{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 100, "global_step": 478, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0020920502092050207, "grad_norm": 9.085185309257627, "learning_rate": 1.0416666666666666e-08, "logits/chosen": -2.7662220001220703, "logits/rejected": -2.7178127765655518, "logps/chosen": -269.6776123046875, "logps/rejected": -360.6510314941406, "loss": 0.6931, "rewards/accuracies": 0.0, "rewards/chosen": 0.0, "rewards/margins": 0.0, "rewards/rejected": 0.0, "step": 1 }, { "epoch": 0.02092050209205021, "grad_norm": 9.1082652647099, "learning_rate": 1.0416666666666667e-07, "logits/chosen": -2.5920939445495605, "logits/rejected": -2.5626003742218018, "logps/chosen": -264.7879638671875, "logps/rejected": -251.56271362304688, "loss": 0.6931, "rewards/accuracies": 0.4652777910232544, "rewards/chosen": -0.00015384703874588013, "rewards/margins": 0.00021662651852238923, "rewards/rejected": -0.00037047354271635413, "step": 10 }, { "epoch": 0.04184100418410042, "grad_norm": 8.403069183005119, "learning_rate": 2.0833333333333333e-07, "logits/chosen": -2.6531224250793457, "logits/rejected": -2.6051571369171143, "logps/chosen": -281.4217224121094, "logps/rejected": -296.8309020996094, "loss": 0.6923, "rewards/accuracies": 0.5, "rewards/chosen": 0.0006401228019967675, "rewards/margins": 0.0008313875878229737, "rewards/rejected": -0.00019126484403386712, "step": 20 }, { "epoch": 0.06276150627615062, "grad_norm": 8.308611367087158, "learning_rate": 3.1249999999999997e-07, "logits/chosen": -2.6641345024108887, "logits/rejected": -2.591866970062256, "logps/chosen": -299.44696044921875, "logps/rejected": -263.68865966796875, "loss": 0.6886, "rewards/accuracies": 0.625, "rewards/chosen": 0.012143267318606377, "rewards/margins": 0.009724361822009087, "rewards/rejected": 0.0024189057294279337, "step": 30 }, { "epoch": 0.08368200836820083, "grad_norm": 8.158936380689568, "learning_rate": 4.1666666666666667e-07, "logits/chosen": -2.556858539581299, "logits/rejected": -2.521756172180176, "logps/chosen": -263.13677978515625, "logps/rejected": -245.13369750976562, "loss": 0.6778, "rewards/accuracies": 0.71875, "rewards/chosen": 0.039376698434352875, "rewards/margins": 0.04692060127854347, "rewards/rejected": -0.007543901912868023, "step": 40 }, { "epoch": 0.10460251046025104, "grad_norm": 8.442774298195335, "learning_rate": 4.999733114418725e-07, "logits/chosen": -2.4983766078948975, "logits/rejected": -2.458930730819702, "logps/chosen": -279.6597595214844, "logps/rejected": -297.4883117675781, "loss": 0.6615, "rewards/accuracies": 0.612500011920929, "rewards/chosen": 0.0010005368385463953, "rewards/margins": 0.048170365393161774, "rewards/rejected": -0.04716982692480087, "step": 50 }, { "epoch": 0.12552301255230125, "grad_norm": 11.320329514836741, "learning_rate": 4.990398100856366e-07, "logits/chosen": -2.514209270477295, "logits/rejected": -2.4425783157348633, "logps/chosen": -273.6862487792969, "logps/rejected": -312.9122314453125, "loss": 0.6389, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -0.020198455080389977, "rewards/margins": 0.12713830173015594, "rewards/rejected": -0.14733675122261047, "step": 60 }, { "epoch": 0.14644351464435146, "grad_norm": 10.658450428160506, "learning_rate": 4.967775735898179e-07, "logits/chosen": -2.3650529384613037, "logits/rejected": -2.3627445697784424, "logps/chosen": -270.68896484375, "logps/rejected": -283.2526550292969, "loss": 0.6157, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -0.05055179074406624, "rewards/margins": 0.21437402069568634, "rewards/rejected": -0.26492583751678467, "step": 70 }, { "epoch": 0.16736401673640167, "grad_norm": 18.344270948626654, "learning_rate": 4.931986719649298e-07, "logits/chosen": -2.5614020824432373, "logits/rejected": -2.4784748554229736, "logps/chosen": -336.35430908203125, "logps/rejected": -303.9096374511719, "loss": 0.6025, "rewards/accuracies": 0.6875, "rewards/chosen": -0.1354917585849762, "rewards/margins": 0.26337119936943054, "rewards/rejected": -0.39886292815208435, "step": 80 }, { "epoch": 0.18828451882845187, "grad_norm": 19.12478751671118, "learning_rate": 4.883222001996351e-07, "logits/chosen": -2.134124755859375, "logits/rejected": -2.0814406871795654, "logps/chosen": -305.53082275390625, "logps/rejected": -345.31109619140625, "loss": 0.5741, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -0.4640178680419922, "rewards/margins": 0.46487826108932495, "rewards/rejected": -0.9288961291313171, "step": 90 }, { "epoch": 0.20920502092050208, "grad_norm": 17.356342239295923, "learning_rate": 4.821741763807186e-07, "logits/chosen": -1.3313448429107666, "logits/rejected": -1.1910110712051392, "logps/chosen": -308.7679138183594, "logps/rejected": -361.08013916015625, "loss": 0.5709, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.4839816987514496, "rewards/margins": 0.5560156106948853, "rewards/rejected": -1.0399973392486572, "step": 100 }, { "epoch": 0.20920502092050208, "eval_logits/chosen": -1.2465178966522217, "eval_logits/rejected": -1.0884699821472168, "eval_logps/chosen": -296.7220153808594, "eval_logps/rejected": -345.89581298828125, "eval_loss": 0.5765112042427063, "eval_rewards/accuracies": 0.73046875, "eval_rewards/chosen": -0.34146958589553833, "eval_rewards/margins": 0.49064701795578003, "eval_rewards/rejected": -0.8321166634559631, "eval_runtime": 40.3807, "eval_samples_per_second": 49.529, "eval_steps_per_second": 0.792, "step": 100 }, { "epoch": 0.2301255230125523, "grad_norm": 17.47373189781474, "learning_rate": 4.747874028753375e-07, "logits/chosen": -0.9215749502182007, "logits/rejected": -0.43307366967201233, "logps/chosen": -360.1551818847656, "logps/rejected": -366.24737548828125, "loss": 0.5682, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -0.5557556748390198, "rewards/margins": 0.528251588344574, "rewards/rejected": -1.0840072631835938, "step": 110 }, { "epoch": 0.2510460251046025, "grad_norm": 20.093025180061602, "learning_rate": 4.662012913161997e-07, "logits/chosen": -0.3439692556858063, "logits/rejected": -0.04505181312561035, "logps/chosen": -336.732177734375, "logps/rejected": -372.484619140625, "loss": 0.5504, "rewards/accuracies": 0.75, "rewards/chosen": -0.5553595423698425, "rewards/margins": 0.5710638761520386, "rewards/rejected": -1.1264234781265259, "step": 120 }, { "epoch": 0.2719665271966527, "grad_norm": 23.028382893640803, "learning_rate": 4.5646165232345103e-07, "logits/chosen": 0.32941126823425293, "logits/rejected": 0.7188884019851685, "logps/chosen": -367.50909423828125, "logps/rejected": -419.0060119628906, "loss": 0.535, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.8557965159416199, "rewards/margins": 0.6497102379798889, "rewards/rejected": -1.5055067539215088, "step": 130 }, { "epoch": 0.2928870292887029, "grad_norm": 18.677447832810206, "learning_rate": 4.456204510851956e-07, "logits/chosen": 0.21762561798095703, "logits/rejected": 0.6349714398384094, "logps/chosen": -381.5950927734375, "logps/rejected": -431.7596740722656, "loss": 0.534, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.6563121676445007, "rewards/margins": 0.6548308730125427, "rewards/rejected": -1.311143159866333, "step": 140 }, { "epoch": 0.3138075313807531, "grad_norm": 27.137772429027354, "learning_rate": 4.337355301007335e-07, "logits/chosen": 0.3337453007698059, "logits/rejected": 0.7608418464660645, "logps/chosen": -367.8205261230469, "logps/rejected": -415.28045654296875, "loss": 0.5402, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -0.8457827568054199, "rewards/margins": 0.5450310707092285, "rewards/rejected": -1.3908138275146484, "step": 150 }, { "epoch": 0.33472803347280333, "grad_norm": 26.991586965942954, "learning_rate": 4.2087030056579986e-07, "logits/chosen": 0.3724555969238281, "logits/rejected": 1.1069540977478027, "logps/chosen": -358.0096130371094, "logps/rejected": -428.70361328125, "loss": 0.5324, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -0.8497047424316406, "rewards/margins": 0.8634451031684875, "rewards/rejected": -1.7131500244140625, "step": 160 }, { "epoch": 0.35564853556485354, "grad_norm": 25.592749003897193, "learning_rate": 4.070934040463998e-07, "logits/chosen": 0.5049823522567749, "logits/rejected": 0.9404481649398804, "logps/chosen": -348.8735046386719, "logps/rejected": -392.8125915527344, "loss": 0.5363, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -0.9956283569335938, "rewards/margins": 0.620826244354248, "rewards/rejected": -1.6164544820785522, "step": 170 }, { "epoch": 0.37656903765690375, "grad_norm": 22.372745978564954, "learning_rate": 3.9247834624635404e-07, "logits/chosen": 0.6127285361289978, "logits/rejected": 1.2147839069366455, "logps/chosen": -340.34906005859375, "logps/rejected": -381.84661865234375, "loss": 0.5064, "rewards/accuracies": 0.75, "rewards/chosen": -0.8403685688972473, "rewards/margins": 0.7025831937789917, "rewards/rejected": -1.5429518222808838, "step": 180 }, { "epoch": 0.39748953974895396, "grad_norm": 29.66314474064595, "learning_rate": 3.7710310482256523e-07, "logits/chosen": 0.3294413983821869, "logits/rejected": 0.9074909090995789, "logps/chosen": -347.7164611816406, "logps/rejected": -403.9298400878906, "loss": 0.5209, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.7645334601402283, "rewards/margins": 0.67705237865448, "rewards/rejected": -1.441585898399353, "step": 190 }, { "epoch": 0.41841004184100417, "grad_norm": 58.26403433971581, "learning_rate": 3.610497133404795e-07, "logits/chosen": 0.549667477607727, "logits/rejected": 1.1599102020263672, "logps/chosen": -346.0694580078125, "logps/rejected": -415.4115295410156, "loss": 0.5427, "rewards/accuracies": 0.75, "rewards/chosen": -0.8615196943283081, "rewards/margins": 0.7718194723129272, "rewards/rejected": -1.6333389282226562, "step": 200 }, { "epoch": 0.41841004184100417, "eval_logits/chosen": 0.18153175711631775, "eval_logits/rejected": 0.9058605432510376, "eval_logps/chosen": -336.09857177734375, "eval_logps/rejected": -416.4310607910156, "eval_loss": 0.5255736708641052, "eval_rewards/accuracies": 0.76953125, "eval_rewards/chosen": -0.7352346181869507, "eval_rewards/margins": 0.8022347688674927, "eval_rewards/rejected": -1.5374693870544434, "eval_runtime": 40.3301, "eval_samples_per_second": 49.591, "eval_steps_per_second": 0.793, "step": 200 }, { "epoch": 0.4393305439330544, "grad_norm": 23.975315777665053, "learning_rate": 3.4440382358952115e-07, "logits/chosen": 0.23435361683368683, "logits/rejected": 1.2575860023498535, "logps/chosen": -365.595703125, "logps/rejected": -398.31182861328125, "loss": 0.5374, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -0.759240448474884, "rewards/margins": 0.711215615272522, "rewards/rejected": -1.4704558849334717, "step": 210 }, { "epoch": 0.4602510460251046, "grad_norm": 33.35111797134312, "learning_rate": 3.272542485937368e-07, "logits/chosen": 0.5184121131896973, "logits/rejected": 1.4770405292510986, "logps/chosen": -347.60296630859375, "logps/rejected": -407.5792541503906, "loss": 0.5233, "rewards/accuracies": 0.78125, "rewards/chosen": -0.8224745988845825, "rewards/margins": 0.7966977953910828, "rewards/rejected": -1.61917245388031, "step": 220 }, { "epoch": 0.4811715481171548, "grad_norm": 27.065916482787443, "learning_rate": 3.096924887558854e-07, "logits/chosen": 0.49935898184776306, "logits/rejected": 1.3911898136138916, "logps/chosen": -331.6248474121094, "logps/rejected": -421.27374267578125, "loss": 0.536, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -0.8267892003059387, "rewards/margins": 0.9082108736038208, "rewards/rejected": -1.7350002527236938, "step": 230 }, { "epoch": 0.502092050209205, "grad_norm": 22.856393344097995, "learning_rate": 2.9181224366319943e-07, "logits/chosen": 0.2771614193916321, "logits/rejected": 0.9905085563659668, "logps/chosen": -338.65069580078125, "logps/rejected": -394.545654296875, "loss": 0.5024, "rewards/accuracies": 0.78125, "rewards/chosen": -0.7491301894187927, "rewards/margins": 0.7059187293052673, "rewards/rejected": -1.4550487995147705, "step": 240 }, { "epoch": 0.5230125523012552, "grad_norm": 26.034393008774543, "learning_rate": 2.7370891215954565e-07, "logits/chosen": 0.8687924146652222, "logits/rejected": 1.8204635381698608, "logps/chosen": -394.45635986328125, "logps/rejected": -441.63739013671875, "loss": 0.4997, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -0.9626559019088745, "rewards/margins": 0.858391284942627, "rewards/rejected": -1.821047067642212, "step": 250 }, { "epoch": 0.5439330543933054, "grad_norm": 32.298336541514864, "learning_rate": 2.55479083351317e-07, "logits/chosen": 0.8348854184150696, "logits/rejected": 1.8038705587387085, "logps/chosen": -395.5223693847656, "logps/rejected": -442.0123596191406, "loss": 0.5091, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -1.0121302604675293, "rewards/margins": 0.8687250018119812, "rewards/rejected": -1.8808553218841553, "step": 260 }, { "epoch": 0.5648535564853556, "grad_norm": 23.028145780758834, "learning_rate": 2.3722002126275822e-07, "logits/chosen": 0.5836859941482544, "logits/rejected": 1.3243392705917358, "logps/chosen": -365.3858337402344, "logps/rejected": -417.47735595703125, "loss": 0.5181, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -0.8838497996330261, "rewards/margins": 0.7144038081169128, "rewards/rejected": -1.5982534885406494, "step": 270 }, { "epoch": 0.5857740585774058, "grad_norm": 24.69890342328338, "learning_rate": 2.19029145890313e-07, "logits/chosen": 0.8636754155158997, "logits/rejected": 1.8474575281143188, "logps/chosen": -362.3246154785156, "logps/rejected": -422.9788513183594, "loss": 0.5317, "rewards/accuracies": 0.71875, "rewards/chosen": -1.0066673755645752, "rewards/margins": 0.8012599945068359, "rewards/rejected": -1.8079273700714111, "step": 280 }, { "epoch": 0.606694560669456, "grad_norm": 23.579877718409463, "learning_rate": 2.0100351342479216e-07, "logits/chosen": 0.9099249839782715, "logits/rejected": 1.71127188205719, "logps/chosen": -365.8359375, "logps/rejected": -429.7777404785156, "loss": 0.5117, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -1.1478241682052612, "rewards/margins": 0.7301602363586426, "rewards/rejected": -1.8779844045639038, "step": 290 }, { "epoch": 0.6276150627615062, "grad_norm": 28.794362121773794, "learning_rate": 1.8323929841460178e-07, "logits/chosen": 0.5150594115257263, "logits/rejected": 1.7005367279052734, "logps/chosen": -400.82708740234375, "logps/rejected": -440.55181884765625, "loss": 0.4892, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.9979560971260071, "rewards/margins": 0.8070657849311829, "rewards/rejected": -1.80502188205719, "step": 300 }, { "epoch": 0.6276150627615062, "eval_logits/chosen": 0.5828189253807068, "eval_logits/rejected": 1.3891677856445312, "eval_logps/chosen": -351.67193603515625, "eval_logps/rejected": -444.7821960449219, "eval_loss": 0.5082466006278992, "eval_rewards/accuracies": 0.76953125, "eval_rewards/chosen": -0.8909686207771301, "eval_rewards/margins": 0.9300119280815125, "eval_rewards/rejected": -1.820980429649353, "eval_runtime": 40.346, "eval_samples_per_second": 49.571, "eval_steps_per_second": 0.793, "step": 300 }, { "epoch": 0.6485355648535565, "grad_norm": 33.12731092406144, "learning_rate": 1.6583128063291573e-07, "logits/chosen": 0.7290142774581909, "logits/rejected": 1.4886902570724487, "logps/chosen": -397.9615478515625, "logps/rejected": -434.9144592285156, "loss": 0.4935, "rewards/accuracies": 0.75, "rewards/chosen": -1.0262950658798218, "rewards/margins": 0.7377363443374634, "rewards/rejected": -1.764031171798706, "step": 310 }, { "epoch": 0.6694560669456067, "grad_norm": 33.809250573756664, "learning_rate": 1.488723393865766e-07, "logits/chosen": 0.8037419319152832, "logits/rejected": 1.8283920288085938, "logps/chosen": -393.73876953125, "logps/rejected": -423.55279541015625, "loss": 0.4925, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -1.0004233121871948, "rewards/margins": 0.7867400050163269, "rewards/rejected": -1.7871633768081665, "step": 320 }, { "epoch": 0.6903765690376569, "grad_norm": 29.294631844662618, "learning_rate": 1.3245295796480788e-07, "logits/chosen": 0.5138927698135376, "logits/rejected": 1.4465028047561646, "logps/chosen": -355.19622802734375, "logps/rejected": -424.10052490234375, "loss": 0.4984, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.9297013282775879, "rewards/margins": 0.7162156701087952, "rewards/rejected": -1.6459171772003174, "step": 330 }, { "epoch": 0.7112970711297071, "grad_norm": 31.084849867717274, "learning_rate": 1.1666074087171627e-07, "logits/chosen": 0.8492459058761597, "logits/rejected": 1.7086740732192993, "logps/chosen": -376.3741455078125, "logps/rejected": -468.2669982910156, "loss": 0.5005, "rewards/accuracies": 0.78125, "rewards/chosen": -0.8633196949958801, "rewards/margins": 1.0182921886444092, "rewards/rejected": -1.8816118240356445, "step": 340 }, { "epoch": 0.7322175732217573, "grad_norm": 24.550583619455697, "learning_rate": 1.0157994641835734e-07, "logits/chosen": 0.9272977113723755, "logits/rejected": 1.8347976207733154, "logps/chosen": -351.08538818359375, "logps/rejected": -420.5586853027344, "loss": 0.4773, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -0.9292421340942383, "rewards/margins": 0.8692865371704102, "rewards/rejected": -1.7985286712646484, "step": 350 }, { "epoch": 0.7531380753138075, "grad_norm": 23.786028174418703, "learning_rate": 8.729103716819111e-08, "logits/chosen": 0.772596001625061, "logits/rejected": 1.983689308166504, "logps/chosen": -400.6702575683594, "logps/rejected": -450.78521728515625, "loss": 0.511, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -0.9801367521286011, "rewards/margins": 0.9152604341506958, "rewards/rejected": -1.8953969478607178, "step": 360 }, { "epoch": 0.7740585774058577, "grad_norm": 28.031131650473906, "learning_rate": 7.387025063449081e-08, "logits/chosen": 1.3180190324783325, "logits/rejected": 2.0549569129943848, "logps/chosen": -387.903076171875, "logps/rejected": -421.43560791015625, "loss": 0.5062, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -1.1257370710372925, "rewards/margins": 0.7061508893966675, "rewards/rejected": -1.83188796043396, "step": 370 }, { "epoch": 0.7949790794979079, "grad_norm": 25.874688989134494, "learning_rate": 6.138919252022435e-08, "logits/chosen": 1.4991798400878906, "logits/rejected": 2.2097418308258057, "logps/chosen": -361.59136962890625, "logps/rejected": -474.5025939941406, "loss": 0.5045, "rewards/accuracies": 0.78125, "rewards/chosen": -1.2211620807647705, "rewards/margins": 0.9370111227035522, "rewards/rejected": -2.1581733226776123, "step": 380 }, { "epoch": 0.8158995815899581, "grad_norm": 25.013956798098185, "learning_rate": 4.991445467064689e-08, "logits/chosen": 0.9765421152114868, "logits/rejected": 1.5557312965393066, "logps/chosen": -405.4990234375, "logps/rejected": -466.1669006347656, "loss": 0.4919, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -0.9986322522163391, "rewards/margins": 0.8234176635742188, "rewards/rejected": -1.822049856185913, "step": 390 }, { "epoch": 0.8368200836820083, "grad_norm": 25.325590503420294, "learning_rate": 3.9507259776993954e-08, "logits/chosen": 1.118195652961731, "logits/rejected": 2.1151347160339355, "logps/chosen": -369.673095703125, "logps/rejected": -455.16949462890625, "loss": 0.5037, "rewards/accuracies": 0.7749999761581421, "rewards/chosen": -0.991197943687439, "rewards/margins": 0.8906410336494446, "rewards/rejected": -1.8818390369415283, "step": 400 }, { "epoch": 0.8368200836820083, "eval_logits/chosen": 0.7959027290344238, "eval_logits/rejected": 1.6106007099151611, "eval_logps/chosen": -346.2210998535156, "eval_logps/rejected": -441.496826171875, "eval_loss": 0.5030511617660522, "eval_rewards/accuracies": 0.78515625, "eval_rewards/chosen": -0.8364605903625488, "eval_rewards/margins": 0.9516662955284119, "eval_rewards/rejected": -1.788127064704895, "eval_runtime": 40.3046, "eval_samples_per_second": 49.622, "eval_steps_per_second": 0.794, "step": 400 }, { "epoch": 0.8577405857740585, "grad_norm": 25.826488801828866, "learning_rate": 3.022313472693447e-08, "logits/chosen": 0.8946113586425781, "logits/rejected": 2.032439708709717, "logps/chosen": -391.67340087890625, "logps/rejected": -448.6036682128906, "loss": 0.5091, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -0.9272828102111816, "rewards/margins": 0.8706603050231934, "rewards/rejected": -1.797943115234375, "step": 410 }, { "epoch": 0.8786610878661087, "grad_norm": 24.611045682765628, "learning_rate": 2.2111614344599684e-08, "logits/chosen": 0.7512882947921753, "logits/rejected": 1.5819581747055054, "logps/chosen": -391.5676574707031, "logps/rejected": -447.89959716796875, "loss": 0.4949, "rewards/accuracies": 0.6812499761581421, "rewards/chosen": -0.9452153444290161, "rewards/margins": 0.789088249206543, "rewards/rejected": -1.7343034744262695, "step": 420 }, { "epoch": 0.899581589958159, "grad_norm": 24.57477424595926, "learning_rate": 1.521597710086439e-08, "logits/chosen": 1.212896704673767, "logits/rejected": 1.9935243129730225, "logps/chosen": -380.76373291015625, "logps/rejected": -443.2093811035156, "loss": 0.4764, "rewards/accuracies": 0.78125, "rewards/chosen": -1.0168489217758179, "rewards/margins": 0.8378502130508423, "rewards/rejected": -1.8546991348266602, "step": 430 }, { "epoch": 0.9205020920502092, "grad_norm": 29.526622672975254, "learning_rate": 9.57301420397924e-09, "logits/chosen": 0.9320109486579895, "logits/rejected": 1.8708562850952148, "logps/chosen": -375.6236877441406, "logps/rejected": -440.24822998046875, "loss": 0.4967, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -0.9130943417549133, "rewards/margins": 0.7984671592712402, "rewards/rejected": -1.7115614414215088, "step": 440 }, { "epoch": 0.9414225941422594, "grad_norm": 25.409754325281586, "learning_rate": 5.212833302556258e-09, "logits/chosen": 0.8921103477478027, "logits/rejected": 1.6804935932159424, "logps/chosen": -392.9017333984375, "logps/rejected": -487.7478942871094, "loss": 0.4971, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -0.9917811155319214, "rewards/margins": 0.8150280117988586, "rewards/rejected": -1.8068091869354248, "step": 450 }, { "epoch": 0.9623430962343096, "grad_norm": 27.931385213072396, "learning_rate": 2.158697848236607e-09, "logits/chosen": 1.0774990320205688, "logits/rejected": 1.940333366394043, "logps/chosen": -366.88958740234375, "logps/rejected": -417.73590087890625, "loss": 0.5033, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.9498540759086609, "rewards/margins": 0.801009476184845, "rewards/rejected": -1.7508634328842163, "step": 460 }, { "epoch": 0.9832635983263598, "grad_norm": 26.97146711470604, "learning_rate": 4.269029751107489e-10, "logits/chosen": 0.9106823801994324, "logits/rejected": 1.9309165477752686, "logps/chosen": -368.4771728515625, "logps/rejected": -455.9451599121094, "loss": 0.4933, "rewards/accuracies": 0.7562500238418579, "rewards/chosen": -0.8981329202651978, "rewards/margins": 0.8376766443252563, "rewards/rejected": -1.735809564590454, "step": 470 }, { "epoch": 1.0, "step": 478, "total_flos": 0.0, "train_loss": 0.5395244165444474, "train_runtime": 4445.7731, "train_samples_per_second": 13.751, "train_steps_per_second": 0.108 } ], "logging_steps": 10, "max_steps": 478, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 100, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }