phi-2-gpo-renew2-b0.01-log-i0 / trainer_state.json
BraylonDash's picture
Model save
dca0b81 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 100,
"global_step": 3821,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.3054830287206268e-08,
"logits/chosen": 0.9550814628601074,
"logits/rejected": 1.0664727687835693,
"logps/chosen": -190.47879028320312,
"logps/rejected": -177.6958770751953,
"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": 1.3054830287206266e-07,
"logits/chosen": 1.0215020179748535,
"logits/rejected": 1.073843240737915,
"logps/chosen": -277.8812561035156,
"logps/rejected": -268.32220458984375,
"loss": 0.6931,
"rewards/accuracies": 0.4097222089767456,
"rewards/chosen": 0.00012883776798844337,
"rewards/margins": 0.0005512596690095961,
"rewards/rejected": -0.00042242190102115273,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 2.610966057441253e-07,
"logits/chosen": 1.054603099822998,
"logits/rejected": 1.0355112552642822,
"logps/chosen": -258.0608215332031,
"logps/rejected": -219.5281219482422,
"loss": 0.6932,
"rewards/accuracies": 0.40625,
"rewards/chosen": -0.0007742593297734857,
"rewards/margins": -0.0006406827596947551,
"rewards/rejected": -0.00013357654097490013,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 3.9164490861618804e-07,
"logits/chosen": 0.9782761335372925,
"logits/rejected": 0.9957435727119446,
"logps/chosen": -234.4627685546875,
"logps/rejected": -216.38687133789062,
"loss": 0.6932,
"rewards/accuracies": 0.48124998807907104,
"rewards/chosen": -0.0007474010926671326,
"rewards/margins": 0.00048109880299307406,
"rewards/rejected": -0.0012284999247640371,
"step": 30
},
{
"epoch": 0.01,
"learning_rate": 5.221932114882506e-07,
"logits/chosen": 1.060667634010315,
"logits/rejected": 1.0616825819015503,
"logps/chosen": -269.3723449707031,
"logps/rejected": -236.52392578125,
"loss": 0.6931,
"rewards/accuracies": 0.4312500059604645,
"rewards/chosen": -0.0005697375163435936,
"rewards/margins": -0.00030056110699661076,
"rewards/rejected": -0.0002691763802431524,
"step": 40
},
{
"epoch": 0.01,
"learning_rate": 6.527415143603135e-07,
"logits/chosen": 1.0123419761657715,
"logits/rejected": 1.0498076677322388,
"logps/chosen": -245.165771484375,
"logps/rejected": -241.9037322998047,
"loss": 0.6932,
"rewards/accuracies": 0.4937500059604645,
"rewards/chosen": -0.0007764647016301751,
"rewards/margins": -0.00020735012367367744,
"rewards/rejected": -0.0005691145197488368,
"step": 50
},
{
"epoch": 0.02,
"learning_rate": 7.832898172323761e-07,
"logits/chosen": 0.9758418202400208,
"logits/rejected": 1.0930532217025757,
"logps/chosen": -283.73480224609375,
"logps/rejected": -234.12576293945312,
"loss": 0.6932,
"rewards/accuracies": 0.46875,
"rewards/chosen": -4.083226667717099e-05,
"rewards/margins": -0.00016547185077797621,
"rewards/rejected": 0.00012463955499697477,
"step": 60
},
{
"epoch": 0.02,
"learning_rate": 9.138381201044387e-07,
"logits/chosen": 1.005788803100586,
"logits/rejected": 1.0820422172546387,
"logps/chosen": -271.99468994140625,
"logps/rejected": -231.10446166992188,
"loss": 0.6931,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": 4.1856186726363376e-05,
"rewards/margins": 0.0007076783804222941,
"rewards/rejected": -0.0006658221827819943,
"step": 70
},
{
"epoch": 0.02,
"learning_rate": 1.0443864229765013e-06,
"logits/chosen": 1.0230119228363037,
"logits/rejected": 1.0622894763946533,
"logps/chosen": -283.88946533203125,
"logps/rejected": -261.6778869628906,
"loss": 0.6932,
"rewards/accuracies": 0.4437499940395355,
"rewards/chosen": -0.000257449341006577,
"rewards/margins": -8.570156205678359e-05,
"rewards/rejected": -0.00017174780077766627,
"step": 80
},
{
"epoch": 0.02,
"learning_rate": 1.1749347258485642e-06,
"logits/chosen": 1.043121337890625,
"logits/rejected": 1.0927269458770752,
"logps/chosen": -278.48974609375,
"logps/rejected": -235.72775268554688,
"loss": 0.6932,
"rewards/accuracies": 0.3687500059604645,
"rewards/chosen": -0.0013169237645342946,
"rewards/margins": -0.0013213430065661669,
"rewards/rejected": 4.419172000780236e-06,
"step": 90
},
{
"epoch": 0.03,
"learning_rate": 1.305483028720627e-06,
"logits/chosen": 0.991162121295929,
"logits/rejected": 1.0666497945785522,
"logps/chosen": -237.25302124023438,
"logps/rejected": -218.53671264648438,
"loss": 0.6931,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": 0.00023656387929804623,
"rewards/margins": 0.0011323514627292752,
"rewards/rejected": -0.0008957876125350595,
"step": 100
},
{
"epoch": 0.03,
"eval_logits/chosen": 0.9709606766700745,
"eval_logits/rejected": 1.0631499290466309,
"eval_logps/chosen": -277.5757751464844,
"eval_logps/rejected": -243.87449645996094,
"eval_loss": 0.6931320428848267,
"eval_rewards/accuracies": 0.4514999985694885,
"eval_rewards/chosen": -0.00028967749676667154,
"eval_rewards/margins": 0.00029621709836646914,
"eval_rewards/rejected": -0.0005858945660293102,
"eval_runtime": 539.9229,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 100
},
{
"epoch": 0.03,
"learning_rate": 1.4360313315926894e-06,
"logits/chosen": 0.9960800409317017,
"logits/rejected": 1.081016182899475,
"logps/chosen": -283.5791015625,
"logps/rejected": -250.1773681640625,
"loss": 0.6931,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": 0.0004133354814257473,
"rewards/margins": 0.0019074681913480163,
"rewards/rejected": -0.0014941326808184385,
"step": 110
},
{
"epoch": 0.03,
"learning_rate": 1.5665796344647521e-06,
"logits/chosen": 1.0297129154205322,
"logits/rejected": 1.075050950050354,
"logps/chosen": -227.7956085205078,
"logps/rejected": -234.15097045898438,
"loss": 0.6931,
"rewards/accuracies": 0.44999998807907104,
"rewards/chosen": 0.00013152281462680548,
"rewards/margins": 0.0006020874716341496,
"rewards/rejected": -0.00047056470066308975,
"step": 120
},
{
"epoch": 0.03,
"learning_rate": 1.6971279373368146e-06,
"logits/chosen": 1.0477956533432007,
"logits/rejected": 1.093898892402649,
"logps/chosen": -282.6871643066406,
"logps/rejected": -239.27749633789062,
"loss": 0.6931,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": 0.00015316384087782353,
"rewards/margins": 0.0016984669491648674,
"rewards/rejected": -0.0015453032683581114,
"step": 130
},
{
"epoch": 0.04,
"learning_rate": 1.8276762402088774e-06,
"logits/chosen": 1.026609182357788,
"logits/rejected": 1.0265687704086304,
"logps/chosen": -263.9731750488281,
"logps/rejected": -237.11538696289062,
"loss": 0.6931,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": 0.0007792095420882106,
"rewards/margins": 0.0024368534795939922,
"rewards/rejected": -0.0016576439375057817,
"step": 140
},
{
"epoch": 0.04,
"learning_rate": 1.9582245430809403e-06,
"logits/chosen": 1.007411241531372,
"logits/rejected": 1.0271480083465576,
"logps/chosen": -262.67718505859375,
"logps/rejected": -235.0521240234375,
"loss": 0.693,
"rewards/accuracies": 0.53125,
"rewards/chosen": 0.00037160428473725915,
"rewards/margins": 0.002256640698760748,
"rewards/rejected": -0.0018850360065698624,
"step": 150
},
{
"epoch": 0.04,
"learning_rate": 2.0887728459530026e-06,
"logits/chosen": 0.9555414915084839,
"logits/rejected": 1.0848571062088013,
"logps/chosen": -258.272705078125,
"logps/rejected": -240.0525360107422,
"loss": 0.6931,
"rewards/accuracies": 0.48124998807907104,
"rewards/chosen": 0.00037167343543842435,
"rewards/margins": 0.0017876753117889166,
"rewards/rejected": -0.0014160019345581532,
"step": 160
},
{
"epoch": 0.04,
"learning_rate": 2.2193211488250653e-06,
"logits/chosen": 0.9961720705032349,
"logits/rejected": 1.0719153881072998,
"logps/chosen": -268.48541259765625,
"logps/rejected": -218.2855987548828,
"loss": 0.693,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": 0.0007700158166699111,
"rewards/margins": 0.0027975619304925203,
"rewards/rejected": -0.0020275460556149483,
"step": 170
},
{
"epoch": 0.05,
"learning_rate": 2.3498694516971284e-06,
"logits/chosen": 0.9853116869926453,
"logits/rejected": 1.033042073249817,
"logps/chosen": -272.58697509765625,
"logps/rejected": -237.8570556640625,
"loss": 0.693,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.000967078551184386,
"rewards/margins": 0.0026892449241131544,
"rewards/rejected": -0.0017221663147211075,
"step": 180
},
{
"epoch": 0.05,
"learning_rate": 2.4804177545691907e-06,
"logits/chosen": 0.9925041198730469,
"logits/rejected": 1.0097901821136475,
"logps/chosen": -269.4550476074219,
"logps/rejected": -235.5220489501953,
"loss": 0.693,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": 0.0016971088480204344,
"rewards/margins": 0.0036631212569773197,
"rewards/rejected": -0.0019660124089568853,
"step": 190
},
{
"epoch": 0.05,
"learning_rate": 2.610966057441254e-06,
"logits/chosen": 1.0215815305709839,
"logits/rejected": 1.1321773529052734,
"logps/chosen": -278.18402099609375,
"logps/rejected": -249.7351531982422,
"loss": 0.693,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.003087539691478014,
"rewards/margins": 0.004473397042602301,
"rewards/rejected": -0.0013858575839549303,
"step": 200
},
{
"epoch": 0.05,
"eval_logits/chosen": 0.9704993367195129,
"eval_logits/rejected": 1.0632352828979492,
"eval_logps/chosen": -277.26611328125,
"eval_logps/rejected": -243.9904022216797,
"eval_loss": 0.6929376721382141,
"eval_rewards/accuracies": 0.5885000228881836,
"eval_rewards/chosen": 0.0028066388331353664,
"eval_rewards/margins": 0.004551402758806944,
"eval_rewards/rejected": -0.0017447640420868993,
"eval_runtime": 540.2748,
"eval_samples_per_second": 3.702,
"eval_steps_per_second": 0.925,
"step": 200
},
{
"epoch": 0.05,
"learning_rate": 2.741514360313316e-06,
"logits/chosen": 1.0122153759002686,
"logits/rejected": 1.091338038444519,
"logps/chosen": -260.8949279785156,
"logps/rejected": -233.3127899169922,
"loss": 0.693,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": 0.0018235554452985525,
"rewards/margins": 0.004228769801557064,
"rewards/rejected": -0.0024052143562585115,
"step": 210
},
{
"epoch": 0.06,
"learning_rate": 2.872062663185379e-06,
"logits/chosen": 1.041244387626648,
"logits/rejected": 1.1119762659072876,
"logps/chosen": -277.6720886230469,
"logps/rejected": -243.11624145507812,
"loss": 0.693,
"rewards/accuracies": 0.5625,
"rewards/chosen": 0.00286296010017395,
"rewards/margins": 0.004079398699104786,
"rewards/rejected": -0.0012164388317614794,
"step": 220
},
{
"epoch": 0.06,
"learning_rate": 3.0026109660574416e-06,
"logits/chosen": 1.0305936336517334,
"logits/rejected": 1.0384341478347778,
"logps/chosen": -268.91888427734375,
"logps/rejected": -275.10919189453125,
"loss": 0.693,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": 0.004722142592072487,
"rewards/margins": 0.005156674422323704,
"rewards/rejected": -0.0004345317429397255,
"step": 230
},
{
"epoch": 0.06,
"learning_rate": 3.1331592689295043e-06,
"logits/chosen": 1.0291810035705566,
"logits/rejected": 1.046808123588562,
"logps/chosen": -272.0259704589844,
"logps/rejected": -231.6946258544922,
"loss": 0.6928,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.007824023254215717,
"rewards/margins": 0.009780755266547203,
"rewards/rejected": -0.001956732477992773,
"step": 240
},
{
"epoch": 0.07,
"learning_rate": 3.263707571801567e-06,
"logits/chosen": 0.9261485934257507,
"logits/rejected": 1.0824263095855713,
"logps/chosen": -262.42010498046875,
"logps/rejected": -207.1998291015625,
"loss": 0.6928,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": 0.004594790283590555,
"rewards/margins": 0.006828789599239826,
"rewards/rejected": -0.0022339997813105583,
"step": 250
},
{
"epoch": 0.07,
"learning_rate": 3.3942558746736293e-06,
"logits/chosen": 1.0177654027938843,
"logits/rejected": 1.0079872608184814,
"logps/chosen": -255.89602661132812,
"logps/rejected": -249.2772979736328,
"loss": 0.6927,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": 0.004748777486383915,
"rewards/margins": 0.007689561694860458,
"rewards/rejected": -0.0029407842084765434,
"step": 260
},
{
"epoch": 0.07,
"learning_rate": 3.524804177545692e-06,
"logits/chosen": 0.9201077222824097,
"logits/rejected": 1.0668845176696777,
"logps/chosen": -250.4193115234375,
"logps/rejected": -225.25808715820312,
"loss": 0.6926,
"rewards/accuracies": 0.5625,
"rewards/chosen": 0.00446278415620327,
"rewards/margins": 0.009258858859539032,
"rewards/rejected": -0.004796075168997049,
"step": 270
},
{
"epoch": 0.07,
"learning_rate": 3.6553524804177547e-06,
"logits/chosen": 0.9566753506660461,
"logits/rejected": 0.9955002069473267,
"logps/chosen": -262.57159423828125,
"logps/rejected": -244.9207763671875,
"loss": 0.6926,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": 0.007930249907076359,
"rewards/margins": 0.012794476933777332,
"rewards/rejected": -0.0048642284236848354,
"step": 280
},
{
"epoch": 0.08,
"learning_rate": 3.7859007832898174e-06,
"logits/chosen": 0.9608441591262817,
"logits/rejected": 1.0510733127593994,
"logps/chosen": -258.68780517578125,
"logps/rejected": -229.01083374023438,
"loss": 0.6925,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.008841058239340782,
"rewards/margins": 0.01509904582053423,
"rewards/rejected": -0.006257989443838596,
"step": 290
},
{
"epoch": 0.08,
"learning_rate": 3.9164490861618806e-06,
"logits/chosen": 0.927672266960144,
"logits/rejected": 1.0490531921386719,
"logps/chosen": -257.1780090332031,
"logps/rejected": -227.1814422607422,
"loss": 0.6926,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": 0.008853832259774208,
"rewards/margins": 0.014651511795818806,
"rewards/rejected": -0.00579767907038331,
"step": 300
},
{
"epoch": 0.08,
"eval_logits/chosen": 0.9545422196388245,
"eval_logits/rejected": 1.0488145351409912,
"eval_logps/chosen": -276.5484924316406,
"eval_logps/rejected": -244.36415100097656,
"eval_loss": 0.6924864649772644,
"eval_rewards/accuracies": 0.6259999871253967,
"eval_rewards/chosen": 0.009982902556657791,
"eval_rewards/margins": 0.015465173870325089,
"eval_rewards/rejected": -0.005482269451022148,
"eval_runtime": 540.3904,
"eval_samples_per_second": 3.701,
"eval_steps_per_second": 0.925,
"step": 300
},
{
"epoch": 0.08,
"learning_rate": 4.046997389033943e-06,
"logits/chosen": 0.9173529744148254,
"logits/rejected": 1.0776686668395996,
"logps/chosen": -256.60821533203125,
"logps/rejected": -226.8739013671875,
"loss": 0.6924,
"rewards/accuracies": 0.65625,
"rewards/chosen": 0.012835139408707619,
"rewards/margins": 0.01847982034087181,
"rewards/rejected": -0.005644683726131916,
"step": 310
},
{
"epoch": 0.08,
"learning_rate": 4.177545691906005e-06,
"logits/chosen": 0.8929821252822876,
"logits/rejected": 1.021347165107727,
"logps/chosen": -282.7056579589844,
"logps/rejected": -256.8482360839844,
"loss": 0.6927,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": 0.010254921391606331,
"rewards/margins": 0.0137600377202034,
"rewards/rejected": -0.003505116328597069,
"step": 320
},
{
"epoch": 0.09,
"learning_rate": 4.308093994778068e-06,
"logits/chosen": 0.9728350639343262,
"logits/rejected": 1.0812106132507324,
"logps/chosen": -278.6542663574219,
"logps/rejected": -243.5943603515625,
"loss": 0.6924,
"rewards/accuracies": 0.59375,
"rewards/chosen": 0.005831174552440643,
"rewards/margins": 0.016604231670498848,
"rewards/rejected": -0.01077305804938078,
"step": 330
},
{
"epoch": 0.09,
"learning_rate": 4.4386422976501306e-06,
"logits/chosen": 0.9642572402954102,
"logits/rejected": 0.9961212873458862,
"logps/chosen": -273.8861389160156,
"logps/rejected": -268.4089050292969,
"loss": 0.6924,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": 0.007403290830552578,
"rewards/margins": 0.017763126641511917,
"rewards/rejected": -0.010359834879636765,
"step": 340
},
{
"epoch": 0.09,
"learning_rate": 4.569190600522193e-06,
"logits/chosen": 0.9777683019638062,
"logits/rejected": 0.9659522771835327,
"logps/chosen": -284.44354248046875,
"logps/rejected": -250.377685546875,
"loss": 0.6922,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.009272046387195587,
"rewards/margins": 0.019789326936006546,
"rewards/rejected": -0.010517279617488384,
"step": 350
},
{
"epoch": 0.09,
"learning_rate": 4.699738903394257e-06,
"logits/chosen": 0.926777184009552,
"logits/rejected": 1.0163743495941162,
"logps/chosen": -301.6736755371094,
"logps/rejected": -258.0189514160156,
"loss": 0.6925,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.009693610481917858,
"rewards/margins": 0.02282491698861122,
"rewards/rejected": -0.013131308369338512,
"step": 360
},
{
"epoch": 0.1,
"learning_rate": 4.8302872062663196e-06,
"logits/chosen": 0.9656028747558594,
"logits/rejected": 0.953553318977356,
"logps/chosen": -304.98760986328125,
"logps/rejected": -260.2730407714844,
"loss": 0.6923,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.00983491726219654,
"rewards/margins": 0.022675124928355217,
"rewards/rejected": -0.012840206734836102,
"step": 370
},
{
"epoch": 0.1,
"learning_rate": 4.9608355091383814e-06,
"logits/chosen": 0.942939281463623,
"logits/rejected": 1.044382929801941,
"logps/chosen": -255.0367431640625,
"logps/rejected": -223.09664916992188,
"loss": 0.6917,
"rewards/accuracies": 0.65625,
"rewards/chosen": 0.012080366723239422,
"rewards/margins": 0.02917185053229332,
"rewards/rejected": -0.017091484740376472,
"step": 380
},
{
"epoch": 0.1,
"learning_rate": 4.9999488562447675e-06,
"logits/chosen": 0.9582707285881042,
"logits/rejected": 0.9803364872932434,
"logps/chosen": -298.5652770996094,
"logps/rejected": -255.8463592529297,
"loss": 0.6923,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.01266313623636961,
"rewards/margins": 0.030872434377670288,
"rewards/rejected": -0.018209297209978104,
"step": 390
},
{
"epoch": 0.1,
"learning_rate": 4.999698361256577e-06,
"logits/chosen": 0.9647480249404907,
"logits/rejected": 0.9665771722793579,
"logps/chosen": -278.61871337890625,
"logps/rejected": -262.8523254394531,
"loss": 0.6916,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.0069937086664140224,
"rewards/margins": 0.03327130153775215,
"rewards/rejected": -0.02627759613096714,
"step": 400
},
{
"epoch": 0.1,
"eval_logits/chosen": 0.8978338837623596,
"eval_logits/rejected": 0.992965817451477,
"eval_logps/chosen": -276.977783203125,
"eval_logps/rejected": -246.2157440185547,
"eval_loss": 0.6920285820960999,
"eval_rewards/accuracies": 0.6340000033378601,
"eval_rewards/chosen": 0.005690301302820444,
"eval_rewards/margins": 0.029688764363527298,
"eval_rewards/rejected": -0.02399846352636814,
"eval_runtime": 540.0105,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 400
},
{
"epoch": 0.11,
"learning_rate": 4.999239142174581e-06,
"logits/chosen": 0.8835350871086121,
"logits/rejected": 0.8935839533805847,
"logps/chosen": -293.0201721191406,
"logps/rejected": -244.9386444091797,
"loss": 0.6921,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": 0.005041834898293018,
"rewards/margins": 0.02935781143605709,
"rewards/rejected": -0.024315981194376945,
"step": 410
},
{
"epoch": 0.11,
"learning_rate": 4.99857123734344e-06,
"logits/chosen": 0.9205878376960754,
"logits/rejected": 1.0264567136764526,
"logps/chosen": -278.33697509765625,
"logps/rejected": -245.725341796875,
"loss": 0.6923,
"rewards/accuracies": 0.59375,
"rewards/chosen": 0.0019535294268280268,
"rewards/margins": 0.029373669996857643,
"rewards/rejected": -0.02742014452815056,
"step": 420
},
{
"epoch": 0.11,
"learning_rate": 4.997694702533016e-06,
"logits/chosen": 0.8263881802558899,
"logits/rejected": 0.9619684219360352,
"logps/chosen": -264.7477111816406,
"logps/rejected": -259.3697509765625,
"loss": 0.6921,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.009586657397449017,
"rewards/margins": 0.028161650523543358,
"rewards/rejected": -0.0377482995390892,
"step": 430
},
{
"epoch": 0.12,
"learning_rate": 4.996609610933713e-06,
"logits/chosen": 0.9048721194267273,
"logits/rejected": 0.8793436288833618,
"logps/chosen": -283.28668212890625,
"logps/rejected": -261.3004150390625,
"loss": 0.6921,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.013652992434799671,
"rewards/margins": 0.024908630177378654,
"rewards/rejected": -0.0385616198182106,
"step": 440
},
{
"epoch": 0.12,
"learning_rate": 4.995316053150366e-06,
"logits/chosen": 0.8724738359451294,
"logits/rejected": 0.9701471328735352,
"logps/chosen": -262.9942321777344,
"logps/rejected": -236.6386260986328,
"loss": 0.6925,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.015233024954795837,
"rewards/margins": 0.0235325675457716,
"rewards/rejected": -0.038765594363212585,
"step": 450
},
{
"epoch": 0.12,
"learning_rate": 4.9938141371946815e-06,
"logits/chosen": 0.910510241985321,
"logits/rejected": 1.033053994178772,
"logps/chosen": -252.5099334716797,
"logps/rejected": -252.48379516601562,
"loss": 0.6917,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.006133816204965115,
"rewards/margins": 0.03705804795026779,
"rewards/rejected": -0.04319187253713608,
"step": 460
},
{
"epoch": 0.12,
"learning_rate": 4.992103988476206e-06,
"logits/chosen": 0.8955548405647278,
"logits/rejected": 0.9851361513137817,
"logps/chosen": -296.1971435546875,
"logps/rejected": -249.4320526123047,
"loss": 0.6917,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.00848749466240406,
"rewards/margins": 0.03199051320552826,
"rewards/rejected": -0.04047800973057747,
"step": 470
},
{
"epoch": 0.13,
"learning_rate": 4.990185749791866e-06,
"logits/chosen": 0.8434032201766968,
"logits/rejected": 0.9481671452522278,
"logps/chosen": -243.55819702148438,
"logps/rejected": -213.0118865966797,
"loss": 0.6919,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.011666789650917053,
"rewards/margins": 0.036795832216739655,
"rewards/rejected": -0.04846261814236641,
"step": 480
},
{
"epoch": 0.13,
"learning_rate": 4.9880595813140395e-06,
"logits/chosen": 0.8387807011604309,
"logits/rejected": 0.9744217991828918,
"logps/chosen": -276.4521484375,
"logps/rejected": -238.837890625,
"loss": 0.6918,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.016785232350230217,
"rewards/margins": 0.03324298933148384,
"rewards/rejected": -0.05002821609377861,
"step": 490
},
{
"epoch": 0.13,
"learning_rate": 4.985725660577184e-06,
"logits/chosen": 0.9428791999816895,
"logits/rejected": 0.9261376261711121,
"logps/chosen": -245.7830047607422,
"logps/rejected": -235.5664825439453,
"loss": 0.6913,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.022965628653764725,
"rewards/margins": 0.034700777381658554,
"rewards/rejected": -0.05766640976071358,
"step": 500
},
{
"epoch": 0.13,
"eval_logits/chosen": 0.8238934874534607,
"eval_logits/rejected": 0.9187954068183899,
"eval_logps/chosen": -280.7515869140625,
"eval_logps/rejected": -250.6851348876953,
"eval_loss": 0.6917084455490112,
"eval_rewards/accuracies": 0.6309999823570251,
"eval_rewards/chosen": -0.03204774856567383,
"eval_rewards/margins": 0.03664441406726837,
"eval_rewards/rejected": -0.0686921626329422,
"eval_runtime": 540.183,
"eval_samples_per_second": 3.702,
"eval_steps_per_second": 0.926,
"step": 500
},
{
"epoch": 0.13,
"learning_rate": 4.983184182463009e-06,
"logits/chosen": 0.8044353723526001,
"logits/rejected": 0.970533549785614,
"logps/chosen": -285.99761962890625,
"logps/rejected": -224.1236114501953,
"loss": 0.6915,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.026267999783158302,
"rewards/margins": 0.038414839655160904,
"rewards/rejected": -0.06468284130096436,
"step": 510
},
{
"epoch": 0.14,
"learning_rate": 4.980435359184203e-06,
"logits/chosen": 0.8804060220718384,
"logits/rejected": 0.9580115079879761,
"logps/chosen": -290.8780822753906,
"logps/rejected": -256.63543701171875,
"loss": 0.6914,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.026573490351438522,
"rewards/margins": 0.04471993073821068,
"rewards/rejected": -0.0712934285402298,
"step": 520
},
{
"epoch": 0.14,
"learning_rate": 4.9774794202667236e-06,
"logits/chosen": 0.863343358039856,
"logits/rejected": 0.9443706274032593,
"logps/chosen": -268.26751708984375,
"logps/rejected": -238.9235076904297,
"loss": 0.6908,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.022545838728547096,
"rewards/margins": 0.04438090696930885,
"rewards/rejected": -0.0669267401099205,
"step": 530
},
{
"epoch": 0.14,
"learning_rate": 4.974316612530615e-06,
"logits/chosen": 0.8318734169006348,
"logits/rejected": 0.8924848437309265,
"logps/chosen": -244.3855438232422,
"logps/rejected": -226.19100952148438,
"loss": 0.6914,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.028137261047959328,
"rewards/margins": 0.04323118180036545,
"rewards/rejected": -0.07136844098567963,
"step": 540
},
{
"epoch": 0.14,
"learning_rate": 4.970947200069416e-06,
"logits/chosen": 0.8746344447135925,
"logits/rejected": 0.9145771861076355,
"logps/chosen": -266.9718322753906,
"logps/rejected": -231.31741333007812,
"loss": 0.6915,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.04218477010726929,
"rewards/margins": 0.05346935987472534,
"rewards/rejected": -0.09565412998199463,
"step": 550
},
{
"epoch": 0.15,
"learning_rate": 4.967371464228096e-06,
"logits/chosen": 0.9077315330505371,
"logits/rejected": 0.9803347587585449,
"logps/chosen": -288.12908935546875,
"logps/rejected": -266.33343505859375,
"loss": 0.6922,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.044233765453100204,
"rewards/margins": 0.0387597382068634,
"rewards/rejected": -0.0829935073852539,
"step": 560
},
{
"epoch": 0.15,
"learning_rate": 4.963589703579569e-06,
"logits/chosen": 0.8836727142333984,
"logits/rejected": 1.0670572519302368,
"logps/chosen": -271.2341003417969,
"logps/rejected": -226.1123046875,
"loss": 0.6903,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.03294859081506729,
"rewards/margins": 0.06488404422998428,
"rewards/rejected": -0.09783263504505157,
"step": 570
},
{
"epoch": 0.15,
"learning_rate": 4.9596022338997615e-06,
"logits/chosen": 0.9106415510177612,
"logits/rejected": 0.8857673406600952,
"logps/chosen": -260.3927307128906,
"logps/rejected": -232.7233428955078,
"loss": 0.6919,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.0389188714325428,
"rewards/margins": 0.03832856938242912,
"rewards/rejected": -0.07724744826555252,
"step": 580
},
{
"epoch": 0.15,
"learning_rate": 4.955409388141243e-06,
"logits/chosen": 0.9676392674446106,
"logits/rejected": 0.9093373417854309,
"logps/chosen": -271.92193603515625,
"logps/rejected": -235.1454620361328,
"loss": 0.6912,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.04100045561790466,
"rewards/margins": 0.03780464082956314,
"rewards/rejected": -0.0788050964474678,
"step": 590
},
{
"epoch": 0.16,
"learning_rate": 4.951011516405429e-06,
"logits/chosen": 0.9376009702682495,
"logits/rejected": 0.9335840344429016,
"logps/chosen": -242.3965301513672,
"logps/rejected": -265.5517578125,
"loss": 0.6916,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.0576152577996254,
"rewards/margins": 0.03630609065294266,
"rewards/rejected": -0.09392134845256805,
"step": 600
},
{
"epoch": 0.16,
"eval_logits/chosen": 0.8586252927780151,
"eval_logits/rejected": 0.9506573677062988,
"eval_logps/chosen": -283.59686279296875,
"eval_logps/rejected": -254.26144409179688,
"eval_loss": 0.6914932131767273,
"eval_rewards/accuracies": 0.6215000152587891,
"eval_rewards/chosen": -0.06050081178545952,
"eval_rewards/margins": 0.0439542680978775,
"eval_rewards/rejected": -0.10445508360862732,
"eval_runtime": 539.8865,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 600
},
{
"epoch": 0.16,
"learning_rate": 4.946408985913344e-06,
"logits/chosen": 0.9270865321159363,
"logits/rejected": 0.946628212928772,
"logps/chosen": -250.1840057373047,
"logps/rejected": -228.7745361328125,
"loss": 0.6922,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.06987909972667694,
"rewards/margins": 0.03838449344038963,
"rewards/rejected": -0.10826359689235687,
"step": 610
},
{
"epoch": 0.16,
"learning_rate": 4.941602180974958e-06,
"logits/chosen": 0.9000824093818665,
"logits/rejected": 0.9378656148910522,
"logps/chosen": -292.17010498046875,
"logps/rejected": -237.016357421875,
"loss": 0.6917,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.06269699335098267,
"rewards/margins": 0.043883226811885834,
"rewards/rejected": -0.1065802201628685,
"step": 620
},
{
"epoch": 0.16,
"learning_rate": 4.936591502957101e-06,
"logits/chosen": 0.7903264760971069,
"logits/rejected": 0.9245864152908325,
"logps/chosen": -301.8384704589844,
"logps/rejected": -256.8984069824219,
"loss": 0.6926,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.04840216785669327,
"rewards/margins": 0.0328446589410305,
"rewards/rejected": -0.08124681562185287,
"step": 630
},
{
"epoch": 0.17,
"learning_rate": 4.931377370249946e-06,
"logits/chosen": 0.8983148336410522,
"logits/rejected": 0.9722617268562317,
"logps/chosen": -278.6772155761719,
"logps/rejected": -235.2743377685547,
"loss": 0.6919,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.04875911772251129,
"rewards/margins": 0.03910985589027405,
"rewards/rejected": -0.08786897361278534,
"step": 640
},
{
"epoch": 0.17,
"learning_rate": 4.925960218232073e-06,
"logits/chosen": 0.9069989323616028,
"logits/rejected": 1.027419924736023,
"logps/chosen": -308.1993103027344,
"logps/rejected": -285.44903564453125,
"loss": 0.6914,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.05254577472805977,
"rewards/margins": 0.04926187917590141,
"rewards/rejected": -0.10180766880512238,
"step": 650
},
{
"epoch": 0.17,
"learning_rate": 4.920340499234116e-06,
"logits/chosen": 0.8621240854263306,
"logits/rejected": 0.9766268730163574,
"logps/chosen": -288.5992431640625,
"logps/rejected": -216.592041015625,
"loss": 0.692,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.0491311252117157,
"rewards/margins": 0.03728429228067398,
"rewards/rejected": -0.08641541749238968,
"step": 660
},
{
"epoch": 0.18,
"learning_rate": 4.914518682500995e-06,
"logits/chosen": 0.9439069032669067,
"logits/rejected": 0.9264806509017944,
"logps/chosen": -282.1919250488281,
"logps/rejected": -278.38037109375,
"loss": 0.6915,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.05234139412641525,
"rewards/margins": 0.04387987032532692,
"rewards/rejected": -0.09622126072645187,
"step": 670
},
{
"epoch": 0.18,
"learning_rate": 4.9084952541527315e-06,
"logits/chosen": 0.8950139880180359,
"logits/rejected": 0.9192444682121277,
"logps/chosen": -258.887939453125,
"logps/rejected": -246.53256225585938,
"loss": 0.6914,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.03944612294435501,
"rewards/margins": 0.04930661618709564,
"rewards/rejected": -0.08875273913145065,
"step": 680
},
{
"epoch": 0.18,
"learning_rate": 4.902270717143858e-06,
"logits/chosen": 0.8779371380805969,
"logits/rejected": 0.9450112581253052,
"logps/chosen": -288.85736083984375,
"logps/rejected": -251.94943237304688,
"loss": 0.6912,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.032757535576820374,
"rewards/margins": 0.05075981095433235,
"rewards/rejected": -0.08351735770702362,
"step": 690
},
{
"epoch": 0.18,
"learning_rate": 4.895845591221427e-06,
"logits/chosen": 0.7992750406265259,
"logits/rejected": 0.9617929458618164,
"logps/chosen": -251.58767700195312,
"logps/rejected": -232.5711669921875,
"loss": 0.6911,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.04521554335951805,
"rewards/margins": 0.0445394404232502,
"rewards/rejected": -0.08975498378276825,
"step": 700
},
{
"epoch": 0.18,
"eval_logits/chosen": 0.8818045854568481,
"eval_logits/rejected": 0.9764631390571594,
"eval_logps/chosen": -281.1485595703125,
"eval_logps/rejected": -251.7943878173828,
"eval_loss": 0.6914020776748657,
"eval_rewards/accuracies": 0.6259999871253967,
"eval_rewards/chosen": -0.03601725026965141,
"eval_rewards/margins": 0.04376746341586113,
"eval_rewards/rejected": -0.07978471368551254,
"eval_runtime": 539.8363,
"eval_samples_per_second": 3.705,
"eval_steps_per_second": 0.926,
"step": 700
},
{
"epoch": 0.19,
"learning_rate": 4.8892204128816e-06,
"logits/chosen": 0.9707541465759277,
"logits/rejected": 1.0011179447174072,
"logps/chosen": -286.44696044921875,
"logps/rejected": -283.7610778808594,
"loss": 0.6911,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.0358760841190815,
"rewards/margins": 0.044750213623046875,
"rewards/rejected": -0.08062629401683807,
"step": 710
},
{
"epoch": 0.19,
"learning_rate": 4.882395735324864e-06,
"logits/chosen": 0.8825947046279907,
"logits/rejected": 0.8657267689704895,
"logps/chosen": -280.93768310546875,
"logps/rejected": -255.46212768554688,
"loss": 0.6914,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.026189198717474937,
"rewards/margins": 0.043071091175079346,
"rewards/rejected": -0.06926029175519943,
"step": 720
},
{
"epoch": 0.19,
"learning_rate": 4.87537212840983e-06,
"logits/chosen": 0.9044111371040344,
"logits/rejected": 0.9175283312797546,
"logps/chosen": -301.9758605957031,
"logps/rejected": -256.6850280761719,
"loss": 0.6915,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.024754587560892105,
"rewards/margins": 0.04776642844080925,
"rewards/rejected": -0.07252101600170135,
"step": 730
},
{
"epoch": 0.19,
"learning_rate": 4.8681501786056545e-06,
"logits/chosen": 0.8817905187606812,
"logits/rejected": 0.9648914337158203,
"logps/chosen": -273.3789367675781,
"logps/rejected": -221.35986328125,
"loss": 0.6916,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.03455664590001106,
"rewards/margins": 0.03378739953041077,
"rewards/rejected": -0.06834404170513153,
"step": 740
},
{
"epoch": 0.2,
"learning_rate": 4.860730488943068e-06,
"logits/chosen": 0.9233236312866211,
"logits/rejected": 0.9309977293014526,
"logps/chosen": -278.74212646484375,
"logps/rejected": -244.9453582763672,
"loss": 0.6907,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.013511193916201591,
"rewards/margins": 0.05613558739423752,
"rewards/rejected": -0.06964678317308426,
"step": 750
},
{
"epoch": 0.2,
"learning_rate": 4.853113678964022e-06,
"logits/chosen": 0.9726675152778625,
"logits/rejected": 0.9908739924430847,
"logps/chosen": -249.2692108154297,
"logps/rejected": -234.45663452148438,
"loss": 0.6908,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.023977819830179214,
"rewards/margins": 0.054052967578172684,
"rewards/rejected": -0.0780307799577713,
"step": 760
},
{
"epoch": 0.2,
"learning_rate": 4.845300384669958e-06,
"logits/chosen": 0.9504178166389465,
"logits/rejected": 0.9623494148254395,
"logps/chosen": -265.9614562988281,
"logps/rejected": -223.534423828125,
"loss": 0.6918,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.033301644027233124,
"rewards/margins": 0.033827196806669235,
"rewards/rejected": -0.06712885200977325,
"step": 770
},
{
"epoch": 0.2,
"learning_rate": 4.837291258468701e-06,
"logits/chosen": 0.8907458186149597,
"logits/rejected": 0.9131177663803101,
"logps/chosen": -279.34661865234375,
"logps/rejected": -248.4618377685547,
"loss": 0.6911,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.04006670415401459,
"rewards/margins": 0.05590524524450302,
"rewards/rejected": -0.09597194939851761,
"step": 780
},
{
"epoch": 0.21,
"learning_rate": 4.829086969119984e-06,
"logits/chosen": 0.8844520449638367,
"logits/rejected": 0.9553192257881165,
"logps/chosen": -248.71533203125,
"logps/rejected": -241.25350952148438,
"loss": 0.6923,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.039020974189043045,
"rewards/margins": 0.04393316060304642,
"rewards/rejected": -0.08295414596796036,
"step": 790
},
{
"epoch": 0.21,
"learning_rate": 4.820688201679605e-06,
"logits/chosen": 0.9063314199447632,
"logits/rejected": 0.9573804140090942,
"logps/chosen": -261.8778381347656,
"logps/rejected": -234.04788208007812,
"loss": 0.6915,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.04035869985818863,
"rewards/margins": 0.04505294933915138,
"rewards/rejected": -0.08541164547204971,
"step": 800
},
{
"epoch": 0.21,
"eval_logits/chosen": 0.9022059440612793,
"eval_logits/rejected": 0.9965441226959229,
"eval_logps/chosen": -281.8776550292969,
"eval_logps/rejected": -252.8778533935547,
"eval_loss": 0.6913056373596191,
"eval_rewards/accuracies": 0.6240000128746033,
"eval_rewards/chosen": -0.04330845922231674,
"eval_rewards/margins": 0.04731076583266258,
"eval_rewards/rejected": -0.09061922132968903,
"eval_runtime": 540.1978,
"eval_samples_per_second": 3.702,
"eval_steps_per_second": 0.926,
"step": 800
},
{
"epoch": 0.21,
"learning_rate": 4.8120956574422315e-06,
"logits/chosen": 0.9879693984985352,
"logits/rejected": 0.9654221534729004,
"logps/chosen": -284.70416259765625,
"logps/rejected": -282.3924255371094,
"loss": 0.6916,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.04387697950005531,
"rewards/margins": 0.05213203281164169,
"rewards/rejected": -0.0960090160369873,
"step": 810
},
{
"epoch": 0.21,
"learning_rate": 4.803310053882831e-06,
"logits/chosen": 0.9151199460029602,
"logits/rejected": 0.9490255117416382,
"logps/chosen": -214.4105987548828,
"logps/rejected": -204.9471435546875,
"loss": 0.6918,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.04438359662890434,
"rewards/margins": 0.04097510129213333,
"rewards/rejected": -0.08535870164632797,
"step": 820
},
{
"epoch": 0.22,
"learning_rate": 4.794332124596775e-06,
"logits/chosen": 0.910225510597229,
"logits/rejected": 0.9843104481697083,
"logps/chosen": -279.525634765625,
"logps/rejected": -255.632568359375,
"loss": 0.6919,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.060014404356479645,
"rewards/margins": 0.03720748424530029,
"rewards/rejected": -0.09722188115119934,
"step": 830
},
{
"epoch": 0.22,
"learning_rate": 4.785162619238575e-06,
"logits/chosen": 0.9961544275283813,
"logits/rejected": 1.0286822319030762,
"logps/chosen": -268.55023193359375,
"logps/rejected": -252.0181121826172,
"loss": 0.6915,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.07715798169374466,
"rewards/margins": 0.04793631657958031,
"rewards/rejected": -0.12509429454803467,
"step": 840
},
{
"epoch": 0.22,
"learning_rate": 4.775802303459288e-06,
"logits/chosen": 0.9325528144836426,
"logits/rejected": 1.0585540533065796,
"logps/chosen": -295.1412658691406,
"logps/rejected": -265.9173889160156,
"loss": 0.6911,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.0818694531917572,
"rewards/margins": 0.049801088869571686,
"rewards/rejected": -0.13167054951190948,
"step": 850
},
{
"epoch": 0.23,
"learning_rate": 4.766251958842589e-06,
"logits/chosen": 0.9676458239555359,
"logits/rejected": 1.0127732753753662,
"logps/chosen": -307.82403564453125,
"logps/rejected": -271.78009033203125,
"loss": 0.6909,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.0727236270904541,
"rewards/margins": 0.0467851385474205,
"rewards/rejected": -0.1195087656378746,
"step": 860
},
{
"epoch": 0.23,
"learning_rate": 4.7565123828395066e-06,
"logits/chosen": 0.9034944772720337,
"logits/rejected": 1.0843673944473267,
"logps/chosen": -318.55096435546875,
"logps/rejected": -294.28131103515625,
"loss": 0.6914,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.050767261534929276,
"rewards/margins": 0.045433562248945236,
"rewards/rejected": -0.09620082378387451,
"step": 870
},
{
"epoch": 0.23,
"learning_rate": 4.746584388701831e-06,
"logits/chosen": 0.990381121635437,
"logits/rejected": 1.1196520328521729,
"logps/chosen": -261.2209777832031,
"logps/rejected": -227.89700317382812,
"loss": 0.6888,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.05034313350915909,
"rewards/margins": 0.06942330300807953,
"rewards/rejected": -0.11976643651723862,
"step": 880
},
{
"epoch": 0.23,
"learning_rate": 4.736468805414218e-06,
"logits/chosen": 0.9858494997024536,
"logits/rejected": 1.1457973718643188,
"logps/chosen": -296.84906005859375,
"logps/rejected": -270.0372009277344,
"loss": 0.6899,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.04343515634536743,
"rewards/margins": 0.07933736592531204,
"rewards/rejected": -0.12277251482009888,
"step": 890
},
{
"epoch": 0.24,
"learning_rate": 4.7261664776249595e-06,
"logits/chosen": 0.9974620938301086,
"logits/rejected": 0.9908777475357056,
"logps/chosen": -273.75848388671875,
"logps/rejected": -256.10833740234375,
"loss": 0.691,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.05218503996729851,
"rewards/margins": 0.057306624948978424,
"rewards/rejected": -0.10949166119098663,
"step": 900
},
{
"epoch": 0.24,
"eval_logits/chosen": 0.9266276955604553,
"eval_logits/rejected": 1.0206482410430908,
"eval_logps/chosen": -282.83209228515625,
"eval_logps/rejected": -254.36526489257812,
"eval_loss": 0.6912428140640259,
"eval_rewards/accuracies": 0.6244999766349792,
"eval_rewards/chosen": -0.052852813154459,
"eval_rewards/margins": 0.052640702575445175,
"eval_rewards/rejected": -0.10549352318048477,
"eval_runtime": 539.8724,
"eval_samples_per_second": 3.705,
"eval_steps_per_second": 0.926,
"step": 900
},
{
"epoch": 0.24,
"learning_rate": 4.715678265575463e-06,
"logits/chosen": 1.0043308734893799,
"logits/rejected": 1.0745335817337036,
"logps/chosen": -312.68194580078125,
"logps/rejected": -295.7378845214844,
"loss": 0.6918,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.05200989916920662,
"rewards/margins": 0.03657294064760208,
"rewards/rejected": -0.088582843542099,
"step": 910
},
{
"epoch": 0.24,
"learning_rate": 4.705005045028415e-06,
"logits/chosen": 0.9822233319282532,
"logits/rejected": 1.0417166948318481,
"logps/chosen": -336.60919189453125,
"logps/rejected": -289.87689208984375,
"loss": 0.6908,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.04931178689002991,
"rewards/margins": 0.06655998528003693,
"rewards/rejected": -0.11587176471948624,
"step": 920
},
{
"epoch": 0.24,
"learning_rate": 4.694147707194659e-06,
"logits/chosen": 0.9825722575187683,
"logits/rejected": 1.0669372081756592,
"logps/chosen": -291.48394775390625,
"logps/rejected": -246.3392333984375,
"loss": 0.6914,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.06181565672159195,
"rewards/margins": 0.05421454459428787,
"rewards/rejected": -0.11603017896413803,
"step": 930
},
{
"epoch": 0.25,
"learning_rate": 4.683107158658782e-06,
"logits/chosen": 0.9451677203178406,
"logits/rejected": 0.9427906274795532,
"logps/chosen": -267.2729797363281,
"logps/rejected": -239.00460815429688,
"loss": 0.6903,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.062487781047821045,
"rewards/margins": 0.05825795605778694,
"rewards/rejected": -0.12074574083089828,
"step": 940
},
{
"epoch": 0.25,
"learning_rate": 4.671884321303407e-06,
"logits/chosen": 0.9309576749801636,
"logits/rejected": 0.9632508158683777,
"logps/chosen": -292.92864990234375,
"logps/rejected": -279.8366394042969,
"loss": 0.6911,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.08061058819293976,
"rewards/margins": 0.05397840216755867,
"rewards/rejected": -0.13458898663520813,
"step": 950
},
{
"epoch": 0.25,
"learning_rate": 4.660480132232224e-06,
"logits/chosen": 0.8813568353652954,
"logits/rejected": 0.9471953511238098,
"logps/chosen": -284.23333740234375,
"logps/rejected": -256.72662353515625,
"loss": 0.6908,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.08392615616321564,
"rewards/margins": 0.04842451959848404,
"rewards/rejected": -0.13235066831111908,
"step": 960
},
{
"epoch": 0.25,
"learning_rate": 4.6488955436917414e-06,
"logits/chosen": 0.9631437063217163,
"logits/rejected": 1.0291635990142822,
"logps/chosen": -284.9555358886719,
"logps/rejected": -221.7299041748047,
"loss": 0.6916,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.07446320354938507,
"rewards/margins": 0.03399290144443512,
"rewards/rejected": -0.10845611244440079,
"step": 970
},
{
"epoch": 0.26,
"learning_rate": 4.6371315229917644e-06,
"logits/chosen": 0.9003992080688477,
"logits/rejected": 0.9652150273323059,
"logps/chosen": -267.9747619628906,
"logps/rejected": -250.9159393310547,
"loss": 0.6915,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.05370105430483818,
"rewards/margins": 0.05308745428919792,
"rewards/rejected": -0.1067885160446167,
"step": 980
},
{
"epoch": 0.26,
"learning_rate": 4.625189052424638e-06,
"logits/chosen": 0.9419862627983093,
"logits/rejected": 1.0019476413726807,
"logps/chosen": -249.2467803955078,
"logps/rejected": -216.28857421875,
"loss": 0.6914,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.04485908895730972,
"rewards/margins": 0.03853844106197357,
"rewards/rejected": -0.0833975300192833,
"step": 990
},
{
"epoch": 0.26,
"learning_rate": 4.613069129183218e-06,
"logits/chosen": 0.9215824007987976,
"logits/rejected": 0.9359531402587891,
"logps/chosen": -265.0127868652344,
"logps/rejected": -257.2760314941406,
"loss": 0.6913,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.047946006059646606,
"rewards/margins": 0.042474087327718735,
"rewards/rejected": -0.09042008966207504,
"step": 1000
},
{
"epoch": 0.26,
"eval_logits/chosen": 0.9215968251228333,
"eval_logits/rejected": 1.01704740524292,
"eval_logps/chosen": -281.5216064453125,
"eval_logps/rejected": -252.86395263671875,
"eval_loss": 0.6912136673927307,
"eval_rewards/accuracies": 0.6290000081062317,
"eval_rewards/chosen": -0.03974788263440132,
"eval_rewards/margins": 0.0507323183119297,
"eval_rewards/rejected": -0.09048020094633102,
"eval_runtime": 539.8968,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 1000
},
{
"epoch": 0.26,
"learning_rate": 4.600772765277607e-06,
"logits/chosen": 0.9124513864517212,
"logits/rejected": 1.0127254724502563,
"logps/chosen": -300.19659423828125,
"logps/rejected": -253.8302764892578,
"loss": 0.6915,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.04729009419679642,
"rewards/margins": 0.055154770612716675,
"rewards/rejected": -0.1024448499083519,
"step": 1010
},
{
"epoch": 0.27,
"learning_rate": 4.588300987450652e-06,
"logits/chosen": 0.9808125495910645,
"logits/rejected": 1.0599421262741089,
"logps/chosen": -282.46038818359375,
"logps/rejected": -267.24188232421875,
"loss": 0.6912,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.05007319524884224,
"rewards/margins": 0.05987462401390076,
"rewards/rejected": -0.1099478155374527,
"step": 1020
},
{
"epoch": 0.27,
"learning_rate": 4.5756548370922136e-06,
"logits/chosen": 0.9697202444076538,
"logits/rejected": 1.0168477296829224,
"logps/chosen": -282.2762145996094,
"logps/rejected": -255.68844604492188,
"loss": 0.6921,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.044876083731651306,
"rewards/margins": 0.03766594082117081,
"rewards/rejected": -0.08254201710224152,
"step": 1030
},
{
"epoch": 0.27,
"learning_rate": 4.562835370152206e-06,
"logits/chosen": 0.9049477577209473,
"logits/rejected": 1.021236538887024,
"logps/chosen": -253.36471557617188,
"logps/rejected": -239.316650390625,
"loss": 0.692,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.038831353187561035,
"rewards/margins": 0.035190094262361526,
"rewards/rejected": -0.07402144372463226,
"step": 1040
},
{
"epoch": 0.27,
"learning_rate": 4.54984365705243e-06,
"logits/chosen": 1.0183782577514648,
"logits/rejected": 1.024595022201538,
"logps/chosen": -254.61691284179688,
"logps/rejected": -231.1280975341797,
"loss": 0.6925,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.027654901146888733,
"rewards/margins": 0.042562730610370636,
"rewards/rejected": -0.07021763920783997,
"step": 1050
},
{
"epoch": 0.28,
"learning_rate": 4.536680782597191e-06,
"logits/chosen": 0.9354788661003113,
"logits/rejected": 1.0331499576568604,
"logps/chosen": -297.9024963378906,
"logps/rejected": -269.0444030761719,
"loss": 0.6908,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.016384651884436607,
"rewards/margins": 0.05180732160806656,
"rewards/rejected": -0.06819198280572891,
"step": 1060
},
{
"epoch": 0.28,
"learning_rate": 4.523347845882718e-06,
"logits/chosen": 1.014983057975769,
"logits/rejected": 1.0802810192108154,
"logps/chosen": -258.166015625,
"logps/rejected": -207.4542999267578,
"loss": 0.6907,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.028195038437843323,
"rewards/margins": 0.05233887955546379,
"rewards/rejected": -0.08053391426801682,
"step": 1070
},
{
"epoch": 0.28,
"learning_rate": 4.50984596020539e-06,
"logits/chosen": 0.9456683397293091,
"logits/rejected": 1.018128752708435,
"logps/chosen": -300.35211181640625,
"logps/rejected": -264.9364318847656,
"loss": 0.6921,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.030218075960874557,
"rewards/margins": 0.046960409730672836,
"rewards/rejected": -0.0771784856915474,
"step": 1080
},
{
"epoch": 0.29,
"learning_rate": 4.4961762529687745e-06,
"logits/chosen": 0.9062995910644531,
"logits/rejected": 1.0201383829116821,
"logps/chosen": -305.39776611328125,
"logps/rejected": -233.7451171875,
"loss": 0.6919,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.04302235692739487,
"rewards/margins": 0.04250651225447655,
"rewards/rejected": -0.08552887290716171,
"step": 1090
},
{
"epoch": 0.29,
"learning_rate": 4.482339865589492e-06,
"logits/chosen": 0.8466746211051941,
"logits/rejected": 1.021051049232483,
"logps/chosen": -266.606689453125,
"logps/rejected": -235.42123413085938,
"loss": 0.6912,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.04704426974058151,
"rewards/margins": 0.05677234008908272,
"rewards/rejected": -0.10381660610437393,
"step": 1100
},
{
"epoch": 0.29,
"eval_logits/chosen": 0.9243915677070618,
"eval_logits/rejected": 1.018973708152771,
"eval_logps/chosen": -283.05096435546875,
"eval_logps/rejected": -253.97817993164062,
"eval_loss": 0.6911666989326477,
"eval_rewards/accuracies": 0.625,
"eval_rewards/chosen": -0.05504164099693298,
"eval_rewards/margins": 0.046581096947193146,
"eval_rewards/rejected": -0.10162272304296494,
"eval_runtime": 539.981,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 1100
},
{
"epoch": 0.29,
"learning_rate": 4.468337953401909e-06,
"logits/chosen": 0.9825204610824585,
"logits/rejected": 1.0412126779556274,
"logps/chosen": -299.93157958984375,
"logps/rejected": -247.61044311523438,
"loss": 0.6906,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.04361564666032791,
"rewards/margins": 0.049379851669073105,
"rewards/rejected": -0.09299550205469131,
"step": 1110
},
{
"epoch": 0.29,
"learning_rate": 4.45417168556166e-06,
"logits/chosen": 0.9352057576179504,
"logits/rejected": 1.0108729600906372,
"logps/chosen": -261.39959716796875,
"logps/rejected": -262.29864501953125,
"loss": 0.6908,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.03935537114739418,
"rewards/margins": 0.043966446071863174,
"rewards/rejected": -0.08332181721925735,
"step": 1120
},
{
"epoch": 0.3,
"learning_rate": 4.439842244948036e-06,
"logits/chosen": 0.9064655303955078,
"logits/rejected": 0.9808236956596375,
"logps/chosen": -262.05462646484375,
"logps/rejected": -245.01651000976562,
"loss": 0.6909,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.042961589992046356,
"rewards/margins": 0.04068412259221077,
"rewards/rejected": -0.08364571630954742,
"step": 1130
},
{
"epoch": 0.3,
"learning_rate": 4.425350828065204e-06,
"logits/chosen": 0.9776175618171692,
"logits/rejected": 0.9820224642753601,
"logps/chosen": -282.891357421875,
"logps/rejected": -266.565673828125,
"loss": 0.6908,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.034526508301496506,
"rewards/margins": 0.04868306592106819,
"rewards/rejected": -0.0832095816731453,
"step": 1140
},
{
"epoch": 0.3,
"learning_rate": 4.410698644942303e-06,
"logits/chosen": 0.9672806859016418,
"logits/rejected": 1.0234858989715576,
"logps/chosen": -281.4849548339844,
"logps/rejected": -285.44293212890625,
"loss": 0.6915,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.032487623393535614,
"rewards/margins": 0.037405431270599365,
"rewards/rejected": -0.06989306956529617,
"step": 1150
},
{
"epoch": 0.3,
"learning_rate": 4.395886919032406e-06,
"logits/chosen": 0.9714674949645996,
"logits/rejected": 1.0344150066375732,
"logps/chosen": -274.4117126464844,
"logps/rejected": -232.0000762939453,
"loss": 0.6913,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.031604982912540436,
"rewards/margins": 0.046002503484487534,
"rewards/rejected": -0.07760748267173767,
"step": 1160
},
{
"epoch": 0.31,
"learning_rate": 4.380916887110366e-06,
"logits/chosen": 0.9162635803222656,
"logits/rejected": 1.0360467433929443,
"logps/chosen": -251.49404907226562,
"logps/rejected": -223.4456329345703,
"loss": 0.6902,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.031815849244594574,
"rewards/margins": 0.059083469212055206,
"rewards/rejected": -0.09089931845664978,
"step": 1170
},
{
"epoch": 0.31,
"learning_rate": 4.365789799169539e-06,
"logits/chosen": 0.977371096611023,
"logits/rejected": 0.9612863659858704,
"logps/chosen": -313.4932556152344,
"logps/rejected": -262.97113037109375,
"loss": 0.6915,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.05341174453496933,
"rewards/margins": 0.05502920225262642,
"rewards/rejected": -0.10844095051288605,
"step": 1180
},
{
"epoch": 0.31,
"learning_rate": 4.350506918317416e-06,
"logits/chosen": 0.9038007855415344,
"logits/rejected": 0.8652862310409546,
"logps/chosen": -306.91912841796875,
"logps/rejected": -275.84429931640625,
"loss": 0.6925,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.041279759258031845,
"rewards/margins": 0.039026908576488495,
"rewards/rejected": -0.08030666410923004,
"step": 1190
},
{
"epoch": 0.31,
"learning_rate": 4.335069520670149e-06,
"logits/chosen": 0.9440900683403015,
"logits/rejected": 1.0484023094177246,
"logps/chosen": -281.1288757324219,
"logps/rejected": -216.98269653320312,
"loss": 0.6902,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.047745514661073685,
"rewards/margins": 0.06614203751087189,
"rewards/rejected": -0.11388754844665527,
"step": 1200
},
{
"epoch": 0.31,
"eval_logits/chosen": 0.9163612127304077,
"eval_logits/rejected": 1.0101075172424316,
"eval_logps/chosen": -283.2487487792969,
"eval_logps/rejected": -254.8289031982422,
"eval_loss": 0.6911502480506897,
"eval_rewards/accuracies": 0.6230000257492065,
"eval_rewards/chosen": -0.057019300758838654,
"eval_rewards/margins": 0.0531105101108551,
"eval_rewards/rejected": -0.11012981832027435,
"eval_runtime": 539.827,
"eval_samples_per_second": 3.705,
"eval_steps_per_second": 0.926,
"step": 1200
},
{
"epoch": 0.32,
"learning_rate": 4.319478895246e-06,
"logits/chosen": 0.9531642198562622,
"logits/rejected": 0.9094236493110657,
"logps/chosen": -278.7569274902344,
"logps/rejected": -242.00259399414062,
"loss": 0.6912,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.04623215273022652,
"rewards/margins": 0.04663626849651337,
"rewards/rejected": -0.09286841005086899,
"step": 1210
},
{
"epoch": 0.32,
"learning_rate": 4.303736343857704e-06,
"logits/chosen": 1.053362488746643,
"logits/rejected": 1.00510573387146,
"logps/chosen": -268.8019714355469,
"logps/rejected": -252.3562774658203,
"loss": 0.6912,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.047048769891262054,
"rewards/margins": 0.0653461441397667,
"rewards/rejected": -0.11239492893218994,
"step": 1220
},
{
"epoch": 0.32,
"learning_rate": 4.287843181003772e-06,
"logits/chosen": 0.9648410677909851,
"logits/rejected": 1.080945611000061,
"logps/chosen": -280.63470458984375,
"logps/rejected": -238.6513214111328,
"loss": 0.6903,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.029626190662384033,
"rewards/margins": 0.06667675822973251,
"rewards/rejected": -0.09630295634269714,
"step": 1230
},
{
"epoch": 0.32,
"learning_rate": 4.27180073375873e-06,
"logits/chosen": 0.9461795687675476,
"logits/rejected": 0.9696234464645386,
"logps/chosen": -280.750244140625,
"logps/rejected": -253.75149536132812,
"loss": 0.6907,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.02992023155093193,
"rewards/margins": 0.06105700135231018,
"rewards/rejected": -0.09097723662853241,
"step": 1240
},
{
"epoch": 0.33,
"learning_rate": 4.255610341662304e-06,
"logits/chosen": 0.9404398798942566,
"logits/rejected": 1.006503701210022,
"logps/chosen": -306.0481262207031,
"logps/rejected": -267.99591064453125,
"loss": 0.6916,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.015811875462532043,
"rewards/margins": 0.061067551374435425,
"rewards/rejected": -0.07687942683696747,
"step": 1250
},
{
"epoch": 0.33,
"learning_rate": 4.2392733566075764e-06,
"logits/chosen": 0.9584512710571289,
"logits/rejected": 1.0292441844940186,
"logps/chosen": -243.0128173828125,
"logps/rejected": -224.4785614013672,
"loss": 0.6905,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.021735265851020813,
"rewards/margins": 0.06055579334497452,
"rewards/rejected": -0.08229105174541473,
"step": 1260
},
{
"epoch": 0.33,
"learning_rate": 4.2227911427280975e-06,
"logits/chosen": 0.9586073160171509,
"logits/rejected": 1.008643388748169,
"logps/chosen": -286.1580505371094,
"logps/rejected": -270.8837585449219,
"loss": 0.6918,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.023701880127191544,
"rewards/margins": 0.04824981838464737,
"rewards/rejected": -0.07195170223712921,
"step": 1270
},
{
"epoch": 0.33,
"learning_rate": 4.206165076283983e-06,
"logits/chosen": 0.9629222750663757,
"logits/rejected": 1.0080350637435913,
"logps/chosen": -238.8936309814453,
"logps/rejected": -228.6975860595703,
"loss": 0.6913,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.022201109677553177,
"rewards/margins": 0.04362647980451584,
"rewards/rejected": -0.06582758575677872,
"step": 1280
},
{
"epoch": 0.34,
"learning_rate": 4.189396545546995e-06,
"logits/chosen": 0.9939113855361938,
"logits/rejected": 1.0935704708099365,
"logps/chosen": -289.91900634765625,
"logps/rejected": -280.5440673828125,
"loss": 0.6917,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.020857712253928185,
"rewards/margins": 0.0328826978802681,
"rewards/rejected": -0.05374041199684143,
"step": 1290
},
{
"epoch": 0.34,
"learning_rate": 4.172486950684627e-06,
"logits/chosen": 1.0020853281021118,
"logits/rejected": 1.0316828489303589,
"logps/chosen": -279.2798767089844,
"logps/rejected": -262.09442138671875,
"loss": 0.6912,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.027787720784544945,
"rewards/margins": 0.048239342868328094,
"rewards/rejected": -0.07602706551551819,
"step": 1300
},
{
"epoch": 0.34,
"eval_logits/chosen": 0.9400979280471802,
"eval_logits/rejected": 1.0356999635696411,
"eval_logps/chosen": -279.8863830566406,
"eval_logps/rejected": -251.13418579101562,
"eval_loss": 0.6911138296127319,
"eval_rewards/accuracies": 0.6129999756813049,
"eval_rewards/chosen": -0.023395679891109467,
"eval_rewards/margins": 0.04978705570101738,
"eval_rewards/rejected": -0.07318273931741714,
"eval_runtime": 540.0054,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 1300
},
{
"epoch": 0.34,
"learning_rate": 4.155437703643182e-06,
"logits/chosen": 0.8979179263114929,
"logits/rejected": 0.9356774091720581,
"logps/chosen": -284.64422607421875,
"logps/rejected": -251.070556640625,
"loss": 0.6909,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.017777901142835617,
"rewards/margins": 0.06165579706430435,
"rewards/rejected": -0.07943369448184967,
"step": 1310
},
{
"epoch": 0.35,
"learning_rate": 4.138250228029882e-06,
"logits/chosen": 0.9831737279891968,
"logits/rejected": 1.103745460510254,
"logps/chosen": -233.90310668945312,
"logps/rejected": -216.5464324951172,
"loss": 0.6901,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.01819593645632267,
"rewards/margins": 0.0664977878332138,
"rewards/rejected": -0.08469371497631073,
"step": 1320
},
{
"epoch": 0.35,
"learning_rate": 4.120925958993994e-06,
"logits/chosen": 0.948932945728302,
"logits/rejected": 1.0110113620758057,
"logps/chosen": -296.0926818847656,
"logps/rejected": -249.8265380859375,
"loss": 0.6918,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.02898983657360077,
"rewards/margins": 0.047562919557094574,
"rewards/rejected": -0.07655275613069534,
"step": 1330
},
{
"epoch": 0.35,
"learning_rate": 4.103466343106999e-06,
"logits/chosen": 0.9645043611526489,
"logits/rejected": 0.9807823896408081,
"logps/chosen": -274.97869873046875,
"logps/rejected": -243.57064819335938,
"loss": 0.6912,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.024762103334069252,
"rewards/margins": 0.04700572043657303,
"rewards/rejected": -0.07176782190799713,
"step": 1340
},
{
"epoch": 0.35,
"learning_rate": 4.085872838241797e-06,
"logits/chosen": 0.9856332540512085,
"logits/rejected": 1.0186711549758911,
"logps/chosen": -286.65740966796875,
"logps/rejected": -285.44451904296875,
"loss": 0.691,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.015045429579913616,
"rewards/margins": 0.0600103922188282,
"rewards/rejected": -0.0750558152794838,
"step": 1350
},
{
"epoch": 0.36,
"learning_rate": 4.06814691345098e-06,
"logits/chosen": 0.9584442973136902,
"logits/rejected": 1.0773932933807373,
"logps/chosen": -303.4087829589844,
"logps/rejected": -257.34381103515625,
"loss": 0.692,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": -0.03773612901568413,
"rewards/margins": 0.043736010789871216,
"rewards/rejected": -0.08147214353084564,
"step": 1360
},
{
"epoch": 0.36,
"learning_rate": 4.050290048844171e-06,
"logits/chosen": 0.9791196584701538,
"logits/rejected": 1.037467360496521,
"logps/chosen": -256.17120361328125,
"logps/rejected": -225.8929443359375,
"loss": 0.6917,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.015562976710498333,
"rewards/margins": 0.03945142775774002,
"rewards/rejected": -0.05501440912485123,
"step": 1370
},
{
"epoch": 0.36,
"learning_rate": 4.032303735464422e-06,
"logits/chosen": 0.9337407946586609,
"logits/rejected": 1.0309889316558838,
"logps/chosen": -289.6443176269531,
"logps/rejected": -257.22784423828125,
"loss": 0.6908,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.005114209838211536,
"rewards/margins": 0.05096329376101494,
"rewards/rejected": -0.0560775101184845,
"step": 1380
},
{
"epoch": 0.36,
"learning_rate": 4.014189475163727e-06,
"logits/chosen": 0.9751637578010559,
"logits/rejected": 1.0514873266220093,
"logps/chosen": -266.8908386230469,
"logps/rejected": -240.3486328125,
"loss": 0.6921,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.006348415277898312,
"rewards/margins": 0.04528724402189255,
"rewards/rejected": -0.05163566395640373,
"step": 1390
},
{
"epoch": 0.37,
"learning_rate": 3.995948780477605e-06,
"logits/chosen": 1.0069153308868408,
"logits/rejected": 1.0419646501541138,
"logps/chosen": -287.21966552734375,
"logps/rejected": -246.9211883544922,
"loss": 0.6914,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.008892977610230446,
"rewards/margins": 0.038406871259212494,
"rewards/rejected": -0.04729985073208809,
"step": 1400
},
{
"epoch": 0.37,
"eval_logits/chosen": 0.9341767430305481,
"eval_logits/rejected": 1.0311079025268555,
"eval_logps/chosen": -279.1179504394531,
"eval_logps/rejected": -250.15402221679688,
"eval_loss": 0.6911120414733887,
"eval_rewards/accuracies": 0.6294999718666077,
"eval_rewards/chosen": -0.015711043030023575,
"eval_rewards/margins": 0.04766979068517685,
"eval_rewards/rejected": -0.06338082998991013,
"eval_runtime": 540.2273,
"eval_samples_per_second": 3.702,
"eval_steps_per_second": 0.926,
"step": 1400
},
{
"epoch": 0.37,
"learning_rate": 3.977583174498816e-06,
"logits/chosen": 0.9394723773002625,
"logits/rejected": 0.9873741269111633,
"logps/chosen": -274.50396728515625,
"logps/rejected": -242.7906951904297,
"loss": 0.6919,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.014722615480422974,
"rewards/margins": 0.039573125541210175,
"rewards/rejected": -0.05429573729634285,
"step": 1410
},
{
"epoch": 0.37,
"learning_rate": 3.959094190750172e-06,
"logits/chosen": 0.9101311564445496,
"logits/rejected": 1.0188195705413818,
"logps/chosen": -288.55426025390625,
"logps/rejected": -259.5224304199219,
"loss": 0.6919,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.029792049899697304,
"rewards/margins": 0.036666542291641235,
"rewards/rejected": -0.06645859032869339,
"step": 1420
},
{
"epoch": 0.37,
"learning_rate": 3.9404833730564975e-06,
"logits/chosen": 0.9995678663253784,
"logits/rejected": 1.0337575674057007,
"logps/chosen": -249.91671752929688,
"logps/rejected": -249.8981475830078,
"loss": 0.6915,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.0360335037112236,
"rewards/margins": 0.040712870657444,
"rewards/rejected": -0.076746366918087,
"step": 1430
},
{
"epoch": 0.38,
"learning_rate": 3.921752275415712e-06,
"logits/chosen": 0.9858711361885071,
"logits/rejected": 1.0380220413208008,
"logps/chosen": -242.677490234375,
"logps/rejected": -215.031494140625,
"loss": 0.6907,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.04031269997358322,
"rewards/margins": 0.04829539731144905,
"rewards/rejected": -0.08860810101032257,
"step": 1440
},
{
"epoch": 0.38,
"learning_rate": 3.902902461869079e-06,
"logits/chosen": 0.9379655122756958,
"logits/rejected": 1.0792181491851807,
"logps/chosen": -273.37957763671875,
"logps/rejected": -246.2140350341797,
"loss": 0.6903,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.047497231513261795,
"rewards/margins": 0.0587293803691864,
"rewards/rejected": -0.1062266081571579,
"step": 1450
},
{
"epoch": 0.38,
"learning_rate": 3.883935506370605e-06,
"logits/chosen": 0.916599452495575,
"logits/rejected": 0.9806827306747437,
"logps/chosen": -281.9441833496094,
"logps/rejected": -244.51806640625,
"loss": 0.6919,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.051647938787937164,
"rewards/margins": 0.05136079713702202,
"rewards/rejected": -0.10300873219966888,
"step": 1460
},
{
"epoch": 0.38,
"learning_rate": 3.864852992655617e-06,
"logits/chosen": 0.8767641186714172,
"logits/rejected": 0.9864808320999146,
"logps/chosen": -270.2128601074219,
"logps/rejected": -259.9897766113281,
"loss": 0.691,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.042821671813726425,
"rewards/margins": 0.05382692068815231,
"rewards/rejected": -0.09664861112833023,
"step": 1470
},
{
"epoch": 0.39,
"learning_rate": 3.845656514108516e-06,
"logits/chosen": 0.8953625559806824,
"logits/rejected": 0.9520618319511414,
"logps/chosen": -297.1680603027344,
"logps/rejected": -249.801025390625,
"loss": 0.69,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.038018837571144104,
"rewards/margins": 0.06626948714256287,
"rewards/rejected": -0.10428832471370697,
"step": 1480
},
{
"epoch": 0.39,
"learning_rate": 3.826347673629738e-06,
"logits/chosen": 0.9500657320022583,
"logits/rejected": 0.9770669937133789,
"logps/chosen": -247.0959930419922,
"logps/rejected": -231.38626098632812,
"loss": 0.6908,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.04545672982931137,
"rewards/margins": 0.05727598816156387,
"rewards/rejected": -0.10273271799087524,
"step": 1490
},
{
"epoch": 0.39,
"learning_rate": 3.8069280835019062e-06,
"logits/chosen": 0.9221125841140747,
"logits/rejected": 1.0069925785064697,
"logps/chosen": -325.3990783691406,
"logps/rejected": -256.74163818359375,
"loss": 0.6919,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.03280999884009361,
"rewards/margins": 0.04576558619737625,
"rewards/rejected": -0.07857557386159897,
"step": 1500
},
{
"epoch": 0.39,
"eval_logits/chosen": 0.9161292314529419,
"eval_logits/rejected": 1.013729453086853,
"eval_logps/chosen": -282.56494140625,
"eval_logps/rejected": -254.04412841796875,
"eval_loss": 0.6910136938095093,
"eval_rewards/accuracies": 0.6320000290870667,
"eval_rewards/chosen": -0.05018109828233719,
"eval_rewards/margins": 0.052100956439971924,
"eval_rewards/rejected": -0.10228205472230911,
"eval_runtime": 540.0139,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 1500
},
{
"epoch": 0.4,
"learning_rate": 3.7873993652552077e-06,
"logits/chosen": 0.9163872003555298,
"logits/rejected": 1.0111920833587646,
"logps/chosen": -263.4795227050781,
"logps/rejected": -241.57888793945312,
"loss": 0.6923,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.05680041387677193,
"rewards/margins": 0.021093839779496193,
"rewards/rejected": -0.07789425551891327,
"step": 1510
},
{
"epoch": 0.4,
"learning_rate": 3.7677631495319953e-06,
"logits/chosen": 0.9174981117248535,
"logits/rejected": 0.9849978685379028,
"logps/chosen": -263.1864929199219,
"logps/rejected": -217.2829132080078,
"loss": 0.692,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.049250274896621704,
"rewards/margins": 0.0348433181643486,
"rewards/rejected": -0.0840936005115509,
"step": 1520
},
{
"epoch": 0.4,
"learning_rate": 3.748021075950633e-06,
"logits/chosen": 1.0062494277954102,
"logits/rejected": 0.9941271543502808,
"logps/chosen": -307.7798767089844,
"logps/rejected": -280.7508239746094,
"loss": 0.6914,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.05701036378741264,
"rewards/margins": 0.05211354047060013,
"rewards/rejected": -0.10912390798330307,
"step": 1530
},
{
"epoch": 0.4,
"learning_rate": 3.7281747929685824e-06,
"logits/chosen": 0.926976203918457,
"logits/rejected": 1.0702403783798218,
"logps/chosen": -292.91351318359375,
"logps/rejected": -256.81536865234375,
"loss": 0.6912,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.04857570677995682,
"rewards/margins": 0.041417308151721954,
"rewards/rejected": -0.08999301493167877,
"step": 1540
},
{
"epoch": 0.41,
"learning_rate": 3.7082259577447604e-06,
"logits/chosen": 0.9064332842826843,
"logits/rejected": 1.022146224975586,
"logps/chosen": -324.94122314453125,
"logps/rejected": -255.82431030273438,
"loss": 0.691,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.0545024499297142,
"rewards/margins": 0.05221433565020561,
"rewards/rejected": -0.10671677440404892,
"step": 1550
},
{
"epoch": 0.41,
"learning_rate": 3.6881762360011688e-06,
"logits/chosen": 0.9837790727615356,
"logits/rejected": 1.0159826278686523,
"logps/chosen": -292.13531494140625,
"logps/rejected": -249.9795684814453,
"loss": 0.6908,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.061875708401203156,
"rewards/margins": 0.05055435746908188,
"rewards/rejected": -0.11243007332086563,
"step": 1560
},
{
"epoch": 0.41,
"learning_rate": 3.668027301883802e-06,
"logits/chosen": 0.9538768529891968,
"logits/rejected": 1.0130136013031006,
"logps/chosen": -266.67266845703125,
"logps/rejected": -224.70468139648438,
"loss": 0.6909,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.052473802119493484,
"rewards/margins": 0.05191361904144287,
"rewards/rejected": -0.10438741743564606,
"step": 1570
},
{
"epoch": 0.41,
"learning_rate": 3.64778083782286e-06,
"logits/chosen": 0.9861133694648743,
"logits/rejected": 1.0575048923492432,
"logps/chosen": -270.86236572265625,
"logps/rejected": -278.5152893066406,
"loss": 0.691,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.04235236719250679,
"rewards/margins": 0.05154258757829666,
"rewards/rejected": -0.09389495849609375,
"step": 1580
},
{
"epoch": 0.42,
"learning_rate": 3.627438534392268e-06,
"logits/chosen": 0.9495855569839478,
"logits/rejected": 0.9809403419494629,
"logps/chosen": -264.71380615234375,
"logps/rejected": -214.28939819335938,
"loss": 0.6923,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.03805701807141304,
"rewards/margins": 0.04229050129652023,
"rewards/rejected": -0.08034752309322357,
"step": 1590
},
{
"epoch": 0.42,
"learning_rate": 3.607002090168506e-06,
"logits/chosen": 0.9524203538894653,
"logits/rejected": 1.011755108833313,
"logps/chosen": -262.1990661621094,
"logps/rejected": -238.88052368164062,
"loss": 0.6912,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.04073198884725571,
"rewards/margins": 0.039406824856996536,
"rewards/rejected": -0.08013881742954254,
"step": 1600
},
{
"epoch": 0.42,
"eval_logits/chosen": 0.9319766759872437,
"eval_logits/rejected": 1.0314892530441284,
"eval_logps/chosen": -281.04010009765625,
"eval_logps/rejected": -252.43983459472656,
"eval_loss": 0.6909967660903931,
"eval_rewards/accuracies": 0.6320000290870667,
"eval_rewards/chosen": -0.03493276238441467,
"eval_rewards/margins": 0.051306504756212234,
"eval_rewards/rejected": -0.08623925596475601,
"eval_runtime": 539.9382,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 1600
},
{
"epoch": 0.42,
"learning_rate": 3.586473211588787e-06,
"logits/chosen": 0.9360347986221313,
"logits/rejected": 1.0636814832687378,
"logps/chosen": -301.8039245605469,
"logps/rejected": -245.0256805419922,
"loss": 0.6904,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.02840288355946541,
"rewards/margins": 0.06062694638967514,
"rewards/rejected": -0.08902983367443085,
"step": 1610
},
{
"epoch": 0.42,
"learning_rate": 3.5658536128085623e-06,
"logits/chosen": 0.9372593760490417,
"logits/rejected": 1.0011626482009888,
"logps/chosen": -253.13583374023438,
"logps/rejected": -247.22116088867188,
"loss": 0.691,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.03672366216778755,
"rewards/margins": 0.055417824536561966,
"rewards/rejected": -0.09214149415493011,
"step": 1620
},
{
"epoch": 0.43,
"learning_rate": 3.545145015558399e-06,
"logits/chosen": 0.9559675455093384,
"logits/rejected": 1.0077977180480957,
"logps/chosen": -301.99700927734375,
"logps/rejected": -291.1852111816406,
"loss": 0.6921,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.03201790526509285,
"rewards/margins": 0.047422006726264954,
"rewards/rejected": -0.0794399082660675,
"step": 1630
},
{
"epoch": 0.43,
"learning_rate": 3.5243491490002056e-06,
"logits/chosen": 0.9405361413955688,
"logits/rejected": 0.972905158996582,
"logps/chosen": -277.42578125,
"logps/rejected": -244.67495727539062,
"loss": 0.6913,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.035732053220272064,
"rewards/margins": 0.041307561099529266,
"rewards/rejected": -0.07703961431980133,
"step": 1640
},
{
"epoch": 0.43,
"learning_rate": 3.503467749582857e-06,
"logits/chosen": 0.9268352389335632,
"logits/rejected": 0.9519475102424622,
"logps/chosen": -251.6887664794922,
"logps/rejected": -207.09823608398438,
"loss": 0.6907,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.03763822466135025,
"rewards/margins": 0.05269993096590042,
"rewards/rejected": -0.09033815562725067,
"step": 1650
},
{
"epoch": 0.43,
"learning_rate": 3.4825025608971947e-06,
"logits/chosen": 0.8857590556144714,
"logits/rejected": 1.0210973024368286,
"logps/chosen": -318.9454650878906,
"logps/rejected": -253.1227569580078,
"loss": 0.6905,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.041044797748327255,
"rewards/margins": 0.05614280700683594,
"rewards/rejected": -0.0971876010298729,
"step": 1660
},
{
"epoch": 0.44,
"learning_rate": 3.4614553335304407e-06,
"logits/chosen": 0.9426645040512085,
"logits/rejected": 1.06788969039917,
"logps/chosen": -280.87506103515625,
"logps/rejected": -242.68856811523438,
"loss": 0.6899,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.04844700172543526,
"rewards/margins": 0.06784703582525253,
"rewards/rejected": -0.11629404127597809,
"step": 1670
},
{
"epoch": 0.44,
"learning_rate": 3.4403278249200222e-06,
"logits/chosen": 0.8899833559989929,
"logits/rejected": 0.981528639793396,
"logps/chosen": -261.93939208984375,
"logps/rejected": -244.01416015625,
"loss": 0.6908,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.048227887600660324,
"rewards/margins": 0.04917442053556442,
"rewards/rejected": -0.09740231186151505,
"step": 1680
},
{
"epoch": 0.44,
"learning_rate": 3.4191217992068293e-06,
"logits/chosen": 0.9601287841796875,
"logits/rejected": 0.9461712837219238,
"logps/chosen": -262.20184326171875,
"logps/rejected": -260.2802429199219,
"loss": 0.6913,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.06563547253608704,
"rewards/margins": 0.05082215741276741,
"rewards/rejected": -0.11645762622356415,
"step": 1690
},
{
"epoch": 0.44,
"learning_rate": 3.3978390270879056e-06,
"logits/chosen": 0.9125478863716125,
"logits/rejected": 1.015061616897583,
"logps/chosen": -299.32171630859375,
"logps/rejected": -256.58953857421875,
"loss": 0.6905,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.04695742204785347,
"rewards/margins": 0.05506708472967148,
"rewards/rejected": -0.10202451795339584,
"step": 1700
},
{
"epoch": 0.44,
"eval_logits/chosen": 0.9100068211555481,
"eval_logits/rejected": 1.0087957382202148,
"eval_logps/chosen": -282.8432922363281,
"eval_logps/rejected": -254.7029571533203,
"eval_loss": 0.69096839427948,
"eval_rewards/accuracies": 0.6324999928474426,
"eval_rewards/chosen": -0.052964530885219574,
"eval_rewards/margins": 0.05590558797121048,
"eval_rewards/rejected": -0.10887012630701065,
"eval_runtime": 540.5566,
"eval_samples_per_second": 3.7,
"eval_steps_per_second": 0.925,
"step": 1700
},
{
"epoch": 0.45,
"learning_rate": 3.3764812856685995e-06,
"logits/chosen": 0.944310188293457,
"logits/rejected": 0.9725696444511414,
"logps/chosen": -294.048095703125,
"logps/rejected": -265.73284912109375,
"loss": 0.6913,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.04054544121026993,
"rewards/margins": 0.06353868544101715,
"rewards/rejected": -0.10408411920070648,
"step": 1710
},
{
"epoch": 0.45,
"learning_rate": 3.3550503583141726e-06,
"logits/chosen": 0.9270051717758179,
"logits/rejected": 0.9650331735610962,
"logps/chosen": -327.187744140625,
"logps/rejected": -308.1183166503906,
"loss": 0.6904,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.045968521386384964,
"rewards/margins": 0.062069911509752274,
"rewards/rejected": -0.10803844034671783,
"step": 1720
},
{
"epoch": 0.45,
"learning_rate": 3.3335480345008907e-06,
"logits/chosen": 0.9014407992362976,
"logits/rejected": 0.9876381754875183,
"logps/chosen": -266.403564453125,
"logps/rejected": -229.73080444335938,
"loss": 0.6902,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.04243239760398865,
"rewards/margins": 0.06432008743286133,
"rewards/rejected": -0.10675249993801117,
"step": 1730
},
{
"epoch": 0.46,
"learning_rate": 3.3119761096666055e-06,
"logits/chosen": 1.0035514831542969,
"logits/rejected": 0.9860905408859253,
"logps/chosen": -317.1201171875,
"logps/rejected": -265.30499267578125,
"loss": 0.6917,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.06333385407924652,
"rewards/margins": 0.047281377017498016,
"rewards/rejected": -0.11061523109674454,
"step": 1740
},
{
"epoch": 0.46,
"learning_rate": 3.290336385060832e-06,
"logits/chosen": 0.9631811380386353,
"logits/rejected": 0.9473791122436523,
"logps/chosen": -288.9410095214844,
"logps/rejected": -267.47454833984375,
"loss": 0.6903,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.049429457634687424,
"rewards/margins": 0.06482435762882233,
"rewards/rejected": -0.11425381898880005,
"step": 1750
},
{
"epoch": 0.46,
"learning_rate": 3.268630667594348e-06,
"logits/chosen": 0.9454625844955444,
"logits/rejected": 1.029585838317871,
"logps/chosen": -284.25531005859375,
"logps/rejected": -266.0755920410156,
"loss": 0.6917,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.05219440534710884,
"rewards/margins": 0.05088215321302414,
"rewards/rejected": -0.10307655483484268,
"step": 1760
},
{
"epoch": 0.46,
"learning_rate": 3.2468607696883147e-06,
"logits/chosen": 0.9423101544380188,
"logits/rejected": 1.0101561546325684,
"logps/chosen": -302.5003356933594,
"logps/rejected": -258.0522155761719,
"loss": 0.6902,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.05451619625091553,
"rewards/margins": 0.06200449541211128,
"rewards/rejected": -0.11652068793773651,
"step": 1770
},
{
"epoch": 0.47,
"learning_rate": 3.225028509122944e-06,
"logits/chosen": 0.9246463775634766,
"logits/rejected": 1.057897925376892,
"logps/chosen": -307.1044921875,
"logps/rejected": -260.0653076171875,
"loss": 0.6913,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.025903914123773575,
"rewards/margins": 0.06586969643831253,
"rewards/rejected": -0.0917736142873764,
"step": 1780
},
{
"epoch": 0.47,
"learning_rate": 3.2031357088857083e-06,
"logits/chosen": 0.9973335266113281,
"logits/rejected": 1.0170247554779053,
"logps/chosen": -275.74005126953125,
"logps/rejected": -261.8125915527344,
"loss": 0.6908,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.047602783888578415,
"rewards/margins": 0.05812246724963188,
"rewards/rejected": -0.1057252511382103,
"step": 1790
},
{
"epoch": 0.47,
"learning_rate": 3.181184197019127e-06,
"logits/chosen": 1.0013173818588257,
"logits/rejected": 0.9759401082992554,
"logps/chosen": -283.0948791503906,
"logps/rejected": -236.4706268310547,
"loss": 0.6901,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.04060991853475571,
"rewards/margins": 0.06453616917133331,
"rewards/rejected": -0.10514608770608902,
"step": 1800
},
{
"epoch": 0.47,
"eval_logits/chosen": 0.9323698282241821,
"eval_logits/rejected": 1.0313962697982788,
"eval_logps/chosen": -281.6338195800781,
"eval_logps/rejected": -253.65231323242188,
"eval_loss": 0.6909690499305725,
"eval_rewards/accuracies": 0.6225000023841858,
"eval_rewards/chosen": -0.0408700592815876,
"eval_rewards/margins": 0.057493917644023895,
"eval_rewards/rejected": -0.0983639732003212,
"eval_runtime": 539.9632,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 1800
},
{
"epoch": 0.47,
"learning_rate": 3.159175806468126e-06,
"logits/chosen": 0.9391202926635742,
"logits/rejected": 1.0263153314590454,
"logps/chosen": -284.8543395996094,
"logps/rejected": -277.1647644042969,
"loss": 0.6914,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.03517382591962814,
"rewards/margins": 0.05179664492607117,
"rewards/rejected": -0.08697047084569931,
"step": 1810
},
{
"epoch": 0.48,
"learning_rate": 3.1371123749269804e-06,
"logits/chosen": 0.9998887777328491,
"logits/rejected": 1.085648536682129,
"logps/chosen": -282.2349548339844,
"logps/rejected": -252.2509765625,
"loss": 0.6907,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.029559914022684097,
"rewards/margins": 0.06976927816867828,
"rewards/rejected": -0.09932918846607208,
"step": 1820
},
{
"epoch": 0.48,
"learning_rate": 3.114995744685877e-06,
"logits/chosen": 0.9946446418762207,
"logits/rejected": 0.9664584398269653,
"logps/chosen": -258.7174072265625,
"logps/rejected": -240.670654296875,
"loss": 0.6902,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.02138775773346424,
"rewards/margins": 0.07551835477352142,
"rewards/rejected": -0.09690611809492111,
"step": 1830
},
{
"epoch": 0.48,
"learning_rate": 3.0928277624770743e-06,
"logits/chosen": 0.9557411074638367,
"logits/rejected": 1.0262891054153442,
"logps/chosen": -286.01458740234375,
"logps/rejected": -232.46139526367188,
"loss": 0.6891,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.01390999834984541,
"rewards/margins": 0.07762787491083145,
"rewards/rejected": -0.09153787791728973,
"step": 1840
},
{
"epoch": 0.48,
"learning_rate": 3.070610279320708e-06,
"logits/chosen": 0.9258459806442261,
"logits/rejected": 1.0731465816497803,
"logps/chosen": -286.8509826660156,
"logps/rejected": -233.0972137451172,
"loss": 0.6921,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.032274287194013596,
"rewards/margins": 0.0481577143073082,
"rewards/rejected": -0.0804319903254509,
"step": 1850
},
{
"epoch": 0.49,
"learning_rate": 3.0483451503702264e-06,
"logits/chosen": 0.9777040481567383,
"logits/rejected": 1.0403920412063599,
"logps/chosen": -288.07568359375,
"logps/rejected": -264.2755432128906,
"loss": 0.6916,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.023307165130972862,
"rewards/margins": 0.048550479114055634,
"rewards/rejected": -0.07185763865709305,
"step": 1860
},
{
"epoch": 0.49,
"learning_rate": 3.0260342347574916e-06,
"logits/chosen": 0.9937931895256042,
"logits/rejected": 1.0553154945373535,
"logps/chosen": -299.058837890625,
"logps/rejected": -248.8879852294922,
"loss": 0.6899,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.036436520516872406,
"rewards/margins": 0.07167023420333862,
"rewards/rejected": -0.10810675472021103,
"step": 1870
},
{
"epoch": 0.49,
"learning_rate": 3.0036793954375358e-06,
"logits/chosen": 0.9339237213134766,
"logits/rejected": 1.0194135904312134,
"logps/chosen": -279.79034423828125,
"logps/rejected": -242.7287139892578,
"loss": 0.691,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.015874430537223816,
"rewards/margins": 0.0718928873538971,
"rewards/rejected": -0.08776732534170151,
"step": 1880
},
{
"epoch": 0.49,
"learning_rate": 2.981282499033009e-06,
"logits/chosen": 0.9355725049972534,
"logits/rejected": 0.9969936609268188,
"logps/chosen": -294.20538330078125,
"logps/rejected": -244.11160278320312,
"loss": 0.6908,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.026097718626260757,
"rewards/margins": 0.05047178268432617,
"rewards/rejected": -0.07656950503587723,
"step": 1890
},
{
"epoch": 0.5,
"learning_rate": 2.9588454156783163e-06,
"logits/chosen": 1.0147713422775269,
"logits/rejected": 0.9884480237960815,
"logps/chosen": -278.02313232421875,
"logps/rejected": -251.70596313476562,
"loss": 0.6902,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.02444344013929367,
"rewards/margins": 0.054150182753801346,
"rewards/rejected": -0.07859362661838531,
"step": 1900
},
{
"epoch": 0.5,
"eval_logits/chosen": 0.9226279258728027,
"eval_logits/rejected": 1.0212137699127197,
"eval_logps/chosen": -280.8078308105469,
"eval_logps/rejected": -252.7657470703125,
"eval_loss": 0.6909632682800293,
"eval_rewards/accuracies": 0.6215000152587891,
"eval_rewards/chosen": -0.03261038661003113,
"eval_rewards/margins": 0.05688786879181862,
"eval_rewards/rejected": -0.08949825167655945,
"eval_runtime": 539.8982,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 1900
},
{
"epoch": 0.5,
"learning_rate": 2.9363700188634597e-06,
"logits/chosen": 0.9406150579452515,
"logits/rejected": 1.0022201538085938,
"logps/chosen": -272.09759521484375,
"logps/rejected": -249.77279663085938,
"loss": 0.6921,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.0340249240398407,
"rewards/margins": 0.041387807577848434,
"rewards/rejected": -0.07541273534297943,
"step": 1910
},
{
"epoch": 0.5,
"learning_rate": 2.9138581852776053e-06,
"logits/chosen": 0.9434909820556641,
"logits/rejected": 1.056489109992981,
"logps/chosen": -277.2883605957031,
"logps/rejected": -262.8563537597656,
"loss": 0.6904,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.031123792752623558,
"rewards/margins": 0.06892909109592438,
"rewards/rejected": -0.10005287826061249,
"step": 1920
},
{
"epoch": 0.51,
"learning_rate": 2.8913117946523805e-06,
"logits/chosen": 1.0239412784576416,
"logits/rejected": 0.993172824382782,
"logps/chosen": -261.37689208984375,
"logps/rejected": -221.52090454101562,
"loss": 0.6916,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.042845502495765686,
"rewards/margins": 0.056726813316345215,
"rewards/rejected": -0.0995723158121109,
"step": 1930
},
{
"epoch": 0.51,
"learning_rate": 2.8687327296049126e-06,
"logits/chosen": 0.9209731817245483,
"logits/rejected": 1.0005519390106201,
"logps/chosen": -281.11822509765625,
"logps/rejected": -259.76007080078125,
"loss": 0.6912,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.028859639540314674,
"rewards/margins": 0.047795943915843964,
"rewards/rejected": -0.07665558159351349,
"step": 1940
},
{
"epoch": 0.51,
"learning_rate": 2.8461228754806376e-06,
"logits/chosen": 0.958030104637146,
"logits/rejected": 1.0400612354278564,
"logps/chosen": -250.44058227539062,
"logps/rejected": -242.2317352294922,
"loss": 0.691,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.0336209237575531,
"rewards/margins": 0.05756276845932007,
"rewards/rejected": -0.09118369966745377,
"step": 1950
},
{
"epoch": 0.51,
"learning_rate": 2.823484120195865e-06,
"logits/chosen": 0.9849356412887573,
"logits/rejected": 0.9542206525802612,
"logps/chosen": -209.1316375732422,
"logps/rejected": -223.1927947998047,
"loss": 0.6909,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.030019324272871017,
"rewards/margins": 0.057300496846437454,
"rewards/rejected": -0.08731982111930847,
"step": 1960
},
{
"epoch": 0.52,
"learning_rate": 2.8008183540801486e-06,
"logits/chosen": 0.8806187510490417,
"logits/rejected": 0.982339084148407,
"logps/chosen": -288.1579895019531,
"logps/rejected": -268.0417785644531,
"loss": 0.6917,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.03183884546160698,
"rewards/margins": 0.042079776525497437,
"rewards/rejected": -0.07391862571239471,
"step": 1970
},
{
"epoch": 0.52,
"learning_rate": 2.7781274697184353e-06,
"logits/chosen": 0.9701333045959473,
"logits/rejected": 1.0392358303070068,
"logps/chosen": -274.5201110839844,
"logps/rejected": -255.364501953125,
"loss": 0.69,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.03297718986868858,
"rewards/margins": 0.0656973272562027,
"rewards/rejected": -0.09867450594902039,
"step": 1980
},
{
"epoch": 0.52,
"learning_rate": 2.7554133617930397e-06,
"logits/chosen": 0.9770027995109558,
"logits/rejected": 0.9862260818481445,
"logps/chosen": -272.98602294921875,
"logps/rejected": -268.11175537109375,
"loss": 0.692,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.037066780030727386,
"rewards/margins": 0.03600457310676575,
"rewards/rejected": -0.07307135313749313,
"step": 1990
},
{
"epoch": 0.52,
"learning_rate": 2.7326779269254363e-06,
"logits/chosen": 0.9106483459472656,
"logits/rejected": 1.0019176006317139,
"logps/chosen": -254.740234375,
"logps/rejected": -229.75634765625,
"loss": 0.6919,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.026663145050406456,
"rewards/margins": 0.03192012384533882,
"rewards/rejected": -0.05858327075839043,
"step": 2000
},
{
"epoch": 0.52,
"eval_logits/chosen": 0.9258546829223633,
"eval_logits/rejected": 1.025207757949829,
"eval_logps/chosen": -279.9319763183594,
"eval_logps/rejected": -251.49105834960938,
"eval_loss": 0.6909723281860352,
"eval_rewards/accuracies": 0.6274999976158142,
"eval_rewards/chosen": -0.02385157160460949,
"eval_rewards/margins": 0.05289952829480171,
"eval_rewards/rejected": -0.07675110548734665,
"eval_runtime": 540.2889,
"eval_samples_per_second": 3.702,
"eval_steps_per_second": 0.925,
"step": 2000
},
{
"epoch": 0.53,
"learning_rate": 2.7099230635178954e-06,
"logits/chosen": 0.9473625421524048,
"logits/rejected": 1.0636024475097656,
"logps/chosen": -250.43350219726562,
"logps/rejected": -213.73043823242188,
"loss": 0.6897,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.016012001782655716,
"rewards/margins": 0.0778764933347702,
"rewards/rejected": -0.09388849884271622,
"step": 2010
},
{
"epoch": 0.53,
"learning_rate": 2.6871506715949608e-06,
"logits/chosen": 1.0584747791290283,
"logits/rejected": 1.0787056684494019,
"logps/chosen": -277.7005615234375,
"logps/rejected": -263.6285400390625,
"loss": 0.6897,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.028144020587205887,
"rewards/margins": 0.07585910707712173,
"rewards/rejected": -0.10400311648845673,
"step": 2020
},
{
"epoch": 0.53,
"learning_rate": 2.6643626526448063e-06,
"logits/chosen": 0.9738826751708984,
"logits/rejected": 1.0668303966522217,
"logps/chosen": -245.468994140625,
"logps/rejected": -252.04458618164062,
"loss": 0.6901,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.026749875396490097,
"rewards/margins": 0.06751126796007156,
"rewards/rejected": -0.09426114708185196,
"step": 2030
},
{
"epoch": 0.53,
"learning_rate": 2.6415609094604562e-06,
"logits/chosen": 0.9280556440353394,
"logits/rejected": 1.0651742219924927,
"logps/chosen": -304.1370849609375,
"logps/rejected": -259.7930603027344,
"loss": 0.6903,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.03546469286084175,
"rewards/margins": 0.06755216419696808,
"rewards/rejected": -0.10301685333251953,
"step": 2040
},
{
"epoch": 0.54,
"learning_rate": 2.618747345980904e-06,
"logits/chosen": 0.908758282661438,
"logits/rejected": 1.0054810047149658,
"logps/chosen": -289.89154052734375,
"logps/rejected": -250.53231811523438,
"loss": 0.6905,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.028398990631103516,
"rewards/margins": 0.06537959724664688,
"rewards/rejected": -0.093778595328331,
"step": 2050
},
{
"epoch": 0.54,
"learning_rate": 2.595923867132136e-06,
"logits/chosen": 1.0136375427246094,
"logits/rejected": 1.071046233177185,
"logps/chosen": -227.84829711914062,
"logps/rejected": -231.29788208007812,
"loss": 0.6912,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.025034338235855103,
"rewards/margins": 0.06069794297218323,
"rewards/rejected": -0.08573228865861893,
"step": 2060
},
{
"epoch": 0.54,
"learning_rate": 2.5730923786680672e-06,
"logits/chosen": 0.9761725664138794,
"logits/rejected": 1.0631458759307861,
"logps/chosen": -275.0975341796875,
"logps/rejected": -230.28170776367188,
"loss": 0.6909,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.03186946362257004,
"rewards/margins": 0.056202489882707596,
"rewards/rejected": -0.08807194232940674,
"step": 2070
},
{
"epoch": 0.54,
"learning_rate": 2.5502547870114137e-06,
"logits/chosen": 0.9701007604598999,
"logits/rejected": 1.0082242488861084,
"logps/chosen": -231.7493133544922,
"logps/rejected": -252.67343139648438,
"loss": 0.6909,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.02439451590180397,
"rewards/margins": 0.04928570240736008,
"rewards/rejected": -0.07368021458387375,
"step": 2080
},
{
"epoch": 0.55,
"learning_rate": 2.527412999094507e-06,
"logits/chosen": 0.9784774780273438,
"logits/rejected": 1.0191878080368042,
"logps/chosen": -283.6234130859375,
"logps/rejected": -256.2926940917969,
"loss": 0.6911,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.03525594249367714,
"rewards/margins": 0.05471722036600113,
"rewards/rejected": -0.08997315913438797,
"step": 2090
},
{
"epoch": 0.55,
"learning_rate": 2.504568922200064e-06,
"logits/chosen": 0.9242954254150391,
"logits/rejected": 0.9784539937973022,
"logps/chosen": -265.85198974609375,
"logps/rejected": -248.41354370117188,
"loss": 0.6919,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.03961525112390518,
"rewards/margins": 0.04367566108703613,
"rewards/rejected": -0.08329091221094131,
"step": 2100
},
{
"epoch": 0.55,
"eval_logits/chosen": 0.9476714730262756,
"eval_logits/rejected": 1.0475825071334839,
"eval_logps/chosen": -281.3605651855469,
"eval_logps/rejected": -253.0793914794922,
"eval_loss": 0.690941333770752,
"eval_rewards/accuracies": 0.6345000267028809,
"eval_rewards/chosen": -0.03813740611076355,
"eval_rewards/margins": 0.054497238248586655,
"eval_rewards/rejected": -0.0926346406340599,
"eval_runtime": 539.933,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 2100
},
{
"epoch": 0.55,
"learning_rate": 2.4817244638019333e-06,
"logits/chosen": 0.9688129425048828,
"logits/rejected": 1.0826255083084106,
"logps/chosen": -259.4082336425781,
"logps/rejected": -252.29434204101562,
"loss": 0.6898,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.03624454885721207,
"rewards/margins": 0.06441988795995712,
"rewards/rejected": -0.10066443681716919,
"step": 2110
},
{
"epoch": 0.55,
"learning_rate": 2.4588815314058155e-06,
"logits/chosen": 0.9535354375839233,
"logits/rejected": 0.9888173937797546,
"logps/chosen": -308.61053466796875,
"logps/rejected": -290.3708801269531,
"loss": 0.692,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.033887576311826706,
"rewards/margins": 0.038659535348415375,
"rewards/rejected": -0.07254711538553238,
"step": 2120
},
{
"epoch": 0.56,
"learning_rate": 2.4360420323899922e-06,
"logits/chosen": 0.9049111604690552,
"logits/rejected": 1.0025997161865234,
"logps/chosen": -300.1152648925781,
"logps/rejected": -246.1627197265625,
"loss": 0.69,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.03963657096028328,
"rewards/margins": 0.07257360219955444,
"rewards/rejected": -0.11221016943454742,
"step": 2130
},
{
"epoch": 0.56,
"learning_rate": 2.4132078738460585e-06,
"logits/chosen": 0.985696017742157,
"logits/rejected": 0.9788693189620972,
"logps/chosen": -273.555419921875,
"logps/rejected": -260.4004211425781,
"loss": 0.6922,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.04467379301786423,
"rewards/margins": 0.04357798025012016,
"rewards/rejected": -0.08825178444385529,
"step": 2140
},
{
"epoch": 0.56,
"learning_rate": 2.3903809624196826e-06,
"logits/chosen": 0.989804744720459,
"logits/rejected": 1.0961401462554932,
"logps/chosen": -260.736328125,
"logps/rejected": -231.63491821289062,
"loss": 0.691,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.051245905458927155,
"rewards/margins": 0.05609399080276489,
"rewards/rejected": -0.10733989626169205,
"step": 2150
},
{
"epoch": 0.57,
"learning_rate": 2.3675632041513978e-06,
"logits/chosen": 1.0626736879348755,
"logits/rejected": 1.1053330898284912,
"logps/chosen": -243.4201202392578,
"logps/rejected": -225.9866180419922,
"loss": 0.6922,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.052421581000089645,
"rewards/margins": 0.037004150450229645,
"rewards/rejected": -0.08942572772502899,
"step": 2160
},
{
"epoch": 0.57,
"learning_rate": 2.3447565043174533e-06,
"logits/chosen": 0.9754394292831421,
"logits/rejected": 0.9671268463134766,
"logps/chosen": -273.8443603515625,
"logps/rejected": -249.9600372314453,
"loss": 0.6911,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.043098267167806625,
"rewards/margins": 0.04836183041334152,
"rewards/rejected": -0.09146009385585785,
"step": 2170
},
{
"epoch": 0.57,
"learning_rate": 2.321962767270724e-06,
"logits/chosen": 1.0004395246505737,
"logits/rejected": 1.0548702478408813,
"logps/chosen": -258.23388671875,
"logps/rejected": -254.6455841064453,
"loss": 0.6909,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.042513586580753326,
"rewards/margins": 0.06627936661243439,
"rewards/rejected": -0.10879294574260712,
"step": 2180
},
{
"epoch": 0.57,
"learning_rate": 2.299183896281692e-06,
"logits/chosen": 0.932357907295227,
"logits/rejected": 1.1255356073379517,
"logps/chosen": -272.53533935546875,
"logps/rejected": -232.40127563476562,
"loss": 0.6905,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.040888115763664246,
"rewards/margins": 0.07338190823793411,
"rewards/rejected": -0.11427001655101776,
"step": 2190
},
{
"epoch": 0.58,
"learning_rate": 2.2764217933795297e-06,
"logits/chosen": 0.9959739446640015,
"logits/rejected": 0.9297063946723938,
"logps/chosen": -290.0731201171875,
"logps/rejected": -259.72259521484375,
"loss": 0.6917,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.048611197620630264,
"rewards/margins": 0.04705143719911575,
"rewards/rejected": -0.09566263109445572,
"step": 2200
},
{
"epoch": 0.58,
"eval_logits/chosen": 0.9399436116218567,
"eval_logits/rejected": 1.0406934022903442,
"eval_logps/chosen": -281.7611389160156,
"eval_logps/rejected": -253.66929626464844,
"eval_loss": 0.690939724445343,
"eval_rewards/accuracies": 0.6324999928474426,
"eval_rewards/chosen": -0.042143091559410095,
"eval_rewards/margins": 0.05639072135090828,
"eval_rewards/rejected": -0.09853381663560867,
"eval_runtime": 539.9748,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 2200
},
{
"epoch": 0.58,
"learning_rate": 2.2536783591932786e-06,
"logits/chosen": 1.089950680732727,
"logits/rejected": 0.9905519485473633,
"logps/chosen": -274.2284851074219,
"logps/rejected": -254.03823852539062,
"loss": 0.6907,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.03807093948125839,
"rewards/margins": 0.06290793418884277,
"rewards/rejected": -0.10097887367010117,
"step": 2210
},
{
"epoch": 0.58,
"learning_rate": 2.230955492793149e-06,
"logits/chosen": 0.9911603927612305,
"logits/rejected": 1.117432951927185,
"logps/chosen": -274.63616943359375,
"logps/rejected": -236.0727996826172,
"loss": 0.6904,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.029746342450380325,
"rewards/margins": 0.05460263043642044,
"rewards/rejected": -0.08434897661209106,
"step": 2220
},
{
"epoch": 0.58,
"learning_rate": 2.208255091531947e-06,
"logits/chosen": 0.9844030141830444,
"logits/rejected": 1.0057882070541382,
"logps/chosen": -267.25860595703125,
"logps/rejected": -221.1670684814453,
"loss": 0.6913,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.02886761724948883,
"rewards/margins": 0.06318513303995132,
"rewards/rejected": -0.09205274283885956,
"step": 2230
},
{
"epoch": 0.59,
"learning_rate": 2.1855790508866435e-06,
"logits/chosen": 0.9558299779891968,
"logits/rejected": 1.0252538919448853,
"logps/chosen": -292.6085510253906,
"logps/rejected": -274.9212951660156,
"loss": 0.6913,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.02827531099319458,
"rewards/margins": 0.041763827204704285,
"rewards/rejected": -0.07003913819789886,
"step": 2240
},
{
"epoch": 0.59,
"learning_rate": 2.162929264300107e-06,
"logits/chosen": 1.0085655450820923,
"logits/rejected": 1.0989354848861694,
"logps/chosen": -255.2216796875,
"logps/rejected": -226.0516815185547,
"loss": 0.691,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.036384742707014084,
"rewards/margins": 0.05588601902127266,
"rewards/rejected": -0.09227076172828674,
"step": 2250
},
{
"epoch": 0.59,
"learning_rate": 2.1403076230230006e-06,
"logits/chosen": 0.9903377294540405,
"logits/rejected": 1.0552234649658203,
"logps/chosen": -270.6020202636719,
"logps/rejected": -208.5309295654297,
"loss": 0.6921,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.04310330003499985,
"rewards/margins": 0.04129331558942795,
"rewards/rejected": -0.0843966156244278,
"step": 2260
},
{
"epoch": 0.59,
"learning_rate": 2.11771601595586e-06,
"logits/chosen": 0.9834343791007996,
"logits/rejected": 1.0014784336090088,
"logps/chosen": -259.8085632324219,
"logps/rejected": -261.5582580566406,
"loss": 0.6916,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.023330455645918846,
"rewards/margins": 0.04660937935113907,
"rewards/rejected": -0.06993982940912247,
"step": 2270
},
{
"epoch": 0.6,
"learning_rate": 2.0951563294913737e-06,
"logits/chosen": 0.9085767865180969,
"logits/rejected": 1.0599794387817383,
"logps/chosen": -264.39208984375,
"logps/rejected": -222.37948608398438,
"loss": 0.6916,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.03771457076072693,
"rewards/margins": 0.04855852574110031,
"rewards/rejected": -0.08627309650182724,
"step": 2280
},
{
"epoch": 0.6,
"learning_rate": 2.0726304473568693e-06,
"logits/chosen": 0.9768549799919128,
"logits/rejected": 1.0272681713104248,
"logps/chosen": -250.63037109375,
"logps/rejected": -270.17938232421875,
"loss": 0.6915,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.03126218914985657,
"rewards/margins": 0.057791270315647125,
"rewards/rejected": -0.08905345946550369,
"step": 2290
},
{
"epoch": 0.6,
"learning_rate": 2.050140250457023e-06,
"logits/chosen": 0.9690292477607727,
"logits/rejected": 0.9458838701248169,
"logps/chosen": -250.7799072265625,
"logps/rejected": -240.507568359375,
"loss": 0.6909,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.026477226987481117,
"rewards/margins": 0.053064655512571335,
"rewards/rejected": -0.079541876912117,
"step": 2300
},
{
"epoch": 0.6,
"eval_logits/chosen": 0.9399006366729736,
"eval_logits/rejected": 1.0408267974853516,
"eval_logps/chosen": -280.7285461425781,
"eval_logps/rejected": -252.42715454101562,
"eval_loss": 0.6909221410751343,
"eval_rewards/accuracies": 0.6334999799728394,
"eval_rewards/chosen": -0.03181701526045799,
"eval_rewards/margins": 0.05429535731673241,
"eval_rewards/rejected": -0.0861123651266098,
"eval_runtime": 540.0931,
"eval_samples_per_second": 3.703,
"eval_steps_per_second": 0.926,
"step": 2300
},
{
"epoch": 0.6,
"learning_rate": 2.0276876167168042e-06,
"logits/chosen": 0.9707075357437134,
"logits/rejected": 1.1204993724822998,
"logps/chosen": -285.9052734375,
"logps/rejected": -265.517333984375,
"loss": 0.6905,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.02710430696606636,
"rewards/margins": 0.0629369467496872,
"rewards/rejected": -0.09004124999046326,
"step": 2310
},
{
"epoch": 0.61,
"learning_rate": 2.0052744209246682e-06,
"logits/chosen": 0.9497090578079224,
"logits/rejected": 1.0253454446792603,
"logps/chosen": -259.4183044433594,
"logps/rejected": -253.4176483154297,
"loss": 0.691,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.03496091067790985,
"rewards/margins": 0.045159030705690384,
"rewards/rejected": -0.08011993765830994,
"step": 2320
},
{
"epoch": 0.61,
"learning_rate": 1.9829025345760127e-06,
"logits/chosen": 0.9671627283096313,
"logits/rejected": 1.0244706869125366,
"logps/chosen": -316.0784912109375,
"logps/rejected": -272.86224365234375,
"loss": 0.691,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.03136713430285454,
"rewards/margins": 0.05638208985328674,
"rewards/rejected": -0.08774922788143158,
"step": 2330
},
{
"epoch": 0.61,
"learning_rate": 1.9605738257169115e-06,
"logits/chosen": 1.038586974143982,
"logits/rejected": 1.0072470903396606,
"logps/chosen": -250.32870483398438,
"logps/rejected": -251.5189971923828,
"loss": 0.691,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.04340856149792671,
"rewards/margins": 0.0459757074713707,
"rewards/rejected": -0.08938425779342651,
"step": 2340
},
{
"epoch": 0.62,
"learning_rate": 1.9382901587881275e-06,
"logits/chosen": 0.9293440580368042,
"logits/rejected": 1.0626415014266968,
"logps/chosen": -260.1195373535156,
"logps/rejected": -246.67117309570312,
"loss": 0.6912,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.03792440518736839,
"rewards/margins": 0.0486772395670414,
"rewards/rejected": -0.08660164475440979,
"step": 2350
},
{
"epoch": 0.62,
"learning_rate": 1.916053394469437e-06,
"logits/chosen": 0.9554189443588257,
"logits/rejected": 1.0219160318374634,
"logps/chosen": -227.74856567382812,
"logps/rejected": -232.154296875,
"loss": 0.691,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.03027189150452614,
"rewards/margins": 0.05248479172587395,
"rewards/rejected": -0.08275668323040009,
"step": 2360
},
{
"epoch": 0.62,
"learning_rate": 1.8938653895242604e-06,
"logits/chosen": 0.9694328308105469,
"logits/rejected": 0.998115062713623,
"logps/chosen": -242.61392211914062,
"logps/rejected": -225.9537811279297,
"loss": 0.691,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.01270161010324955,
"rewards/margins": 0.06611243635416031,
"rewards/rejected": -0.07881404459476471,
"step": 2370
},
{
"epoch": 0.62,
"learning_rate": 1.8717279966446267e-06,
"logits/chosen": 0.939893901348114,
"logits/rejected": 1.0379140377044678,
"logps/chosen": -271.5564270019531,
"logps/rejected": -256.105712890625,
"loss": 0.69,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.024293258786201477,
"rewards/margins": 0.05607147887349129,
"rewards/rejected": -0.08036474138498306,
"step": 2380
},
{
"epoch": 0.63,
"learning_rate": 1.8496430642964698e-06,
"logits/chosen": 0.9116802215576172,
"logits/rejected": 1.035742998123169,
"logps/chosen": -255.07760620117188,
"logps/rejected": -239.1924285888672,
"loss": 0.6922,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.019214514642953873,
"rewards/margins": 0.05538605526089668,
"rewards/rejected": -0.07460056990385056,
"step": 2390
},
{
"epoch": 0.63,
"learning_rate": 1.827612436565286e-06,
"logits/chosen": 0.926386833190918,
"logits/rejected": 0.9954707026481628,
"logps/chosen": -254.49209594726562,
"logps/rejected": -233.5257568359375,
"loss": 0.6903,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.027828719466924667,
"rewards/margins": 0.059567712247371674,
"rewards/rejected": -0.08739643543958664,
"step": 2400
},
{
"epoch": 0.63,
"eval_logits/chosen": 0.9198330044746399,
"eval_logits/rejected": 1.021889328956604,
"eval_logps/chosen": -280.510009765625,
"eval_logps/rejected": -252.31210327148438,
"eval_loss": 0.6909388303756714,
"eval_rewards/accuracies": 0.6359999775886536,
"eval_rewards/chosen": -0.029631877318024635,
"eval_rewards/margins": 0.05532996356487274,
"eval_rewards/rejected": -0.08496184647083282,
"eval_runtime": 539.9384,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 2400
},
{
"epoch": 0.63,
"learning_rate": 1.8056379530021492e-06,
"logits/chosen": 0.8990974426269531,
"logits/rejected": 1.0346615314483643,
"logps/chosen": -256.46343994140625,
"logps/rejected": -225.152099609375,
"loss": 0.691,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.029439201578497887,
"rewards/margins": 0.060266874730587006,
"rewards/rejected": -0.08970607817173004,
"step": 2410
},
{
"epoch": 0.63,
"learning_rate": 1.7837214484701154e-06,
"logits/chosen": 0.9923946261405945,
"logits/rejected": 1.0339380502700806,
"logps/chosen": -273.4414367675781,
"logps/rejected": -264.6556091308594,
"loss": 0.6892,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.029256407171487808,
"rewards/margins": 0.07676726579666138,
"rewards/rejected": -0.10602366924285889,
"step": 2420
},
{
"epoch": 0.64,
"learning_rate": 1.7618647529910043e-06,
"logits/chosen": 0.9024986028671265,
"logits/rejected": 1.0799829959869385,
"logps/chosen": -260.97979736328125,
"logps/rejected": -240.7161102294922,
"loss": 0.6897,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.025985723361372948,
"rewards/margins": 0.06577527523040771,
"rewards/rejected": -0.09176099300384521,
"step": 2430
},
{
"epoch": 0.64,
"learning_rate": 1.7400696915925996e-06,
"logits/chosen": 0.9433887600898743,
"logits/rejected": 1.0706889629364014,
"logps/chosen": -278.13543701171875,
"logps/rejected": -270.07415771484375,
"loss": 0.6916,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.01627534069120884,
"rewards/margins": 0.07549260556697845,
"rewards/rejected": -0.09176793694496155,
"step": 2440
},
{
"epoch": 0.64,
"learning_rate": 1.718338084156254e-06,
"logits/chosen": 0.9400017857551575,
"logits/rejected": 0.9848964810371399,
"logps/chosen": -277.2012634277344,
"logps/rejected": -248.524169921875,
"loss": 0.6908,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.023953277617692947,
"rewards/margins": 0.05445709824562073,
"rewards/rejected": -0.07841037213802338,
"step": 2450
},
{
"epoch": 0.64,
"learning_rate": 1.6966717452649372e-06,
"logits/chosen": 0.9121321439743042,
"logits/rejected": 0.9821128845214844,
"logps/chosen": -280.4564208984375,
"logps/rejected": -272.4418640136719,
"loss": 0.6909,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.022658145055174828,
"rewards/margins": 0.058780424296855927,
"rewards/rejected": -0.0814385712146759,
"step": 2460
},
{
"epoch": 0.65,
"learning_rate": 1.6750724840517103e-06,
"logits/chosen": 0.9408855438232422,
"logits/rejected": 1.041475534439087,
"logps/chosen": -295.4039306640625,
"logps/rejected": -253.6959686279297,
"loss": 0.6892,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.02255399525165558,
"rewards/margins": 0.0683935284614563,
"rewards/rejected": -0.09094752371311188,
"step": 2470
},
{
"epoch": 0.65,
"learning_rate": 1.6535421040486686e-06,
"logits/chosen": 0.9445545077323914,
"logits/rejected": 0.9898989796638489,
"logps/chosen": -257.03387451171875,
"logps/rejected": -232.8082275390625,
"loss": 0.6907,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.022778628394007683,
"rewards/margins": 0.05287964269518852,
"rewards/rejected": -0.07565827667713165,
"step": 2480
},
{
"epoch": 0.65,
"learning_rate": 1.6320824030363458e-06,
"logits/chosen": 0.8453947901725769,
"logits/rejected": 1.0863592624664307,
"logps/chosen": -268.630615234375,
"logps/rejected": -246.2313690185547,
"loss": 0.691,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.030399400740861893,
"rewards/margins": 0.0645337849855423,
"rewards/rejected": -0.09493318945169449,
"step": 2490
},
{
"epoch": 0.65,
"learning_rate": 1.6106951728936028e-06,
"logits/chosen": 0.9887116551399231,
"logits/rejected": 1.091596007347107,
"logps/chosen": -232.3385772705078,
"logps/rejected": -221.32968139648438,
"loss": 0.6908,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.03866693750023842,
"rewards/margins": 0.05171254277229309,
"rewards/rejected": -0.09037948399782181,
"step": 2500
},
{
"epoch": 0.65,
"eval_logits/chosen": 0.9195509552955627,
"eval_logits/rejected": 1.0213452577590942,
"eval_logps/chosen": -281.2753601074219,
"eval_logps/rejected": -253.40109252929688,
"eval_loss": 0.6909087896347046,
"eval_rewards/accuracies": 0.6330000162124634,
"eval_rewards/chosen": -0.037285856902599335,
"eval_rewards/margins": 0.05856594070792198,
"eval_rewards/rejected": -0.09585181623697281,
"eval_runtime": 539.985,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 2500
},
{
"epoch": 0.66,
"learning_rate": 1.5893821994479996e-06,
"logits/chosen": 1.0217519998550415,
"logits/rejected": 1.1035264730453491,
"logps/chosen": -280.2012939453125,
"logps/rejected": -262.41033935546875,
"loss": 0.6902,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.03267248719930649,
"rewards/margins": 0.07917577773332596,
"rewards/rejected": -0.11184825003147125,
"step": 2510
},
{
"epoch": 0.66,
"learning_rate": 1.5681452623266868e-06,
"logits/chosen": 0.9456014633178711,
"logits/rejected": 1.0302622318267822,
"logps/chosen": -300.8179931640625,
"logps/rejected": -262.069580078125,
"loss": 0.6915,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.03674257546663284,
"rewards/margins": 0.056660883128643036,
"rewards/rejected": -0.09340345859527588,
"step": 2520
},
{
"epoch": 0.66,
"learning_rate": 1.5469861348078014e-06,
"logits/chosen": 0.9554840326309204,
"logits/rejected": 0.9924384951591492,
"logps/chosen": -268.2906188964844,
"logps/rejected": -242.49649047851562,
"loss": 0.6904,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.032142072916030884,
"rewards/margins": 0.056136567145586014,
"rewards/rejected": -0.0882786363363266,
"step": 2530
},
{
"epoch": 0.66,
"learning_rate": 1.5259065836724035e-06,
"logits/chosen": 0.9138981103897095,
"logits/rejected": 0.9927932620048523,
"logps/chosen": -269.4898986816406,
"logps/rejected": -257.8581848144531,
"loss": 0.6908,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.026234816759824753,
"rewards/margins": 0.050769805908203125,
"rewards/rejected": -0.07700462639331818,
"step": 2540
},
{
"epoch": 0.67,
"learning_rate": 1.5049083690569456e-06,
"logits/chosen": 0.8760209083557129,
"logits/rejected": 0.9947643280029297,
"logps/chosen": -277.1868591308594,
"logps/rejected": -236.7860870361328,
"loss": 0.6896,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.045319218188524246,
"rewards/margins": 0.06617378443479538,
"rewards/rejected": -0.11149300634860992,
"step": 2550
},
{
"epoch": 0.67,
"learning_rate": 1.4839932443063057e-06,
"logits/chosen": 1.0274193286895752,
"logits/rejected": 1.0255438089370728,
"logps/chosen": -241.68191528320312,
"logps/rejected": -215.37057495117188,
"loss": 0.6908,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.046610768884420395,
"rewards/margins": 0.05322499945759773,
"rewards/rejected": -0.09983576834201813,
"step": 2560
},
{
"epoch": 0.67,
"learning_rate": 1.4631629558273803e-06,
"logits/chosen": 0.9781502485275269,
"logits/rejected": 1.0266082286834717,
"logps/chosen": -291.5887756347656,
"logps/rejected": -256.538818359375,
"loss": 0.6903,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.028579484671354294,
"rewards/margins": 0.07505873590707779,
"rewards/rejected": -0.10363821685314178,
"step": 2570
},
{
"epoch": 0.68,
"learning_rate": 1.4424192429432657e-06,
"logits/chosen": 0.9894530177116394,
"logits/rejected": 1.038171648979187,
"logps/chosen": -308.27691650390625,
"logps/rejected": -225.60122680664062,
"loss": 0.6904,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.03209725767374039,
"rewards/margins": 0.06519778817892075,
"rewards/rejected": -0.09729506075382233,
"step": 2580
},
{
"epoch": 0.68,
"learning_rate": 1.421763837748016e-06,
"logits/chosen": 0.9771683812141418,
"logits/rejected": 1.058801531791687,
"logps/chosen": -268.15179443359375,
"logps/rejected": -273.706298828125,
"loss": 0.6905,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.05141326040029526,
"rewards/margins": 0.058643460273742676,
"rewards/rejected": -0.11005672067403793,
"step": 2590
},
{
"epoch": 0.68,
"learning_rate": 1.401198464962021e-06,
"logits/chosen": 1.0130704641342163,
"logits/rejected": 0.9349973797798157,
"logps/chosen": -273.2847900390625,
"logps/rejected": -240.21829223632812,
"loss": 0.6907,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.0360480472445488,
"rewards/margins": 0.05611228942871094,
"rewards/rejected": -0.09216034412384033,
"step": 2600
},
{
"epoch": 0.68,
"eval_logits/chosen": 0.9161260724067688,
"eval_logits/rejected": 1.017266035079956,
"eval_logps/chosen": -281.78839111328125,
"eval_logps/rejected": -254.0473175048828,
"eval_loss": 0.6909086108207703,
"eval_rewards/accuracies": 0.6294999718666077,
"eval_rewards/chosen": -0.042415801435709,
"eval_rewards/margins": 0.05989806354045868,
"eval_rewards/rejected": -0.10231386125087738,
"eval_runtime": 540.1872,
"eval_samples_per_second": 3.702,
"eval_steps_per_second": 0.926,
"step": 2600
},
{
"epoch": 0.68,
"learning_rate": 1.3807248417879896e-06,
"logits/chosen": 0.932266354560852,
"logits/rejected": 0.9696345329284668,
"logps/chosen": -259.15472412109375,
"logps/rejected": -220.0342559814453,
"loss": 0.6914,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.02964337170124054,
"rewards/margins": 0.05693807080388069,
"rewards/rejected": -0.08658144623041153,
"step": 2610
},
{
"epoch": 0.69,
"learning_rate": 1.3603446777675665e-06,
"logits/chosen": 0.9445293545722961,
"logits/rejected": 0.9707993268966675,
"logps/chosen": -260.50518798828125,
"logps/rejected": -241.0626983642578,
"loss": 0.6909,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.038475148379802704,
"rewards/margins": 0.05124124884605408,
"rewards/rejected": -0.08971639722585678,
"step": 2620
},
{
"epoch": 0.69,
"learning_rate": 1.3400596746385817e-06,
"logits/chosen": 0.9341602325439453,
"logits/rejected": 0.9812172055244446,
"logps/chosen": -291.263671875,
"logps/rejected": -256.05816650390625,
"loss": 0.6898,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.03991129249334335,
"rewards/margins": 0.06936169415712357,
"rewards/rejected": -0.10927299410104752,
"step": 2630
},
{
"epoch": 0.69,
"learning_rate": 1.3198715261929587e-06,
"logits/chosen": 0.9518648386001587,
"logits/rejected": 1.0701024532318115,
"logps/chosen": -322.59869384765625,
"logps/rejected": -246.1614532470703,
"loss": 0.6916,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.032660938799381256,
"rewards/margins": 0.045336611568927765,
"rewards/rejected": -0.07799754291772842,
"step": 2640
},
{
"epoch": 0.69,
"learning_rate": 1.2997819181352823e-06,
"logits/chosen": 0.9637743234634399,
"logits/rejected": 0.9643661379814148,
"logps/chosen": -299.9573669433594,
"logps/rejected": -267.3519592285156,
"loss": 0.6909,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.03847068175673485,
"rewards/margins": 0.05295687913894653,
"rewards/rejected": -0.09142756462097168,
"step": 2650
},
{
"epoch": 0.7,
"learning_rate": 1.2797925279420454e-06,
"logits/chosen": 0.9999778866767883,
"logits/rejected": 1.0599424839019775,
"logps/chosen": -291.417724609375,
"logps/rejected": -258.858642578125,
"loss": 0.6911,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.04292257875204086,
"rewards/margins": 0.045674268156290054,
"rewards/rejected": -0.08859684318304062,
"step": 2660
},
{
"epoch": 0.7,
"learning_rate": 1.2599050247215764e-06,
"logits/chosen": 0.9801284670829773,
"logits/rejected": 1.0555846691131592,
"logps/chosen": -286.49017333984375,
"logps/rejected": -262.8829650878906,
"loss": 0.69,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.02869715727865696,
"rewards/margins": 0.07526004314422607,
"rewards/rejected": -0.10395719856023788,
"step": 2670
},
{
"epoch": 0.7,
"learning_rate": 1.2401210690746705e-06,
"logits/chosen": 0.9294122457504272,
"logits/rejected": 1.0954596996307373,
"logps/chosen": -264.6773986816406,
"logps/rejected": -241.242431640625,
"loss": 0.6916,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.03240332007408142,
"rewards/margins": 0.04624287039041519,
"rewards/rejected": -0.07864619046449661,
"step": 2680
},
{
"epoch": 0.7,
"learning_rate": 1.2204423129559306e-06,
"logits/chosen": 0.8595544695854187,
"logits/rejected": 1.0383957624435425,
"logps/chosen": -296.6278076171875,
"logps/rejected": -270.6478576660156,
"loss": 0.6912,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.030149826779961586,
"rewards/margins": 0.045799605548381805,
"rewards/rejected": -0.07594943791627884,
"step": 2690
},
{
"epoch": 0.71,
"learning_rate": 1.20087039953583e-06,
"logits/chosen": 0.9707738757133484,
"logits/rejected": 0.9931126832962036,
"logps/chosen": -281.4847412109375,
"logps/rejected": -259.38934326171875,
"loss": 0.6905,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.023116273805499077,
"rewards/margins": 0.05793018266558647,
"rewards/rejected": -0.0810464546084404,
"step": 2700
},
{
"epoch": 0.71,
"eval_logits/chosen": 0.9119435548782349,
"eval_logits/rejected": 1.0139191150665283,
"eval_logps/chosen": -281.0736083984375,
"eval_logps/rejected": -253.19638061523438,
"eval_loss": 0.6908898949623108,
"eval_rewards/accuracies": 0.6309999823570251,
"eval_rewards/chosen": -0.03526770696043968,
"eval_rewards/margins": 0.05853661522269249,
"eval_rewards/rejected": -0.09380432963371277,
"eval_runtime": 540.0121,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 2700
},
{
"epoch": 0.71,
"learning_rate": 1.181406963063507e-06,
"logits/chosen": 0.9695969820022583,
"logits/rejected": 0.9412325024604797,
"logps/chosen": -277.9027099609375,
"logps/rejected": -233.14767456054688,
"loss": 0.6915,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.029822606593370438,
"rewards/margins": 0.05514361336827278,
"rewards/rejected": -0.08496621251106262,
"step": 2710
},
{
"epoch": 0.71,
"learning_rate": 1.1620536287303052e-06,
"logits/chosen": 0.9639630317687988,
"logits/rejected": 1.0453459024429321,
"logps/chosen": -261.4854736328125,
"logps/rejected": -243.4965057373047,
"loss": 0.6913,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.037622541189193726,
"rewards/margins": 0.05851644277572632,
"rewards/rejected": -0.09613899141550064,
"step": 2720
},
{
"epoch": 0.71,
"learning_rate": 1.1428120125340717e-06,
"logits/chosen": 1.0106990337371826,
"logits/rejected": 0.9288986325263977,
"logps/chosen": -273.11346435546875,
"logps/rejected": -243.2933807373047,
"loss": 0.6903,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.03193587437272072,
"rewards/margins": 0.055615246295928955,
"rewards/rejected": -0.08755112439393997,
"step": 2730
},
{
"epoch": 0.72,
"learning_rate": 1.123683721144223e-06,
"logits/chosen": 0.9593019485473633,
"logits/rejected": 0.9824401140213013,
"logps/chosen": -273.22357177734375,
"logps/rejected": -241.69149780273438,
"loss": 0.6893,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.03742024675011635,
"rewards/margins": 0.07019098103046417,
"rewards/rejected": -0.10761122405529022,
"step": 2740
},
{
"epoch": 0.72,
"learning_rate": 1.1046703517675848e-06,
"logits/chosen": 0.9543115496635437,
"logits/rejected": 0.9525305032730103,
"logps/chosen": -247.79141235351562,
"logps/rejected": -250.76815795898438,
"loss": 0.6909,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.03152688220143318,
"rewards/margins": 0.05581844598054886,
"rewards/rejected": -0.08734532445669174,
"step": 2750
},
{
"epoch": 0.72,
"learning_rate": 1.085773492015028e-06,
"logits/chosen": 0.9415119290351868,
"logits/rejected": 1.0495812892913818,
"logps/chosen": -288.77117919921875,
"logps/rejected": -236.2522735595703,
"loss": 0.69,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.034685637801885605,
"rewards/margins": 0.07457654178142548,
"rewards/rejected": -0.10926218330860138,
"step": 2760
},
{
"epoch": 0.72,
"learning_rate": 1.0669947197689034e-06,
"logits/chosen": 0.9325952529907227,
"logits/rejected": 0.9842671155929565,
"logps/chosen": -230.1918487548828,
"logps/rejected": -243.35086059570312,
"loss": 0.6915,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": -0.04897845536470413,
"rewards/margins": 0.04100048914551735,
"rewards/rejected": -0.08997894823551178,
"step": 2770
},
{
"epoch": 0.73,
"learning_rate": 1.048335603051291e-06,
"logits/chosen": 0.8943045735359192,
"logits/rejected": 1.0183279514312744,
"logps/chosen": -272.97314453125,
"logps/rejected": -255.63125610351562,
"loss": 0.6906,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.03204040229320526,
"rewards/margins": 0.0708232969045639,
"rewards/rejected": -0.10286370664834976,
"step": 2780
},
{
"epoch": 0.73,
"learning_rate": 1.0297976998930665e-06,
"logits/chosen": 0.897107720375061,
"logits/rejected": 0.9336503744125366,
"logps/chosen": -278.9725341796875,
"logps/rejected": -243.04238891601562,
"loss": 0.6913,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.027060145512223244,
"rewards/margins": 0.05337075516581535,
"rewards/rejected": -0.08043090254068375,
"step": 2790
},
{
"epoch": 0.73,
"learning_rate": 1.0113825582038078e-06,
"logits/chosen": 0.9530097246170044,
"logits/rejected": 1.0400608777999878,
"logps/chosen": -271.3522033691406,
"logps/rejected": -225.7040557861328,
"loss": 0.692,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.03952925279736519,
"rewards/margins": 0.05476094409823418,
"rewards/rejected": -0.09429020434617996,
"step": 2800
},
{
"epoch": 0.73,
"eval_logits/chosen": 0.9141185283660889,
"eval_logits/rejected": 1.0163326263427734,
"eval_logps/chosen": -280.8155517578125,
"eval_logps/rejected": -252.75259399414062,
"eval_loss": 0.6908916234970093,
"eval_rewards/accuracies": 0.6305000185966492,
"eval_rewards/chosen": -0.0326874740421772,
"eval_rewards/margins": 0.05667929723858833,
"eval_rewards/rejected": -0.08936676383018494,
"eval_runtime": 540.3199,
"eval_samples_per_second": 3.702,
"eval_steps_per_second": 0.925,
"step": 2800
},
{
"epoch": 0.74,
"learning_rate": 9.930917156425477e-07,
"logits/chosen": 0.9315062761306763,
"logits/rejected": 1.0539835691452026,
"logps/chosen": -247.2057342529297,
"logps/rejected": -225.90493774414062,
"loss": 0.6906,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.02216990664601326,
"rewards/margins": 0.07055743038654327,
"rewards/rejected": -0.09272731840610504,
"step": 2810
},
{
"epoch": 0.74,
"learning_rate": 9.749266994893756e-07,
"logits/chosen": 0.9194700121879578,
"logits/rejected": 1.008080244064331,
"logps/chosen": -242.8909454345703,
"logps/rejected": -255.8279266357422,
"loss": 0.6914,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.033027276396751404,
"rewards/margins": 0.043912582099437714,
"rewards/rejected": -0.07693986594676971,
"step": 2820
},
{
"epoch": 0.74,
"learning_rate": 9.56889026517913e-07,
"logits/chosen": 0.95648193359375,
"logits/rejected": 1.0707709789276123,
"logps/chosen": -272.9548645019531,
"logps/rejected": -241.94235229492188,
"loss": 0.6899,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.030579229816794395,
"rewards/margins": 0.06354820728302002,
"rewards/rejected": -0.09412743896245956,
"step": 2830
},
{
"epoch": 0.74,
"learning_rate": 9.389802028686617e-07,
"logits/chosen": 0.9589303135871887,
"logits/rejected": 0.9868467450141907,
"logps/chosen": -219.3035125732422,
"logps/rejected": -192.64849853515625,
"loss": 0.6904,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.0318605974316597,
"rewards/margins": 0.05495452880859375,
"rewards/rejected": -0.08681513369083405,
"step": 2840
},
{
"epoch": 0.75,
"learning_rate": 9.212017239232427e-07,
"logits/chosen": 0.9364684820175171,
"logits/rejected": 1.0579065084457397,
"logps/chosen": -272.2196350097656,
"logps/rejected": -280.43341064453125,
"loss": 0.6894,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.020410938188433647,
"rewards/margins": 0.0782955139875412,
"rewards/rejected": -0.0987064465880394,
"step": 2850
},
{
"epoch": 0.75,
"learning_rate": 9.03555074179533e-07,
"logits/chosen": 0.9313532114028931,
"logits/rejected": 1.04340398311615,
"logps/chosen": -267.8077087402344,
"logps/rejected": -246.677001953125,
"loss": 0.6908,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.024608857929706573,
"rewards/margins": 0.06871498376131058,
"rewards/rejected": -0.09332384169101715,
"step": 2860
},
{
"epoch": 0.75,
"learning_rate": 8.860417271277067e-07,
"logits/chosen": 0.9149921536445618,
"logits/rejected": 1.0074148178100586,
"logps/chosen": -243.07144165039062,
"logps/rejected": -241.79171752929688,
"loss": 0.6902,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.029673978686332703,
"rewards/margins": 0.0581536665558815,
"rewards/rejected": -0.0878276452422142,
"step": 2870
},
{
"epoch": 0.75,
"learning_rate": 8.686631451272029e-07,
"logits/chosen": 0.9447315335273743,
"logits/rejected": 0.9529320001602173,
"logps/chosen": -269.72869873046875,
"logps/rejected": -222.9635467529297,
"loss": 0.6911,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.03748570382595062,
"rewards/margins": 0.05406556650996208,
"rewards/rejected": -0.0915512815117836,
"step": 2880
},
{
"epoch": 0.76,
"learning_rate": 8.514207792846168e-07,
"logits/chosen": 1.0064308643341064,
"logits/rejected": 1.0328409671783447,
"logps/chosen": -253.03274536132812,
"logps/rejected": -257.581298828125,
"loss": 0.69,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.03226761519908905,
"rewards/margins": 0.06126902252435684,
"rewards/rejected": -0.09353663772344589,
"step": 2890
},
{
"epoch": 0.76,
"learning_rate": 8.343160693325356e-07,
"logits/chosen": 0.9581842422485352,
"logits/rejected": 1.0024080276489258,
"logps/chosen": -233.0767364501953,
"logps/rejected": -227.9841766357422,
"loss": 0.6906,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.035527393221855164,
"rewards/margins": 0.056871771812438965,
"rewards/rejected": -0.09239916503429413,
"step": 2900
},
{
"epoch": 0.76,
"eval_logits/chosen": 0.9098308682441711,
"eval_logits/rejected": 1.0122586488723755,
"eval_logps/chosen": -280.8845520019531,
"eval_logps/rejected": -252.85272216796875,
"eval_loss": 0.6908985376358032,
"eval_rewards/accuracies": 0.6294999718666077,
"eval_rewards/chosen": -0.033377815037965775,
"eval_rewards/margins": 0.0569901280105114,
"eval_rewards/rejected": -0.09036794304847717,
"eval_runtime": 540.038,
"eval_samples_per_second": 3.703,
"eval_steps_per_second": 0.926,
"step": 2900
},
{
"epoch": 0.76,
"learning_rate": 8.173504435093174e-07,
"logits/chosen": 0.9396654963493347,
"logits/rejected": 0.9836663007736206,
"logps/chosen": -286.25970458984375,
"logps/rejected": -269.1986083984375,
"loss": 0.6916,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.03581177443265915,
"rewards/margins": 0.04660804197192192,
"rewards/rejected": -0.08241982758045197,
"step": 2910
},
{
"epoch": 0.76,
"learning_rate": 8.00525318439836e-07,
"logits/chosen": 0.9897071123123169,
"logits/rejected": 1.0295822620391846,
"logps/chosen": -272.0080871582031,
"logps/rejected": -232.94650268554688,
"loss": 0.6908,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.035867296159267426,
"rewards/margins": 0.05984731763601303,
"rewards/rejected": -0.09571461379528046,
"step": 2920
},
{
"epoch": 0.77,
"learning_rate": 7.838420990171927e-07,
"logits/chosen": 0.9099165797233582,
"logits/rejected": 1.017539381980896,
"logps/chosen": -313.5694885253906,
"logps/rejected": -268.1796569824219,
"loss": 0.6903,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.026468079537153244,
"rewards/margins": 0.05932525545358658,
"rewards/rejected": -0.08579333126544952,
"step": 2930
},
{
"epoch": 0.77,
"learning_rate": 7.673021782854084e-07,
"logits/chosen": 0.9644147753715515,
"logits/rejected": 0.9656912088394165,
"logps/chosen": -250.9320831298828,
"logps/rejected": -264.95233154296875,
"loss": 0.6905,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.03290482237935066,
"rewards/margins": 0.0584709532558918,
"rewards/rejected": -0.09137578308582306,
"step": 2940
},
{
"epoch": 0.77,
"learning_rate": 7.509069373231039e-07,
"logits/chosen": 0.9676412343978882,
"logits/rejected": 0.9627419710159302,
"logps/chosen": -294.781982421875,
"logps/rejected": -271.5765380859375,
"loss": 0.6905,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.022627348080277443,
"rewards/margins": 0.06841419637203217,
"rewards/rejected": -0.09104155004024506,
"step": 2950
},
{
"epoch": 0.77,
"learning_rate": 7.346577451281822e-07,
"logits/chosen": 0.9224090576171875,
"logits/rejected": 1.0375105142593384,
"logps/chosen": -251.53451538085938,
"logps/rejected": -233.12393188476562,
"loss": 0.6909,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.04297469183802605,
"rewards/margins": 0.05884693190455437,
"rewards/rejected": -0.10182162374258041,
"step": 2960
},
{
"epoch": 0.78,
"learning_rate": 7.185559585035138e-07,
"logits/chosen": 0.9037634134292603,
"logits/rejected": 0.9682399034500122,
"logps/chosen": -286.7387390136719,
"logps/rejected": -260.4252014160156,
"loss": 0.6901,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.023535221815109253,
"rewards/margins": 0.057094089686870575,
"rewards/rejected": -0.08062931150197983,
"step": 2970
},
{
"epoch": 0.78,
"learning_rate": 7.026029219436504e-07,
"logits/chosen": 0.9559614062309265,
"logits/rejected": 1.007653832435608,
"logps/chosen": -306.8362121582031,
"logps/rejected": -272.37249755859375,
"loss": 0.6904,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.029121745377779007,
"rewards/margins": 0.0643700435757637,
"rewards/rejected": -0.09349179267883301,
"step": 2980
},
{
"epoch": 0.78,
"learning_rate": 6.867999675225523e-07,
"logits/chosen": 0.9615996479988098,
"logits/rejected": 0.9802320599555969,
"logps/chosen": -297.98504638671875,
"logps/rejected": -250.20980834960938,
"loss": 0.6911,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.02488793060183525,
"rewards/margins": 0.05079926922917366,
"rewards/rejected": -0.07568720728158951,
"step": 2990
},
{
"epoch": 0.79,
"learning_rate": 6.711484147823663e-07,
"logits/chosen": 0.9520236849784851,
"logits/rejected": 0.9852927923202515,
"logps/chosen": -270.9884033203125,
"logps/rejected": -242.1695098876953,
"loss": 0.6904,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.029557928442955017,
"rewards/margins": 0.06538344919681549,
"rewards/rejected": -0.0949413850903511,
"step": 3000
},
{
"epoch": 0.79,
"eval_logits/chosen": 0.9123407602310181,
"eval_logits/rejected": 1.0147464275360107,
"eval_logps/chosen": -280.66253662109375,
"eval_logps/rejected": -252.71670532226562,
"eval_loss": 0.6908931732177734,
"eval_rewards/accuracies": 0.6294999718666077,
"eval_rewards/chosen": -0.031157268211245537,
"eval_rewards/margins": 0.05785065144300461,
"eval_rewards/rejected": -0.0890079066157341,
"eval_runtime": 540.0515,
"eval_samples_per_second": 3.703,
"eval_steps_per_second": 0.926,
"step": 3000
},
{
"epoch": 0.79,
"learning_rate": 6.556495706232413e-07,
"logits/chosen": 0.9562872052192688,
"logits/rejected": 1.0361840724945068,
"logps/chosen": -274.87127685546875,
"logps/rejected": -239.40811157226562,
"loss": 0.6906,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.02163364365696907,
"rewards/margins": 0.06655408442020416,
"rewards/rejected": -0.08818772435188293,
"step": 3010
},
{
"epoch": 0.79,
"learning_rate": 6.403047291942057e-07,
"logits/chosen": 0.9342554211616516,
"logits/rejected": 0.9498234987258911,
"logps/chosen": -278.9698486328125,
"logps/rejected": -264.5009765625,
"loss": 0.6916,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.016571203246712685,
"rewards/margins": 0.055479150265455246,
"rewards/rejected": -0.07205035537481308,
"step": 3020
},
{
"epoch": 0.79,
"learning_rate": 6.251151717851023e-07,
"logits/chosen": 0.9194384813308716,
"logits/rejected": 0.9668914079666138,
"logps/chosen": -291.434814453125,
"logps/rejected": -275.8680114746094,
"loss": 0.6891,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.03377396613359451,
"rewards/margins": 0.07892084121704102,
"rewards/rejected": -0.11269481480121613,
"step": 3030
},
{
"epoch": 0.8,
"learning_rate": 6.100821667196041e-07,
"logits/chosen": 0.9412601590156555,
"logits/rejected": 1.0217000246047974,
"logps/chosen": -304.5139465332031,
"logps/rejected": -248.54312133789062,
"loss": 0.6907,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.01636696420609951,
"rewards/margins": 0.06269043684005737,
"rewards/rejected": -0.07905739545822144,
"step": 3040
},
{
"epoch": 0.8,
"learning_rate": 5.952069692493062e-07,
"logits/chosen": 0.88300621509552,
"logits/rejected": 1.0030475854873657,
"logps/chosen": -301.0616149902344,
"logps/rejected": -266.23419189453125,
"loss": 0.6905,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.011049589142203331,
"rewards/margins": 0.07019064575433731,
"rewards/rejected": -0.0812402293086052,
"step": 3050
},
{
"epoch": 0.8,
"learning_rate": 5.80490821448918e-07,
"logits/chosen": 0.9647413492202759,
"logits/rejected": 0.9922765493392944,
"logps/chosen": -222.5147705078125,
"logps/rejected": -206.96414184570312,
"loss": 0.6909,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.03366454318165779,
"rewards/margins": 0.058607954531908035,
"rewards/rejected": -0.09227249771356583,
"step": 3060
},
{
"epoch": 0.8,
"learning_rate": 5.659349521125459e-07,
"logits/chosen": 0.9332239031791687,
"logits/rejected": 1.0312001705169678,
"logps/chosen": -238.71188354492188,
"logps/rejected": -243.7023468017578,
"loss": 0.6903,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.033622294664382935,
"rewards/margins": 0.06797768920660019,
"rewards/rejected": -0.10159997642040253,
"step": 3070
},
{
"epoch": 0.81,
"learning_rate": 5.5154057665109e-07,
"logits/chosen": 0.9673851132392883,
"logits/rejected": 1.029404878616333,
"logps/chosen": -280.47100830078125,
"logps/rejected": -243.7905731201172,
"loss": 0.6917,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.02676049992442131,
"rewards/margins": 0.06190108135342598,
"rewards/rejected": -0.08866159617900848,
"step": 3080
},
{
"epoch": 0.81,
"learning_rate": 5.373088969907586e-07,
"logits/chosen": 0.9254564046859741,
"logits/rejected": 1.013877272605896,
"logps/chosen": -260.7253723144531,
"logps/rejected": -259.11224365234375,
"loss": 0.691,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.024502191692590714,
"rewards/margins": 0.059731971472501755,
"rewards/rejected": -0.08423416316509247,
"step": 3090
},
{
"epoch": 0.81,
"learning_rate": 5.23241101472709e-07,
"logits/chosen": 0.9964572191238403,
"logits/rejected": 0.9820792078971863,
"logps/chosen": -255.88662719726562,
"logps/rejected": -244.15097045898438,
"loss": 0.6905,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.02560483291745186,
"rewards/margins": 0.0563865527510643,
"rewards/rejected": -0.08199138939380646,
"step": 3100
},
{
"epoch": 0.81,
"eval_logits/chosen": 0.9147237539291382,
"eval_logits/rejected": 1.0175365209579468,
"eval_logps/chosen": -280.5528869628906,
"eval_logps/rejected": -252.58462524414062,
"eval_loss": 0.6908957958221436,
"eval_rewards/accuracies": 0.6330000162124634,
"eval_rewards/chosen": -0.03006073087453842,
"eval_rewards/margins": 0.057626351714134216,
"eval_rewards/rejected": -0.08768707513809204,
"eval_runtime": 540.1337,
"eval_samples_per_second": 3.703,
"eval_steps_per_second": 0.926,
"step": 3100
},
{
"epoch": 0.81,
"learning_rate": 5.09338364753818e-07,
"logits/chosen": 0.901587963104248,
"logits/rejected": 1.0238076448440552,
"logps/chosen": -293.3055114746094,
"logps/rejected": -265.13775634765625,
"loss": 0.6899,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.022513773292303085,
"rewards/margins": 0.08927594870328903,
"rewards/rejected": -0.11178971827030182,
"step": 3110
},
{
"epoch": 0.82,
"learning_rate": 4.956018477086005e-07,
"logits/chosen": 1.040078043937683,
"logits/rejected": 1.0295056104660034,
"logps/chosen": -282.08648681640625,
"logps/rejected": -230.9876251220703,
"loss": 0.6916,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.026005476713180542,
"rewards/margins": 0.054637789726257324,
"rewards/rejected": -0.08064327389001846,
"step": 3120
},
{
"epoch": 0.82,
"learning_rate": 4.820326973322764e-07,
"logits/chosen": 0.9570829272270203,
"logits/rejected": 1.007840871810913,
"logps/chosen": -296.0901794433594,
"logps/rejected": -222.00198364257812,
"loss": 0.6906,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.015858953818678856,
"rewards/margins": 0.0702957734465599,
"rewards/rejected": -0.08615472912788391,
"step": 3130
},
{
"epoch": 0.82,
"learning_rate": 4.686320466449981e-07,
"logits/chosen": 0.9432961344718933,
"logits/rejected": 0.9931074976921082,
"logps/chosen": -237.6160430908203,
"logps/rejected": -222.57870483398438,
"loss": 0.6902,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.024945750832557678,
"rewards/margins": 0.07617698609828949,
"rewards/rejected": -0.10112272202968597,
"step": 3140
},
{
"epoch": 0.82,
"learning_rate": 4.554010145972418e-07,
"logits/chosen": 0.9430766105651855,
"logits/rejected": 1.0184309482574463,
"logps/chosen": -286.0646667480469,
"logps/rejected": -229.7809600830078,
"loss": 0.6911,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.017250075936317444,
"rewards/margins": 0.05181562900543213,
"rewards/rejected": -0.06906570494174957,
"step": 3150
},
{
"epoch": 0.83,
"learning_rate": 4.4234070597637455e-07,
"logits/chosen": 0.8948880434036255,
"logits/rejected": 1.0683404207229614,
"logps/chosen": -271.1024169921875,
"logps/rejected": -254.7742462158203,
"loss": 0.6914,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.032865189015865326,
"rewards/margins": 0.04334001615643501,
"rewards/rejected": -0.07620520889759064,
"step": 3160
},
{
"epoch": 0.83,
"learning_rate": 4.2945221131440783e-07,
"logits/chosen": 0.9620285034179688,
"logits/rejected": 1.0088149309158325,
"logps/chosen": -246.1608123779297,
"logps/rejected": -221.75363159179688,
"loss": 0.6903,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.027930408716201782,
"rewards/margins": 0.05746666342020035,
"rewards/rejected": -0.08539707213640213,
"step": 3170
},
{
"epoch": 0.83,
"learning_rate": 4.167366067969381e-07,
"logits/chosen": 0.895652174949646,
"logits/rejected": 0.9522021412849426,
"logps/chosen": -272.8707580566406,
"logps/rejected": -259.20855712890625,
"loss": 0.6915,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.0206812284886837,
"rewards/margins": 0.05796490237116814,
"rewards/rejected": -0.07864613831043243,
"step": 3180
},
{
"epoch": 0.83,
"learning_rate": 4.041949541732826e-07,
"logits/chosen": 0.9851544499397278,
"logits/rejected": 1.0554238557815552,
"logps/chosen": -296.3244323730469,
"logps/rejected": -245.6664581298828,
"loss": 0.6915,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.02956976927816868,
"rewards/margins": 0.04805067181587219,
"rewards/rejected": -0.07762044668197632,
"step": 3190
},
{
"epoch": 0.84,
"learning_rate": 3.9182830066782614e-07,
"logits/chosen": 0.9625126123428345,
"logits/rejected": 0.9740635752677917,
"logps/chosen": -282.29425048828125,
"logps/rejected": -246.2125244140625,
"loss": 0.6919,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.03583105653524399,
"rewards/margins": 0.048236239701509476,
"rewards/rejected": -0.08406729251146317,
"step": 3200
},
{
"epoch": 0.84,
"eval_logits/chosen": 0.9153636693954468,
"eval_logits/rejected": 1.0176055431365967,
"eval_logps/chosen": -280.5576477050781,
"eval_logps/rejected": -252.59996032714844,
"eval_loss": 0.6908729076385498,
"eval_rewards/accuracies": 0.6305000185966492,
"eval_rewards/chosen": -0.03010854683816433,
"eval_rewards/margins": 0.057731661945581436,
"eval_rewards/rejected": -0.08784020692110062,
"eval_runtime": 540.1227,
"eval_samples_per_second": 3.703,
"eval_steps_per_second": 0.926,
"step": 3200
},
{
"epoch": 0.84,
"learning_rate": 3.796376788925771e-07,
"logits/chosen": 0.9004298448562622,
"logits/rejected": 1.0678789615631104,
"logps/chosen": -247.7801513671875,
"logps/rejected": -225.3026580810547,
"loss": 0.6903,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.03286176174879074,
"rewards/margins": 0.062369298189878464,
"rewards/rejected": -0.0952310562133789,
"step": 3210
},
{
"epoch": 0.84,
"learning_rate": 3.676241067609465e-07,
"logits/chosen": 0.9522799253463745,
"logits/rejected": 0.9462326169013977,
"logps/chosen": -242.72482299804688,
"logps/rejected": -237.6994171142578,
"loss": 0.6895,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.030536144971847534,
"rewards/margins": 0.06319095939397812,
"rewards/rejected": -0.09372710436582565,
"step": 3220
},
{
"epoch": 0.85,
"learning_rate": 3.5578858740274976e-07,
"logits/chosen": 0.9089903831481934,
"logits/rejected": 1.0164532661437988,
"logps/chosen": -300.99224853515625,
"logps/rejected": -260.6708984375,
"loss": 0.6911,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.021878289058804512,
"rewards/margins": 0.06206582114100456,
"rewards/rejected": -0.08394411206245422,
"step": 3230
},
{
"epoch": 0.85,
"learning_rate": 3.44132109080447e-07,
"logits/chosen": 0.9136902093887329,
"logits/rejected": 1.064955234527588,
"logps/chosen": -260.19952392578125,
"logps/rejected": -237.27487182617188,
"loss": 0.6915,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.03714042156934738,
"rewards/margins": 0.04816558212041855,
"rewards/rejected": -0.08530601114034653,
"step": 3240
},
{
"epoch": 0.85,
"learning_rate": 3.3265564510662344e-07,
"logits/chosen": 0.9555643200874329,
"logits/rejected": 1.0349557399749756,
"logps/chosen": -254.3995361328125,
"logps/rejected": -219.08493041992188,
"loss": 0.6908,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.02656802162528038,
"rewards/margins": 0.052377671003341675,
"rewards/rejected": -0.07894569635391235,
"step": 3250
},
{
"epoch": 0.85,
"learning_rate": 3.213601537627195e-07,
"logits/chosen": 0.9827576875686646,
"logits/rejected": 0.9898948669433594,
"logps/chosen": -261.27630615234375,
"logps/rejected": -252.62423706054688,
"loss": 0.6912,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.031087204813957214,
"rewards/margins": 0.05801438167691231,
"rewards/rejected": -0.08910159021615982,
"step": 3260
},
{
"epoch": 0.86,
"learning_rate": 3.1024657821901063e-07,
"logits/chosen": 0.9491473436355591,
"logits/rejected": 0.972917914390564,
"logps/chosen": -255.64334106445312,
"logps/rejected": -237.9709930419922,
"loss": 0.6909,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.025073718279600143,
"rewards/margins": 0.06804076582193375,
"rewards/rejected": -0.09311448037624359,
"step": 3270
},
{
"epoch": 0.86,
"learning_rate": 2.9931584645585654e-07,
"logits/chosen": 1.007514238357544,
"logits/rejected": 1.0354502201080322,
"logps/chosen": -285.310302734375,
"logps/rejected": -281.1996154785156,
"loss": 0.6916,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.02509785071015358,
"rewards/margins": 0.07314437627792358,
"rewards/rejected": -0.09824222326278687,
"step": 3280
},
{
"epoch": 0.86,
"learning_rate": 2.885688711862136e-07,
"logits/chosen": 0.9445541501045227,
"logits/rejected": 0.9918941259384155,
"logps/chosen": -257.47137451171875,
"logps/rejected": -222.67733764648438,
"loss": 0.6903,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.025721009820699692,
"rewards/margins": 0.0427216961979866,
"rewards/rejected": -0.068442702293396,
"step": 3290
},
{
"epoch": 0.86,
"learning_rate": 2.7800654977942486e-07,
"logits/chosen": 0.9048360586166382,
"logits/rejected": 1.0405280590057373,
"logps/chosen": -294.37847900390625,
"logps/rejected": -225.70297241210938,
"loss": 0.69,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.02359098568558693,
"rewards/margins": 0.06500183045864105,
"rewards/rejected": -0.08859282732009888,
"step": 3300
},
{
"epoch": 0.86,
"eval_logits/chosen": 0.9186109900474548,
"eval_logits/rejected": 1.0211718082427979,
"eval_logps/chosen": -280.2095947265625,
"eval_logps/rejected": -252.20504760742188,
"eval_loss": 0.6908652782440186,
"eval_rewards/accuracies": 0.6284999847412109,
"eval_rewards/chosen": -0.02662779949605465,
"eval_rewards/margins": 0.057263679802417755,
"eval_rewards/rejected": -0.08389147371053696,
"eval_runtime": 540.4002,
"eval_samples_per_second": 3.701,
"eval_steps_per_second": 0.925,
"step": 3300
},
{
"epoch": 0.87,
"learning_rate": 2.6762976418628797e-07,
"logits/chosen": 0.9340456128120422,
"logits/rejected": 0.9448652267456055,
"logps/chosen": -252.797119140625,
"logps/rejected": -231.66714477539062,
"loss": 0.6914,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.022808540612459183,
"rewards/margins": 0.05553862452507019,
"rewards/rejected": -0.07834717631340027,
"step": 3310
},
{
"epoch": 0.87,
"learning_rate": 2.5743938086541354e-07,
"logits/chosen": 0.9630203247070312,
"logits/rejected": 1.0022109746932983,
"logps/chosen": -297.3249816894531,
"logps/rejected": -251.38656616210938,
"loss": 0.6895,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.015477204695343971,
"rewards/margins": 0.08188708126544952,
"rewards/rejected": -0.09736428409814835,
"step": 3320
},
{
"epoch": 0.87,
"learning_rate": 2.4743625071087574e-07,
"logits/chosen": 0.9089611768722534,
"logits/rejected": 1.0690263509750366,
"logps/chosen": -269.80987548828125,
"logps/rejected": -246.6451873779297,
"loss": 0.69,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.031479865312576294,
"rewards/margins": 0.0678834319114685,
"rewards/rejected": -0.0993632897734642,
"step": 3330
},
{
"epoch": 0.87,
"learning_rate": 2.3762120898116498e-07,
"logits/chosen": 0.9765059351921082,
"logits/rejected": 1.011185884475708,
"logps/chosen": -267.643798828125,
"logps/rejected": -244.47586059570312,
"loss": 0.6913,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.02102813683450222,
"rewards/margins": 0.03919995576143265,
"rewards/rejected": -0.060228098183870316,
"step": 3340
},
{
"epoch": 0.88,
"learning_rate": 2.2799507522944048e-07,
"logits/chosen": 0.9900724291801453,
"logits/rejected": 1.0086749792099,
"logps/chosen": -296.67889404296875,
"logps/rejected": -239.91189575195312,
"loss": 0.6925,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.028908709064126015,
"rewards/margins": 0.030585547909140587,
"rewards/rejected": -0.0594942569732666,
"step": 3350
},
{
"epoch": 0.88,
"learning_rate": 2.1855865323510056e-07,
"logits/chosen": 0.9651070833206177,
"logits/rejected": 1.0190773010253906,
"logps/chosen": -290.01953125,
"logps/rejected": -266.74298095703125,
"loss": 0.6905,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.017114948481321335,
"rewards/margins": 0.06442873179912567,
"rewards/rejected": -0.08154366910457611,
"step": 3360
},
{
"epoch": 0.88,
"learning_rate": 2.0931273093666575e-07,
"logits/chosen": 1.0086729526519775,
"logits/rejected": 1.0279282331466675,
"logps/chosen": -267.48614501953125,
"logps/rejected": -265.8891296386719,
"loss": 0.6907,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.02099434658885002,
"rewards/margins": 0.061572205275297165,
"rewards/rejected": -0.08256654441356659,
"step": 3370
},
{
"epoch": 0.88,
"learning_rate": 2.002580803659873e-07,
"logits/chosen": 1.0000813007354736,
"logits/rejected": 1.0106067657470703,
"logps/chosen": -293.6463623046875,
"logps/rejected": -262.5312805175781,
"loss": 0.6916,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.036777563393116,
"rewards/margins": 0.04339155554771423,
"rewards/rejected": -0.08016912639141083,
"step": 3380
},
{
"epoch": 0.89,
"learning_rate": 1.913954575837826e-07,
"logits/chosen": 1.0087939500808716,
"logits/rejected": 1.079708456993103,
"logps/chosen": -292.1297912597656,
"logps/rejected": -263.822509765625,
"loss": 0.6904,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.028077151626348495,
"rewards/margins": 0.06453180313110352,
"rewards/rejected": -0.09260895103216171,
"step": 3390
},
{
"epoch": 0.89,
"learning_rate": 1.827256026165028e-07,
"logits/chosen": 1.0100739002227783,
"logits/rejected": 1.02089524269104,
"logps/chosen": -272.09161376953125,
"logps/rejected": -239.74560546875,
"loss": 0.689,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.02850997820496559,
"rewards/margins": 0.07207761704921722,
"rewards/rejected": -0.10058760643005371,
"step": 3400
},
{
"epoch": 0.89,
"eval_logits/chosen": 0.9202005863189697,
"eval_logits/rejected": 1.0222870111465454,
"eval_logps/chosen": -280.4384460449219,
"eval_logps/rejected": -252.48487854003906,
"eval_loss": 0.6908697485923767,
"eval_rewards/accuracies": 0.628000020980835,
"eval_rewards/chosen": -0.028916185721755028,
"eval_rewards/margins": 0.057773273438215256,
"eval_rewards/rejected": -0.08668945729732513,
"eval_runtime": 540.2108,
"eval_samples_per_second": 3.702,
"eval_steps_per_second": 0.926,
"step": 3400
},
{
"epoch": 0.89,
"learning_rate": 1.7424923939454274e-07,
"logits/chosen": 1.0115320682525635,
"logits/rejected": 1.0266879796981812,
"logps/chosen": -282.57763671875,
"logps/rejected": -256.1681213378906,
"loss": 0.6902,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.031413815915584564,
"rewards/margins": 0.056411970406770706,
"rewards/rejected": -0.08782579004764557,
"step": 3410
},
{
"epoch": 0.9,
"learning_rate": 1.6596707569179304e-07,
"logits/chosen": 0.9267918467521667,
"logits/rejected": 0.988551139831543,
"logps/chosen": -267.50616455078125,
"logps/rejected": -228.9858856201172,
"loss": 0.6906,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.022525835782289505,
"rewards/margins": 0.055544476956129074,
"rewards/rejected": -0.07807030528783798,
"step": 3420
},
{
"epoch": 0.9,
"learning_rate": 1.578798030665385e-07,
"logits/chosen": 0.9413129687309265,
"logits/rejected": 0.9737070798873901,
"logps/chosen": -287.58221435546875,
"logps/rejected": -231.7978973388672,
"loss": 0.6916,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.031874872744083405,
"rewards/margins": 0.03842931613326073,
"rewards/rejected": -0.07030418515205383,
"step": 3430
},
{
"epoch": 0.9,
"learning_rate": 1.499880968037165e-07,
"logits/chosen": 0.9008700251579285,
"logits/rejected": 1.0831358432769775,
"logps/chosen": -283.8552551269531,
"logps/rejected": -254.0209503173828,
"loss": 0.6905,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.02647927962243557,
"rewards/margins": 0.07812824845314026,
"rewards/rejected": -0.10460753738880157,
"step": 3440
},
{
"epoch": 0.9,
"learning_rate": 1.4229261585852805e-07,
"logits/chosen": 0.9461116790771484,
"logits/rejected": 1.0905543565750122,
"logps/chosen": -259.1544494628906,
"logps/rejected": -243.8948211669922,
"loss": 0.6902,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.03435998409986496,
"rewards/margins": 0.07146745920181274,
"rewards/rejected": -0.1058274507522583,
"step": 3450
},
{
"epoch": 0.91,
"learning_rate": 1.3479400280141886e-07,
"logits/chosen": 0.9727323651313782,
"logits/rejected": 0.9646995663642883,
"logps/chosen": -240.6188507080078,
"logps/rejected": -220.088623046875,
"loss": 0.6921,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.021371301263570786,
"rewards/margins": 0.04031342267990112,
"rewards/rejected": -0.06168472766876221,
"step": 3460
},
{
"epoch": 0.91,
"learning_rate": 1.2749288376442044e-07,
"logits/chosen": 1.0163447856903076,
"logits/rejected": 1.0307289361953735,
"logps/chosen": -249.5280303955078,
"logps/rejected": -242.83419799804688,
"loss": 0.6904,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.033246301114559174,
"rewards/margins": 0.05100921913981438,
"rewards/rejected": -0.08425550907850266,
"step": 3470
},
{
"epoch": 0.91,
"learning_rate": 1.203898683888713e-07,
"logits/chosen": 0.9769922494888306,
"logits/rejected": 1.0272125005722046,
"logps/chosen": -255.9679718017578,
"logps/rejected": -241.6696319580078,
"loss": 0.6905,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.036718789488077164,
"rewards/margins": 0.06190019100904465,
"rewards/rejected": -0.09861898422241211,
"step": 3480
},
{
"epoch": 0.91,
"learning_rate": 1.1348554977451132e-07,
"logits/chosen": 1.0243645906448364,
"logits/rejected": 1.0316869020462036,
"logps/chosen": -243.71621704101562,
"logps/rejected": -253.5388946533203,
"loss": 0.6902,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.03732157126069069,
"rewards/margins": 0.06564446538686752,
"rewards/rejected": -0.10296603292226791,
"step": 3490
},
{
"epoch": 0.92,
"learning_rate": 1.0678050442995802e-07,
"logits/chosen": 0.9439903497695923,
"logits/rejected": 0.9749709963798523,
"logps/chosen": -262.73333740234375,
"logps/rejected": -220.5067138671875,
"loss": 0.6901,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.03468143194913864,
"rewards/margins": 0.0676378607749939,
"rewards/rejected": -0.10231930017471313,
"step": 3500
},
{
"epoch": 0.92,
"eval_logits/chosen": 0.9216287136077881,
"eval_logits/rejected": 1.0239145755767822,
"eval_logps/chosen": -280.4474792480469,
"eval_logps/rejected": -252.50457763671875,
"eval_loss": 0.6908650398254395,
"eval_rewards/accuracies": 0.6259999871253967,
"eval_rewards/chosen": -0.029006626456975937,
"eval_rewards/margins": 0.05787980556488037,
"eval_rewards/rejected": -0.0868864357471466,
"eval_runtime": 539.9895,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 3500
},
{
"epoch": 0.92,
"learning_rate": 1.0027529222456755e-07,
"logits/chosen": 0.9812003374099731,
"logits/rejected": 0.9633600115776062,
"logps/chosen": -264.53521728515625,
"logps/rejected": -254.2129364013672,
"loss": 0.6916,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.03213541954755783,
"rewards/margins": 0.0566454641520977,
"rewards/rejected": -0.08878089487552643,
"step": 3510
},
{
"epoch": 0.92,
"learning_rate": 9.397045634168766e-08,
"logits/chosen": 0.9166300892829895,
"logits/rejected": 1.0722475051879883,
"logps/chosen": -275.48828125,
"logps/rejected": -235.26626586914062,
"loss": 0.69,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.016414564102888107,
"rewards/margins": 0.06454362720251083,
"rewards/rejected": -0.08095818758010864,
"step": 3520
},
{
"epoch": 0.92,
"learning_rate": 8.78665232332998e-08,
"logits/chosen": 0.9330304265022278,
"logits/rejected": 1.0522311925888062,
"logps/chosen": -260.5693054199219,
"logps/rejected": -258.9266052246094,
"loss": 0.691,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.025154178962111473,
"rewards/margins": 0.05904405564069748,
"rewards/rejected": -0.08419822156429291,
"step": 3530
},
{
"epoch": 0.93,
"learning_rate": 8.196400257606208e-08,
"logits/chosen": 0.9113238453865051,
"logits/rejected": 0.9963987469673157,
"logps/chosen": -279.2100524902344,
"logps/rejected": -249.4517059326172,
"loss": 0.6921,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.020202763378620148,
"rewards/margins": 0.04745917767286301,
"rewards/rejected": -0.06766194850206375,
"step": 3540
},
{
"epoch": 0.93,
"learning_rate": 7.626338722875076e-08,
"logits/chosen": 1.0007197856903076,
"logits/rejected": 0.9505215883255005,
"logps/chosen": -280.51666259765625,
"logps/rejected": -254.579833984375,
"loss": 0.6908,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.03505931794643402,
"rewards/margins": 0.04930321127176285,
"rewards/rejected": -0.08436252176761627,
"step": 3550
},
{
"epoch": 0.93,
"learning_rate": 7.076515319110688e-08,
"logits/chosen": 1.0191797018051147,
"logits/rejected": 1.0009005069732666,
"logps/chosen": -292.7381591796875,
"logps/rejected": -252.88143920898438,
"loss": 0.6903,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.029474809765815735,
"rewards/margins": 0.05774076655507088,
"rewards/rejected": -0.08721558004617691,
"step": 3560
},
{
"epoch": 0.93,
"learning_rate": 6.54697595640899e-08,
"logits/chosen": 1.0553741455078125,
"logits/rejected": 0.9938759803771973,
"logps/chosen": -280.03265380859375,
"logps/rejected": -261.9696044921875,
"loss": 0.6904,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.024510715156793594,
"rewards/margins": 0.05735497549176216,
"rewards/rejected": -0.08186569064855576,
"step": 3570
},
{
"epoch": 0.94,
"learning_rate": 6.037764851154426e-08,
"logits/chosen": 0.9185417890548706,
"logits/rejected": 1.0824509859085083,
"logps/chosen": -279.9969787597656,
"logps/rejected": -240.0286865234375,
"loss": 0.6912,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.029702965170145035,
"rewards/margins": 0.05176641792058945,
"rewards/rejected": -0.08146937936544418,
"step": 3580
},
{
"epoch": 0.94,
"learning_rate": 5.548924522327748e-08,
"logits/chosen": 0.894769012928009,
"logits/rejected": 1.0935579538345337,
"logps/chosen": -275.4918518066406,
"logps/rejected": -258.45953369140625,
"loss": 0.6911,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.02207362651824951,
"rewards/margins": 0.06469397246837616,
"rewards/rejected": -0.08676759898662567,
"step": 3590
},
{
"epoch": 0.94,
"learning_rate": 5.0804957879556915e-08,
"logits/chosen": 0.9377716183662415,
"logits/rejected": 1.0579009056091309,
"logps/chosen": -318.527587890625,
"logps/rejected": -260.9480285644531,
"loss": 0.6914,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.03672604635357857,
"rewards/margins": 0.05514238029718399,
"rewards/rejected": -0.09186841547489166,
"step": 3600
},
{
"epoch": 0.94,
"eval_logits/chosen": 0.9221275448799133,
"eval_logits/rejected": 1.0243616104125977,
"eval_logps/chosen": -280.42578125,
"eval_logps/rejected": -252.4630584716797,
"eval_loss": 0.6908705830574036,
"eval_rewards/accuracies": 0.6290000081062317,
"eval_rewards/chosen": -0.02878967486321926,
"eval_rewards/margins": 0.057681918144226074,
"eval_rewards/rejected": -0.08647158741950989,
"eval_runtime": 539.9286,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 3600
},
{
"epoch": 0.94,
"learning_rate": 4.632517761702815e-08,
"logits/chosen": 0.9917756915092468,
"logits/rejected": 1.043176293373108,
"logps/chosen": -285.3194274902344,
"logps/rejected": -251.86471557617188,
"loss": 0.6901,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.020223677158355713,
"rewards/margins": 0.0789574682712555,
"rewards/rejected": -0.0991811528801918,
"step": 3610
},
{
"epoch": 0.95,
"learning_rate": 4.205027849605359e-08,
"logits/chosen": 0.9845743179321289,
"logits/rejected": 0.9572548866271973,
"logps/chosen": -231.9497833251953,
"logps/rejected": -235.6708984375,
"loss": 0.6913,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.024526074528694153,
"rewards/margins": 0.0476609468460083,
"rewards/rejected": -0.07218702882528305,
"step": 3620
},
{
"epoch": 0.95,
"learning_rate": 3.798061746947995e-08,
"logits/chosen": 0.9714505076408386,
"logits/rejected": 0.9893890619277954,
"logps/chosen": -345.9156188964844,
"logps/rejected": -263.2041931152344,
"loss": 0.6918,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.022486504167318344,
"rewards/margins": 0.052516452968120575,
"rewards/rejected": -0.07500295341014862,
"step": 3630
},
{
"epoch": 0.95,
"learning_rate": 3.411653435283158e-08,
"logits/chosen": 0.9488525390625,
"logits/rejected": 1.0411800146102905,
"logps/chosen": -294.39727783203125,
"logps/rejected": -257.3688659667969,
"loss": 0.6915,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.03506433218717575,
"rewards/margins": 0.0475606694817543,
"rewards/rejected": -0.08262500911951065,
"step": 3640
},
{
"epoch": 0.96,
"learning_rate": 3.04583517959367e-08,
"logits/chosen": 0.9405366778373718,
"logits/rejected": 0.9672233462333679,
"logps/chosen": -299.8280334472656,
"logps/rejected": -250.7571258544922,
"loss": 0.6905,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.011996401473879814,
"rewards/margins": 0.08034516870975494,
"rewards/rejected": -0.09234156459569931,
"step": 3650
},
{
"epoch": 0.96,
"learning_rate": 2.7006375255985984e-08,
"logits/chosen": 0.9549415707588196,
"logits/rejected": 1.0070356130599976,
"logps/chosen": -241.44503784179688,
"logps/rejected": -215.5791778564453,
"loss": 0.6908,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.029812267050147057,
"rewards/margins": 0.04664590209722519,
"rewards/rejected": -0.0764581710100174,
"step": 3660
},
{
"epoch": 0.96,
"learning_rate": 2.3760892972027328e-08,
"logits/chosen": 0.8861829042434692,
"logits/rejected": 1.0260193347930908,
"logps/chosen": -290.62060546875,
"logps/rejected": -238.04080200195312,
"loss": 0.6907,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.025474613532423973,
"rewards/margins": 0.061755161732435226,
"rewards/rejected": -0.08722977340221405,
"step": 3670
},
{
"epoch": 0.96,
"learning_rate": 2.072217594089765e-08,
"logits/chosen": 0.9960481524467468,
"logits/rejected": 0.9741401672363281,
"logps/chosen": -247.71484375,
"logps/rejected": -232.7592315673828,
"loss": 0.6918,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.025015806779265404,
"rewards/margins": 0.04509962350130081,
"rewards/rejected": -0.07011543214321136,
"step": 3680
},
{
"epoch": 0.97,
"learning_rate": 1.789047789459375e-08,
"logits/chosen": 0.9169954061508179,
"logits/rejected": 0.9940811991691589,
"logps/chosen": -279.7234191894531,
"logps/rejected": -257.5582580566406,
"loss": 0.6901,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.029348302632570267,
"rewards/margins": 0.05934641510248184,
"rewards/rejected": -0.08869470655918121,
"step": 3690
},
{
"epoch": 0.97,
"learning_rate": 1.5266035279088708e-08,
"logits/chosen": 0.9306381940841675,
"logits/rejected": 0.9920527338981628,
"logps/chosen": -320.2879333496094,
"logps/rejected": -246.09512329101562,
"loss": 0.6914,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.02664271369576454,
"rewards/margins": 0.05591448396444321,
"rewards/rejected": -0.08255720138549805,
"step": 3700
},
{
"epoch": 0.97,
"eval_logits/chosen": 0.9215983152389526,
"eval_logits/rejected": 1.0239914655685425,
"eval_logps/chosen": -280.43499755859375,
"eval_logps/rejected": -252.45912170410156,
"eval_loss": 0.6908671855926514,
"eval_rewards/accuracies": 0.6320000290870667,
"eval_rewards/chosen": -0.028881965205073357,
"eval_rewards/margins": 0.05755016952753067,
"eval_rewards/rejected": -0.08643212914466858,
"eval_runtime": 540.2057,
"eval_samples_per_second": 3.702,
"eval_steps_per_second": 0.926,
"step": 3700
},
{
"epoch": 0.97,
"learning_rate": 1.2849067234584623e-08,
"logits/chosen": 0.9069113731384277,
"logits/rejected": 0.9997833371162415,
"logps/chosen": -258.52178955078125,
"logps/rejected": -240.258544921875,
"loss": 0.6913,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.02280072309076786,
"rewards/margins": 0.054703257977962494,
"rewards/rejected": -0.0775039792060852,
"step": 3710
},
{
"epoch": 0.97,
"learning_rate": 1.0639775577218625e-08,
"logits/chosen": 0.9457674026489258,
"logits/rejected": 0.987481415271759,
"logps/chosen": -288.9521484375,
"logps/rejected": -266.4629821777344,
"loss": 0.6911,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.019102510064840317,
"rewards/margins": 0.06657516956329346,
"rewards/rejected": -0.08567767590284348,
"step": 3720
},
{
"epoch": 0.98,
"learning_rate": 8.638344782207486e-09,
"logits/chosen": 0.9037330746650696,
"logits/rejected": 1.0098841190338135,
"logps/chosen": -303.5415954589844,
"logps/rejected": -282.108642578125,
"loss": 0.6918,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.038687679916620255,
"rewards/margins": 0.04073931276798248,
"rewards/rejected": -0.07942698895931244,
"step": 3730
},
{
"epoch": 0.98,
"learning_rate": 6.84494196844715e-09,
"logits/chosen": 0.9645431637763977,
"logits/rejected": 1.0531866550445557,
"logps/chosen": -265.5387268066406,
"logps/rejected": -245.14968872070312,
"loss": 0.6901,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.018353218212723732,
"rewards/margins": 0.05940115451812744,
"rewards/rejected": -0.07775436341762543,
"step": 3740
},
{
"epoch": 0.98,
"learning_rate": 5.259716884556121e-09,
"logits/chosen": 0.9109350442886353,
"logits/rejected": 1.009334683418274,
"logps/chosen": -311.3597412109375,
"logps/rejected": -255.84878540039062,
"loss": 0.691,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.02166152000427246,
"rewards/margins": 0.06039486080408096,
"rewards/rejected": -0.08205638825893402,
"step": 3750
},
{
"epoch": 0.98,
"learning_rate": 3.882801896372967e-09,
"logits/chosen": 0.8763138055801392,
"logits/rejected": 1.017446517944336,
"logps/chosen": -282.8580627441406,
"logps/rejected": -261.3891906738281,
"loss": 0.692,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.024912741035223007,
"rewards/margins": 0.045935411006212234,
"rewards/rejected": -0.07084815204143524,
"step": 3760
},
{
"epoch": 0.99,
"learning_rate": 2.7143119759026614e-09,
"logits/chosen": 0.9283539056777954,
"logits/rejected": 1.0674773454666138,
"logps/chosen": -270.36358642578125,
"logps/rejected": -274.0054626464844,
"loss": 0.6914,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.01020726840943098,
"rewards/margins": 0.06225059553980827,
"rewards/rejected": -0.07245786488056183,
"step": 3770
},
{
"epoch": 0.99,
"learning_rate": 1.754344691717591e-09,
"logits/chosen": 0.8904141187667847,
"logits/rejected": 0.9620451927185059,
"logps/chosen": -264.84930419921875,
"logps/rejected": -218.98477172851562,
"loss": 0.6906,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.012690825387835503,
"rewards/margins": 0.06626447290182114,
"rewards/rejected": -0.07895530760288239,
"step": 3780
},
{
"epoch": 0.99,
"learning_rate": 1.0029802008096335e-09,
"logits/chosen": 1.0099847316741943,
"logits/rejected": 0.9373987913131714,
"logps/chosen": -268.06048583984375,
"logps/rejected": -228.6332550048828,
"loss": 0.6902,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.028048217296600342,
"rewards/margins": 0.06466346234083176,
"rewards/rejected": -0.0927116721868515,
"step": 3790
},
{
"epoch": 0.99,
"learning_rate": 4.602812418974534e-10,
"logits/chosen": 0.8952652812004089,
"logits/rejected": 1.017452597618103,
"logps/chosen": -254.45571899414062,
"logps/rejected": -240.29006958007812,
"loss": 0.6917,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.032994892448186874,
"rewards/margins": 0.04944116994738579,
"rewards/rejected": -0.08243606984615326,
"step": 3800
},
{
"epoch": 0.99,
"eval_logits/chosen": 0.9220536351203918,
"eval_logits/rejected": 1.0245850086212158,
"eval_logps/chosen": -280.4203796386719,
"eval_logps/rejected": -252.4789581298828,
"eval_loss": 0.690862774848938,
"eval_rewards/accuracies": 0.6320000290870667,
"eval_rewards/chosen": -0.028735652565956116,
"eval_rewards/margins": 0.057894736528396606,
"eval_rewards/rejected": -0.08663039654493332,
"eval_runtime": 539.9529,
"eval_samples_per_second": 3.704,
"eval_steps_per_second": 0.926,
"step": 3800
},
{
"epoch": 1.0,
"learning_rate": 1.2629313018819312e-10,
"logits/chosen": 0.895196795463562,
"logits/rejected": 1.000282645225525,
"logps/chosen": -302.4236145019531,
"logps/rejected": -285.1495361328125,
"loss": 0.6914,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.016779940575361252,
"rewards/margins": 0.05169288069009781,
"rewards/rejected": -0.06847281754016876,
"step": 3810
},
{
"epoch": 1.0,
"learning_rate": 1.0437535929996855e-12,
"logits/chosen": 0.982556164264679,
"logits/rejected": 0.9328534007072449,
"logps/chosen": -285.391845703125,
"logps/rejected": -266.6493225097656,
"loss": 0.6902,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.026478338986635208,
"rewards/margins": 0.06039903312921524,
"rewards/rejected": -0.08687736093997955,
"step": 3820
},
{
"epoch": 1.0,
"step": 3821,
"total_flos": 0.0,
"train_loss": 0.691200782103989,
"train_runtime": 55857.7878,
"train_samples_per_second": 1.094,
"train_steps_per_second": 0.068
}
],
"logging_steps": 10,
"max_steps": 3821,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 100,
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}