327.1845397949219, + "loss": 0.6501, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.3730572760105133, + "rewards/margins": 0.20047850906848907, + "rewards/rejected": -0.5735357999801636, + "step": 7590 + }, + { + "epoch": 0.99, + "learning_rate": 4.6014739467997725e-10, + "logits/chosen": -2.221858501434326, + "logits/rejected": -2.188326358795166, + "logps/chosen": -307.19915771484375, + "logps/rejected": -311.16961669921875, + "loss": 0.6322, + "rewards/accuracies": 0.612500011920929, + "rewards/chosen": -0.2230357676744461, + "rewards/margins": 0.23147812485694885, + "rewards/rejected": -0.45451393723487854, + "step": 7600 + }, + { + "epoch": 1.0, + "learning_rate": 2.671184785033032e-10, + "logits/chosen": -2.3816375732421875, + "logits/rejected": -2.2962653636932373, + "logps/chosen": -336.73284912109375, + "logps/rejected": -351.965576171875, + "loss": 0.5929, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.20788922905921936, + "rewards/margins": 0.30553072690963745, + "rewards/rejected": -0.5134199857711792, + "step": 7610 + }, + { + "epoch": 1.0, + "learning_rate": 1.2625640403302054e-10, + "logits/chosen": -2.318190813064575, + "logits/rejected": -2.2393438816070557, + "logps/chosen": -313.6557922363281, + "logps/rejected": -356.33367919921875, + "loss": 0.5491, + "rewards/accuracies": 0.7749999761581421, + "rewards/chosen": -0.13495388627052307, + "rewards/margins": 0.42312994599342346, + "rewards/rejected": -0.5580838322639465, + "step": 7620 + }, + { + "epoch": 1.0, + "learning_rate": 3.756411091515588e-11, + "logits/chosen": -2.2948296070098877, + "logits/rejected": -2.2274951934814453, + "logps/chosen": -351.2091064453125, + "logps/rejected": -353.2593688964844, + "loss": 0.5941, + "rewards/accuracies": 0.699999988079071, + "rewards/chosen": -0.17898276448249817, + "rewards/margins": 0.32828378677368164, + "rewards/rejected": -0.5072665810585022, + "step": 7630 + }, + { + "epoch": 1.0, + "learning_rate": 1.0434500657963143e-12, + "logits/chosen": -2.1720404624938965, + "logits/rejected": -2.13431715965271, + "logps/chosen": -302.1468505859375, + "logps/rejected": -350.88897705078125, + "loss": 0.5555, + "rewards/accuracies": 0.737500011920929, + "rewards/chosen": -0.21987417340278625, + "rewards/margins": 0.3745797574520111, + "rewards/rejected": -0.5944539308547974, + "step": 7640 + }, + { + "epoch": 1.0, + "step": 7642, + "total_flos": 0.0, + "train_loss": 0.6084560017191406, + "train_runtime": 59476.7244, + "train_samples_per_second": 1.028, + "train_steps_per_second": 0.128 + } + ], + "logging_steps": 10, + "max_steps": 7642, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 500, + "total_flos": 0.0, + "train_batch_size": 4, + "trial_name": null, + "trial_params": null +}