diff --git "a/checkpoint-500/trainer_state.json" "b/checkpoint-500/trainer_state.json" new file mode 100644--- /dev/null +++ "b/checkpoint-500/trainer_state.json" @@ -0,0 +1,7021 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.6543432030099787, + "eval_steps": 500, + "global_step": 500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0, + "learning_rate": 5.000000000000001e-07, + "logits/chosen": -1.977054476737976, + "logits/rejected": -2.017892599105835, + "logps/chosen": -169.97320556640625, + "logps/rejected": -186.7821807861328, + "loss": 0.7082, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.016062308102846146, + "rewards/margins": -0.02699420601129532, + "rewards/rejected": 0.010931899771094322, + "step": 1 + }, + { + "epoch": 0.0, + "learning_rate": 1.0000000000000002e-06, + "logits/chosen": -1.8305251598358154, + "logits/rejected": -1.8582998514175415, + "logps/chosen": -155.8516082763672, + "logps/rejected": -165.23692321777344, + "loss": 0.6929, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.0017804148374125361, + "rewards/margins": 0.004515504464507103, + "rewards/rejected": -0.0027350913733243942, + "step": 2 + }, + { + "epoch": 0.0, + "learning_rate": 1.5e-06, + "logits/chosen": -1.7455682754516602, + "logits/rejected": -1.7944730520248413, + "logps/chosen": -158.9869842529297, + "logps/rejected": -179.4861602783203, + "loss": 0.6886, + "rewards/accuracies": 0.6875, + "rewards/chosen": 0.00722665898501873, + "rewards/margins": 0.01148004550486803, + "rewards/rejected": -0.004253389313817024, + "step": 3 + }, + { + "epoch": 0.01, + "learning_rate": 2.0000000000000003e-06, + "logits/chosen": -1.8648240566253662, + "logits/rejected": -1.8583375215530396, + "logps/chosen": -186.27041625976562, + "logps/rejected": -174.98153686523438, + "loss": 0.6867, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.013353967107832432, + "rewards/margins": 0.014140583574771881, + "rewards/rejected": -0.0007866150699555874, + "step": 4 + }, + { + "epoch": 0.01, + "learning_rate": 2.5e-06, + "logits/chosen": -1.5987751483917236, + "logits/rejected": -1.602742075920105, + "logps/chosen": -166.3009796142578, + "logps/rejected": -182.16493225097656, + "loss": 0.6973, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.01031036488711834, + "rewards/margins": -0.007737827021628618, + "rewards/rejected": -0.0025725378654897213, + "step": 5 + }, + { + "epoch": 0.01, + "learning_rate": 3e-06, + "logits/chosen": -1.4330253601074219, + "logits/rejected": -1.4852710962295532, + "logps/chosen": -170.77926635742188, + "logps/rejected": -203.38731384277344, + "loss": 0.6967, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.0022835731506347656, + "rewards/margins": -0.005300428252667189, + "rewards/rejected": 0.003016853705048561, + "step": 6 + }, + { + "epoch": 0.01, + "learning_rate": 3.5000000000000004e-06, + "logits/chosen": -1.5559855699539185, + "logits/rejected": -1.5930315256118774, + "logps/chosen": -200.5897979736328, + "logps/rejected": -216.58615112304688, + "loss": 0.6832, + "rewards/accuracies": 0.4375, + "rewards/chosen": 0.015020323917269707, + "rewards/margins": 0.02334442362189293, + "rewards/rejected": -0.008324098773300648, + "step": 7 + }, + { + "epoch": 0.01, + "learning_rate": 4.000000000000001e-06, + "logits/chosen": -1.7643598318099976, + "logits/rejected": -1.7637088298797607, + "logps/chosen": -166.2308349609375, + "logps/rejected": -167.15399169921875, + "loss": 0.6973, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.03954277187585831, + "rewards/margins": -0.005299141630530357, + "rewards/rejected": -0.0342436321079731, + "step": 8 + }, + { + "epoch": 0.01, + "learning_rate": 4.5e-06, + "logits/chosen": -2.035048246383667, + "logits/rejected": -2.1090240478515625, + "logps/chosen": -192.5548095703125, + "logps/rejected": -188.7212371826172, + "loss": 0.6788, + "rewards/accuracies": 0.625, + "rewards/chosen": 0.019565440714359283, + "rewards/margins": 0.030419450253248215, + "rewards/rejected": -0.010854003950953484, + "step": 9 + }, + { + "epoch": 0.01, + "learning_rate": 5e-06, + "logits/chosen": -1.78806471824646, + "logits/rejected": -1.771653175354004, + "logps/chosen": -169.3682403564453, + "logps/rejected": -163.67990112304688, + "loss": 0.6861, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.0037752394564449787, + "rewards/margins": 0.01528622955083847, + "rewards/rejected": -0.01906147226691246, + "step": 10 + }, + { + "epoch": 0.01, + "learning_rate": 5.500000000000001e-06, + "logits/chosen": -1.6899704933166504, + "logits/rejected": -1.7716753482818604, + "logps/chosen": -197.17576599121094, + "logps/rejected": -232.6337127685547, + "loss": 0.7158, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.06386594474315643, + "rewards/margins": -0.04272947832942009, + "rewards/rejected": -0.02113647386431694, + "step": 11 + }, + { + "epoch": 0.02, + "learning_rate": 6e-06, + "logits/chosen": -1.8938390016555786, + "logits/rejected": -1.8334659337997437, + "logps/chosen": -150.29385375976562, + "logps/rejected": -159.55947875976562, + "loss": 0.6895, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.0028306012973189354, + "rewards/margins": 0.01067290361970663, + "rewards/rejected": -0.00784230325371027, + "step": 12 + }, + { + "epoch": 0.02, + "learning_rate": 6.5000000000000004e-06, + "logits/chosen": -1.87659752368927, + "logits/rejected": -1.8801605701446533, + "logps/chosen": -205.460693359375, + "logps/rejected": -202.0462188720703, + "loss": 0.6984, + "rewards/accuracies": 0.4375, + "rewards/chosen": 0.0009463531896471977, + "rewards/margins": -0.007184028625488281, + "rewards/rejected": 0.008130382746458054, + "step": 13 + }, + { + "epoch": 0.02, + "learning_rate": 7.000000000000001e-06, + "logits/chosen": -1.9238759279251099, + "logits/rejected": -2.0003201961517334, + "logps/chosen": -184.6726531982422, + "logps/rejected": -177.62509155273438, + "loss": 0.6972, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.01391973439604044, + "rewards/margins": -0.004122593905776739, + "rewards/rejected": -0.009797144681215286, + "step": 14 + }, + { + "epoch": 0.02, + "learning_rate": 7.5e-06, + "logits/chosen": -1.569742202758789, + "logits/rejected": -1.5442943572998047, + "logps/chosen": -171.07496643066406, + "logps/rejected": -170.74981689453125, + "loss": 0.6992, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.020387031137943268, + "rewards/margins": -0.008263109251856804, + "rewards/rejected": -0.012123920023441315, + "step": 15 + }, + { + "epoch": 0.02, + "learning_rate": 8.000000000000001e-06, + "logits/chosen": -1.7774536609649658, + "logits/rejected": -1.8797023296356201, + "logps/chosen": -173.0279541015625, + "logps/rejected": -187.43557739257812, + "loss": 0.6999, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.03834056854248047, + "rewards/margins": -0.01055521797388792, + "rewards/rejected": -0.027785349637269974, + "step": 16 + }, + { + "epoch": 0.02, + "learning_rate": 8.500000000000002e-06, + "logits/chosen": -1.7320338487625122, + "logits/rejected": -1.6945910453796387, + "logps/chosen": -185.27392578125, + "logps/rejected": -189.87738037109375, + "loss": 0.6808, + "rewards/accuracies": 0.6875, + "rewards/chosen": 0.007571219466626644, + "rewards/margins": 0.025459958240389824, + "rewards/rejected": -0.017888737842440605, + "step": 17 + }, + { + "epoch": 0.02, + "learning_rate": 9e-06, + "logits/chosen": -1.7604248523712158, + "logits/rejected": -1.7776434421539307, + "logps/chosen": -196.05404663085938, + "logps/rejected": -190.14569091796875, + "loss": 0.6914, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.02323136478662491, + "rewards/margins": 0.00760660320520401, + "rewards/rejected": -0.03083796612918377, + "step": 18 + }, + { + "epoch": 0.02, + "learning_rate": 9.5e-06, + "logits/chosen": -1.59357488155365, + "logits/rejected": -1.5590327978134155, + "logps/chosen": -213.9495391845703, + "logps/rejected": -218.56654357910156, + "loss": 0.7222, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.05001959949731827, + "rewards/margins": -0.05468587949872017, + "rewards/rejected": 0.004666280932724476, + "step": 19 + }, + { + "epoch": 0.03, + "learning_rate": 1e-05, + "logits/chosen": -1.752557635307312, + "logits/rejected": -1.7027242183685303, + "logps/chosen": -213.31336975097656, + "logps/rejected": -204.09646606445312, + "loss": 0.6887, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.017501091584563255, + "rewards/margins": 0.010245682671666145, + "rewards/rejected": -0.027746770530939102, + "step": 20 + }, + { + "epoch": 0.03, + "learning_rate": 1.05e-05, + "logits/chosen": -1.9553875923156738, + "logits/rejected": -1.9184911251068115, + "logps/chosen": -175.35333251953125, + "logps/rejected": -180.54550170898438, + "loss": 0.655, + "rewards/accuracies": 0.9375, + "rewards/chosen": 0.021092725917696953, + "rewards/margins": 0.08075069636106491, + "rewards/rejected": -0.059657976031303406, + "step": 21 + }, + { + "epoch": 0.03, + "learning_rate": 1.1000000000000001e-05, + "logits/chosen": -1.8159900903701782, + "logits/rejected": -1.771599531173706, + "logps/chosen": -185.33059692382812, + "logps/rejected": -209.6474609375, + "loss": 0.6913, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.017464350908994675, + "rewards/margins": 0.006189251318573952, + "rewards/rejected": -0.023653600364923477, + "step": 22 + }, + { + "epoch": 0.03, + "learning_rate": 1.1500000000000002e-05, + "logits/chosen": -1.8995972871780396, + "logits/rejected": -1.9293156862258911, + "logps/chosen": -178.39755249023438, + "logps/rejected": -211.63937377929688, + "loss": 0.7191, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.05782761424779892, + "rewards/margins": -0.04498009383678436, + "rewards/rejected": -0.012847519479691982, + "step": 23 + }, + { + "epoch": 0.03, + "learning_rate": 1.2e-05, + "logits/chosen": -1.8924778699874878, + "logits/rejected": -1.8979182243347168, + "logps/chosen": -167.22109985351562, + "logps/rejected": -176.33663940429688, + "loss": 0.7028, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.02767309918999672, + "rewards/margins": -0.016704557463526726, + "rewards/rejected": -0.010968542657792568, + "step": 24 + }, + { + "epoch": 0.03, + "learning_rate": 1.25e-05, + "logits/chosen": -1.840967059135437, + "logits/rejected": -1.797666311264038, + "logps/chosen": -175.57766723632812, + "logps/rejected": -148.11917114257812, + "loss": 0.6967, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.01812169887125492, + "rewards/margins": -0.003334569279104471, + "rewards/rejected": -0.014787126332521439, + "step": 25 + }, + { + "epoch": 0.03, + "learning_rate": 1.3000000000000001e-05, + "logits/chosen": -1.8270690441131592, + "logits/rejected": -1.7811157703399658, + "logps/chosen": -165.993896484375, + "logps/rejected": -165.7574005126953, + "loss": 0.7151, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.06210968643426895, + "rewards/margins": -0.040364596992731094, + "rewards/rejected": -0.02174508571624756, + "step": 26 + }, + { + "epoch": 0.04, + "learning_rate": 1.3500000000000001e-05, + "logits/chosen": -1.877970576286316, + "logits/rejected": -1.9722176790237427, + "logps/chosen": -148.4381103515625, + "logps/rejected": -173.12579345703125, + "loss": 0.6835, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.022617867216467857, + "rewards/margins": 0.021304797381162643, + "rewards/rejected": -0.04392266273498535, + "step": 27 + }, + { + "epoch": 0.04, + "learning_rate": 1.4000000000000001e-05, + "logits/chosen": -1.922934889793396, + "logits/rejected": -1.9657589197158813, + "logps/chosen": -170.66554260253906, + "logps/rejected": -176.7419891357422, + "loss": 0.7316, + "rewards/accuracies": 0.1875, + "rewards/chosen": -0.10023985058069229, + "rewards/margins": -0.0725136250257492, + "rewards/rejected": -0.027726221829652786, + "step": 28 + }, + { + "epoch": 0.04, + "learning_rate": 1.45e-05, + "logits/chosen": -1.9069832563400269, + "logits/rejected": -1.8908956050872803, + "logps/chosen": -193.16102600097656, + "logps/rejected": -193.22003173828125, + "loss": 0.6847, + "rewards/accuracies": 0.3125, + "rewards/chosen": 0.005608892068266869, + "rewards/margins": 0.020983649417757988, + "rewards/rejected": -0.01537475548684597, + "step": 29 + }, + { + "epoch": 0.04, + "learning_rate": 1.5e-05, + "logits/chosen": -1.7016417980194092, + "logits/rejected": -1.7221649885177612, + "logps/chosen": -188.40786743164062, + "logps/rejected": -175.92909240722656, + "loss": 0.7016, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.09070225059986115, + "rewards/margins": -0.013078359887003899, + "rewards/rejected": -0.0776238888502121, + "step": 30 + }, + { + "epoch": 0.04, + "learning_rate": 1.55e-05, + "logits/chosen": -1.9492610692977905, + "logits/rejected": -1.9104335308074951, + "logps/chosen": -180.67147827148438, + "logps/rejected": -184.8843994140625, + "loss": 0.6708, + "rewards/accuracies": 0.625, + "rewards/chosen": 0.019913675263524055, + "rewards/margins": 0.049428701400756836, + "rewards/rejected": -0.02951502799987793, + "step": 31 + }, + { + "epoch": 0.04, + "learning_rate": 1.6000000000000003e-05, + "logits/chosen": -1.8524138927459717, + "logits/rejected": -1.9383589029312134, + "logps/chosen": -159.52560424804688, + "logps/rejected": -170.27255249023438, + "loss": 0.7146, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.03338008001446724, + "rewards/margins": -0.03860168159008026, + "rewards/rejected": 0.00522160530090332, + "step": 32 + }, + { + "epoch": 0.04, + "learning_rate": 1.65e-05, + "logits/chosen": -1.7788478136062622, + "logits/rejected": -1.835086703300476, + "logps/chosen": -180.18177795410156, + "logps/rejected": -206.07110595703125, + "loss": 0.7059, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.05509199947118759, + "rewards/margins": -0.023075008764863014, + "rewards/rejected": -0.03201699256896973, + "step": 33 + }, + { + "epoch": 0.04, + "learning_rate": 1.7000000000000003e-05, + "logits/chosen": -1.8998254537582397, + "logits/rejected": -1.9138494729995728, + "logps/chosen": -194.03167724609375, + "logps/rejected": -203.6524658203125, + "loss": 0.6915, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.05050516128540039, + "rewards/margins": 0.007271335460245609, + "rewards/rejected": -0.057776499539613724, + "step": 34 + }, + { + "epoch": 0.05, + "learning_rate": 1.75e-05, + "logits/chosen": -1.9758315086364746, + "logits/rejected": -2.0455610752105713, + "logps/chosen": -153.28883361816406, + "logps/rejected": -162.89920043945312, + "loss": 0.7122, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.04079794883728027, + "rewards/margins": -0.034585997462272644, + "rewards/rejected": -0.006211946718394756, + "step": 35 + }, + { + "epoch": 0.05, + "learning_rate": 1.8e-05, + "logits/chosen": -1.597517967224121, + "logits/rejected": -1.631009817123413, + "logps/chosen": -160.9354248046875, + "logps/rejected": -168.85618591308594, + "loss": 0.6918, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.07598643749952316, + "rewards/margins": 0.005950784310698509, + "rewards/rejected": -0.08193722367286682, + "step": 36 + }, + { + "epoch": 0.05, + "learning_rate": 1.85e-05, + "logits/chosen": -1.7368295192718506, + "logits/rejected": -1.7420881986618042, + "logps/chosen": -169.4617462158203, + "logps/rejected": -184.20599365234375, + "loss": 0.6948, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.0290069580078125, + "rewards/margins": -0.0007306085899472237, + "rewards/rejected": -0.0282763484865427, + "step": 37 + }, + { + "epoch": 0.05, + "learning_rate": 1.9e-05, + "logits/chosen": -1.687361717224121, + "logits/rejected": -1.7176148891448975, + "logps/chosen": -174.06297302246094, + "logps/rejected": -192.69715881347656, + "loss": 0.6973, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.09517102688550949, + "rewards/margins": -0.0029970891773700714, + "rewards/rejected": -0.09217393398284912, + "step": 38 + }, + { + "epoch": 0.05, + "learning_rate": 1.9500000000000003e-05, + "logits/chosen": -1.8842296600341797, + "logits/rejected": -1.9039793014526367, + "logps/chosen": -162.37741088867188, + "logps/rejected": -176.11697387695312, + "loss": 0.7109, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.13322168588638306, + "rewards/margins": -0.03214216232299805, + "rewards/rejected": -0.10107951611280441, + "step": 39 + }, + { + "epoch": 0.05, + "learning_rate": 2e-05, + "logits/chosen": -1.6896815299987793, + "logits/rejected": -1.7052747011184692, + "logps/chosen": -155.33111572265625, + "logps/rejected": -159.064208984375, + "loss": 0.6846, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.04464542865753174, + "rewards/margins": 0.025189755484461784, + "rewards/rejected": -0.06983518600463867, + "step": 40 + }, + { + "epoch": 0.05, + "learning_rate": 2.05e-05, + "logits/chosen": -1.7584484815597534, + "logits/rejected": -1.6827234029769897, + "logps/chosen": -161.47369384765625, + "logps/rejected": -166.0779571533203, + "loss": 0.6856, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.07150936126708984, + "rewards/margins": 0.019259024411439896, + "rewards/rejected": -0.09076839685440063, + "step": 41 + }, + { + "epoch": 0.05, + "learning_rate": 2.1e-05, + "logits/chosen": -2.0230112075805664, + "logits/rejected": -1.9628382921218872, + "logps/chosen": -160.32073974609375, + "logps/rejected": -182.98248291015625, + "loss": 0.6857, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.09749487042427063, + "rewards/margins": 0.01751875691115856, + "rewards/rejected": -0.11501362919807434, + "step": 42 + }, + { + "epoch": 0.06, + "learning_rate": 2.15e-05, + "logits/chosen": -1.8469973802566528, + "logits/rejected": -1.814754843711853, + "logps/chosen": -194.247314453125, + "logps/rejected": -204.80491638183594, + "loss": 0.7142, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.15520897507667542, + "rewards/margins": -0.03922419250011444, + "rewards/rejected": -0.11598476767539978, + "step": 43 + }, + { + "epoch": 0.06, + "learning_rate": 2.2000000000000003e-05, + "logits/chosen": -1.9133589267730713, + "logits/rejected": -1.9711928367614746, + "logps/chosen": -186.54978942871094, + "logps/rejected": -189.23045349121094, + "loss": 0.7238, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.1358685940504074, + "rewards/margins": -0.05334620922803879, + "rewards/rejected": -0.08252239227294922, + "step": 44 + }, + { + "epoch": 0.06, + "learning_rate": 2.25e-05, + "logits/chosen": -1.6977447271347046, + "logits/rejected": -1.7883001565933228, + "logps/chosen": -178.20858764648438, + "logps/rejected": -201.03770446777344, + "loss": 0.6465, + "rewards/accuracies": 0.8125, + "rewards/chosen": -0.053537700325250626, + "rewards/margins": 0.1018117368221283, + "rewards/rejected": -0.15534944832324982, + "step": 45 + }, + { + "epoch": 0.06, + "learning_rate": 2.3000000000000003e-05, + "logits/chosen": -1.9271280765533447, + "logits/rejected": -1.9463679790496826, + "logps/chosen": -176.5238800048828, + "logps/rejected": -166.8386993408203, + "loss": 0.7159, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.09149947762489319, + "rewards/margins": -0.04024248570203781, + "rewards/rejected": -0.05125699192285538, + "step": 46 + }, + { + "epoch": 0.06, + "learning_rate": 2.35e-05, + "logits/chosen": -1.7680522203445435, + "logits/rejected": -1.668277621269226, + "logps/chosen": -189.94366455078125, + "logps/rejected": -177.8812255859375, + "loss": 0.6647, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.08811293542385101, + "rewards/margins": 0.06799888610839844, + "rewards/rejected": -0.15611180663108826, + "step": 47 + }, + { + "epoch": 0.06, + "learning_rate": 2.4e-05, + "logits/chosen": -1.5737125873565674, + "logits/rejected": -1.626237154006958, + "logps/chosen": -177.7281036376953, + "logps/rejected": -185.34068298339844, + "loss": 0.7038, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.12158739566802979, + "rewards/margins": -0.004673934541642666, + "rewards/rejected": -0.11691347509622574, + "step": 48 + }, + { + "epoch": 0.06, + "learning_rate": 2.45e-05, + "logits/chosen": -1.487099528312683, + "logits/rejected": -1.5200517177581787, + "logps/chosen": -152.00973510742188, + "logps/rejected": -163.13150024414062, + "loss": 0.7211, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.12497053295373917, + "rewards/margins": -0.05033881962299347, + "rewards/rejected": -0.0746317207813263, + "step": 49 + }, + { + "epoch": 0.07, + "learning_rate": 2.5e-05, + "logits/chosen": -1.751338243484497, + "logits/rejected": -1.7326526641845703, + "logps/chosen": -180.03758239746094, + "logps/rejected": -191.03634643554688, + "loss": 0.7029, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.12404017895460129, + "rewards/margins": -0.013475272804498672, + "rewards/rejected": -0.11056490242481232, + "step": 50 + }, + { + "epoch": 0.07, + "learning_rate": 2.5500000000000003e-05, + "logits/chosen": -1.9426243305206299, + "logits/rejected": -1.9601188898086548, + "logps/chosen": -158.4928741455078, + "logps/rejected": -164.06317138671875, + "loss": 0.7142, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.12027917802333832, + "rewards/margins": -0.028645988553762436, + "rewards/rejected": -0.09163318574428558, + "step": 51 + }, + { + "epoch": 0.07, + "learning_rate": 2.6000000000000002e-05, + "logits/chosen": -1.5965681076049805, + "logits/rejected": -1.597716212272644, + "logps/chosen": -195.06454467773438, + "logps/rejected": -193.9747314453125, + "loss": 0.6905, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.14499236643314362, + "rewards/margins": 0.02140347845852375, + "rewards/rejected": -0.16639582812786102, + "step": 52 + }, + { + "epoch": 0.07, + "learning_rate": 2.6500000000000004e-05, + "logits/chosen": -1.9194509983062744, + "logits/rejected": -1.883784294128418, + "logps/chosen": -161.39295959472656, + "logps/rejected": -167.61610412597656, + "loss": 0.6965, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.11441681534051895, + "rewards/margins": -0.00019459612667560577, + "rewards/rejected": -0.1142222210764885, + "step": 53 + }, + { + "epoch": 0.07, + "learning_rate": 2.7000000000000002e-05, + "logits/chosen": -1.7368502616882324, + "logits/rejected": -1.7216427326202393, + "logps/chosen": -205.18130493164062, + "logps/rejected": -191.90237426757812, + "loss": 0.66, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.1574774831533432, + "rewards/margins": 0.07581701874732971, + "rewards/rejected": -0.23329448699951172, + "step": 54 + }, + { + "epoch": 0.07, + "learning_rate": 2.7500000000000004e-05, + "logits/chosen": -2.0303847789764404, + "logits/rejected": -2.0456559658050537, + "logps/chosen": -169.33453369140625, + "logps/rejected": -166.40707397460938, + "loss": 0.6479, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.14774441719055176, + "rewards/margins": 0.1020236536860466, + "rewards/rejected": -0.24976806342601776, + "step": 55 + }, + { + "epoch": 0.07, + "learning_rate": 2.8000000000000003e-05, + "logits/chosen": -2.035883903503418, + "logits/rejected": -1.9933511018753052, + "logps/chosen": -190.82186889648438, + "logps/rejected": -190.29147338867188, + "loss": 0.7023, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.11226066946983337, + "rewards/margins": -0.01058507151901722, + "rewards/rejected": -0.1016756072640419, + "step": 56 + }, + { + "epoch": 0.07, + "learning_rate": 2.8499999999999998e-05, + "logits/chosen": -1.9773008823394775, + "logits/rejected": -1.935595989227295, + "logps/chosen": -180.8785400390625, + "logps/rejected": -193.48155212402344, + "loss": 0.6739, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.2835652828216553, + "rewards/margins": 0.048453718423843384, + "rewards/rejected": -0.33201897144317627, + "step": 57 + }, + { + "epoch": 0.08, + "learning_rate": 2.9e-05, + "logits/chosen": -1.681299090385437, + "logits/rejected": -1.6699227094650269, + "logps/chosen": -165.8355255126953, + "logps/rejected": -191.2743377685547, + "loss": 0.7356, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.17172232270240784, + "rewards/margins": -0.07655029743909836, + "rewards/rejected": -0.09517201036214828, + "step": 58 + }, + { + "epoch": 0.08, + "learning_rate": 2.95e-05, + "logits/chosen": -1.71488356590271, + "logits/rejected": -1.7937383651733398, + "logps/chosen": -177.51776123046875, + "logps/rejected": -188.60760498046875, + "loss": 0.6358, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.1568491905927658, + "rewards/margins": 0.12602630257606506, + "rewards/rejected": -0.2828754782676697, + "step": 59 + }, + { + "epoch": 0.08, + "learning_rate": 3e-05, + "logits/chosen": -2.0625619888305664, + "logits/rejected": -2.1217410564422607, + "logps/chosen": -190.26431274414062, + "logps/rejected": -215.28648376464844, + "loss": 0.7777, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.3512558341026306, + "rewards/margins": -0.13770633935928345, + "rewards/rejected": -0.21354950964450836, + "step": 60 + }, + { + "epoch": 0.08, + "learning_rate": 3.05e-05, + "logits/chosen": -1.5000535249710083, + "logits/rejected": -1.508121371269226, + "logps/chosen": -167.42588806152344, + "logps/rejected": -151.4751739501953, + "loss": 0.692, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.15655343234539032, + "rewards/margins": 0.01617264747619629, + "rewards/rejected": -0.1727260947227478, + "step": 61 + }, + { + "epoch": 0.08, + "learning_rate": 3.1e-05, + "logits/chosen": -1.628549337387085, + "logits/rejected": -1.7045596837997437, + "logps/chosen": -206.2198944091797, + "logps/rejected": -188.40350341796875, + "loss": 0.7688, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.2638259530067444, + "rewards/margins": -0.12546539306640625, + "rewards/rejected": -0.13836055994033813, + "step": 62 + }, + { + "epoch": 0.08, + "learning_rate": 3.15e-05, + "logits/chosen": -1.6634955406188965, + "logits/rejected": -1.641862154006958, + "logps/chosen": -173.11769104003906, + "logps/rejected": -179.14816284179688, + "loss": 0.6446, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.1516265571117401, + "rewards/margins": 0.10875654965639114, + "rewards/rejected": -0.26038309931755066, + "step": 63 + }, + { + "epoch": 0.08, + "learning_rate": 3.2000000000000005e-05, + "logits/chosen": -1.90239417552948, + "logits/rejected": -1.846576452255249, + "logps/chosen": -185.16522216796875, + "logps/rejected": -187.98654174804688, + "loss": 0.696, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.26641684770584106, + "rewards/margins": 0.004999059252440929, + "rewards/rejected": -0.27141591906547546, + "step": 64 + }, + { + "epoch": 0.09, + "learning_rate": 3.2500000000000004e-05, + "logits/chosen": -1.9015129804611206, + "logits/rejected": -1.9043163061141968, + "logps/chosen": -166.89393615722656, + "logps/rejected": -185.8280487060547, + "loss": 0.7096, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.3341715931892395, + "rewards/margins": -0.005350928753614426, + "rewards/rejected": -0.32882067561149597, + "step": 65 + }, + { + "epoch": 0.09, + "learning_rate": 3.3e-05, + "logits/chosen": -1.7935004234313965, + "logits/rejected": -1.8295319080352783, + "logps/chosen": -145.04159545898438, + "logps/rejected": -152.42388916015625, + "loss": 0.7515, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.3907526433467865, + "rewards/margins": -0.10265941917896271, + "rewards/rejected": -0.2880932092666626, + "step": 66 + }, + { + "epoch": 0.09, + "learning_rate": 3.35e-05, + "logits/chosen": -1.6959490776062012, + "logits/rejected": -1.680245280265808, + "logps/chosen": -165.76852416992188, + "logps/rejected": -161.328857421875, + "loss": 0.7212, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.3491905629634857, + "rewards/margins": -0.03104216605424881, + "rewards/rejected": -0.3181484043598175, + "step": 67 + }, + { + "epoch": 0.09, + "learning_rate": 3.4000000000000007e-05, + "logits/chosen": -1.5831743478775024, + "logits/rejected": -1.5631486177444458, + "logps/chosen": -182.02952575683594, + "logps/rejected": -183.06907653808594, + "loss": 0.6674, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.22372691333293915, + "rewards/margins": 0.07622986286878586, + "rewards/rejected": -0.2999567687511444, + "step": 68 + }, + { + "epoch": 0.09, + "learning_rate": 3.45e-05, + "logits/chosen": -1.7518221139907837, + "logits/rejected": -1.7015327215194702, + "logps/chosen": -181.39324951171875, + "logps/rejected": -181.88720703125, + "loss": 0.6668, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.3097344934940338, + "rewards/margins": 0.0728234276175499, + "rewards/rejected": -0.3825579285621643, + "step": 69 + }, + { + "epoch": 0.09, + "learning_rate": 3.5e-05, + "logits/chosen": -1.8494486808776855, + "logits/rejected": -1.80762779712677, + "logps/chosen": -181.36915588378906, + "logps/rejected": -186.561279296875, + "loss": 0.7449, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.47243672609329224, + "rewards/margins": -0.07020688056945801, + "rewards/rejected": -0.40222981572151184, + "step": 70 + }, + { + "epoch": 0.09, + "learning_rate": 3.55e-05, + "logits/chosen": -1.701026201248169, + "logits/rejected": -1.6742680072784424, + "logps/chosen": -157.6632843017578, + "logps/rejected": -161.87045288085938, + "loss": 0.6544, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.27613064646720886, + "rewards/margins": 0.10120917111635208, + "rewards/rejected": -0.37733981013298035, + "step": 71 + }, + { + "epoch": 0.09, + "learning_rate": 3.6e-05, + "logits/chosen": -1.7675864696502686, + "logits/rejected": -1.8016642332077026, + "logps/chosen": -185.2778778076172, + "logps/rejected": -212.45452880859375, + "loss": 0.6621, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.3811972141265869, + "rewards/margins": 0.07623375207185745, + "rewards/rejected": -0.45743098855018616, + "step": 72 + }, + { + "epoch": 0.1, + "learning_rate": 3.65e-05, + "logits/chosen": -1.8149230480194092, + "logits/rejected": -1.8410683870315552, + "logps/chosen": -167.20913696289062, + "logps/rejected": -197.37989807128906, + "loss": 0.7319, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.48692482709884644, + "rewards/margins": -0.050217654556035995, + "rewards/rejected": -0.43670713901519775, + "step": 73 + }, + { + "epoch": 0.1, + "learning_rate": 3.7e-05, + "logits/chosen": -1.8524658679962158, + "logits/rejected": -1.8538126945495605, + "logps/chosen": -164.4295654296875, + "logps/rejected": -180.02191162109375, + "loss": 0.7363, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.4177883267402649, + "rewards/margins": -0.050662752240896225, + "rewards/rejected": -0.36712557077407837, + "step": 74 + }, + { + "epoch": 0.1, + "learning_rate": 3.7500000000000003e-05, + "logits/chosen": -1.2946817874908447, + "logits/rejected": -1.3125836849212646, + "logps/chosen": -225.81192016601562, + "logps/rejected": -252.005859375, + "loss": 0.7049, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.37055703997612, + "rewards/margins": 0.016286462545394897, + "rewards/rejected": -0.3868435025215149, + "step": 75 + }, + { + "epoch": 0.1, + "learning_rate": 3.8e-05, + "logits/chosen": -1.7864320278167725, + "logits/rejected": -1.8551700115203857, + "logps/chosen": -176.35296630859375, + "logps/rejected": -182.03231811523438, + "loss": 0.7029, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.49798864126205444, + "rewards/margins": 0.02277611568570137, + "rewards/rejected": -0.5207647681236267, + "step": 76 + }, + { + "epoch": 0.1, + "learning_rate": 3.85e-05, + "logits/chosen": -1.7414928674697876, + "logits/rejected": -1.7363300323486328, + "logps/chosen": -191.74154663085938, + "logps/rejected": -199.81724548339844, + "loss": 0.6928, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.4908958375453949, + "rewards/margins": 0.02561158686876297, + "rewards/rejected": -0.5165074467658997, + "step": 77 + }, + { + "epoch": 0.1, + "learning_rate": 3.9000000000000006e-05, + "logits/chosen": -1.835726261138916, + "logits/rejected": -1.840294599533081, + "logps/chosen": -191.46029663085938, + "logps/rejected": -168.4354248046875, + "loss": 0.7986, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.5988375544548035, + "rewards/margins": -0.1457357108592987, + "rewards/rejected": -0.45310184359550476, + "step": 78 + }, + { + "epoch": 0.1, + "learning_rate": 3.9500000000000005e-05, + "logits/chosen": -1.6384897232055664, + "logits/rejected": -1.5781760215759277, + "logps/chosen": -217.35772705078125, + "logps/rejected": -217.51126098632812, + "loss": 0.7354, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6283512115478516, + "rewards/margins": -0.027955979108810425, + "rewards/rejected": -0.6003952026367188, + "step": 79 + }, + { + "epoch": 0.1, + "learning_rate": 4e-05, + "logits/chosen": -1.7660375833511353, + "logits/rejected": -1.7639034986495972, + "logps/chosen": -152.1668243408203, + "logps/rejected": -141.91021728515625, + "loss": 0.7167, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.5395210385322571, + "rewards/margins": -0.012110946699976921, + "rewards/rejected": -0.527410089969635, + "step": 80 + }, + { + "epoch": 0.11, + "learning_rate": 4.05e-05, + "logits/chosen": -1.5262843370437622, + "logits/rejected": -1.534919261932373, + "logps/chosen": -182.58859252929688, + "logps/rejected": -200.82281494140625, + "loss": 0.6912, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6189998388290405, + "rewards/margins": 0.02953934296965599, + "rewards/rejected": -0.6485391855239868, + "step": 81 + }, + { + "epoch": 0.11, + "learning_rate": 4.1e-05, + "logits/chosen": -1.6617029905319214, + "logits/rejected": -1.6961578130722046, + "logps/chosen": -177.42286682128906, + "logps/rejected": -168.7061767578125, + "loss": 0.7955, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.7245805859565735, + "rewards/margins": -0.15311545133590698, + "rewards/rejected": -0.5714651942253113, + "step": 82 + }, + { + "epoch": 0.11, + "learning_rate": 4.15e-05, + "logits/chosen": -1.6676827669143677, + "logits/rejected": -1.5490450859069824, + "logps/chosen": -171.3192596435547, + "logps/rejected": -173.60470581054688, + "loss": 0.7403, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.5414891242980957, + "rewards/margins": -0.05256550386548042, + "rewards/rejected": -0.4889236092567444, + "step": 83 + }, + { + "epoch": 0.11, + "learning_rate": 4.2e-05, + "logits/chosen": -1.8843668699264526, + "logits/rejected": -1.9513225555419922, + "logps/chosen": -172.60548400878906, + "logps/rejected": -164.8987579345703, + "loss": 0.7633, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.5382730960845947, + "rewards/margins": -0.10875138640403748, + "rewards/rejected": -0.42952167987823486, + "step": 84 + }, + { + "epoch": 0.11, + "learning_rate": 4.25e-05, + "logits/chosen": -1.9940603971481323, + "logits/rejected": -1.9973390102386475, + "logps/chosen": -161.12863159179688, + "logps/rejected": -164.32958984375, + "loss": 0.7724, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.5231802463531494, + "rewards/margins": -0.11174039542675018, + "rewards/rejected": -0.41143983602523804, + "step": 85 + }, + { + "epoch": 0.11, + "learning_rate": 4.3e-05, + "logits/chosen": -1.8103983402252197, + "logits/rejected": -1.7585985660552979, + "logps/chosen": -195.51132202148438, + "logps/rejected": -189.4246826171875, + "loss": 0.747, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.7077435255050659, + "rewards/margins": -0.08058477193117142, + "rewards/rejected": -0.6271587610244751, + "step": 86 + }, + { + "epoch": 0.11, + "learning_rate": 4.35e-05, + "logits/chosen": -1.9289149045944214, + "logits/rejected": -1.982129693031311, + "logps/chosen": -170.78253173828125, + "logps/rejected": -172.57882690429688, + "loss": 0.8229, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.6636737585067749, + "rewards/margins": -0.2039366364479065, + "rewards/rejected": -0.4597371816635132, + "step": 87 + }, + { + "epoch": 0.12, + "learning_rate": 4.4000000000000006e-05, + "logits/chosen": -1.875556230545044, + "logits/rejected": -1.8859914541244507, + "logps/chosen": -160.60577392578125, + "logps/rejected": -171.32774353027344, + "loss": 0.6905, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.3266763389110565, + "rewards/margins": 0.03434550017118454, + "rewards/rejected": -0.36102184653282166, + "step": 88 + }, + { + "epoch": 0.12, + "learning_rate": 4.4500000000000004e-05, + "logits/chosen": -1.7824954986572266, + "logits/rejected": -1.7617722749710083, + "logps/chosen": -177.59042358398438, + "logps/rejected": -200.9052276611328, + "loss": 0.6799, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.4084916114807129, + "rewards/margins": 0.0701964944601059, + "rewards/rejected": -0.47868812084198, + "step": 89 + }, + { + "epoch": 0.12, + "learning_rate": 4.5e-05, + "logits/chosen": -1.7419114112854004, + "logits/rejected": -1.7621021270751953, + "logps/chosen": -185.32742309570312, + "logps/rejected": -171.69085693359375, + "loss": 0.8305, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.3024221956729889, + "rewards/margins": -0.20739878714084625, + "rewards/rejected": -0.09502339363098145, + "step": 90 + }, + { + "epoch": 0.12, + "learning_rate": 4.55e-05, + "logits/chosen": -1.83467435836792, + "logits/rejected": -1.8328973054885864, + "logps/chosen": -207.63388061523438, + "logps/rejected": -210.3101043701172, + "loss": 0.6628, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.3415074646472931, + "rewards/margins": 0.07163538783788681, + "rewards/rejected": -0.4131428897380829, + "step": 91 + }, + { + "epoch": 0.12, + "learning_rate": 4.600000000000001e-05, + "logits/chosen": -1.6864506006240845, + "logits/rejected": -1.7266168594360352, + "logps/chosen": -180.49522399902344, + "logps/rejected": -198.40599060058594, + "loss": 0.6523, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.27834925055503845, + "rewards/margins": 0.10665541142225266, + "rewards/rejected": -0.3850046396255493, + "step": 92 + }, + { + "epoch": 0.12, + "learning_rate": 4.6500000000000005e-05, + "logits/chosen": -1.4600155353546143, + "logits/rejected": -1.4545408487319946, + "logps/chosen": -191.01947021484375, + "logps/rejected": -180.70298767089844, + "loss": 0.7165, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.1802830994129181, + "rewards/margins": -0.027431445196270943, + "rewards/rejected": -0.1528516411781311, + "step": 93 + }, + { + "epoch": 0.12, + "learning_rate": 4.7e-05, + "logits/chosen": -1.8134263753890991, + "logits/rejected": -1.8928412199020386, + "logps/chosen": -184.84503173828125, + "logps/rejected": -193.90377807617188, + "loss": 0.6644, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.3032827377319336, + "rewards/margins": 0.0916454941034317, + "rewards/rejected": -0.3949282467365265, + "step": 94 + }, + { + "epoch": 0.12, + "learning_rate": 4.75e-05, + "logits/chosen": -1.8253490924835205, + "logits/rejected": -1.810834527015686, + "logps/chosen": -208.2128448486328, + "logps/rejected": -175.4803009033203, + "loss": 0.7444, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.5496629476547241, + "rewards/margins": -0.08057989180088043, + "rewards/rejected": -0.4690830111503601, + "step": 95 + }, + { + "epoch": 0.13, + "learning_rate": 4.8e-05, + "logits/chosen": -1.908506989479065, + "logits/rejected": -1.9602031707763672, + "logps/chosen": -184.815185546875, + "logps/rejected": -176.07138061523438, + "loss": 0.6982, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.3538370728492737, + "rewards/margins": 0.013028910383582115, + "rewards/rejected": -0.36686599254608154, + "step": 96 + }, + { + "epoch": 0.13, + "learning_rate": 4.85e-05, + "logits/chosen": -1.7022223472595215, + "logits/rejected": -1.6424753665924072, + "logps/chosen": -198.0614776611328, + "logps/rejected": -215.3161163330078, + "loss": 0.7737, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.2290499359369278, + "rewards/margins": -0.12730106711387634, + "rewards/rejected": -0.10174884647130966, + "step": 97 + }, + { + "epoch": 0.13, + "learning_rate": 4.9e-05, + "logits/chosen": -2.018606662750244, + "logits/rejected": -2.027151584625244, + "logps/chosen": -167.92147827148438, + "logps/rejected": -166.90982055664062, + "loss": 0.6961, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.3297194838523865, + "rewards/margins": 0.02775608003139496, + "rewards/rejected": -0.35747551918029785, + "step": 98 + }, + { + "epoch": 0.13, + "learning_rate": 4.9500000000000004e-05, + "logits/chosen": -1.7240030765533447, + "logits/rejected": -1.7241712808609009, + "logps/chosen": -180.00389099121094, + "logps/rejected": -189.3558349609375, + "loss": 0.7948, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.4574447572231293, + "rewards/margins": -0.12154103070497513, + "rewards/rejected": -0.33590370416641235, + "step": 99 + }, + { + "epoch": 0.13, + "learning_rate": 5e-05, + "logits/chosen": -1.7809938192367554, + "logits/rejected": -1.9036985635757446, + "logps/chosen": -157.1652069091797, + "logps/rejected": -173.80288696289062, + "loss": 0.6839, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.2328178882598877, + "rewards/margins": 0.05165515094995499, + "rewards/rejected": -0.2844730615615845, + "step": 100 + }, + { + "epoch": 0.13, + "learning_rate": 4.999997432392803e-05, + "logits/chosen": -1.9480023384094238, + "logits/rejected": -1.9346106052398682, + "logps/chosen": -197.60128784179688, + "logps/rejected": -193.88124084472656, + "loss": 0.6427, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.26800671219825745, + "rewards/margins": 0.13769717514514923, + "rewards/rejected": -0.40570390224456787, + "step": 101 + }, + { + "epoch": 0.13, + "learning_rate": 4.9999897295764844e-05, + "logits/chosen": -2.048476457595825, + "logits/rejected": -2.0353472232818604, + "logps/chosen": -180.3016357421875, + "logps/rejected": -180.51600646972656, + "loss": 0.6366, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.35791218280792236, + "rewards/margins": 0.15290379524230957, + "rewards/rejected": -0.5108159780502319, + "step": 102 + }, + { + "epoch": 0.13, + "learning_rate": 4.9999768915668665e-05, + "logits/chosen": -1.9134316444396973, + "logits/rejected": -1.8900353908538818, + "logps/chosen": -152.71189880371094, + "logps/rejected": -153.96690368652344, + "loss": 0.6941, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.18337132036685944, + "rewards/margins": 0.03930587321519852, + "rewards/rejected": -0.22267718613147736, + "step": 103 + }, + { + "epoch": 0.14, + "learning_rate": 4.999958918390321e-05, + "logits/chosen": -1.8933653831481934, + "logits/rejected": -1.8493753671646118, + "logps/chosen": -188.6655731201172, + "logps/rejected": -183.1341552734375, + "loss": 0.7168, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.38441047072410583, + "rewards/margins": -0.010426240041851997, + "rewards/rejected": -0.3739842474460602, + "step": 104 + }, + { + "epoch": 0.14, + "learning_rate": 4.999935810083766e-05, + "logits/chosen": -1.7264684438705444, + "logits/rejected": -1.694319486618042, + "logps/chosen": -156.22084045410156, + "logps/rejected": -152.80894470214844, + "loss": 0.6763, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.28453487157821655, + "rewards/margins": 0.052657999098300934, + "rewards/rejected": -0.3371928334236145, + "step": 105 + }, + { + "epoch": 0.14, + "learning_rate": 4.999907566694667e-05, + "logits/chosen": -1.8885689973831177, + "logits/rejected": -1.9364815950393677, + "logps/chosen": -167.39117431640625, + "logps/rejected": -191.052001953125, + "loss": 0.6963, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.21066321432590485, + "rewards/margins": 0.03920959681272507, + "rewards/rejected": -0.2498728483915329, + "step": 106 + }, + { + "epoch": 0.14, + "learning_rate": 4.9998741882810384e-05, + "logits/chosen": -1.7698872089385986, + "logits/rejected": -1.7441281080245972, + "logps/chosen": -178.68572998046875, + "logps/rejected": -174.1964569091797, + "loss": 0.7459, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.24953651428222656, + "rewards/margins": -0.08057431131601334, + "rewards/rejected": -0.16896219551563263, + "step": 107 + }, + { + "epoch": 0.14, + "learning_rate": 4.999835674911443e-05, + "logits/chosen": -1.812888741493225, + "logits/rejected": -1.7789666652679443, + "logps/chosen": -228.23631286621094, + "logps/rejected": -203.85357666015625, + "loss": 0.6838, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.10872795432806015, + "rewards/margins": 0.05488254129886627, + "rewards/rejected": -0.16361048817634583, + "step": 108 + }, + { + "epoch": 0.14, + "learning_rate": 4.999792026664991e-05, + "logits/chosen": -1.6739952564239502, + "logits/rejected": -1.6701843738555908, + "logps/chosen": -203.80810546875, + "logps/rejected": -211.82334899902344, + "loss": 0.6853, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.3305152356624603, + "rewards/margins": 0.05953298509120941, + "rewards/rejected": -0.39004823565483093, + "step": 109 + }, + { + "epoch": 0.14, + "learning_rate": 4.9997432436313384e-05, + "logits/chosen": -1.6255407333374023, + "logits/rejected": -1.586310625076294, + "logps/chosen": -165.31040954589844, + "logps/rejected": -186.821044921875, + "loss": 0.646, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.3459918200969696, + "rewards/margins": 0.1395949125289917, + "rewards/rejected": -0.4855867028236389, + "step": 110 + }, + { + "epoch": 0.15, + "learning_rate": 4.99968932591069e-05, + "logits/chosen": -1.8984012603759766, + "logits/rejected": -1.8569310903549194, + "logps/chosen": -182.6356201171875, + "logps/rejected": -176.14752197265625, + "loss": 0.7585, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.3858329653739929, + "rewards/margins": -0.05486001819372177, + "rewards/rejected": -0.33097296953201294, + "step": 111 + }, + { + "epoch": 0.15, + "learning_rate": 4.999630273613799e-05, + "logits/chosen": -1.8298226594924927, + "logits/rejected": -1.812253475189209, + "logps/chosen": -163.0826416015625, + "logps/rejected": -200.62098693847656, + "loss": 0.7425, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.2680785357952118, + "rewards/margins": -0.04629645124077797, + "rewards/rejected": -0.22178205847740173, + "step": 112 + }, + { + "epoch": 0.15, + "learning_rate": 4.999566086861961e-05, + "logits/chosen": -1.6931625604629517, + "logits/rejected": -1.7089687585830688, + "logps/chosen": -147.44491577148438, + "logps/rejected": -150.4454345703125, + "loss": 0.7146, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.224314883351326, + "rewards/margins": 0.05584639310836792, + "rewards/rejected": -0.2801613211631775, + "step": 113 + }, + { + "epoch": 0.15, + "learning_rate": 4.999496765787024e-05, + "logits/chosen": -1.7311415672302246, + "logits/rejected": -1.612362027168274, + "logps/chosen": -195.65594482421875, + "logps/rejected": -194.56130981445312, + "loss": 0.7017, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.28545114398002625, + "rewards/margins": 0.020438771694898605, + "rewards/rejected": -0.30588990449905396, + "step": 114 + }, + { + "epoch": 0.15, + "learning_rate": 4.9994223105313774e-05, + "logits/chosen": -1.9310710430145264, + "logits/rejected": -1.9658077955245972, + "logps/chosen": -179.00779724121094, + "logps/rejected": -181.42135620117188, + "loss": 0.6605, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.2686367928981781, + "rewards/margins": 0.10176892578601837, + "rewards/rejected": -0.37040573358535767, + "step": 115 + }, + { + "epoch": 0.15, + "learning_rate": 4.9993427212479606e-05, + "logits/chosen": -1.7969621419906616, + "logits/rejected": -1.8055509328842163, + "logps/chosen": -176.26036071777344, + "logps/rejected": -171.2470703125, + "loss": 0.6656, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.33693569898605347, + "rewards/margins": 0.08709227293729782, + "rewards/rejected": -0.4240279793739319, + "step": 116 + }, + { + "epoch": 0.15, + "learning_rate": 4.999257998100254e-05, + "logits/chosen": -1.5152311325073242, + "logits/rejected": -1.553884506225586, + "logps/chosen": -179.60372924804688, + "logps/rejected": -164.34481811523438, + "loss": 0.7563, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.3611801564693451, + "rewards/margins": -0.08427368104457855, + "rewards/rejected": -0.27690649032592773, + "step": 117 + }, + { + "epoch": 0.15, + "learning_rate": 4.999168141262289e-05, + "logits/chosen": -1.7877562046051025, + "logits/rejected": -1.7855968475341797, + "logps/chosen": -165.04220581054688, + "logps/rejected": -179.1771240234375, + "loss": 0.7999, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.3364974856376648, + "rewards/margins": -0.1316596120595932, + "rewards/rejected": -0.20483790338039398, + "step": 118 + }, + { + "epoch": 0.16, + "learning_rate": 4.9990731509186376e-05, + "logits/chosen": -1.4499574899673462, + "logits/rejected": -1.4440600872039795, + "logps/chosen": -183.58778381347656, + "logps/rejected": -198.231689453125, + "loss": 0.7216, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.3034099340438843, + "rewards/margins": -0.00025239214301109314, + "rewards/rejected": -0.3031575381755829, + "step": 119 + }, + { + "epoch": 0.16, + "learning_rate": 4.998973027264419e-05, + "logits/chosen": -1.6583937406539917, + "logits/rejected": -1.689673900604248, + "logps/chosen": -187.60508728027344, + "logps/rejected": -212.7592315673828, + "loss": 0.6404, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.3053140342235565, + "rewards/margins": 0.17152619361877441, + "rewards/rejected": -0.47684019804000854, + "step": 120 + }, + { + "epoch": 0.16, + "learning_rate": 4.998867770505295e-05, + "logits/chosen": -1.6554609537124634, + "logits/rejected": -1.6350326538085938, + "logps/chosen": -181.46209716796875, + "logps/rejected": -173.83157348632812, + "loss": 0.726, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.24238371849060059, + "rewards/margins": 0.002255776897072792, + "rewards/rejected": -0.24463950097560883, + "step": 121 + }, + { + "epoch": 0.16, + "learning_rate": 4.9987573808574726e-05, + "logits/chosen": -1.8908902406692505, + "logits/rejected": -1.858853816986084, + "logps/chosen": -174.7786865234375, + "logps/rejected": -177.30545043945312, + "loss": 0.7269, + "rewards/accuracies": 0.625, + "rewards/chosen": 0.0712527185678482, + "rewards/margins": -0.02764507755637169, + "rewards/rejected": 0.09889779984951019, + "step": 122 + }, + { + "epoch": 0.16, + "learning_rate": 4.9986418585477016e-05, + "logits/chosen": -1.7081693410873413, + "logits/rejected": -1.7438371181488037, + "logps/chosen": -191.28123474121094, + "logps/rejected": -197.5807647705078, + "loss": 0.6859, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.2962351143360138, + "rewards/margins": 0.06765662133693695, + "rewards/rejected": -0.36389175057411194, + "step": 123 + }, + { + "epoch": 0.16, + "learning_rate": 4.998521203813274e-05, + "logits/chosen": -1.805833101272583, + "logits/rejected": -1.7511969804763794, + "logps/chosen": -166.08702087402344, + "logps/rejected": -159.9141082763672, + "loss": 0.6387, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.16753454506397247, + "rewards/margins": 0.16043387353420258, + "rewards/rejected": -0.32796838879585266, + "step": 124 + }, + { + "epoch": 0.16, + "learning_rate": 4.9983954169020256e-05, + "logits/chosen": -1.5241700410842896, + "logits/rejected": -1.6004612445831299, + "logps/chosen": -189.8046875, + "logps/rejected": -198.48001098632812, + "loss": 0.7803, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.35226038098335266, + "rewards/margins": -0.09872373938560486, + "rewards/rejected": -0.2535366714000702, + "step": 125 + }, + { + "epoch": 0.16, + "learning_rate": 4.9982644980723334e-05, + "logits/chosen": -1.3276153802871704, + "logits/rejected": -1.372768759727478, + "logps/chosen": -179.91940307617188, + "logps/rejected": -180.60751342773438, + "loss": 0.7245, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.4333783686161041, + "rewards/margins": -0.014091454446315765, + "rewards/rejected": -0.41928690671920776, + "step": 126 + }, + { + "epoch": 0.17, + "learning_rate": 4.998128447593117e-05, + "logits/chosen": -1.5195338726043701, + "logits/rejected": -1.4305707216262817, + "logps/chosen": -179.16712951660156, + "logps/rejected": -162.93356323242188, + "loss": 0.7394, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.3149658441543579, + "rewards/margins": -0.04299226403236389, + "rewards/rejected": -0.2719736099243164, + "step": 127 + }, + { + "epoch": 0.17, + "learning_rate": 4.997987265743834e-05, + "logits/chosen": -1.8512688875198364, + "logits/rejected": -1.791621446609497, + "logps/chosen": -171.71990966796875, + "logps/rejected": -177.26954650878906, + "loss": 0.7217, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.2529350221157074, + "rewards/margins": -0.008312180638313293, + "rewards/rejected": -0.2446228414773941, + "step": 128 + }, + { + "epoch": 0.17, + "learning_rate": 4.997840952814484e-05, + "logits/chosen": -1.8096928596496582, + "logits/rejected": -1.7790553569793701, + "logps/chosen": -174.4140167236328, + "logps/rejected": -176.01708984375, + "loss": 0.7997, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.24751965701580048, + "rewards/margins": -0.1494641900062561, + "rewards/rejected": -0.09805545210838318, + "step": 129 + }, + { + "epoch": 0.17, + "learning_rate": 4.9976895091056075e-05, + "logits/chosen": -1.7724186182022095, + "logits/rejected": -1.777117371559143, + "logps/chosen": -171.97897338867188, + "logps/rejected": -197.2012939453125, + "loss": 0.6544, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.3243841528892517, + "rewards/margins": 0.10846509784460068, + "rewards/rejected": -0.432849258184433, + "step": 130 + }, + { + "epoch": 0.17, + "learning_rate": 4.9975329349282826e-05, + "logits/chosen": -1.6474465131759644, + "logits/rejected": -1.632018804550171, + "logps/chosen": -176.29461669921875, + "logps/rejected": -191.7507781982422, + "loss": 0.7379, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.5088250637054443, + "rewards/margins": -0.030252262949943542, + "rewards/rejected": -0.4785728454589844, + "step": 131 + }, + { + "epoch": 0.17, + "learning_rate": 4.9973712306041256e-05, + "logits/chosen": -2.033618927001953, + "logits/rejected": -2.0342512130737305, + "logps/chosen": -198.94796752929688, + "logps/rejected": -192.28482055664062, + "loss": 0.826, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.34583836793899536, + "rewards/margins": -0.18120835721492767, + "rewards/rejected": -0.1646299660205841, + "step": 132 + }, + { + "epoch": 0.17, + "learning_rate": 4.997204396465292e-05, + "logits/chosen": -1.5918235778808594, + "logits/rejected": -1.5706799030303955, + "logps/chosen": -157.1723175048828, + "logps/rejected": -182.7075958251953, + "loss": 0.8134, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.5003759860992432, + "rewards/margins": -0.07385056465864182, + "rewards/rejected": -0.42652541399002075, + "step": 133 + }, + { + "epoch": 0.18, + "learning_rate": 4.997032432854472e-05, + "logits/chosen": -1.6958372592926025, + "logits/rejected": -1.7072336673736572, + "logps/chosen": -182.0128173828125, + "logps/rejected": -202.56802368164062, + "loss": 0.52, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.08301500976085663, + "rewards/margins": 0.4688724875450134, + "rewards/rejected": -0.5518875122070312, + "step": 134 + }, + { + "epoch": 0.18, + "learning_rate": 4.996855340124894e-05, + "logits/chosen": -1.7832672595977783, + "logits/rejected": -1.767440915107727, + "logps/chosen": -193.7747344970703, + "logps/rejected": -197.20289611816406, + "loss": 0.7392, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.39458268880844116, + "rewards/margins": -0.038821250200271606, + "rewards/rejected": -0.35576146841049194, + "step": 135 + }, + { + "epoch": 0.18, + "learning_rate": 4.996673118640323e-05, + "logits/chosen": -1.7633535861968994, + "logits/rejected": -1.7420529127120972, + "logps/chosen": -154.8060760498047, + "logps/rejected": -170.86996459960938, + "loss": 0.6108, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.15521948039531708, + "rewards/margins": 0.2389230877161026, + "rewards/rejected": -0.3941425681114197, + "step": 136 + }, + { + "epoch": 0.18, + "learning_rate": 4.996485768775055e-05, + "logits/chosen": -1.7529706954956055, + "logits/rejected": -1.7141683101654053, + "logps/chosen": -233.09747314453125, + "logps/rejected": -237.85906982421875, + "loss": 0.7094, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.4181414544582367, + "rewards/margins": 0.051600366830825806, + "rewards/rejected": -0.4697418212890625, + "step": 137 + }, + { + "epoch": 0.18, + "learning_rate": 4.996293290913926e-05, + "logits/chosen": -1.6701140403747559, + "logits/rejected": -1.684870719909668, + "logps/chosen": -192.6127166748047, + "logps/rejected": -188.32798767089844, + "loss": 0.5773, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.0851680114865303, + "rewards/margins": 0.31171905994415283, + "rewards/rejected": -0.39688706398010254, + "step": 138 + }, + { + "epoch": 0.18, + "learning_rate": 4.9960956854522986e-05, + "logits/chosen": -1.7777843475341797, + "logits/rejected": -1.73331618309021, + "logps/chosen": -199.59889221191406, + "logps/rejected": -174.27069091796875, + "loss": 0.785, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.5378831028938293, + "rewards/margins": -0.13611721992492676, + "rewards/rejected": -0.4017658531665802, + "step": 139 + }, + { + "epoch": 0.18, + "learning_rate": 4.995892952796074e-05, + "logits/chosen": -1.7307249307632446, + "logits/rejected": -1.7584043741226196, + "logps/chosen": -184.14364624023438, + "logps/rejected": -196.90878295898438, + "loss": 0.6489, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.4751961827278137, + "rewards/margins": 0.1301158368587494, + "rewards/rejected": -0.6053119897842407, + "step": 140 + }, + { + "epoch": 0.18, + "learning_rate": 4.995685093361682e-05, + "logits/chosen": -1.770018219947815, + "logits/rejected": -1.832364797592163, + "logps/chosen": -183.0137939453125, + "logps/rejected": -204.83804321289062, + "loss": 0.7497, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.4263853132724762, + "rewards/margins": -0.025808706879615784, + "rewards/rejected": -0.4005766212940216, + "step": 141 + }, + { + "epoch": 0.19, + "learning_rate": 4.9954721075760824e-05, + "logits/chosen": -1.7604196071624756, + "logits/rejected": -1.7073791027069092, + "logps/chosen": -191.46923828125, + "logps/rejected": -201.73545837402344, + "loss": 0.8181, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.5646533370018005, + "rewards/margins": -0.1740642488002777, + "rewards/rejected": -0.3905891180038452, + "step": 142 + }, + { + "epoch": 0.19, + "learning_rate": 4.995253995876767e-05, + "logits/chosen": -1.5798883438110352, + "logits/rejected": -1.5167430639266968, + "logps/chosen": -235.8765106201172, + "logps/rejected": -222.53204345703125, + "loss": 0.7115, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.5441933870315552, + "rewards/margins": 0.09400075674057007, + "rewards/rejected": -0.6381941437721252, + "step": 143 + }, + { + "epoch": 0.19, + "learning_rate": 4.995030758711756e-05, + "logits/chosen": -1.932177186012268, + "logits/rejected": -1.8748353719711304, + "logps/chosen": -182.81915283203125, + "logps/rejected": -170.09844970703125, + "loss": 0.6448, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.48413196206092834, + "rewards/margins": 0.19957152009010315, + "rewards/rejected": -0.6837034821510315, + "step": 144 + }, + { + "epoch": 0.19, + "learning_rate": 4.994802396539598e-05, + "logits/chosen": -1.8008129596710205, + "logits/rejected": -1.7928048372268677, + "logps/chosen": -202.68096923828125, + "logps/rejected": -199.8201904296875, + "loss": 0.8172, + "rewards/accuracies": 0.5625, + "rewards/chosen": -1.0157395601272583, + "rewards/margins": -0.10564298927783966, + "rewards/rejected": -0.9100965857505798, + "step": 145 + }, + { + "epoch": 0.19, + "learning_rate": 4.994568909829368e-05, + "logits/chosen": -1.7543656826019287, + "logits/rejected": -1.7212865352630615, + "logps/chosen": -201.3524932861328, + "logps/rejected": -218.5485382080078, + "loss": 0.9316, + "rewards/accuracies": 0.1875, + "rewards/chosen": -0.7368103265762329, + "rewards/margins": -0.3777569532394409, + "rewards/rejected": -0.359053373336792, + "step": 146 + }, + { + "epoch": 0.19, + "learning_rate": 4.9943302990606684e-05, + "logits/chosen": -1.7704360485076904, + "logits/rejected": -1.6632460355758667, + "logps/chosen": -187.3475341796875, + "logps/rejected": -180.01144409179688, + "loss": 0.751, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.4657554626464844, + "rewards/margins": 0.006175771355628967, + "rewards/rejected": -0.47193124890327454, + "step": 147 + }, + { + "epoch": 0.19, + "learning_rate": 4.994086564723626e-05, + "logits/chosen": -1.9261763095855713, + "logits/rejected": -1.9572409391403198, + "logps/chosen": -171.56101989746094, + "logps/rejected": -182.58717346191406, + "loss": 0.7349, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.7470525503158569, + "rewards/margins": -0.043389588594436646, + "rewards/rejected": -0.7036629915237427, + "step": 148 + }, + { + "epoch": 0.19, + "learning_rate": 4.9938377073188905e-05, + "logits/chosen": -1.9480628967285156, + "logits/rejected": -2.002164363861084, + "logps/chosen": -197.41912841796875, + "logps/rejected": -184.93325805664062, + "loss": 0.813, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.6927478909492493, + "rewards/margins": -0.12049313634634018, + "rewards/rejected": -0.5722547769546509, + "step": 149 + }, + { + "epoch": 0.2, + "learning_rate": 4.993583727357638e-05, + "logits/chosen": -1.6262449026107788, + "logits/rejected": -1.640842080116272, + "logps/chosen": -205.38461303710938, + "logps/rejected": -213.60650634765625, + "loss": 0.7821, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.7866306304931641, + "rewards/margins": -0.11763662099838257, + "rewards/rejected": -0.6689940094947815, + "step": 150 + }, + { + "epoch": 0.2, + "learning_rate": 4.993324625361565e-05, + "logits/chosen": -1.8480533361434937, + "logits/rejected": -1.8557144403457642, + "logps/chosen": -158.26290893554688, + "logps/rejected": -169.06105041503906, + "loss": 0.7438, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6200124025344849, + "rewards/margins": -0.012443792074918747, + "rewards/rejected": -0.607568621635437, + "step": 151 + }, + { + "epoch": 0.2, + "learning_rate": 4.993060401862888e-05, + "logits/chosen": -1.8685041666030884, + "logits/rejected": -1.8648606538772583, + "logps/chosen": -176.7852020263672, + "logps/rejected": -183.40328979492188, + "loss": 0.6935, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.7892077565193176, + "rewards/margins": 0.08869240432977676, + "rewards/rejected": -0.877900242805481, + "step": 152 + }, + { + "epoch": 0.2, + "learning_rate": 4.9927910574043465e-05, + "logits/chosen": -1.9234154224395752, + "logits/rejected": -1.904573917388916, + "logps/chosen": -159.97625732421875, + "logps/rejected": -152.645263671875, + "loss": 0.7778, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6230319142341614, + "rewards/margins": -0.06979034841060638, + "rewards/rejected": -0.5532415509223938, + "step": 153 + }, + { + "epoch": 0.2, + "learning_rate": 4.992516592539196e-05, + "logits/chosen": -1.6896902322769165, + "logits/rejected": -1.7036737203598022, + "logps/chosen": -148.6313018798828, + "logps/rejected": -164.7644500732422, + "loss": 0.5546, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.29331690073013306, + "rewards/margins": 0.44352981448173523, + "rewards/rejected": -0.7368468046188354, + "step": 154 + }, + { + "epoch": 0.2, + "learning_rate": 4.9922370078312105e-05, + "logits/chosen": -2.013890266418457, + "logits/rejected": -1.9424934387207031, + "logps/chosen": -215.90118408203125, + "logps/rejected": -209.59071350097656, + "loss": 0.4938, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.2655085027217865, + "rewards/margins": 0.5123203992843628, + "rewards/rejected": -0.7778289318084717, + "step": 155 + }, + { + "epoch": 0.2, + "learning_rate": 4.991952303854682e-05, + "logits/chosen": -1.8328962326049805, + "logits/rejected": -1.8138638734817505, + "logps/chosen": -170.13475036621094, + "logps/rejected": -176.11810302734375, + "loss": 0.6684, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.3248033821582794, + "rewards/margins": 0.12224595248699188, + "rewards/rejected": -0.4470493197441101, + "step": 156 + }, + { + "epoch": 0.21, + "learning_rate": 4.9916624811944175e-05, + "logits/chosen": -1.9051162004470825, + "logits/rejected": -1.9407715797424316, + "logps/chosen": -177.2139434814453, + "logps/rejected": -185.92947387695312, + "loss": 0.6297, + "rewards/accuracies": 0.8125, + "rewards/chosen": -0.6396099328994751, + "rewards/margins": 0.1931779384613037, + "rewards/rejected": -0.832787811756134, + "step": 157 + }, + { + "epoch": 0.21, + "learning_rate": 4.991367540445735e-05, + "logits/chosen": -1.7430989742279053, + "logits/rejected": -1.7986749410629272, + "logps/chosen": -199.38021850585938, + "logps/rejected": -195.27647399902344, + "loss": 0.6881, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.8835601210594177, + "rewards/margins": 0.10807879269123077, + "rewards/rejected": -0.9916388392448425, + "step": 158 + }, + { + "epoch": 0.21, + "learning_rate": 4.991067482214471e-05, + "logits/chosen": -1.868577241897583, + "logits/rejected": -1.799201488494873, + "logps/chosen": -177.93130493164062, + "logps/rejected": -164.038330078125, + "loss": 0.7179, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.7006940245628357, + "rewards/margins": 0.020756253972649574, + "rewards/rejected": -0.7214502096176147, + "step": 159 + }, + { + "epoch": 0.21, + "learning_rate": 4.9907623071169686e-05, + "logits/chosen": -1.8050721883773804, + "logits/rejected": -1.7359880208969116, + "logps/chosen": -197.66583251953125, + "logps/rejected": -172.34146118164062, + "loss": 0.6379, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.7713190317153931, + "rewards/margins": 0.19513630867004395, + "rewards/rejected": -0.966455340385437, + "step": 160 + }, + { + "epoch": 0.21, + "learning_rate": 4.990452015780085e-05, + "logits/chosen": -1.74982750415802, + "logits/rejected": -1.719763159751892, + "logps/chosen": -204.21517944335938, + "logps/rejected": -196.62576293945312, + "loss": 0.7434, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.35949403047561646, + "rewards/margins": 0.03506145626306534, + "rewards/rejected": -0.3945554494857788, + "step": 161 + }, + { + "epoch": 0.21, + "learning_rate": 4.9901366088411846e-05, + "logits/chosen": -1.6477172374725342, + "logits/rejected": -1.622018814086914, + "logps/chosen": -207.30174255371094, + "logps/rejected": -229.1028289794922, + "loss": 0.9207, + "rewards/accuracies": 0.4375, + "rewards/chosen": -1.1269769668579102, + "rewards/margins": -0.2236841320991516, + "rewards/rejected": -0.9032928943634033, + "step": 162 + }, + { + "epoch": 0.21, + "learning_rate": 4.98981608694814e-05, + "logits/chosen": -1.8223265409469604, + "logits/rejected": -1.8162957429885864, + "logps/chosen": -171.70675659179688, + "logps/rejected": -171.83108520507812, + "loss": 0.9174, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.8722136616706848, + "rewards/margins": -0.21637174487113953, + "rewards/rejected": -0.6558419466018677, + "step": 163 + }, + { + "epoch": 0.21, + "learning_rate": 4.9894904507593316e-05, + "logits/chosen": -1.9424258470535278, + "logits/rejected": -1.828155279159546, + "logps/chosen": -196.1715545654297, + "logps/rejected": -192.08624267578125, + "loss": 0.7506, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.5839821696281433, + "rewards/margins": 0.0038854647427797318, + "rewards/rejected": -0.5878676772117615, + "step": 164 + }, + { + "epoch": 0.22, + "learning_rate": 4.989159700943643e-05, + "logits/chosen": -1.623518705368042, + "logits/rejected": -1.679386854171753, + "logps/chosen": -175.6849365234375, + "logps/rejected": -179.52059936523438, + "loss": 0.9491, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.7352603673934937, + "rewards/margins": -0.2515270411968231, + "rewards/rejected": -0.4837333559989929, + "step": 165 + }, + { + "epoch": 0.22, + "learning_rate": 4.988823838180464e-05, + "logits/chosen": -1.997894048690796, + "logits/rejected": -1.9666211605072021, + "logps/chosen": -167.2881622314453, + "logps/rejected": -183.72474670410156, + "loss": 0.6663, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.5555349588394165, + "rewards/margins": 0.17635540664196014, + "rewards/rejected": -0.7318904399871826, + "step": 166 + }, + { + "epoch": 0.22, + "learning_rate": 4.988482863159684e-05, + "logits/chosen": -1.8912848234176636, + "logits/rejected": -1.968542218208313, + "logps/chosen": -174.55911254882812, + "logps/rejected": -169.36610412597656, + "loss": 0.7905, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.227503702044487, + "rewards/margins": -0.08394889533519745, + "rewards/rejected": -0.14355483651161194, + "step": 167 + }, + { + "epoch": 0.22, + "learning_rate": 4.988136776581696e-05, + "logits/chosen": -2.151402711868286, + "logits/rejected": -2.186088800430298, + "logps/chosen": -157.3328094482422, + "logps/rejected": -150.9196319580078, + "loss": 0.6989, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.29456794261932373, + "rewards/margins": 0.04511295258998871, + "rewards/rejected": -0.33968091011047363, + "step": 168 + }, + { + "epoch": 0.22, + "learning_rate": 4.9877855791573915e-05, + "logits/chosen": -1.8164244890213013, + "logits/rejected": -1.8539032936096191, + "logps/chosen": -188.2481689453125, + "logps/rejected": -171.02090454101562, + "loss": 0.9341, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.5969239473342896, + "rewards/margins": -0.33455953001976013, + "rewards/rejected": -0.2623644471168518, + "step": 169 + }, + { + "epoch": 0.22, + "learning_rate": 4.9874292716081595e-05, + "logits/chosen": -1.7419726848602295, + "logits/rejected": -1.7444337606430054, + "logps/chosen": -169.46658325195312, + "logps/rejected": -173.33348083496094, + "loss": 0.6817, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.4356076121330261, + "rewards/margins": 0.14688673615455627, + "rewards/rejected": -0.5824943780899048, + "step": 170 + }, + { + "epoch": 0.22, + "learning_rate": 4.9870678546658865e-05, + "logits/chosen": -1.6884466409683228, + "logits/rejected": -1.7295485734939575, + "logps/chosen": -160.8187713623047, + "logps/rejected": -176.16746520996094, + "loss": 0.9101, + "rewards/accuracies": 0.125, + "rewards/chosen": -0.6096308827400208, + "rewards/margins": -0.28187263011932373, + "rewards/rejected": -0.3277583122253418, + "step": 171 + }, + { + "epoch": 0.23, + "learning_rate": 4.9867013290729535e-05, + "logits/chosen": -1.932777762413025, + "logits/rejected": -1.8993282318115234, + "logps/chosen": -198.2738037109375, + "logps/rejected": -197.27252197265625, + "loss": 0.7183, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.19178181886672974, + "rewards/margins": 0.006053738296031952, + "rewards/rejected": -0.1978355497121811, + "step": 172 + }, + { + "epoch": 0.23, + "learning_rate": 4.986329695582237e-05, + "logits/chosen": -2.03489351272583, + "logits/rejected": -2.076827049255371, + "logps/chosen": -179.49679565429688, + "logps/rejected": -177.2965545654297, + "loss": 0.7798, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6112560629844666, + "rewards/margins": -0.09085651487112045, + "rewards/rejected": -0.5203995704650879, + "step": 173 + }, + { + "epoch": 0.23, + "learning_rate": 4.985952954957103e-05, + "logits/chosen": -1.8877630233764648, + "logits/rejected": -1.8290894031524658, + "logps/chosen": -233.42221069335938, + "logps/rejected": -228.90179443359375, + "loss": 0.7997, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.4796496629714966, + "rewards/margins": 0.0218522846698761, + "rewards/rejected": -0.5015019774436951, + "step": 174 + }, + { + "epoch": 0.23, + "learning_rate": 4.985571107971408e-05, + "logits/chosen": -1.8358758687973022, + "logits/rejected": -1.8217942714691162, + "logps/chosen": -173.35556030273438, + "logps/rejected": -176.17031860351562, + "loss": 0.6678, + "rewards/accuracies": 0.625, + "rewards/chosen": 0.03331407904624939, + "rewards/margins": 0.18146347999572754, + "rewards/rejected": -0.14814940094947815, + "step": 175 + }, + { + "epoch": 0.23, + "learning_rate": 4.9851841554095e-05, + "logits/chosen": -1.9195314645767212, + "logits/rejected": -1.9220951795578003, + "logps/chosen": -233.7957305908203, + "logps/rejected": -216.81581115722656, + "loss": 0.5707, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.0034320950508117676, + "rewards/margins": 0.3490698039531708, + "rewards/rejected": -0.35250189900398254, + "step": 176 + }, + { + "epoch": 0.23, + "learning_rate": 4.9847920980662134e-05, + "logits/chosen": -1.598222255706787, + "logits/rejected": -1.591965913772583, + "logps/chosen": -226.130615234375, + "logps/rejected": -236.63760375976562, + "loss": 0.6588, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.06091824173927307, + "rewards/margins": 0.13529105484485626, + "rewards/rejected": -0.07437281310558319, + "step": 177 + }, + { + "epoch": 0.23, + "learning_rate": 4.984394936746865e-05, + "logits/chosen": -1.7949796915054321, + "logits/rejected": -1.805631399154663, + "logps/chosen": -232.09498596191406, + "logps/rejected": -235.5016632080078, + "loss": 0.709, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.5284535884857178, + "rewards/margins": 0.1780376434326172, + "rewards/rejected": -0.706491231918335, + "step": 178 + }, + { + "epoch": 0.23, + "learning_rate": 4.98399267226726e-05, + "logits/chosen": -2.0285134315490723, + "logits/rejected": -2.060884714126587, + "logps/chosen": -183.62449645996094, + "logps/rejected": -172.28872680664062, + "loss": 0.7795, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.42201852798461914, + "rewards/margins": -0.029273340478539467, + "rewards/rejected": -0.3927451968193054, + "step": 179 + }, + { + "epoch": 0.24, + "learning_rate": 4.9835853054536846e-05, + "logits/chosen": -1.7763793468475342, + "logits/rejected": -1.8531831502914429, + "logps/chosen": -182.07516479492188, + "logps/rejected": -205.60504150390625, + "loss": 0.9431, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.2773244380950928, + "rewards/margins": -0.36518311500549316, + "rewards/rejected": 0.08785867691040039, + "step": 180 + }, + { + "epoch": 0.24, + "learning_rate": 4.9831728371429046e-05, + "logits/chosen": -1.8723288774490356, + "logits/rejected": -1.9752717018127441, + "logps/chosen": -200.94346618652344, + "logps/rejected": -210.53933715820312, + "loss": 0.7466, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.15453863143920898, + "rewards/margins": -0.037006717175245285, + "rewards/rejected": -0.117531917989254, + "step": 181 + }, + { + "epoch": 0.24, + "learning_rate": 4.982755268182164e-05, + "logits/chosen": -1.7186784744262695, + "logits/rejected": -1.752870798110962, + "logps/chosen": -168.73394775390625, + "logps/rejected": -177.32054138183594, + "loss": 0.7175, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.1720157265663147, + "rewards/margins": -0.019572071731090546, + "rewards/rejected": -0.15244367718696594, + "step": 182 + }, + { + "epoch": 0.24, + "learning_rate": 4.982332599429187e-05, + "logits/chosen": -1.9437085390090942, + "logits/rejected": -1.9051276445388794, + "logps/chosen": -168.2744903564453, + "logps/rejected": -158.18289184570312, + "loss": 0.8465, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.40080204606056213, + "rewards/margins": -0.24280087649822235, + "rewards/rejected": -0.15800118446350098, + "step": 183 + }, + { + "epoch": 0.24, + "learning_rate": 4.981904831752171e-05, + "logits/chosen": -1.9985157251358032, + "logits/rejected": -1.9973095655441284, + "logps/chosen": -182.0140380859375, + "logps/rejected": -169.61883544921875, + "loss": 0.6244, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.13748487830162048, + "rewards/margins": 0.31921717524528503, + "rewards/rejected": -0.4567020535469055, + "step": 184 + }, + { + "epoch": 0.24, + "learning_rate": 4.981471966029787e-05, + "logits/chosen": -1.9101628065109253, + "logits/rejected": -1.9257937669754028, + "logps/chosen": -192.24288940429688, + "logps/rejected": -188.32144165039062, + "loss": 0.6148, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.2866635024547577, + "rewards/margins": 0.2923651337623596, + "rewards/rejected": -0.00570157915353775, + "step": 185 + }, + { + "epoch": 0.24, + "learning_rate": 4.981034003151178e-05, + "logits/chosen": -1.8881627321243286, + "logits/rejected": -1.9550986289978027, + "logps/chosen": -195.71372985839844, + "logps/rejected": -210.1089324951172, + "loss": 0.8689, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.2077367603778839, + "rewards/margins": -0.24497191607952118, + "rewards/rejected": 0.03723515570163727, + "step": 186 + }, + { + "epoch": 0.24, + "learning_rate": 4.980590944015958e-05, + "logits/chosen": -1.7422230243682861, + "logits/rejected": -1.7998031377792358, + "logps/chosen": -216.39398193359375, + "logps/rejected": -222.33880615234375, + "loss": 0.7159, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.11723195761442184, + "rewards/margins": 0.26700979471206665, + "rewards/rejected": -0.14977779984474182, + "step": 187 + }, + { + "epoch": 0.25, + "learning_rate": 4.98014278953421e-05, + "logits/chosen": -1.8071612119674683, + "logits/rejected": -1.8180828094482422, + "logps/chosen": -176.4374542236328, + "logps/rejected": -189.68438720703125, + "loss": 0.8578, + "rewards/accuracies": 0.3125, + "rewards/chosen": 0.02567705512046814, + "rewards/margins": -0.21896688640117645, + "rewards/rejected": 0.2446439117193222, + "step": 188 + }, + { + "epoch": 0.25, + "learning_rate": 4.979689540626479e-05, + "logits/chosen": -1.7617331743240356, + "logits/rejected": -1.8094758987426758, + "logps/chosen": -158.13441467285156, + "logps/rejected": -177.64797973632812, + "loss": 0.5662, + "rewards/accuracies": 0.75, + "rewards/chosen": 0.3027116656303406, + "rewards/margins": 0.35941970348358154, + "rewards/rejected": -0.05670810118317604, + "step": 189 + }, + { + "epoch": 0.25, + "learning_rate": 4.9792311982237774e-05, + "logits/chosen": -1.528577208518982, + "logits/rejected": -1.5104334354400635, + "logps/chosen": -164.3386993408203, + "logps/rejected": -167.84840393066406, + "loss": 0.7506, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.04946248233318329, + "rewards/margins": -0.02245260775089264, + "rewards/rejected": -0.027009889483451843, + "step": 190 + }, + { + "epoch": 0.25, + "learning_rate": 4.9787677632675825e-05, + "logits/chosen": -1.86935555934906, + "logits/rejected": -1.8780128955841064, + "logps/chosen": -167.26119995117188, + "logps/rejected": -187.39584350585938, + "loss": 0.6634, + "rewards/accuracies": 0.6875, + "rewards/chosen": 0.09376392513513565, + "rewards/margins": 0.16032886505126953, + "rewards/rejected": -0.06656493991613388, + "step": 191 + }, + { + "epoch": 0.25, + "learning_rate": 4.978299236709826e-05, + "logits/chosen": -1.79931640625, + "logits/rejected": -1.7840967178344727, + "logps/chosen": -173.83987426757812, + "logps/rejected": -174.4556427001953, + "loss": 0.8509, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.17128746211528778, + "rewards/margins": -0.21691852807998657, + "rewards/rejected": 0.04563106596469879, + "step": 192 + }, + { + "epoch": 0.25, + "learning_rate": 4.977825619512904e-05, + "logits/chosen": -1.9714162349700928, + "logits/rejected": -1.9255212545394897, + "logps/chosen": -216.35391235351562, + "logps/rejected": -213.25570678710938, + "loss": 0.787, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.07267741113901138, + "rewards/margins": -0.07426212728023529, + "rewards/rejected": 0.0015847217291593552, + "step": 193 + }, + { + "epoch": 0.25, + "learning_rate": 4.977346912649666e-05, + "logits/chosen": -1.8389030694961548, + "logits/rejected": -1.872612476348877, + "logps/chosen": -199.1173095703125, + "logps/rejected": -170.51205444335938, + "loss": 0.6682, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.054506294429302216, + "rewards/margins": 0.1410951465368271, + "rewards/rejected": -0.1956014782190323, + "step": 194 + }, + { + "epoch": 0.26, + "learning_rate": 4.9768631171034175e-05, + "logits/chosen": -1.6105570793151855, + "logits/rejected": -1.655145525932312, + "logps/chosen": -183.67723083496094, + "logps/rejected": -176.943115234375, + "loss": 0.8043, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.09822743386030197, + "rewards/margins": -0.16278451681137085, + "rewards/rejected": 0.06455708295106888, + "step": 195 + }, + { + "epoch": 0.26, + "learning_rate": 4.9763742338679145e-05, + "logits/chosen": -1.589104413986206, + "logits/rejected": -1.544286847114563, + "logps/chosen": -188.53167724609375, + "logps/rejected": -190.18521118164062, + "loss": 0.8959, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.08583441376686096, + "rewards/margins": -0.30193185806274414, + "rewards/rejected": 0.21609747409820557, + "step": 196 + }, + { + "epoch": 0.26, + "learning_rate": 4.975880263947367e-05, + "logits/chosen": -1.5240559577941895, + "logits/rejected": -1.5749508142471313, + "logps/chosen": -173.70506286621094, + "logps/rejected": -174.54681396484375, + "loss": 0.708, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.030573375523090363, + "rewards/margins": 0.030332941561937332, + "rewards/rejected": 0.00024041905999183655, + "step": 197 + }, + { + "epoch": 0.26, + "learning_rate": 4.9753812083564304e-05, + "logits/chosen": -1.879758596420288, + "logits/rejected": -1.8182477951049805, + "logps/chosen": -186.97525024414062, + "logps/rejected": -159.67259216308594, + "loss": 0.7133, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.18908704817295074, + "rewards/margins": 0.015400439500808716, + "rewards/rejected": 0.17368660867214203, + "step": 198 + }, + { + "epoch": 0.26, + "learning_rate": 4.974877068120208e-05, + "logits/chosen": -1.8003509044647217, + "logits/rejected": -1.8137016296386719, + "logps/chosen": -180.23757934570312, + "logps/rejected": -191.359375, + "loss": 0.7853, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.0054572634398937225, + "rewards/margins": -0.046278372406959534, + "rewards/rejected": 0.05173564702272415, + "step": 199 + }, + { + "epoch": 0.26, + "learning_rate": 4.974367844274248e-05, + "logits/chosen": -1.7015259265899658, + "logits/rejected": -1.6682251691818237, + "logps/chosen": -155.98707580566406, + "logps/rejected": -146.8739776611328, + "loss": 0.7147, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.09909596294164658, + "rewards/margins": 0.09801648557186127, + "rewards/rejected": 0.001079469919204712, + "step": 200 + }, + { + "epoch": 0.26, + "learning_rate": 4.973853537864538e-05, + "logits/chosen": -1.9078797101974487, + "logits/rejected": -1.8878577947616577, + "logps/chosen": -213.63034057617188, + "logps/rejected": -196.01133728027344, + "loss": 0.7176, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.07858332991600037, + "rewards/margins": 0.04805755987763405, + "rewards/rejected": -0.12664087116718292, + "step": 201 + }, + { + "epoch": 0.26, + "learning_rate": 4.973334149947508e-05, + "logits/chosen": -1.8100643157958984, + "logits/rejected": -1.8916385173797607, + "logps/chosen": -154.20033264160156, + "logps/rejected": -180.57833862304688, + "loss": 0.9816, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.11480588465929031, + "rewards/margins": -0.3577505350112915, + "rewards/rejected": 0.24294468760490417, + "step": 202 + }, + { + "epoch": 0.27, + "learning_rate": 4.972809681590026e-05, + "logits/chosen": -1.6414841413497925, + "logits/rejected": -1.6490445137023926, + "logps/chosen": -186.58531188964844, + "logps/rejected": -188.83343505859375, + "loss": 0.7131, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.0989757627248764, + "rewards/margins": 0.02306460589170456, + "rewards/rejected": 0.07591113448143005, + "step": 203 + }, + { + "epoch": 0.27, + "learning_rate": 4.972280133869396e-05, + "logits/chosen": -1.7980338335037231, + "logits/rejected": -1.8528972864151, + "logps/chosen": -194.47787475585938, + "logps/rejected": -186.64306640625, + "loss": 0.7408, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.09773577749729156, + "rewards/margins": 0.03546319156885147, + "rewards/rejected": 0.06227259710431099, + "step": 204 + }, + { + "epoch": 0.27, + "learning_rate": 4.971745507873352e-05, + "logits/chosen": -1.8588396310806274, + "logits/rejected": -1.7567483186721802, + "logps/chosen": -181.69345092773438, + "logps/rejected": -187.5961151123047, + "loss": 0.8451, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.1902497261762619, + "rewards/margins": -0.21404078602790833, + "rewards/rejected": 0.023791024461388588, + "step": 205 + }, + { + "epoch": 0.27, + "learning_rate": 4.971205804700063e-05, + "logits/chosen": -1.8144092559814453, + "logits/rejected": -1.8621768951416016, + "logps/chosen": -143.86961364746094, + "logps/rejected": -168.99295043945312, + "loss": 0.8074, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.11274349689483643, + "rewards/margins": -0.13580113649368286, + "rewards/rejected": 0.023057660087943077, + "step": 206 + }, + { + "epoch": 0.27, + "learning_rate": 4.970661025458125e-05, + "logits/chosen": -1.573486089706421, + "logits/rejected": -1.5931901931762695, + "logps/chosen": -157.45480346679688, + "logps/rejected": -169.86550903320312, + "loss": 0.7474, + "rewards/accuracies": 0.375, + "rewards/chosen": 0.007056853733956814, + "rewards/margins": -0.06664810329675674, + "rewards/rejected": 0.07370495796203613, + "step": 207 + }, + { + "epoch": 0.27, + "learning_rate": 4.9701111712665625e-05, + "logits/chosen": -2.0106117725372314, + "logits/rejected": -1.9366306066513062, + "logps/chosen": -190.40353393554688, + "logps/rejected": -176.56820678710938, + "loss": 0.6494, + "rewards/accuracies": 0.4375, + "rewards/chosen": 0.12490460276603699, + "rewards/margins": 0.2138100564479828, + "rewards/rejected": -0.08890549838542938, + "step": 208 + }, + { + "epoch": 0.27, + "learning_rate": 4.969556243254822e-05, + "logits/chosen": -1.7928516864776611, + "logits/rejected": -1.8632910251617432, + "logps/chosen": -232.3812255859375, + "logps/rejected": -237.18328857421875, + "loss": 0.725, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.20639315247535706, + "rewards/margins": 0.010047540068626404, + "rewards/rejected": 0.19634561240673065, + "step": 209 + }, + { + "epoch": 0.27, + "learning_rate": 4.968996242562774e-05, + "logits/chosen": -1.8890585899353027, + "logits/rejected": -1.8688653707504272, + "logps/chosen": -191.57362365722656, + "logps/rejected": -199.00550842285156, + "loss": 0.7262, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.2404320389032364, + "rewards/margins": -0.03374467045068741, + "rewards/rejected": -0.2066873461008072, + "step": 210 + }, + { + "epoch": 0.28, + "learning_rate": 4.968431170340706e-05, + "logits/chosen": -1.6715140342712402, + "logits/rejected": -1.619262456893921, + "logps/chosen": -178.09326171875, + "logps/rejected": -181.5880126953125, + "loss": 0.7146, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.08340275287628174, + "rewards/margins": 0.0807909369468689, + "rewards/rejected": 0.0026118261739611626, + "step": 211 + }, + { + "epoch": 0.28, + "learning_rate": 4.9678610277493275e-05, + "logits/chosen": -1.6335276365280151, + "logits/rejected": -1.6984977722167969, + "logps/chosen": -181.51470947265625, + "logps/rejected": -188.1268768310547, + "loss": 0.6818, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.13411936163902283, + "rewards/margins": 0.05135034769773483, + "rewards/rejected": 0.0827689841389656, + "step": 212 + }, + { + "epoch": 0.28, + "learning_rate": 4.967285815959759e-05, + "logits/chosen": -1.5702835321426392, + "logits/rejected": -1.6215555667877197, + "logps/chosen": -177.0418701171875, + "logps/rejected": -186.50494384765625, + "loss": 0.6603, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.04579095542430878, + "rewards/margins": 0.188707172870636, + "rewards/rejected": -0.23449814319610596, + "step": 213 + }, + { + "epoch": 0.28, + "learning_rate": 4.9667055361535354e-05, + "logits/chosen": -1.7180269956588745, + "logits/rejected": -1.7135778665542603, + "logps/chosen": -195.20785522460938, + "logps/rejected": -210.96878051757812, + "loss": 0.9002, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.03346429392695427, + "rewards/margins": -0.26186707615852356, + "rewards/rejected": 0.29533132910728455, + "step": 214 + }, + { + "epoch": 0.28, + "learning_rate": 4.9661201895226e-05, + "logits/chosen": -1.7542705535888672, + "logits/rejected": -1.7284282445907593, + "logps/chosen": -173.01751708984375, + "logps/rejected": -157.28419494628906, + "loss": 0.6185, + "rewards/accuracies": 0.75, + "rewards/chosen": 0.02107839845120907, + "rewards/margins": 0.2878818213939667, + "rewards/rejected": -0.26680341362953186, + "step": 215 + }, + { + "epoch": 0.28, + "learning_rate": 4.965529777269306e-05, + "logits/chosen": -1.736549973487854, + "logits/rejected": -1.771423578262329, + "logps/chosen": -158.92160034179688, + "logps/rejected": -163.87283325195312, + "loss": 0.7139, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.006686069071292877, + "rewards/margins": 0.04404951259493828, + "rewards/rejected": -0.0373634397983551, + "step": 216 + }, + { + "epoch": 0.28, + "learning_rate": 4.964934300606411e-05, + "logits/chosen": -1.511568307876587, + "logits/rejected": -1.5145093202590942, + "logps/chosen": -170.07809448242188, + "logps/rejected": -186.4696807861328, + "loss": 0.607, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.2800918221473694, + "rewards/margins": 0.30444180965423584, + "rewards/rejected": -0.02434997633099556, + "step": 217 + }, + { + "epoch": 0.29, + "learning_rate": 4.964333760757074e-05, + "logits/chosen": -1.436962366104126, + "logits/rejected": -1.4119391441345215, + "logps/chosen": -309.4395446777344, + "logps/rejected": -291.1634521484375, + "loss": 0.6898, + "rewards/accuracies": 0.375, + "rewards/chosen": 0.0010376125574111938, + "rewards/margins": 0.03738358989357948, + "rewards/rejected": -0.036345988512039185, + "step": 218 + }, + { + "epoch": 0.29, + "learning_rate": 4.963728158954856e-05, + "logits/chosen": -1.891182541847229, + "logits/rejected": -1.8770077228546143, + "logps/chosen": -162.81988525390625, + "logps/rejected": -169.5299072265625, + "loss": 0.8258, + "rewards/accuracies": 0.4375, + "rewards/chosen": 0.07328500598669052, + "rewards/margins": -0.13723579049110413, + "rewards/rejected": 0.21052080392837524, + "step": 219 + }, + { + "epoch": 0.29, + "learning_rate": 4.963117496443715e-05, + "logits/chosen": -1.8470525741577148, + "logits/rejected": -1.8625476360321045, + "logps/chosen": -166.7591552734375, + "logps/rejected": -194.91290283203125, + "loss": 0.9473, + "rewards/accuracies": 0.1875, + "rewards/chosen": -0.4374409317970276, + "rewards/margins": -0.35452863574028015, + "rewards/rejected": -0.08291231095790863, + "step": 220 + }, + { + "epoch": 0.29, + "learning_rate": 4.9625017744780045e-05, + "logits/chosen": -1.5161206722259521, + "logits/rejected": -1.4952480792999268, + "logps/chosen": -173.3487548828125, + "logps/rejected": -167.93292236328125, + "loss": 0.8109, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.0004083700478076935, + "rewards/margins": -0.171901136636734, + "rewards/rejected": 0.1714927703142166, + "step": 221 + }, + { + "epoch": 0.29, + "learning_rate": 4.96188099432247e-05, + "logits/chosen": -1.617663025856018, + "logits/rejected": -1.6117898225784302, + "logps/chosen": -194.1741180419922, + "logps/rejected": -188.07025146484375, + "loss": 0.7477, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.21349965035915375, + "rewards/margins": 0.016070939600467682, + "rewards/rejected": -0.22957059741020203, + "step": 222 + }, + { + "epoch": 0.29, + "learning_rate": 4.9612551572522464e-05, + "logits/chosen": -1.899290680885315, + "logits/rejected": -1.8964145183563232, + "logps/chosen": -192.67469787597656, + "logps/rejected": -179.21112060546875, + "loss": 1.0079, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.24318143725395203, + "rewards/margins": -0.3611811697483063, + "rewards/rejected": 0.11799970269203186, + "step": 223 + }, + { + "epoch": 0.29, + "learning_rate": 4.960624264552858e-05, + "logits/chosen": -1.6361061334609985, + "logits/rejected": -1.6821738481521606, + "logps/chosen": -194.6638641357422, + "logps/rejected": -179.90225219726562, + "loss": 0.6424, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.19569191336631775, + "rewards/margins": 0.2166380137205124, + "rewards/rejected": -0.020946092903614044, + "step": 224 + }, + { + "epoch": 0.29, + "learning_rate": 4.9599883175202124e-05, + "logits/chosen": -1.522665023803711, + "logits/rejected": -1.4538843631744385, + "logps/chosen": -222.08251953125, + "logps/rejected": -200.12472534179688, + "loss": 0.8261, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.26220959424972534, + "rewards/margins": -0.16530472040176392, + "rewards/rejected": -0.09690490365028381, + "step": 225 + }, + { + "epoch": 0.3, + "learning_rate": 4.9593473174605974e-05, + "logits/chosen": -1.4936704635620117, + "logits/rejected": -1.5721888542175293, + "logps/chosen": -182.029541015625, + "logps/rejected": -204.53567504882812, + "loss": 0.6881, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.3360684812068939, + "rewards/margins": 0.14660386741161346, + "rewards/rejected": -0.4826723635196686, + "step": 226 + }, + { + "epoch": 0.3, + "learning_rate": 4.958701265690685e-05, + "logits/chosen": -1.6544736623764038, + "logits/rejected": -1.672118902206421, + "logps/chosen": -174.0331268310547, + "logps/rejected": -203.7425079345703, + "loss": 0.7058, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.34017136693000793, + "rewards/margins": 0.01865684613585472, + "rewards/rejected": -0.3588281571865082, + "step": 227 + }, + { + "epoch": 0.3, + "learning_rate": 4.958050163537519e-05, + "logits/chosen": -1.8430697917938232, + "logits/rejected": -1.7734485864639282, + "logps/chosen": -208.96421813964844, + "logps/rejected": -212.52711486816406, + "loss": 0.8319, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.6053764820098877, + "rewards/margins": -0.1353476196527481, + "rewards/rejected": -0.4700288772583008, + "step": 228 + }, + { + "epoch": 0.3, + "learning_rate": 4.957394012338519e-05, + "logits/chosen": -1.9725829362869263, + "logits/rejected": -1.9301397800445557, + "logps/chosen": -229.4776153564453, + "logps/rejected": -215.60470581054688, + "loss": 0.7281, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.046676263213157654, + "rewards/margins": 0.04501792788505554, + "rewards/rejected": 0.0016583409160375595, + "step": 229 + }, + { + "epoch": 0.3, + "learning_rate": 4.956732813441477e-05, + "logits/chosen": -1.733205795288086, + "logits/rejected": -1.617655873298645, + "logps/chosen": -174.42959594726562, + "logps/rejected": -154.43499755859375, + "loss": 0.9236, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.30461782217025757, + "rewards/margins": -0.30838024616241455, + "rewards/rejected": 0.0037624058313667774, + "step": 230 + }, + { + "epoch": 0.3, + "learning_rate": 4.956066568204552e-05, + "logits/chosen": -1.6661994457244873, + "logits/rejected": -1.7003294229507446, + "logps/chosen": -179.96853637695312, + "logps/rejected": -187.9959716796875, + "loss": 0.8541, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.47250720858573914, + "rewards/margins": -0.12853975594043732, + "rewards/rejected": -0.3439674377441406, + "step": 231 + }, + { + "epoch": 0.3, + "learning_rate": 4.955395277996268e-05, + "logits/chosen": -1.786563754081726, + "logits/rejected": -1.7648732662200928, + "logps/chosen": -193.4469757080078, + "logps/rejected": -202.775146484375, + "loss": 0.5666, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.182941734790802, + "rewards/margins": 0.3387344479560852, + "rewards/rejected": -0.5216761231422424, + "step": 232 + }, + { + "epoch": 0.3, + "learning_rate": 4.954718944195512e-05, + "logits/chosen": -1.7109639644622803, + "logits/rejected": -1.6762810945510864, + "logps/chosen": -154.25538635253906, + "logps/rejected": -158.89683532714844, + "loss": 0.7073, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.25806286931037903, + "rewards/margins": 0.007834136486053467, + "rewards/rejected": -0.2658970057964325, + "step": 233 + }, + { + "epoch": 0.31, + "learning_rate": 4.954037568191534e-05, + "logits/chosen": -1.7765631675720215, + "logits/rejected": -1.8208155632019043, + "logps/chosen": -189.78883361816406, + "logps/rejected": -214.85067749023438, + "loss": 0.7049, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.3182518482208252, + "rewards/margins": 0.08276516944169998, + "rewards/rejected": -0.4010169804096222, + "step": 234 + }, + { + "epoch": 0.31, + "learning_rate": 4.9533511513839384e-05, + "logits/chosen": -1.6697825193405151, + "logits/rejected": -1.688278079032898, + "logps/chosen": -174.92227172851562, + "logps/rejected": -173.88099670410156, + "loss": 0.579, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.08302205801010132, + "rewards/margins": 0.34601137042045593, + "rewards/rejected": -0.42903345823287964, + "step": 235 + }, + { + "epoch": 0.31, + "learning_rate": 4.9526596951826824e-05, + "logits/chosen": -1.9389506578445435, + "logits/rejected": -1.8745498657226562, + "logps/chosen": -193.29092407226562, + "logps/rejected": -187.18719482421875, + "loss": 0.8118, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6870115995407104, + "rewards/margins": -0.025065027177333832, + "rewards/rejected": -0.6619465351104736, + "step": 236 + }, + { + "epoch": 0.31, + "learning_rate": 4.951963201008076e-05, + "logits/chosen": -1.9140545129776, + "logits/rejected": -1.8272314071655273, + "logps/chosen": -154.29287719726562, + "logps/rejected": -149.56021118164062, + "loss": 0.8096, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.10280251502990723, + "rewards/margins": -0.07113885879516602, + "rewards/rejected": -0.03166365623474121, + "step": 237 + }, + { + "epoch": 0.31, + "learning_rate": 4.951261670290781e-05, + "logits/chosen": -2.0082364082336426, + "logits/rejected": -2.00789737701416, + "logps/chosen": -186.12057495117188, + "logps/rejected": -193.4689483642578, + "loss": 0.7454, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.14050771296024323, + "rewards/margins": 0.02027921937406063, + "rewards/rejected": -0.16078691184520721, + "step": 238 + }, + { + "epoch": 0.31, + "learning_rate": 4.950555104471799e-05, + "logits/chosen": -1.851813554763794, + "logits/rejected": -1.8402018547058105, + "logps/chosen": -157.9239959716797, + "logps/rejected": -147.2781524658203, + "loss": 0.7481, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.21430522203445435, + "rewards/margins": -0.030568838119506836, + "rewards/rejected": -0.1837363839149475, + "step": 239 + }, + { + "epoch": 0.31, + "learning_rate": 4.949843505002477e-05, + "logits/chosen": -1.9467929601669312, + "logits/rejected": -1.976270079612732, + "logps/chosen": -170.30682373046875, + "logps/rejected": -167.61927795410156, + "loss": 0.7485, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.2596588134765625, + "rewards/margins": -0.021375911310315132, + "rewards/rejected": -0.23828287422657013, + "step": 240 + }, + { + "epoch": 0.32, + "learning_rate": 4.9491268733445034e-05, + "logits/chosen": -1.724785327911377, + "logits/rejected": -1.7340233325958252, + "logps/chosen": -204.80548095703125, + "logps/rejected": -209.44329833984375, + "loss": 0.7051, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.09321151673793793, + "rewards/margins": 0.0742495059967041, + "rewards/rejected": -0.16746100783348083, + "step": 241 + }, + { + "epoch": 0.32, + "learning_rate": 4.9484052109698984e-05, + "logits/chosen": -1.7430789470672607, + "logits/rejected": -1.7313511371612549, + "logps/chosen": -181.70632934570312, + "logps/rejected": -162.24334716796875, + "loss": 0.844, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.014489106833934784, + "rewards/margins": -0.029044844210147858, + "rewards/rejected": 0.014555716887116432, + "step": 242 + }, + { + "epoch": 0.32, + "learning_rate": 4.947678519361021e-05, + "logits/chosen": -1.9160277843475342, + "logits/rejected": -1.8753935098648071, + "logps/chosen": -175.2951202392578, + "logps/rejected": -161.4536590576172, + "loss": 0.6499, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.1920948028564453, + "rewards/margins": 0.18029190599918365, + "rewards/rejected": -0.3723866939544678, + "step": 243 + }, + { + "epoch": 0.32, + "learning_rate": 4.946946800010556e-05, + "logits/chosen": -1.788377046585083, + "logits/rejected": -1.804762601852417, + "logps/chosen": -190.5827178955078, + "logps/rejected": -207.48460388183594, + "loss": 0.7442, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.20643703639507294, + "rewards/margins": 0.0016913870349526405, + "rewards/rejected": -0.20812839269638062, + "step": 244 + }, + { + "epoch": 0.32, + "learning_rate": 4.946210054421518e-05, + "logits/chosen": -1.943693995475769, + "logits/rejected": -1.9860758781433105, + "logps/chosen": -162.47232055664062, + "logps/rejected": -187.59640502929688, + "loss": 0.5544, + "rewards/accuracies": 0.8125, + "rewards/chosen": -0.04131259024143219, + "rewards/margins": 0.34646379947662354, + "rewards/rejected": -0.3877764344215393, + "step": 245 + }, + { + "epoch": 0.32, + "learning_rate": 4.945468284107246e-05, + "logits/chosen": -1.7154016494750977, + "logits/rejected": -1.729323387145996, + "logps/chosen": -151.67153930664062, + "logps/rejected": -175.7374725341797, + "loss": 0.7351, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.3676840364933014, + "rewards/margins": -0.05210195109248161, + "rewards/rejected": -0.3155820667743683, + "step": 246 + }, + { + "epoch": 0.32, + "learning_rate": 4.944721490591401e-05, + "logits/chosen": -1.5419683456420898, + "logits/rejected": -1.5722306966781616, + "logps/chosen": -158.3173065185547, + "logps/rejected": -168.21975708007812, + "loss": 0.7106, + "rewards/accuracies": 0.625, + "rewards/chosen": 0.040096037089824677, + "rewards/margins": 0.040343452244997025, + "rewards/rejected": -0.00024740397930145264, + "step": 247 + }, + { + "epoch": 0.32, + "learning_rate": 4.9439696754079595e-05, + "logits/chosen": -1.8851487636566162, + "logits/rejected": -1.927181601524353, + "logps/chosen": -163.97447204589844, + "logps/rejected": -171.12020874023438, + "loss": 0.6697, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.4691685140132904, + "rewards/margins": 0.14997676014900208, + "rewards/rejected": -0.6191452741622925, + "step": 248 + }, + { + "epoch": 0.33, + "learning_rate": 4.9432128401012144e-05, + "logits/chosen": -1.5929148197174072, + "logits/rejected": -1.5544054508209229, + "logps/chosen": -143.14022827148438, + "logps/rejected": -158.71368408203125, + "loss": 0.6763, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.0918157696723938, + "rewards/margins": 0.08131375163793564, + "rewards/rejected": -0.17312952876091003, + "step": 249 + }, + { + "epoch": 0.33, + "learning_rate": 4.9424509862257706e-05, + "logits/chosen": -1.599873423576355, + "logits/rejected": -1.5568993091583252, + "logps/chosen": -197.35276794433594, + "logps/rejected": -228.1996307373047, + "loss": 0.6008, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.22382640838623047, + "rewards/margins": 0.30180901288986206, + "rewards/rejected": -0.5256354808807373, + "step": 250 + }, + { + "epoch": 0.33, + "learning_rate": 4.941684115346541e-05, + "logits/chosen": -1.9682908058166504, + "logits/rejected": -1.9601702690124512, + "logps/chosen": -178.14833068847656, + "logps/rejected": -180.44769287109375, + "loss": 0.6096, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.2652210593223572, + "rewards/margins": 0.3119816184043884, + "rewards/rejected": -0.5772026181221008, + "step": 251 + }, + { + "epoch": 0.33, + "learning_rate": 4.940912229038745e-05, + "logits/chosen": -1.7443188428878784, + "logits/rejected": -1.720470666885376, + "logps/chosen": -176.2379150390625, + "logps/rejected": -166.10626220703125, + "loss": 0.8199, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.5284540057182312, + "rewards/margins": -0.06340186297893524, + "rewards/rejected": -0.46505218744277954, + "step": 252 + }, + { + "epoch": 0.33, + "learning_rate": 4.9401353288879024e-05, + "logits/chosen": -1.8005255460739136, + "logits/rejected": -1.814915657043457, + "logps/chosen": -173.22021484375, + "logps/rejected": -187.9818878173828, + "loss": 0.6487, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.36548957228660583, + "rewards/margins": 0.12811800837516785, + "rewards/rejected": -0.49360761046409607, + "step": 253 + }, + { + "epoch": 0.33, + "learning_rate": 4.9393534164898335e-05, + "logits/chosen": -1.8766534328460693, + "logits/rejected": -1.9305753707885742, + "logps/chosen": -166.86866760253906, + "logps/rejected": -195.80569458007812, + "loss": 0.6937, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.2575337588787079, + "rewards/margins": 0.2505089044570923, + "rewards/rejected": -0.5080426931381226, + "step": 254 + }, + { + "epoch": 0.33, + "learning_rate": 4.9385664934506526e-05, + "logits/chosen": -1.7149075269699097, + "logits/rejected": -1.7573699951171875, + "logps/chosen": -169.73626708984375, + "logps/rejected": -178.3174591064453, + "loss": 0.6638, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.4145450294017792, + "rewards/margins": 0.1643792688846588, + "rewards/rejected": -0.578924298286438, + "step": 255 + }, + { + "epoch": 0.34, + "learning_rate": 4.937774561386768e-05, + "logits/chosen": -1.8144739866256714, + "logits/rejected": -1.8054416179656982, + "logps/chosen": -201.68247985839844, + "logps/rejected": -208.70188903808594, + "loss": 0.7625, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.2262319028377533, + "rewards/margins": -0.053480371832847595, + "rewards/rejected": -0.1727515161037445, + "step": 256 + }, + { + "epoch": 0.34, + "learning_rate": 4.936977621924875e-05, + "logits/chosen": -1.721892237663269, + "logits/rejected": -1.7585711479187012, + "logps/chosen": -193.45179748535156, + "logps/rejected": -200.13726806640625, + "loss": 0.798, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.7297480702400208, + "rewards/margins": -0.1376451551914215, + "rewards/rejected": -0.5921030044555664, + "step": 257 + }, + { + "epoch": 0.34, + "learning_rate": 4.9361756767019564e-05, + "logits/chosen": -1.8132922649383545, + "logits/rejected": -1.8062866926193237, + "logps/chosen": -204.11619567871094, + "logps/rejected": -196.605224609375, + "loss": 0.802, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.10710492730140686, + "rewards/margins": -0.09782031178474426, + "rewards/rejected": -0.009284593164920807, + "step": 258 + }, + { + "epoch": 0.34, + "learning_rate": 4.935368727365276e-05, + "logits/chosen": -1.6550960540771484, + "logits/rejected": -1.6377525329589844, + "logps/chosen": -191.87579345703125, + "logps/rejected": -181.06930541992188, + "loss": 0.7402, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.8034918308258057, + "rewards/margins": -0.04004772752523422, + "rewards/rejected": -0.763444185256958, + "step": 259 + }, + { + "epoch": 0.34, + "learning_rate": 4.934556775572377e-05, + "logits/chosen": -1.9349067211151123, + "logits/rejected": -1.9205700159072876, + "logps/chosen": -173.06373596191406, + "logps/rejected": -173.32766723632812, + "loss": 0.6361, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.24924635887145996, + "rewards/margins": 0.23114144802093506, + "rewards/rejected": -0.480387806892395, + "step": 260 + }, + { + "epoch": 0.34, + "learning_rate": 4.9337398229910784e-05, + "logits/chosen": -1.8233386278152466, + "logits/rejected": -1.8753117322921753, + "logps/chosen": -189.73959350585938, + "logps/rejected": -197.85728454589844, + "loss": 0.7225, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.24001392722129822, + "rewards/margins": 0.08805333077907562, + "rewards/rejected": -0.32806724309921265, + "step": 261 + }, + { + "epoch": 0.34, + "learning_rate": 4.932917871299471e-05, + "logits/chosen": -1.540401816368103, + "logits/rejected": -1.5170302391052246, + "logps/chosen": -205.3408203125, + "logps/rejected": -206.5533905029297, + "loss": 0.8948, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.7126679420471191, + "rewards/margins": -0.22102710604667664, + "rewards/rejected": -0.4916408061981201, + "step": 262 + }, + { + "epoch": 0.34, + "learning_rate": 4.9320909221859134e-05, + "logits/chosen": -1.934309482574463, + "logits/rejected": -1.945433497428894, + "logps/chosen": -170.4419708251953, + "logps/rejected": -165.6936492919922, + "loss": 0.7461, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.23198306560516357, + "rewards/margins": -0.018536821007728577, + "rewards/rejected": -0.213446244597435, + "step": 263 + }, + { + "epoch": 0.35, + "learning_rate": 4.9312589773490304e-05, + "logits/chosen": -2.026982545852661, + "logits/rejected": -1.9359885454177856, + "logps/chosen": -185.02920532226562, + "logps/rejected": -173.5999298095703, + "loss": 0.6839, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.39162588119506836, + "rewards/margins": 0.09019112586975098, + "rewards/rejected": -0.48181700706481934, + "step": 264 + }, + { + "epoch": 0.35, + "learning_rate": 4.930422038497708e-05, + "logits/chosen": -1.9103275537490845, + "logits/rejected": -1.8527649641036987, + "logps/chosen": -167.06378173828125, + "logps/rejected": -153.1953125, + "loss": 0.6232, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.32273101806640625, + "rewards/margins": 0.2372804582118988, + "rewards/rejected": -0.5600115060806274, + "step": 265 + }, + { + "epoch": 0.35, + "learning_rate": 4.92958010735109e-05, + "logits/chosen": -1.9541754722595215, + "logits/rejected": -2.0632808208465576, + "logps/chosen": -181.56781005859375, + "logps/rejected": -199.48483276367188, + "loss": 0.5323, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.1671663373708725, + "rewards/margins": 0.5313636064529419, + "rewards/rejected": -0.6985299587249756, + "step": 266 + }, + { + "epoch": 0.35, + "learning_rate": 4.928733185638575e-05, + "logits/chosen": -1.7843657732009888, + "logits/rejected": -1.8627678155899048, + "logps/chosen": -167.1883087158203, + "logps/rejected": -172.48223876953125, + "loss": 0.7816, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.2200300097465515, + "rewards/margins": -0.0496581606566906, + "rewards/rejected": -0.1703718900680542, + "step": 267 + }, + { + "epoch": 0.35, + "learning_rate": 4.927881275099815e-05, + "logits/chosen": -1.713842511177063, + "logits/rejected": -1.805971622467041, + "logps/chosen": -192.92250061035156, + "logps/rejected": -212.14866638183594, + "loss": 0.6343, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.17002353072166443, + "rewards/margins": 0.28741562366485596, + "rewards/rejected": -0.4574391543865204, + "step": 268 + }, + { + "epoch": 0.35, + "learning_rate": 4.927024377484705e-05, + "logits/chosen": -1.682020664215088, + "logits/rejected": -1.7268104553222656, + "logps/chosen": -156.587158203125, + "logps/rejected": -159.53341674804688, + "loss": 0.6657, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.16453158855438232, + "rewards/margins": 0.15859198570251465, + "rewards/rejected": -0.323123574256897, + "step": 269 + }, + { + "epoch": 0.35, + "learning_rate": 4.9261624945533855e-05, + "logits/chosen": -1.8595139980316162, + "logits/rejected": -1.8612048625946045, + "logps/chosen": -163.1502685546875, + "logps/rejected": -193.11166381835938, + "loss": 0.6646, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.2516610622406006, + "rewards/margins": 0.18862426280975342, + "rewards/rejected": -0.440285325050354, + "step": 270 + }, + { + "epoch": 0.35, + "learning_rate": 4.925295628076241e-05, + "logits/chosen": -1.8986504077911377, + "logits/rejected": -1.951588749885559, + "logps/chosen": -161.01625061035156, + "logps/rejected": -171.39744567871094, + "loss": 0.7413, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.18460389971733093, + "rewards/margins": -0.004900887608528137, + "rewards/rejected": -0.1797029972076416, + "step": 271 + }, + { + "epoch": 0.36, + "learning_rate": 4.9244237798338866e-05, + "logits/chosen": -1.7942367792129517, + "logits/rejected": -1.8609907627105713, + "logps/chosen": -190.5836181640625, + "logps/rejected": -181.18942260742188, + "loss": 0.8213, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.7920368909835815, + "rewards/margins": -0.13006603717803955, + "rewards/rejected": -0.6619707942008972, + "step": 272 + }, + { + "epoch": 0.36, + "learning_rate": 4.923546951617175e-05, + "logits/chosen": -1.7586820125579834, + "logits/rejected": -1.8338139057159424, + "logps/chosen": -161.4608154296875, + "logps/rejected": -174.6810302734375, + "loss": 0.7838, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.15520796179771423, + "rewards/margins": 0.007751762866973877, + "rewards/rejected": -0.1629597246646881, + "step": 273 + }, + { + "epoch": 0.36, + "learning_rate": 4.922665145227187e-05, + "logits/chosen": -1.999558925628662, + "logits/rejected": -1.9667410850524902, + "logps/chosen": -177.38986206054688, + "logps/rejected": -182.5379638671875, + "loss": 0.8666, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.26553717255592346, + "rewards/margins": -0.1137295514345169, + "rewards/rejected": -0.15180760622024536, + "step": 274 + }, + { + "epoch": 0.36, + "learning_rate": 4.9217783624752266e-05, + "logits/chosen": -1.7799978256225586, + "logits/rejected": -1.800316572189331, + "logps/chosen": -189.352783203125, + "logps/rejected": -173.5353546142578, + "loss": 0.7974, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.6156832575798035, + "rewards/margins": -0.09660260379314423, + "rewards/rejected": -0.519080638885498, + "step": 275 + }, + { + "epoch": 0.36, + "learning_rate": 4.920886605182823e-05, + "logits/chosen": -1.7154024839401245, + "logits/rejected": -1.7451473474502563, + "logps/chosen": -162.1999053955078, + "logps/rejected": -177.0426483154297, + "loss": 0.6234, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.3276807367801666, + "rewards/margins": 0.274338036775589, + "rewards/rejected": -0.6020187139511108, + "step": 276 + }, + { + "epoch": 0.36, + "learning_rate": 4.919989875181722e-05, + "logits/chosen": -1.769112467765808, + "logits/rejected": -1.759423851966858, + "logps/chosen": -170.71876525878906, + "logps/rejected": -173.2998809814453, + "loss": 0.7953, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.2965158224105835, + "rewards/margins": 0.05386320501565933, + "rewards/rejected": -0.3503790497779846, + "step": 277 + }, + { + "epoch": 0.36, + "learning_rate": 4.919088174313884e-05, + "logits/chosen": -1.2533071041107178, + "logits/rejected": -1.3566581010818481, + "logps/chosen": -200.62548828125, + "logps/rejected": -190.6791534423828, + "loss": 0.7712, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.7463827729225159, + "rewards/margins": -0.011319484561681747, + "rewards/rejected": -0.7350633144378662, + "step": 278 + }, + { + "epoch": 0.37, + "learning_rate": 4.91818150443148e-05, + "logits/chosen": -1.958874225616455, + "logits/rejected": -1.8372151851654053, + "logps/chosen": -184.42295837402344, + "logps/rejected": -178.30995178222656, + "loss": 0.9, + "rewards/accuracies": 0.1875, + "rewards/chosen": -0.6464177370071411, + "rewards/margins": -0.2897469401359558, + "rewards/rejected": -0.3566707670688629, + "step": 279 + }, + { + "epoch": 0.37, + "learning_rate": 4.917269867396886e-05, + "logits/chosen": -1.731322169303894, + "logits/rejected": -1.8198059797286987, + "logps/chosen": -157.6179962158203, + "logps/rejected": -162.81597900390625, + "loss": 0.7364, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.6854066848754883, + "rewards/margins": 0.03314337879419327, + "rewards/rejected": -0.7185500860214233, + "step": 280 + }, + { + "epoch": 0.37, + "learning_rate": 4.916353265082686e-05, + "logits/chosen": -1.5918539762496948, + "logits/rejected": -1.5675849914550781, + "logps/chosen": -229.21499633789062, + "logps/rejected": -248.54562377929688, + "loss": 0.6943, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.4582862854003906, + "rewards/margins": 0.10020212829113007, + "rewards/rejected": -0.5584883689880371, + "step": 281 + }, + { + "epoch": 0.37, + "learning_rate": 4.9154316993716565e-05, + "logits/chosen": -1.9066779613494873, + "logits/rejected": -1.9495766162872314, + "logps/chosen": -149.90614318847656, + "logps/rejected": -146.74632263183594, + "loss": 0.7492, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.44106772541999817, + "rewards/margins": -0.07195230573415756, + "rewards/rejected": -0.36911541223526, + "step": 282 + }, + { + "epoch": 0.37, + "learning_rate": 4.9145051721567734e-05, + "logits/chosen": -2.0382936000823975, + "logits/rejected": -1.9569990634918213, + "logps/chosen": -166.32168579101562, + "logps/rejected": -162.79673767089844, + "loss": 0.6561, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.3699617087841034, + "rewards/margins": 0.14387893676757812, + "rewards/rejected": -0.5138406157493591, + "step": 283 + }, + { + "epoch": 0.37, + "learning_rate": 4.913573685341205e-05, + "logits/chosen": -1.474026083946228, + "logits/rejected": -1.5045154094696045, + "logps/chosen": -229.38619995117188, + "logps/rejected": -236.83164978027344, + "loss": 0.8663, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.9608985781669617, + "rewards/margins": -0.17912134528160095, + "rewards/rejected": -0.7817772626876831, + "step": 284 + }, + { + "epoch": 0.37, + "learning_rate": 4.9126372408383025e-05, + "logits/chosen": -1.8263182640075684, + "logits/rejected": -1.8128119707107544, + "logps/chosen": -190.57106018066406, + "logps/rejected": -223.3215789794922, + "loss": 0.7163, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.5443102717399597, + "rewards/margins": 0.06481970846652985, + "rewards/rejected": -0.6091300249099731, + "step": 285 + }, + { + "epoch": 0.37, + "learning_rate": 4.911695840571605e-05, + "logits/chosen": -1.7644751071929932, + "logits/rejected": -1.8120529651641846, + "logps/chosen": -205.31759643554688, + "logps/rejected": -199.3420867919922, + "loss": 0.8917, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.6110701560974121, + "rewards/margins": -0.2775843143463135, + "rewards/rejected": -0.33348581194877625, + "step": 286 + }, + { + "epoch": 0.38, + "learning_rate": 4.910749486474828e-05, + "logits/chosen": -1.6636385917663574, + "logits/rejected": -1.5808690786361694, + "logps/chosen": -176.58572387695312, + "logps/rejected": -195.53761291503906, + "loss": 0.7043, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.6308318972587585, + "rewards/margins": 0.1960841417312622, + "rewards/rejected": -0.8269160985946655, + "step": 287 + }, + { + "epoch": 0.38, + "learning_rate": 4.909798180491865e-05, + "logits/chosen": -1.946243166923523, + "logits/rejected": -2.0143542289733887, + "logps/chosen": -178.8447265625, + "logps/rejected": -188.88853454589844, + "loss": 0.796, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.4359901547431946, + "rewards/margins": -0.07123897969722748, + "rewards/rejected": -0.3647511601448059, + "step": 288 + }, + { + "epoch": 0.38, + "learning_rate": 4.9088419245767803e-05, + "logits/chosen": -2.0332159996032715, + "logits/rejected": -1.9962480068206787, + "logps/chosen": -181.74835205078125, + "logps/rejected": -198.86822509765625, + "loss": 0.8508, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.9367510676383972, + "rewards/margins": -0.1084718257188797, + "rewards/rejected": -0.8282791972160339, + "step": 289 + }, + { + "epoch": 0.38, + "learning_rate": 4.907880720693804e-05, + "logits/chosen": -2.006517171859741, + "logits/rejected": -1.8688626289367676, + "logps/chosen": -176.2784423828125, + "logps/rejected": -171.2430419921875, + "loss": 0.8853, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.8897907733917236, + "rewards/margins": -0.2713664770126343, + "rewards/rejected": -0.6184243559837341, + "step": 290 + }, + { + "epoch": 0.38, + "learning_rate": 4.9069145708173324e-05, + "logits/chosen": -1.914872169494629, + "logits/rejected": -1.9809991121292114, + "logps/chosen": -152.42398071289062, + "logps/rejected": -172.4163360595703, + "loss": 0.7878, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.8594350814819336, + "rewards/margins": -0.11893659830093384, + "rewards/rejected": -0.740498423576355, + "step": 291 + }, + { + "epoch": 0.38, + "learning_rate": 4.9059434769319205e-05, + "logits/chosen": -1.4920971393585205, + "logits/rejected": -1.479290246963501, + "logps/chosen": -221.7364501953125, + "logps/rejected": -217.07138061523438, + "loss": 0.8203, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.8598781824111938, + "rewards/margins": -0.09480879455804825, + "rewards/rejected": -0.7650693655014038, + "step": 292 + }, + { + "epoch": 0.38, + "learning_rate": 4.904967441032278e-05, + "logits/chosen": -1.6795321702957153, + "logits/rejected": -1.7278181314468384, + "logps/chosen": -176.05941772460938, + "logps/rejected": -176.9445037841797, + "loss": 0.6673, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.6722793579101562, + "rewards/margins": 0.1413556933403015, + "rewards/rejected": -0.8136351108551025, + "step": 293 + }, + { + "epoch": 0.38, + "learning_rate": 4.903986465123266e-05, + "logits/chosen": -1.8752367496490479, + "logits/rejected": -1.815263032913208, + "logps/chosen": -163.13931274414062, + "logps/rejected": -167.08154296875, + "loss": 0.6859, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.587963879108429, + "rewards/margins": 0.14399470388889313, + "rewards/rejected": -0.7319585084915161, + "step": 294 + }, + { + "epoch": 0.39, + "learning_rate": 4.903000551219894e-05, + "logits/chosen": -2.049884796142578, + "logits/rejected": -2.096831798553467, + "logps/chosen": -159.4933624267578, + "logps/rejected": -172.0717010498047, + "loss": 0.7986, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.7208172082901001, + "rewards/margins": -0.12478935718536377, + "rewards/rejected": -0.5960277915000916, + "step": 295 + }, + { + "epoch": 0.39, + "learning_rate": 4.902009701347313e-05, + "logits/chosen": -1.733120322227478, + "logits/rejected": -1.7238028049468994, + "logps/chosen": -207.2619171142578, + "logps/rejected": -196.6807098388672, + "loss": 0.7436, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.46741175651550293, + "rewards/margins": 0.01438647136092186, + "rewards/rejected": -0.4817982316017151, + "step": 296 + }, + { + "epoch": 0.39, + "learning_rate": 4.901013917540814e-05, + "logits/chosen": -2.027272939682007, + "logits/rejected": -1.9993281364440918, + "logps/chosen": -184.6562042236328, + "logps/rejected": -183.0182647705078, + "loss": 0.7772, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.6997889280319214, + "rewards/margins": 0.0013962779194116592, + "rewards/rejected": -0.7011851668357849, + "step": 297 + }, + { + "epoch": 0.39, + "learning_rate": 4.900013201845821e-05, + "logits/chosen": -1.5796035528182983, + "logits/rejected": -1.3949412107467651, + "logps/chosen": -175.8470458984375, + "logps/rejected": -212.80296325683594, + "loss": 0.7523, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.37455612421035767, + "rewards/margins": -0.005637466907501221, + "rewards/rejected": -0.36891865730285645, + "step": 298 + }, + { + "epoch": 0.39, + "learning_rate": 4.899007556317893e-05, + "logits/chosen": -1.9345701932907104, + "logits/rejected": -2.001033306121826, + "logps/chosen": -250.92816162109375, + "logps/rejected": -241.9100341796875, + "loss": 0.7686, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.4188327193260193, + "rewards/margins": -0.09854313731193542, + "rewards/rejected": -0.32028958201408386, + "step": 299 + }, + { + "epoch": 0.39, + "learning_rate": 4.8979969830227086e-05, + "logits/chosen": -1.9917688369750977, + "logits/rejected": -2.0171382427215576, + "logps/chosen": -177.34434509277344, + "logps/rejected": -170.58657836914062, + "loss": 0.8629, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.5179021954536438, + "rewards/margins": -0.09615220129489899, + "rewards/rejected": -0.4217500388622284, + "step": 300 + }, + { + "epoch": 0.39, + "learning_rate": 4.896981484036074e-05, + "logits/chosen": -2.010779619216919, + "logits/rejected": -2.0212411880493164, + "logps/chosen": -190.38487243652344, + "logps/rejected": -189.54226684570312, + "loss": 0.6184, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.45494896173477173, + "rewards/margins": 0.2873493432998657, + "rewards/rejected": -0.7422983050346375, + "step": 301 + }, + { + "epoch": 0.4, + "learning_rate": 4.895961061443911e-05, + "logits/chosen": -1.8409286737442017, + "logits/rejected": -1.7947205305099487, + "logps/chosen": -189.4046173095703, + "logps/rejected": -178.43617248535156, + "loss": 0.8209, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.7244110107421875, + "rewards/margins": -0.0501336008310318, + "rewards/rejected": -0.6742774248123169, + "step": 302 + }, + { + "epoch": 0.4, + "learning_rate": 4.894935717342255e-05, + "logits/chosen": -1.7337801456451416, + "logits/rejected": -1.6998471021652222, + "logps/chosen": -228.77264404296875, + "logps/rejected": -214.67515563964844, + "loss": 0.7953, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.7624901533126831, + "rewards/margins": -0.0503702238202095, + "rewards/rejected": -0.7121198773384094, + "step": 303 + }, + { + "epoch": 0.4, + "learning_rate": 4.8939054538372496e-05, + "logits/chosen": -1.7520679235458374, + "logits/rejected": -1.752877116203308, + "logps/chosen": -186.9480743408203, + "logps/rejected": -195.37429809570312, + "loss": 0.8588, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6405542492866516, + "rewards/margins": -0.206780344247818, + "rewards/rejected": -0.4337739050388336, + "step": 304 + }, + { + "epoch": 0.4, + "learning_rate": 4.8928702730451456e-05, + "logits/chosen": -1.9656537771224976, + "logits/rejected": -2.027155876159668, + "logps/chosen": -199.00657653808594, + "logps/rejected": -212.56494140625, + "loss": 0.7799, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.6149024963378906, + "rewards/margins": -0.018680021166801453, + "rewards/rejected": -0.5962225198745728, + "step": 305 + }, + { + "epoch": 0.4, + "learning_rate": 4.891830177092294e-05, + "logits/chosen": -1.629424810409546, + "logits/rejected": -1.6878198385238647, + "logps/chosen": -170.93138122558594, + "logps/rejected": -171.99090576171875, + "loss": 0.6828, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.673701822757721, + "rewards/margins": 0.13800469040870667, + "rewards/rejected": -0.8117064237594604, + "step": 306 + }, + { + "epoch": 0.4, + "learning_rate": 4.8907851681151396e-05, + "logits/chosen": -1.7744640111923218, + "logits/rejected": -1.855400800704956, + "logps/chosen": -162.42169189453125, + "logps/rejected": -166.64620971679688, + "loss": 0.9846, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.8018124103546143, + "rewards/margins": -0.3769915699958801, + "rewards/rejected": -0.42482078075408936, + "step": 307 + }, + { + "epoch": 0.4, + "learning_rate": 4.889735248260221e-05, + "logits/chosen": -1.889973521232605, + "logits/rejected": -1.907044768333435, + "logps/chosen": -166.85736083984375, + "logps/rejected": -188.35101318359375, + "loss": 0.6128, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.3316620886325836, + "rewards/margins": 0.2465287297964096, + "rewards/rejected": -0.578190803527832, + "step": 308 + }, + { + "epoch": 0.4, + "learning_rate": 4.8886804196841626e-05, + "logits/chosen": -2.047497034072876, + "logits/rejected": -2.0145740509033203, + "logps/chosen": -201.215087890625, + "logps/rejected": -203.3246307373047, + "loss": 0.7635, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.49714764952659607, + "rewards/margins": 0.03132334351539612, + "rewards/rejected": -0.5284709930419922, + "step": 309 + }, + { + "epoch": 0.41, + "learning_rate": 4.887620684553674e-05, + "logits/chosen": -1.831432580947876, + "logits/rejected": -1.826611876487732, + "logps/chosen": -170.79603576660156, + "logps/rejected": -184.5007781982422, + "loss": 0.734, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.7777528166770935, + "rewards/margins": 0.054428160190582275, + "rewards/rejected": -0.832180917263031, + "step": 310 + }, + { + "epoch": 0.41, + "learning_rate": 4.886556045045542e-05, + "logits/chosen": -2.050309896469116, + "logits/rejected": -2.0222089290618896, + "logps/chosen": -186.81155395507812, + "logps/rejected": -177.9705810546875, + "loss": 0.7883, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.6639207601547241, + "rewards/margins": 0.03769933432340622, + "rewards/rejected": -0.7016200423240662, + "step": 311 + }, + { + "epoch": 0.41, + "learning_rate": 4.8854865033466275e-05, + "logits/chosen": -2.076099395751953, + "logits/rejected": -2.0824105739593506, + "logps/chosen": -179.39295959472656, + "logps/rejected": -174.53573608398438, + "loss": 0.7306, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6415868401527405, + "rewards/margins": 0.04155872389674187, + "rewards/rejected": -0.6831455230712891, + "step": 312 + }, + { + "epoch": 0.41, + "learning_rate": 4.88441206165386e-05, + "logits/chosen": -1.6709840297698975, + "logits/rejected": -1.8081105947494507, + "logps/chosen": -166.87539672851562, + "logps/rejected": -183.02621459960938, + "loss": 0.7271, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.518951952457428, + "rewards/margins": 0.06794089823961258, + "rewards/rejected": -0.58689284324646, + "step": 313 + }, + { + "epoch": 0.41, + "learning_rate": 4.8833327221742356e-05, + "logits/chosen": -1.9388179779052734, + "logits/rejected": -1.9303326606750488, + "logps/chosen": -229.31259155273438, + "logps/rejected": -218.38491821289062, + "loss": 0.9369, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.7209200263023376, + "rewards/margins": -0.36736997961997986, + "rewards/rejected": -0.3535500466823578, + "step": 314 + }, + { + "epoch": 0.41, + "learning_rate": 4.88224848712481e-05, + "logits/chosen": -1.8626773357391357, + "logits/rejected": -1.9242452383041382, + "logps/chosen": -190.93325805664062, + "logps/rejected": -220.92437744140625, + "loss": 0.7958, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.9179306030273438, + "rewards/margins": -0.005557693541049957, + "rewards/rejected": -0.9123728275299072, + "step": 315 + }, + { + "epoch": 0.41, + "learning_rate": 4.881159358732694e-05, + "logits/chosen": -1.9244499206542969, + "logits/rejected": -1.9016033411026, + "logps/chosen": -208.19638061523438, + "logps/rejected": -194.61898803710938, + "loss": 0.6003, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.5419009923934937, + "rewards/margins": 0.4854595363140106, + "rewards/rejected": -1.0273605585098267, + "step": 316 + }, + { + "epoch": 0.41, + "learning_rate": 4.8800653392350526e-05, + "logits/chosen": -2.065157413482666, + "logits/rejected": -1.979295015335083, + "logps/chosen": -167.4358367919922, + "logps/rejected": -148.04000854492188, + "loss": 0.79, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6589381098747253, + "rewards/margins": -0.009491220116615295, + "rewards/rejected": -0.6494468450546265, + "step": 317 + }, + { + "epoch": 0.42, + "learning_rate": 4.8789664308790936e-05, + "logits/chosen": -1.8643383979797363, + "logits/rejected": -1.801065444946289, + "logps/chosen": -163.1219482421875, + "logps/rejected": -162.10537719726562, + "loss": 0.6148, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.19017738103866577, + "rewards/margins": 0.2843154966831207, + "rewards/rejected": -0.4744928777217865, + "step": 318 + }, + { + "epoch": 0.42, + "learning_rate": 4.8778626359220715e-05, + "logits/chosen": -1.7437247037887573, + "logits/rejected": -1.704676628112793, + "logps/chosen": -152.5662078857422, + "logps/rejected": -159.8699493408203, + "loss": 0.7078, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.5272018909454346, + "rewards/margins": 0.054309070110321045, + "rewards/rejected": -0.5815109014511108, + "step": 319 + }, + { + "epoch": 0.42, + "learning_rate": 4.8767539566312734e-05, + "logits/chosen": -1.884958028793335, + "logits/rejected": -1.849034070968628, + "logps/chosen": -162.9790802001953, + "logps/rejected": -154.30078125, + "loss": 0.7916, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.5811585187911987, + "rewards/margins": -0.11695164442062378, + "rewards/rejected": -0.46420690417289734, + "step": 320 + }, + { + "epoch": 0.42, + "learning_rate": 4.875640395284023e-05, + "logits/chosen": -1.9072563648223877, + "logits/rejected": -1.8811615705490112, + "logps/chosen": -173.24008178710938, + "logps/rejected": -179.58273315429688, + "loss": 0.7763, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.484822154045105, + "rewards/margins": -0.08394555747509003, + "rewards/rejected": -0.40087658166885376, + "step": 321 + }, + { + "epoch": 0.42, + "learning_rate": 4.874521954167671e-05, + "logits/chosen": -1.8919446468353271, + "logits/rejected": -1.9220166206359863, + "logps/chosen": -158.00631713867188, + "logps/rejected": -158.1962890625, + "loss": 0.6527, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.2469102144241333, + "rewards/margins": 0.14598041772842407, + "rewards/rejected": -0.39289066195487976, + "step": 322 + }, + { + "epoch": 0.42, + "learning_rate": 4.8733986355795905e-05, + "logits/chosen": -2.0576484203338623, + "logits/rejected": -1.957137107849121, + "logps/chosen": -183.29859924316406, + "logps/rejected": -164.59544372558594, + "loss": 0.6596, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.19953250885009766, + "rewards/margins": 0.1992800384759903, + "rewards/rejected": -0.39881253242492676, + "step": 323 + }, + { + "epoch": 0.42, + "learning_rate": 4.8722704418271745e-05, + "logits/chosen": -1.9749754667282104, + "logits/rejected": -1.918540120124817, + "logps/chosen": -167.0876922607422, + "logps/rejected": -155.88064575195312, + "loss": 0.8313, + "rewards/accuracies": 0.1875, + "rewards/chosen": -0.398946613073349, + "rewards/margins": -0.19024060666561127, + "rewards/rejected": -0.20870603621006012, + "step": 324 + }, + { + "epoch": 0.43, + "learning_rate": 4.871137375227829e-05, + "logits/chosen": -1.6178803443908691, + "logits/rejected": -1.73251211643219, + "logps/chosen": -267.42718505859375, + "logps/rejected": -300.83477783203125, + "loss": 0.8896, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.6622705459594727, + "rewards/margins": -0.21010492742061615, + "rewards/rejected": -0.4521656334400177, + "step": 325 + }, + { + "epoch": 0.43, + "learning_rate": 4.869999438108971e-05, + "logits/chosen": -1.888526439666748, + "logits/rejected": -1.8802413940429688, + "logps/chosen": -190.6936798095703, + "logps/rejected": -198.58245849609375, + "loss": 0.8107, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6914330124855042, + "rewards/margins": -0.09704277664422989, + "rewards/rejected": -0.5943902134895325, + "step": 326 + }, + { + "epoch": 0.43, + "learning_rate": 4.8688566328080215e-05, + "logits/chosen": -1.5124105215072632, + "logits/rejected": -1.5067615509033203, + "logps/chosen": -282.9752502441406, + "logps/rejected": -281.37933349609375, + "loss": 0.7879, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.5186710953712463, + "rewards/margins": -0.08422104269266129, + "rewards/rejected": -0.4344501197338104, + "step": 327 + }, + { + "epoch": 0.43, + "learning_rate": 4.867708961672399e-05, + "logits/chosen": -1.8901722431182861, + "logits/rejected": -1.8916290998458862, + "logps/chosen": -194.58010864257812, + "logps/rejected": -206.20706176757812, + "loss": 0.7058, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.2762334942817688, + "rewards/margins": 0.04776221513748169, + "rewards/rejected": -0.3239956796169281, + "step": 328 + }, + { + "epoch": 0.43, + "learning_rate": 4.866556427059519e-05, + "logits/chosen": -1.9781274795532227, + "logits/rejected": -1.991908073425293, + "logps/chosen": -190.02462768554688, + "logps/rejected": -172.36788940429688, + "loss": 0.6999, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.3898717761039734, + "rewards/margins": 0.08414015173912048, + "rewards/rejected": -0.47401195764541626, + "step": 329 + }, + { + "epoch": 0.43, + "learning_rate": 4.865399031336787e-05, + "logits/chosen": -1.7196893692016602, + "logits/rejected": -1.7024250030517578, + "logps/chosen": -183.65408325195312, + "logps/rejected": -183.18309020996094, + "loss": 0.6556, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.18018794059753418, + "rewards/margins": 0.1877795308828354, + "rewards/rejected": -0.3679674565792084, + "step": 330 + }, + { + "epoch": 0.43, + "learning_rate": 4.8642367768815936e-05, + "logits/chosen": -1.82463800907135, + "logits/rejected": -1.9342741966247559, + "logps/chosen": -136.54745483398438, + "logps/rejected": -149.12637329101562, + "loss": 0.7994, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.28317660093307495, + "rewards/margins": -0.111660435795784, + "rewards/rejected": -0.17151619493961334, + "step": 331 + }, + { + "epoch": 0.43, + "learning_rate": 4.863069666081307e-05, + "logits/chosen": -1.9666064977645874, + "logits/rejected": -1.9600831270217896, + "logps/chosen": -160.50009155273438, + "logps/rejected": -172.79641723632812, + "loss": 0.7938, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.3549930453300476, + "rewards/margins": -0.036753974854946136, + "rewards/rejected": -0.3182390332221985, + "step": 332 + }, + { + "epoch": 0.44, + "learning_rate": 4.861897701333274e-05, + "logits/chosen": -1.7229609489440918, + "logits/rejected": -1.7512197494506836, + "logps/chosen": -160.8597869873047, + "logps/rejected": -182.94717407226562, + "loss": 0.7614, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.5264466404914856, + "rewards/margins": -0.05138474702835083, + "rewards/rejected": -0.4750618636608124, + "step": 333 + }, + { + "epoch": 0.44, + "learning_rate": 4.86072088504481e-05, + "logits/chosen": -1.8726041316986084, + "logits/rejected": -1.8346372842788696, + "logps/chosen": -193.1851348876953, + "logps/rejected": -175.16256713867188, + "loss": 0.7167, + "rewards/accuracies": 0.625, + "rewards/chosen": 0.04130769148468971, + "rewards/margins": 0.13345830142498016, + "rewards/rejected": -0.09215061366558075, + "step": 334 + }, + { + "epoch": 0.44, + "learning_rate": 4.859539219633199e-05, + "logits/chosen": -1.6277759075164795, + "logits/rejected": -1.650618553161621, + "logps/chosen": -192.23538208007812, + "logps/rejected": -195.0891876220703, + "loss": 0.9001, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.2764393389225006, + "rewards/margins": -0.1653566211462021, + "rewards/rejected": -0.11108270287513733, + "step": 335 + }, + { + "epoch": 0.44, + "learning_rate": 4.8583527075256804e-05, + "logits/chosen": -1.9752824306488037, + "logits/rejected": -1.976406455039978, + "logps/chosen": -215.763916015625, + "logps/rejected": -222.56468200683594, + "loss": 0.6023, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.3105733394622803, + "rewards/margins": 0.3618737459182739, + "rewards/rejected": -0.6724470853805542, + "step": 336 + }, + { + "epoch": 0.44, + "learning_rate": 4.857161351159454e-05, + "logits/chosen": -1.7939121723175049, + "logits/rejected": -1.8076589107513428, + "logps/chosen": -175.3466033935547, + "logps/rejected": -176.5845947265625, + "loss": 0.6957, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.10015758872032166, + "rewards/margins": 0.10513995587825775, + "rewards/rejected": -0.2052975744009018, + "step": 337 + }, + { + "epoch": 0.44, + "learning_rate": 4.8559651529816664e-05, + "logits/chosen": -1.7438242435455322, + "logits/rejected": -1.7907154560089111, + "logps/chosen": -190.70948791503906, + "logps/rejected": -204.99143981933594, + "loss": 0.6727, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.06472301483154297, + "rewards/margins": 0.1332576423883438, + "rewards/rejected": -0.19798064231872559, + "step": 338 + }, + { + "epoch": 0.44, + "learning_rate": 4.854764115449411e-05, + "logits/chosen": -1.7970941066741943, + "logits/rejected": -1.8352383375167847, + "logps/chosen": -167.88230895996094, + "logps/rejected": -166.8907928466797, + "loss": 0.6771, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.2522534728050232, + "rewards/margins": 0.06870199739933014, + "rewards/rejected": -0.3209554851055145, + "step": 339 + }, + { + "epoch": 0.44, + "learning_rate": 4.853558241029723e-05, + "logits/chosen": -1.9054516553878784, + "logits/rejected": -1.9092512130737305, + "logps/chosen": -173.30734252929688, + "logps/rejected": -189.3043975830078, + "loss": 0.7071, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.2111372947692871, + "rewards/margins": 0.07802311331033707, + "rewards/rejected": -0.2891604006290436, + "step": 340 + }, + { + "epoch": 0.45, + "learning_rate": 4.8523475321995715e-05, + "logits/chosen": -1.6928297281265259, + "logits/rejected": -1.7166639566421509, + "logps/chosen": -163.96253967285156, + "logps/rejected": -157.01870727539062, + "loss": 0.6892, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.023365147411823273, + "rewards/margins": 0.09150275588035583, + "rewards/rejected": -0.06813760101795197, + "step": 341 + }, + { + "epoch": 0.45, + "learning_rate": 4.8511319914458555e-05, + "logits/chosen": -1.5925623178482056, + "logits/rejected": -1.6342376470565796, + "logps/chosen": -169.8107452392578, + "logps/rejected": -174.4763641357422, + "loss": 0.7952, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6270018815994263, + "rewards/margins": -0.09866765886545181, + "rewards/rejected": -0.5283341407775879, + "step": 342 + }, + { + "epoch": 0.45, + "learning_rate": 4.849911621265401e-05, + "logits/chosen": -1.6875545978546143, + "logits/rejected": -1.6878042221069336, + "logps/chosen": -160.625732421875, + "logps/rejected": -186.27923583984375, + "loss": 0.6322, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.33780720829963684, + "rewards/margins": 0.20755568146705627, + "rewards/rejected": -0.5453628301620483, + "step": 343 + }, + { + "epoch": 0.45, + "learning_rate": 4.848686424164953e-05, + "logits/chosen": -1.846010684967041, + "logits/rejected": -1.860701084136963, + "logps/chosen": -181.06878662109375, + "logps/rejected": -178.49310302734375, + "loss": 0.7739, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.3998520076274872, + "rewards/margins": -0.03460027277469635, + "rewards/rejected": -0.36525171995162964, + "step": 344 + }, + { + "epoch": 0.45, + "learning_rate": 4.84745640266117e-05, + "logits/chosen": -1.8941820859909058, + "logits/rejected": -1.8956482410430908, + "logps/chosen": -228.6375732421875, + "logps/rejected": -222.01824951171875, + "loss": 0.7173, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.3104371130466461, + "rewards/margins": 0.0530361533164978, + "rewards/rejected": -0.3634732663631439, + "step": 345 + }, + { + "epoch": 0.45, + "learning_rate": 4.846221559280624e-05, + "logits/chosen": -1.8459906578063965, + "logits/rejected": -1.9024940729141235, + "logps/chosen": -159.93936157226562, + "logps/rejected": -165.43307495117188, + "loss": 0.896, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.23201912641525269, + "rewards/margins": -0.12730032205581665, + "rewards/rejected": -0.10471877455711365, + "step": 346 + }, + { + "epoch": 0.45, + "learning_rate": 4.844981896559787e-05, + "logits/chosen": -2.048933506011963, + "logits/rejected": -2.061128616333008, + "logps/chosen": -181.51718139648438, + "logps/rejected": -185.25296020507812, + "loss": 0.8408, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.32012802362442017, + "rewards/margins": -0.20446370542049408, + "rewards/rejected": -0.1156642958521843, + "step": 347 + }, + { + "epoch": 0.46, + "learning_rate": 4.8437374170450344e-05, + "logits/chosen": -1.9143999814987183, + "logits/rejected": -1.9218837022781372, + "logps/chosen": -182.51480102539062, + "logps/rejected": -182.2652130126953, + "loss": 0.8078, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.6757728457450867, + "rewards/margins": -0.08970170468091965, + "rewards/rejected": -0.5860711932182312, + "step": 348 + }, + { + "epoch": 0.46, + "learning_rate": 4.842488123292632e-05, + "logits/chosen": -1.736176609992981, + "logits/rejected": -1.7229468822479248, + "logps/chosen": -159.5717315673828, + "logps/rejected": -189.70269775390625, + "loss": 0.6977, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.2442682683467865, + "rewards/margins": 0.0333622470498085, + "rewards/rejected": -0.2776305675506592, + "step": 349 + }, + { + "epoch": 0.46, + "learning_rate": 4.8412340178687374e-05, + "logits/chosen": -1.6423401832580566, + "logits/rejected": -1.689012050628662, + "logps/chosen": -151.13458251953125, + "logps/rejected": -175.60679626464844, + "loss": 0.8116, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.33638978004455566, + "rewards/margins": -0.15178707242012024, + "rewards/rejected": -0.18460272252559662, + "step": 350 + }, + { + "epoch": 0.46, + "learning_rate": 4.839975103349391e-05, + "logits/chosen": -1.8096094131469727, + "logits/rejected": -1.8605788946151733, + "logps/chosen": -158.09568786621094, + "logps/rejected": -172.2019805908203, + "loss": 0.8539, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.17084676027297974, + "rewards/margins": -0.16634216904640198, + "rewards/rejected": -0.004504583775997162, + "step": 351 + }, + { + "epoch": 0.46, + "learning_rate": 4.8387113823205096e-05, + "logits/chosen": -1.8915197849273682, + "logits/rejected": -1.9206541776657104, + "logps/chosen": -173.2298583984375, + "logps/rejected": -181.8549346923828, + "loss": 0.6712, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.22103241086006165, + "rewards/margins": 0.12150835990905762, + "rewards/rejected": -0.34254080057144165, + "step": 352 + }, + { + "epoch": 0.46, + "learning_rate": 4.8374428573778864e-05, + "logits/chosen": -1.9554250240325928, + "logits/rejected": -2.0212881565093994, + "logps/chosen": -188.2921142578125, + "logps/rejected": -199.60987854003906, + "loss": 0.7256, + "rewards/accuracies": 0.625, + "rewards/chosen": 0.0011289417743682861, + "rewards/margins": 0.15477600693702698, + "rewards/rejected": -0.1536470353603363, + "step": 353 + }, + { + "epoch": 0.46, + "learning_rate": 4.8361695311271795e-05, + "logits/chosen": -1.626081943511963, + "logits/rejected": -1.5311224460601807, + "logps/chosen": -185.79702758789062, + "logps/rejected": -203.40957641601562, + "loss": 0.8635, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.516851007938385, + "rewards/margins": -0.19476839900016785, + "rewards/rejected": -0.32208263874053955, + "step": 354 + }, + { + "epoch": 0.46, + "learning_rate": 4.83489140618391e-05, + "logits/chosen": -1.7895288467407227, + "logits/rejected": -1.753225564956665, + "logps/chosen": -217.21707153320312, + "logps/rejected": -191.97915649414062, + "loss": 0.7744, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.4739922285079956, + "rewards/margins": -0.06083906441926956, + "rewards/rejected": -0.41315317153930664, + "step": 355 + }, + { + "epoch": 0.47, + "learning_rate": 4.833608485173457e-05, + "logits/chosen": -1.9408211708068848, + "logits/rejected": -1.916908860206604, + "logps/chosen": -210.69586181640625, + "logps/rejected": -233.63662719726562, + "loss": 0.7485, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.3478991985321045, + "rewards/margins": -0.0014675185084342957, + "rewards/rejected": -0.3464316725730896, + "step": 356 + }, + { + "epoch": 0.47, + "learning_rate": 4.8323207707310496e-05, + "logits/chosen": -2.0299384593963623, + "logits/rejected": -2.0607504844665527, + "logps/chosen": -186.0093231201172, + "logps/rejected": -179.7775115966797, + "loss": 0.6698, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.22601480782032013, + "rewards/margins": 0.12358909845352173, + "rewards/rejected": -0.34960389137268066, + "step": 357 + }, + { + "epoch": 0.47, + "learning_rate": 4.831028265501764e-05, + "logits/chosen": -1.7695857286453247, + "logits/rejected": -1.7599815130233765, + "logps/chosen": -161.50375366210938, + "logps/rejected": -168.04541015625, + "loss": 0.6868, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.32897791266441345, + "rewards/margins": 0.10846008360385895, + "rewards/rejected": -0.437438040971756, + "step": 358 + }, + { + "epoch": 0.47, + "learning_rate": 4.829730972140517e-05, + "logits/chosen": -1.8524454832077026, + "logits/rejected": -1.9345438480377197, + "logps/chosen": -149.3410186767578, + "logps/rejected": -160.43539428710938, + "loss": 0.7965, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.08119592815637589, + "rewards/margins": -0.04433928430080414, + "rewards/rejected": -0.03685663640499115, + "step": 359 + }, + { + "epoch": 0.47, + "learning_rate": 4.8284288933120594e-05, + "logits/chosen": -1.8411493301391602, + "logits/rejected": -1.8524004220962524, + "logps/chosen": -175.48532104492188, + "logps/rejected": -182.57582092285156, + "loss": 0.5989, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.46560171246528625, + "rewards/margins": 0.3886818289756775, + "rewards/rejected": -0.8542835712432861, + "step": 360 + }, + { + "epoch": 0.47, + "learning_rate": 4.8271220316909735e-05, + "logits/chosen": -1.687551498413086, + "logits/rejected": -1.722497582435608, + "logps/chosen": -167.26939392089844, + "logps/rejected": -181.43814086914062, + "loss": 0.9314, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.5299715399742126, + "rewards/margins": -0.17927514016628265, + "rewards/rejected": -0.35069650411605835, + "step": 361 + }, + { + "epoch": 0.47, + "learning_rate": 4.825810389961666e-05, + "logits/chosen": -1.916797399520874, + "logits/rejected": -1.9123202562332153, + "logps/chosen": -198.55821228027344, + "logps/rejected": -225.2833709716797, + "loss": 0.8431, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.3632412254810333, + "rewards/margins": -0.22463062405586243, + "rewards/rejected": -0.1386105716228485, + "step": 362 + }, + { + "epoch": 0.48, + "learning_rate": 4.8244939708183596e-05, + "logits/chosen": -1.6189442873001099, + "logits/rejected": -1.6897720098495483, + "logps/chosen": -186.33004760742188, + "logps/rejected": -181.720947265625, + "loss": 0.8053, + "rewards/accuracies": 0.375, + "rewards/chosen": 0.27750128507614136, + "rewards/margins": -0.1327960044145584, + "rewards/rejected": 0.41029733419418335, + "step": 363 + }, + { + "epoch": 0.48, + "learning_rate": 4.823172776965094e-05, + "logits/chosen": -2.003798246383667, + "logits/rejected": -1.9069428443908691, + "logps/chosen": -206.22410583496094, + "logps/rejected": -202.70343017578125, + "loss": 0.7931, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.07843560725450516, + "rewards/margins": 0.052309781312942505, + "rewards/rejected": -0.13074536621570587, + "step": 364 + }, + { + "epoch": 0.48, + "learning_rate": 4.821846811115713e-05, + "logits/chosen": -1.4173839092254639, + "logits/rejected": -1.469193696975708, + "logps/chosen": -252.58900451660156, + "logps/rejected": -266.8291015625, + "loss": 0.7776, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.2866368591785431, + "rewards/margins": -0.01211586594581604, + "rewards/rejected": -0.27452099323272705, + "step": 365 + }, + { + "epoch": 0.48, + "learning_rate": 4.820516075993865e-05, + "logits/chosen": -1.9226216077804565, + "logits/rejected": -1.8626333475112915, + "logps/chosen": -189.990966796875, + "logps/rejected": -212.71446228027344, + "loss": 0.7835, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.2961033880710602, + "rewards/margins": -0.048593662679195404, + "rewards/rejected": -0.24750974774360657, + "step": 366 + }, + { + "epoch": 0.48, + "learning_rate": 4.819180574332994e-05, + "logits/chosen": -2.1082520484924316, + "logits/rejected": -2.0929677486419678, + "logps/chosen": -172.19317626953125, + "logps/rejected": -166.5411376953125, + "loss": 0.7558, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.3426204323768616, + "rewards/margins": 0.029606737196445465, + "rewards/rejected": -0.37222716212272644, + "step": 367 + }, + { + "epoch": 0.48, + "learning_rate": 4.8178403088763355e-05, + "logits/chosen": -1.8143612146377563, + "logits/rejected": -1.8444868326187134, + "logps/chosen": -219.51095581054688, + "logps/rejected": -223.00332641601562, + "loss": 0.7669, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.5216841697692871, + "rewards/margins": -0.06673192977905273, + "rewards/rejected": -0.454952210187912, + "step": 368 + }, + { + "epoch": 0.48, + "learning_rate": 4.8164952823769085e-05, + "logits/chosen": -2.0486927032470703, + "logits/rejected": -1.986092209815979, + "logps/chosen": -179.5719757080078, + "logps/rejected": -176.9426727294922, + "loss": 0.7705, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.12122049927711487, + "rewards/margins": -0.017412271350622177, + "rewards/rejected": -0.1038082093000412, + "step": 369 + }, + { + "epoch": 0.48, + "learning_rate": 4.815145497597514e-05, + "logits/chosen": -1.6093693971633911, + "logits/rejected": -1.6691244840621948, + "logps/chosen": -220.64230346679688, + "logps/rejected": -226.87054443359375, + "loss": 0.739, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.48053669929504395, + "rewards/margins": 0.01233639195561409, + "rewards/rejected": -0.4928731620311737, + "step": 370 + }, + { + "epoch": 0.49, + "learning_rate": 4.8137909573107246e-05, + "logits/chosen": -1.5354715585708618, + "logits/rejected": -1.4911106824874878, + "logps/chosen": -173.0819549560547, + "logps/rejected": -171.91053771972656, + "loss": 0.7285, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.09934857487678528, + "rewards/margins": 0.08546656370162964, + "rewards/rejected": -0.18481513857841492, + "step": 371 + }, + { + "epoch": 0.49, + "learning_rate": 4.812431664298883e-05, + "logits/chosen": -1.8645007610321045, + "logits/rejected": -1.8654489517211914, + "logps/chosen": -174.5074005126953, + "logps/rejected": -174.8561248779297, + "loss": 0.7322, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.34228798747062683, + "rewards/margins": 0.07575173676013947, + "rewards/rejected": -0.4180397093296051, + "step": 372 + }, + { + "epoch": 0.49, + "learning_rate": 4.811067621354094e-05, + "logits/chosen": -1.672195315361023, + "logits/rejected": -1.7209053039550781, + "logps/chosen": -179.9447784423828, + "logps/rejected": -163.33111572265625, + "loss": 0.9538, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.3875757157802582, + "rewards/margins": -0.4043459892272949, + "rewards/rejected": 0.016770271584391594, + "step": 373 + }, + { + "epoch": 0.49, + "learning_rate": 4.8096988312782174e-05, + "logits/chosen": -2.031759023666382, + "logits/rejected": -2.071578025817871, + "logps/chosen": -179.43516540527344, + "logps/rejected": -177.50064086914062, + "loss": 0.8834, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.4397561550140381, + "rewards/margins": -0.1974533647298813, + "rewards/rejected": -0.2423027753829956, + "step": 374 + }, + { + "epoch": 0.49, + "learning_rate": 4.8083252968828665e-05, + "logits/chosen": -1.9421418905258179, + "logits/rejected": -1.922428846359253, + "logps/chosen": -142.18792724609375, + "logps/rejected": -137.76077270507812, + "loss": 0.5961, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.12917383015155792, + "rewards/margins": 0.23805946111679077, + "rewards/rejected": -0.3672332763671875, + "step": 375 + }, + { + "epoch": 0.49, + "learning_rate": 4.8069470209893974e-05, + "logits/chosen": -1.8579202890396118, + "logits/rejected": -1.7998918294906616, + "logps/chosen": -155.51712036132812, + "logps/rejected": -156.0476531982422, + "loss": 0.7183, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.02486548200249672, + "rewards/margins": 0.029943522065877914, + "rewards/rejected": -0.054809004068374634, + "step": 376 + }, + { + "epoch": 0.49, + "learning_rate": 4.8055640064289086e-05, + "logits/chosen": -1.9409297704696655, + "logits/rejected": -1.911987066268921, + "logps/chosen": -245.75701904296875, + "logps/rejected": -248.6239013671875, + "loss": 0.8102, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.4732462763786316, + "rewards/margins": -0.1539333313703537, + "rewards/rejected": -0.3193129599094391, + "step": 377 + }, + { + "epoch": 0.49, + "learning_rate": 4.80417625604223e-05, + "logits/chosen": -1.8907678127288818, + "logits/rejected": -1.8695294857025146, + "logps/chosen": -177.34124755859375, + "logps/rejected": -179.99476623535156, + "loss": 0.6441, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.10845950245857239, + "rewards/margins": 0.19328270852565765, + "rewards/rejected": -0.3017422556877136, + "step": 378 + }, + { + "epoch": 0.5, + "learning_rate": 4.8027837726799205e-05, + "logits/chosen": -1.8234997987747192, + "logits/rejected": -1.8477734327316284, + "logps/chosen": -152.91793823242188, + "logps/rejected": -166.2785186767578, + "loss": 0.7019, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.19078156352043152, + "rewards/margins": 0.07663966715335846, + "rewards/rejected": -0.26742124557495117, + "step": 379 + }, + { + "epoch": 0.5, + "learning_rate": 4.801386559202259e-05, + "logits/chosen": -1.9331165552139282, + "logits/rejected": -1.9343795776367188, + "logps/chosen": -197.23309326171875, + "logps/rejected": -216.14816284179688, + "loss": 0.6427, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.1457238346338272, + "rewards/margins": 0.22981032729148865, + "rewards/rejected": -0.37553414702415466, + "step": 380 + }, + { + "epoch": 0.5, + "learning_rate": 4.799984618479242e-05, + "logits/chosen": -1.7535991668701172, + "logits/rejected": -1.8327221870422363, + "logps/chosen": -170.40121459960938, + "logps/rejected": -194.4297332763672, + "loss": 0.7553, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.2612544000148773, + "rewards/margins": 0.023236550390720367, + "rewards/rejected": -0.2844909727573395, + "step": 381 + }, + { + "epoch": 0.5, + "learning_rate": 4.798577953390577e-05, + "logits/chosen": -1.9297330379486084, + "logits/rejected": -1.8824340105056763, + "logps/chosen": -192.4176788330078, + "logps/rejected": -204.23397827148438, + "loss": 0.7058, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.016188140958547592, + "rewards/margins": 0.1552063524723053, + "rewards/rejected": -0.1713944971561432, + "step": 382 + }, + { + "epoch": 0.5, + "learning_rate": 4.797166566825675e-05, + "logits/chosen": -1.983964204788208, + "logits/rejected": -2.0162336826324463, + "logps/chosen": -165.31370544433594, + "logps/rejected": -175.11459350585938, + "loss": 0.8134, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.43316060304641724, + "rewards/margins": -0.14317026734352112, + "rewards/rejected": -0.2899903357028961, + "step": 383 + }, + { + "epoch": 0.5, + "learning_rate": 4.795750461683644e-05, + "logits/chosen": -1.7382255792617798, + "logits/rejected": -1.7152175903320312, + "logps/chosen": -162.63970947265625, + "logps/rejected": -167.93568420410156, + "loss": 0.8503, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.3506108224391937, + "rewards/margins": -0.19645802676677704, + "rewards/rejected": -0.1541527807712555, + "step": 384 + }, + { + "epoch": 0.5, + "learning_rate": 4.794329640873285e-05, + "logits/chosen": -1.9835039377212524, + "logits/rejected": -1.9371455907821655, + "logps/chosen": -164.90518188476562, + "logps/rejected": -153.89732360839844, + "loss": 0.8365, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.10440421104431152, + "rewards/margins": -0.1423448920249939, + "rewards/rejected": 0.03794068843126297, + "step": 385 + }, + { + "epoch": 0.51, + "learning_rate": 4.7929041073130867e-05, + "logits/chosen": -1.6812117099761963, + "logits/rejected": -1.7789117097854614, + "logps/chosen": -171.525390625, + "logps/rejected": -194.28512573242188, + "loss": 0.6979, + "rewards/accuracies": 0.375, + "rewards/chosen": 0.21003414690494537, + "rewards/margins": 0.1351175457239151, + "rewards/rejected": 0.07491665333509445, + "step": 386 + }, + { + "epoch": 0.51, + "learning_rate": 4.7914738639312165e-05, + "logits/chosen": -1.9188036918640137, + "logits/rejected": -1.9109784364700317, + "logps/chosen": -189.7833251953125, + "logps/rejected": -164.029541015625, + "loss": 0.8337, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.3247818648815155, + "rewards/margins": -0.18027785420417786, + "rewards/rejected": -0.14450398087501526, + "step": 387 + }, + { + "epoch": 0.51, + "learning_rate": 4.790038913665519e-05, + "logits/chosen": -2.0011377334594727, + "logits/rejected": -2.060800552368164, + "logps/chosen": -176.84857177734375, + "logps/rejected": -189.64443969726562, + "loss": 0.7481, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.09154945611953735, + "rewards/margins": -0.03873196616768837, + "rewards/rejected": -0.052817486226558685, + "step": 388 + }, + { + "epoch": 0.51, + "learning_rate": 4.788599259463502e-05, + "logits/chosen": -1.8452333211898804, + "logits/rejected": -1.8431695699691772, + "logps/chosen": -154.12435913085938, + "logps/rejected": -155.32220458984375, + "loss": 0.7097, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.14383897185325623, + "rewards/margins": 0.18797016143798828, + "rewards/rejected": -0.3318091332912445, + "step": 389 + }, + { + "epoch": 0.51, + "learning_rate": 4.787154904282341e-05, + "logits/chosen": -1.3743209838867188, + "logits/rejected": -1.4174141883850098, + "logps/chosen": -177.5303192138672, + "logps/rejected": -211.40457153320312, + "loss": 0.5768, + "rewards/accuracies": 0.6875, + "rewards/chosen": 0.2827177047729492, + "rewards/margins": 0.42551764845848083, + "rewards/rejected": -0.1427999585866928, + "step": 390 + }, + { + "epoch": 0.51, + "learning_rate": 4.7857058510888645e-05, + "logits/chosen": -2.1841466426849365, + "logits/rejected": -2.124525547027588, + "logps/chosen": -246.4276885986328, + "logps/rejected": -244.47906494140625, + "loss": 0.6631, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.038991779088974, + "rewards/margins": 0.20248231291770935, + "rewards/rejected": -0.24147410690784454, + "step": 391 + }, + { + "epoch": 0.51, + "learning_rate": 4.7842521028595526e-05, + "logits/chosen": -1.8262689113616943, + "logits/rejected": -1.8402464389801025, + "logps/chosen": -158.2376708984375, + "logps/rejected": -177.70938110351562, + "loss": 0.7674, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.19041013717651367, + "rewards/margins": 0.02704358845949173, + "rewards/rejected": -0.2174537032842636, + "step": 392 + }, + { + "epoch": 0.51, + "learning_rate": 4.7827936625805284e-05, + "logits/chosen": -1.9320769309997559, + "logits/rejected": -1.9378974437713623, + "logps/chosen": -162.44107055664062, + "logps/rejected": -162.588623046875, + "loss": 0.702, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.22483624517917633, + "rewards/margins": 0.04632706940174103, + "rewards/rejected": 0.1785091906785965, + "step": 393 + }, + { + "epoch": 0.52, + "learning_rate": 4.7813305332475535e-05, + "logits/chosen": -2.023815870285034, + "logits/rejected": -2.1101415157318115, + "logps/chosen": -161.48867797851562, + "logps/rejected": -177.5762939453125, + "loss": 0.7641, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.2686167359352112, + "rewards/margins": -0.034188684076070786, + "rewards/rejected": -0.2344280332326889, + "step": 394 + }, + { + "epoch": 0.52, + "learning_rate": 4.77986271786602e-05, + "logits/chosen": -1.9036113023757935, + "logits/rejected": -1.8756999969482422, + "logps/chosen": -200.5419464111328, + "logps/rejected": -212.63906860351562, + "loss": 0.6506, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.06710982322692871, + "rewards/margins": 0.18568173050880432, + "rewards/rejected": -0.11857189238071442, + "step": 395 + }, + { + "epoch": 0.52, + "learning_rate": 4.778390219450949e-05, + "logits/chosen": -1.8086354732513428, + "logits/rejected": -1.8471179008483887, + "logps/chosen": -152.69277954101562, + "logps/rejected": -143.47000122070312, + "loss": 0.6566, + "rewards/accuracies": 0.6875, + "rewards/chosen": 0.007534712553024292, + "rewards/margins": 0.15669256448745728, + "rewards/rejected": -0.14915785193443298, + "step": 396 + }, + { + "epoch": 0.52, + "learning_rate": 4.776913041026976e-05, + "logits/chosen": -2.1575872898101807, + "logits/rejected": -2.189612627029419, + "logps/chosen": -178.6571044921875, + "logps/rejected": -187.46389770507812, + "loss": 0.8387, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.3838292360305786, + "rewards/margins": -0.1892194300889969, + "rewards/rejected": -0.19460979104042053, + "step": 397 + }, + { + "epoch": 0.52, + "learning_rate": 4.775431185628353e-05, + "logits/chosen": -2.0314245223999023, + "logits/rejected": -2.0670695304870605, + "logps/chosen": -139.46705627441406, + "logps/rejected": -137.02342224121094, + "loss": 0.799, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.3406837582588196, + "rewards/margins": -0.09202456474304199, + "rewards/rejected": -0.24865922331809998, + "step": 398 + }, + { + "epoch": 0.52, + "learning_rate": 4.7739446562989384e-05, + "logits/chosen": -1.7543888092041016, + "logits/rejected": -1.8098934888839722, + "logps/chosen": -171.0781707763672, + "logps/rejected": -192.0042724609375, + "loss": 0.7873, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.4363712966442108, + "rewards/margins": 0.039258234202861786, + "rewards/rejected": -0.4756295382976532, + "step": 399 + }, + { + "epoch": 0.52, + "learning_rate": 4.772453456092191e-05, + "logits/chosen": -1.8036949634552002, + "logits/rejected": -1.8187798261642456, + "logps/chosen": -176.441650390625, + "logps/rejected": -187.6487274169922, + "loss": 0.7976, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.22391349077224731, + "rewards/margins": -0.052816301584243774, + "rewards/rejected": -0.17109718918800354, + "step": 400 + }, + { + "epoch": 0.52, + "learning_rate": 4.7709575880711634e-05, + "logits/chosen": -2.0721547603607178, + "logits/rejected": -2.05245041847229, + "logps/chosen": -181.74232482910156, + "logps/rejected": -187.65249633789062, + "loss": 0.5921, + "rewards/accuracies": 0.5, + "rewards/chosen": 0.2778853476047516, + "rewards/margins": 0.33293941617012024, + "rewards/rejected": -0.05505405738949776, + "step": 401 + }, + { + "epoch": 0.53, + "learning_rate": 4.769457055308497e-05, + "logits/chosen": -2.026765823364258, + "logits/rejected": -2.018843412399292, + "logps/chosen": -197.64260864257812, + "logps/rejected": -179.19317626953125, + "loss": 0.9551, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.30001845955848694, + "rewards/margins": -0.15433457493782043, + "rewards/rejected": -0.1456838697195053, + "step": 402 + }, + { + "epoch": 0.53, + "learning_rate": 4.767951860886415e-05, + "logits/chosen": -1.7545514106750488, + "logits/rejected": -1.7311463356018066, + "logps/chosen": -189.0639190673828, + "logps/rejected": -202.9300537109375, + "loss": 0.7032, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.16693828999996185, + "rewards/margins": 0.15034297108650208, + "rewards/rejected": -0.3172812759876251, + "step": 403 + }, + { + "epoch": 0.53, + "learning_rate": 4.766442007896715e-05, + "logits/chosen": -1.3602584600448608, + "logits/rejected": -1.3138482570648193, + "logps/chosen": -229.68812561035156, + "logps/rejected": -228.81280517578125, + "loss": 0.7063, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.30803146958351135, + "rewards/margins": 0.05981824919581413, + "rewards/rejected": -0.3678497076034546, + "step": 404 + }, + { + "epoch": 0.53, + "learning_rate": 4.764927499440767e-05, + "logits/chosen": -1.2695108652114868, + "logits/rejected": -1.2994788885116577, + "logps/chosen": -178.60507202148438, + "logps/rejected": -209.27049255371094, + "loss": 0.9922, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.590323805809021, + "rewards/margins": -0.39020806550979614, + "rewards/rejected": -0.20011577010154724, + "step": 405 + }, + { + "epoch": 0.53, + "learning_rate": 4.763408338629498e-05, + "logits/chosen": -2.1045475006103516, + "logits/rejected": -2.1285929679870605, + "logps/chosen": -230.02374267578125, + "logps/rejected": -225.12106323242188, + "loss": 0.8737, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.33914846181869507, + "rewards/margins": -0.23797425627708435, + "rewards/rejected": -0.10117418318986893, + "step": 406 + }, + { + "epoch": 0.53, + "learning_rate": 4.761884528583396e-05, + "logits/chosen": -1.4888020753860474, + "logits/rejected": -1.4826213121414185, + "logps/chosen": -221.777587890625, + "logps/rejected": -241.30885314941406, + "loss": 0.7111, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.6717454791069031, + "rewards/margins": 0.279751718044281, + "rewards/rejected": -0.9514971375465393, + "step": 407 + }, + { + "epoch": 0.53, + "learning_rate": 4.760356072432498e-05, + "logits/chosen": -1.8832398653030396, + "logits/rejected": -2.041220188140869, + "logps/chosen": -290.4461364746094, + "logps/rejected": -303.868408203125, + "loss": 0.747, + "rewards/accuracies": 0.5625, + "rewards/chosen": 0.1965964436531067, + "rewards/margins": 0.01566828042268753, + "rewards/rejected": 0.1809280961751938, + "step": 408 + }, + { + "epoch": 0.54, + "learning_rate": 4.7588229733163834e-05, + "logits/chosen": -1.9910494089126587, + "logits/rejected": -2.065354585647583, + "logps/chosen": -189.50331115722656, + "logps/rejected": -204.8780975341797, + "loss": 0.7243, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.7861883640289307, + "rewards/margins": 0.04858472943305969, + "rewards/rejected": -0.834773063659668, + "step": 409 + }, + { + "epoch": 0.54, + "learning_rate": 4.757285234384169e-05, + "logits/chosen": -1.94736909866333, + "logits/rejected": -2.0690090656280518, + "logps/chosen": -183.47666931152344, + "logps/rejected": -199.96681213378906, + "loss": 0.7047, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.5590149164199829, + "rewards/margins": 0.15124721825122833, + "rewards/rejected": -0.7102621793746948, + "step": 410 + }, + { + "epoch": 0.54, + "learning_rate": 4.755742858794503e-05, + "logits/chosen": -2.1221137046813965, + "logits/rejected": -2.069035291671753, + "logps/chosen": -203.06430053710938, + "logps/rejected": -187.47093200683594, + "loss": 0.7289, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.4738054871559143, + "rewards/margins": 0.03721272572875023, + "rewards/rejected": -0.5110181570053101, + "step": 411 + }, + { + "epoch": 0.54, + "learning_rate": 4.754195849715557e-05, + "logits/chosen": -1.9132802486419678, + "logits/rejected": -1.976714849472046, + "logps/chosen": -162.88668823242188, + "logps/rejected": -180.09291076660156, + "loss": 0.658, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.10689959675073624, + "rewards/margins": 0.1677176058292389, + "rewards/rejected": -0.2746172249317169, + "step": 412 + }, + { + "epoch": 0.54, + "learning_rate": 4.75264421032502e-05, + "logits/chosen": -1.8963744640350342, + "logits/rejected": -1.8926461935043335, + "logps/chosen": -199.32736206054688, + "logps/rejected": -198.4470672607422, + "loss": 0.6689, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.6300160884857178, + "rewards/margins": 0.09289233386516571, + "rewards/rejected": -0.7229084968566895, + "step": 413 + }, + { + "epoch": 0.54, + "learning_rate": 4.751087943810093e-05, + "logits/chosen": -1.8022470474243164, + "logits/rejected": -1.797208547592163, + "logps/chosen": -184.97244262695312, + "logps/rejected": -180.94895935058594, + "loss": 0.5437, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.35914158821105957, + "rewards/margins": 0.4712386727333069, + "rewards/rejected": -0.8303802609443665, + "step": 414 + }, + { + "epoch": 0.54, + "learning_rate": 4.749527053367481e-05, + "logits/chosen": -1.7871997356414795, + "logits/rejected": -1.7747814655303955, + "logps/chosen": -189.15310668945312, + "logps/rejected": -178.39805603027344, + "loss": 0.842, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.7546498775482178, + "rewards/margins": -0.17995727062225342, + "rewards/rejected": -0.5746926069259644, + "step": 415 + }, + { + "epoch": 0.54, + "learning_rate": 4.747961542203386e-05, + "logits/chosen": -1.872157335281372, + "logits/rejected": -1.9486424922943115, + "logps/chosen": -172.46145629882812, + "logps/rejected": -195.35421752929688, + "loss": 0.7544, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.42608222365379333, + "rewards/margins": -0.038050394505262375, + "rewards/rejected": -0.38803184032440186, + "step": 416 + }, + { + "epoch": 0.55, + "learning_rate": 4.746391413533503e-05, + "logits/chosen": -1.9934085607528687, + "logits/rejected": -2.0255632400512695, + "logps/chosen": -166.4789276123047, + "logps/rejected": -183.97271728515625, + "loss": 0.6867, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.414784699678421, + "rewards/margins": 0.18602606654167175, + "rewards/rejected": -0.600810706615448, + "step": 417 + }, + { + "epoch": 0.55, + "learning_rate": 4.74481667058301e-05, + "logits/chosen": -1.8942878246307373, + "logits/rejected": -1.87151038646698, + "logps/chosen": -173.86004638671875, + "logps/rejected": -163.0937957763672, + "loss": 0.8824, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.772480309009552, + "rewards/margins": -0.22370155155658722, + "rewards/rejected": -0.5487788319587708, + "step": 418 + }, + { + "epoch": 0.55, + "learning_rate": 4.743237316586564e-05, + "logits/chosen": -1.9378021955490112, + "logits/rejected": -1.9703481197357178, + "logps/chosen": -180.11892700195312, + "logps/rejected": -195.14578247070312, + "loss": 0.7388, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.3675965666770935, + "rewards/margins": -0.02655930444598198, + "rewards/rejected": -0.3410373032093048, + "step": 419 + }, + { + "epoch": 0.55, + "learning_rate": 4.741653354788295e-05, + "logits/chosen": -2.0154881477355957, + "logits/rejected": -2.0120866298675537, + "logps/chosen": -174.45498657226562, + "logps/rejected": -182.1272430419922, + "loss": 0.8029, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.7631839513778687, + "rewards/margins": -0.11559872329235077, + "rewards/rejected": -0.6475852727890015, + "step": 420 + }, + { + "epoch": 0.55, + "learning_rate": 4.7400647884417956e-05, + "logits/chosen": -1.8835885524749756, + "logits/rejected": -2.027597665786743, + "logps/chosen": -169.2918701171875, + "logps/rejected": -167.33514404296875, + "loss": 0.6719, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.33407047390937805, + "rewards/margins": 0.15689513087272644, + "rewards/rejected": -0.49096566438674927, + "step": 421 + }, + { + "epoch": 0.55, + "learning_rate": 4.7384716208101166e-05, + "logits/chosen": -2.028184413909912, + "logits/rejected": -2.0173041820526123, + "logps/chosen": -168.9990692138672, + "logps/rejected": -160.54428100585938, + "loss": 0.7799, + "rewards/accuracies": 0.1875, + "rewards/chosen": -0.5599774122238159, + "rewards/margins": -0.1413966715335846, + "rewards/rejected": -0.4185807704925537, + "step": 422 + }, + { + "epoch": 0.55, + "learning_rate": 4.736873855165762e-05, + "logits/chosen": -1.9846090078353882, + "logits/rejected": -1.9857451915740967, + "logps/chosen": -193.67715454101562, + "logps/rejected": -175.9185333251953, + "loss": 0.6656, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.5016548037528992, + "rewards/margins": 0.22657424211502075, + "rewards/rejected": -0.7282290458679199, + "step": 423 + }, + { + "epoch": 0.55, + "learning_rate": 4.735271494790678e-05, + "logits/chosen": -1.92975914478302, + "logits/rejected": -1.9406872987747192, + "logps/chosen": -181.31509399414062, + "logps/rejected": -167.101806640625, + "loss": 0.7386, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.18490439653396606, + "rewards/margins": 0.0683104544878006, + "rewards/rejected": -0.25321486592292786, + "step": 424 + }, + { + "epoch": 0.56, + "learning_rate": 4.733664542976253e-05, + "logits/chosen": -1.9516609907150269, + "logits/rejected": -2.005047559738159, + "logps/chosen": -167.55972290039062, + "logps/rejected": -168.99807739257812, + "loss": 0.9659, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.3683350086212158, + "rewards/margins": -0.24237681925296783, + "rewards/rejected": -0.12595820426940918, + "step": 425 + }, + { + "epoch": 0.56, + "learning_rate": 4.732053003023301e-05, + "logits/chosen": -1.934274435043335, + "logits/rejected": -1.9798094034194946, + "logps/chosen": -153.78518676757812, + "logps/rejected": -166.25979614257812, + "loss": 0.7058, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.5433434247970581, + "rewards/margins": 0.13497428596019745, + "rewards/rejected": -0.6783177256584167, + "step": 426 + }, + { + "epoch": 0.56, + "learning_rate": 4.730436878242064e-05, + "logits/chosen": -1.9575129747390747, + "logits/rejected": -1.9926663637161255, + "logps/chosen": -153.06951904296875, + "logps/rejected": -174.07875061035156, + "loss": 0.8359, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.4944266080856323, + "rewards/margins": -0.00036665797233581543, + "rewards/rejected": -0.4940599203109741, + "step": 427 + }, + { + "epoch": 0.56, + "learning_rate": 4.7288161719522016e-05, + "logits/chosen": -1.9566092491149902, + "logits/rejected": -1.9228875637054443, + "logps/chosen": -162.54771423339844, + "logps/rejected": -167.0357666015625, + "loss": 0.8701, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.5988922715187073, + "rewards/margins": -0.1607256382703781, + "rewards/rejected": -0.43816661834716797, + "step": 428 + }, + { + "epoch": 0.56, + "learning_rate": 4.727190887482783e-05, + "logits/chosen": -2.212228775024414, + "logits/rejected": -2.238290309906006, + "logps/chosen": -185.6492462158203, + "logps/rejected": -201.63949584960938, + "loss": 0.7027, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.45072147250175476, + "rewards/margins": 0.03795819729566574, + "rewards/rejected": -0.4886796772480011, + "step": 429 + }, + { + "epoch": 0.56, + "learning_rate": 4.725561028172282e-05, + "logits/chosen": -2.08243989944458, + "logits/rejected": -2.100586414337158, + "logps/chosen": -169.82723999023438, + "logps/rejected": -169.41476440429688, + "loss": 0.8011, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.5032753348350525, + "rewards/margins": -0.07662791758775711, + "rewards/rejected": -0.4266473352909088, + "step": 430 + }, + { + "epoch": 0.56, + "learning_rate": 4.7239265973685696e-05, + "logits/chosen": -1.7974039316177368, + "logits/rejected": -1.802499771118164, + "logps/chosen": -166.3297119140625, + "logps/rejected": -184.35450744628906, + "loss": 0.6115, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.3483116924762726, + "rewards/margins": 0.2958502173423767, + "rewards/rejected": -0.6441619396209717, + "step": 431 + }, + { + "epoch": 0.57, + "learning_rate": 4.722287598428907e-05, + "logits/chosen": -1.9482653141021729, + "logits/rejected": -1.9997018575668335, + "logps/chosen": -202.8570098876953, + "logps/rejected": -219.10565185546875, + "loss": 0.6303, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.0477115735411644, + "rewards/margins": 0.21878241002559662, + "rewards/rejected": -0.26649394631385803, + "step": 432 + }, + { + "epoch": 0.57, + "learning_rate": 4.720644034719938e-05, + "logits/chosen": -1.8863980770111084, + "logits/rejected": -1.8580697774887085, + "logps/chosen": -178.37112426757812, + "logps/rejected": -196.09762573242188, + "loss": 0.7045, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.359438419342041, + "rewards/margins": 0.0665070116519928, + "rewards/rejected": -0.4259454607963562, + "step": 433 + }, + { + "epoch": 0.57, + "learning_rate": 4.7189959096176825e-05, + "logits/chosen": -1.962789535522461, + "logits/rejected": -2.0059375762939453, + "logps/chosen": -168.6863250732422, + "logps/rejected": -207.6638946533203, + "loss": 0.7517, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.33223748207092285, + "rewards/margins": 0.007875222712755203, + "rewards/rejected": -0.3401126563549042, + "step": 434 + }, + { + "epoch": 0.57, + "learning_rate": 4.7173432265075334e-05, + "logits/chosen": -2.13173770904541, + "logits/rejected": -2.1997811794281006, + "logps/chosen": -171.5570526123047, + "logps/rejected": -173.00726318359375, + "loss": 0.7831, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.5620677471160889, + "rewards/margins": -0.08540257066488266, + "rewards/rejected": -0.47666510939598083, + "step": 435 + }, + { + "epoch": 0.57, + "learning_rate": 4.7156859887842416e-05, + "logits/chosen": -1.9717164039611816, + "logits/rejected": -1.9659010171890259, + "logps/chosen": -163.4027099609375, + "logps/rejected": -171.51205444335938, + "loss": 0.8767, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.36586710810661316, + "rewards/margins": -0.22375579178333282, + "rewards/rejected": -0.14211128652095795, + "step": 436 + }, + { + "epoch": 0.57, + "learning_rate": 4.714024199851915e-05, + "logits/chosen": -1.9461572170257568, + "logits/rejected": -1.9711329936981201, + "logps/chosen": -173.81141662597656, + "logps/rejected": -172.41860961914062, + "loss": 0.7655, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.5734329223632812, + "rewards/margins": 0.0679081380367279, + "rewards/rejected": -0.6413410305976868, + "step": 437 + }, + { + "epoch": 0.57, + "learning_rate": 4.712357863124013e-05, + "logits/chosen": -2.0299744606018066, + "logits/rejected": -2.055668830871582, + "logps/chosen": -168.48684692382812, + "logps/rejected": -182.92257690429688, + "loss": 0.69, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.44026386737823486, + "rewards/margins": 0.11328400671482086, + "rewards/rejected": -0.5535478591918945, + "step": 438 + }, + { + "epoch": 0.57, + "learning_rate": 4.710686982023332e-05, + "logits/chosen": -2.0356907844543457, + "logits/rejected": -1.9616978168487549, + "logps/chosen": -147.8501739501953, + "logps/rejected": -154.8659210205078, + "loss": 0.7309, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.06124575436115265, + "rewards/margins": 0.03877441585063934, + "rewards/rejected": -0.10002017021179199, + "step": 439 + }, + { + "epoch": 0.58, + "learning_rate": 4.709011559982006e-05, + "logits/chosen": -2.043642997741699, + "logits/rejected": -1.9762914180755615, + "logps/chosen": -193.67340087890625, + "logps/rejected": -186.66543579101562, + "loss": 0.7127, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.34980466961860657, + "rewards/margins": 0.07452677190303802, + "rewards/rejected": -0.4243314862251282, + "step": 440 + }, + { + "epoch": 0.58, + "learning_rate": 4.707331600441495e-05, + "logits/chosen": -2.072479724884033, + "logits/rejected": -2.091381549835205, + "logps/chosen": -191.45059204101562, + "logps/rejected": -176.9473876953125, + "loss": 0.6801, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.029101327061653137, + "rewards/margins": 0.26256757974624634, + "rewards/rejected": -0.2916688919067383, + "step": 441 + }, + { + "epoch": 0.58, + "learning_rate": 4.705647106852581e-05, + "logits/chosen": -1.9268254041671753, + "logits/rejected": -1.8882079124450684, + "logps/chosen": -171.5392303466797, + "logps/rejected": -182.5844268798828, + "loss": 0.7921, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.2353130429983139, + "rewards/margins": -0.0611705407500267, + "rewards/rejected": -0.1741425096988678, + "step": 442 + }, + { + "epoch": 0.58, + "learning_rate": 4.7039580826753564e-05, + "logits/chosen": -2.029810667037964, + "logits/rejected": -2.026019334793091, + "logps/chosen": -169.64918518066406, + "logps/rejected": -186.5985107421875, + "loss": 0.6615, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.34466996788978577, + "rewards/margins": 0.14720463752746582, + "rewards/rejected": -0.491874635219574, + "step": 443 + }, + { + "epoch": 0.58, + "learning_rate": 4.7022645313792235e-05, + "logits/chosen": -1.5730178356170654, + "logits/rejected": -1.5958049297332764, + "logps/chosen": -160.8270263671875, + "logps/rejected": -176.29554748535156, + "loss": 0.7613, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.34809601306915283, + "rewards/margins": 0.23688597977161407, + "rewards/rejected": -0.5849819779396057, + "step": 444 + }, + { + "epoch": 0.58, + "learning_rate": 4.700566456442882e-05, + "logits/chosen": -2.009403705596924, + "logits/rejected": -1.9752486944198608, + "logps/chosen": -178.82701110839844, + "logps/rejected": -180.32125854492188, + "loss": 1.0244, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.6049767136573792, + "rewards/margins": -0.4276657700538635, + "rewards/rejected": -0.1773110032081604, + "step": 445 + }, + { + "epoch": 0.58, + "learning_rate": 4.6988638613543216e-05, + "logits/chosen": -1.7354819774627686, + "logits/rejected": -1.731933355331421, + "logps/chosen": -167.52633666992188, + "logps/rejected": -181.66583251953125, + "loss": 0.789, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.6287019848823547, + "rewards/margins": -0.054200708866119385, + "rewards/rejected": -0.5745012760162354, + "step": 446 + }, + { + "epoch": 0.58, + "learning_rate": 4.6971567496108206e-05, + "logits/chosen": -1.9981721639633179, + "logits/rejected": -2.026167154312134, + "logps/chosen": -212.54891967773438, + "logps/rejected": -215.69253540039062, + "loss": 0.7395, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.4362500011920929, + "rewards/margins": -0.020926453173160553, + "rewards/rejected": -0.4153235852718353, + "step": 447 + }, + { + "epoch": 0.59, + "learning_rate": 4.695445124718931e-05, + "logits/chosen": -2.239379405975342, + "logits/rejected": -2.186093807220459, + "logps/chosen": -189.42547607421875, + "logps/rejected": -181.0236358642578, + "loss": 0.6656, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.3167141079902649, + "rewards/margins": 0.10905791819095612, + "rewards/rejected": -0.4257720112800598, + "step": 448 + }, + { + "epoch": 0.59, + "learning_rate": 4.693728990194479e-05, + "logits/chosen": -2.132059097290039, + "logits/rejected": -2.1224942207336426, + "logps/chosen": -211.412841796875, + "logps/rejected": -197.23439025878906, + "loss": 0.6771, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.29473942518234253, + "rewards/margins": 0.20832209289073944, + "rewards/rejected": -0.5030615329742432, + "step": 449 + }, + { + "epoch": 0.59, + "learning_rate": 4.692008349562551e-05, + "logits/chosen": -2.2113986015319824, + "logits/rejected": -2.1679513454437256, + "logps/chosen": -174.953369140625, + "logps/rejected": -188.64576721191406, + "loss": 0.9728, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.5910927057266235, + "rewards/margins": -0.3634761869907379, + "rewards/rejected": -0.22761650383472443, + "step": 450 + }, + { + "epoch": 0.59, + "learning_rate": 4.690283206357491e-05, + "logits/chosen": -1.984092116355896, + "logits/rejected": -2.0135576725006104, + "logps/chosen": -187.25738525390625, + "logps/rejected": -189.52325439453125, + "loss": 0.672, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.4424164593219757, + "rewards/margins": 0.20770896971225739, + "rewards/rejected": -0.6501253843307495, + "step": 451 + }, + { + "epoch": 0.59, + "learning_rate": 4.6885535641228904e-05, + "logits/chosen": -2.0326945781707764, + "logits/rejected": -2.0409088134765625, + "logps/chosen": -189.7724609375, + "logps/rejected": -196.32164001464844, + "loss": 0.846, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.2984338402748108, + "rewards/margins": -0.14787545800209045, + "rewards/rejected": -0.15055838227272034, + "step": 452 + }, + { + "epoch": 0.59, + "learning_rate": 4.6868194264115833e-05, + "logits/chosen": -1.8356232643127441, + "logits/rejected": -1.860498309135437, + "logps/chosen": -190.26791381835938, + "logps/rejected": -192.42124938964844, + "loss": 0.6865, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.5506845712661743, + "rewards/margins": 0.12794733047485352, + "rewards/rejected": -0.6786318421363831, + "step": 453 + }, + { + "epoch": 0.59, + "learning_rate": 4.685080796785637e-05, + "logits/chosen": -2.0949289798736572, + "logits/rejected": -2.105635643005371, + "logps/chosen": -188.15576171875, + "logps/rejected": -177.27642822265625, + "loss": 0.7395, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.6156455874443054, + "rewards/margins": 0.00026201456785202026, + "rewards/rejected": -0.615907609462738, + "step": 454 + }, + { + "epoch": 0.6, + "learning_rate": 4.683337678816345e-05, + "logits/chosen": -2.062208414077759, + "logits/rejected": -1.9715068340301514, + "logps/chosen": -247.74729919433594, + "logps/rejected": -225.91094970703125, + "loss": 0.8454, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.7724697589874268, + "rewards/margins": -0.1988295614719391, + "rewards/rejected": -0.5736401081085205, + "step": 455 + }, + { + "epoch": 0.6, + "learning_rate": 4.6815900760842236e-05, + "logits/chosen": -1.999656081199646, + "logits/rejected": -2.013665199279785, + "logps/chosen": -189.76425170898438, + "logps/rejected": -200.77801513671875, + "loss": 0.7929, + "rewards/accuracies": 0.5625, + "rewards/chosen": -1.1113797426223755, + "rewards/margins": -0.07439464330673218, + "rewards/rejected": -1.0369850397109985, + "step": 456 + }, + { + "epoch": 0.6, + "learning_rate": 4.679837992178996e-05, + "logits/chosen": -1.962624192237854, + "logits/rejected": -1.8894569873809814, + "logps/chosen": -163.32264709472656, + "logps/rejected": -176.6225128173828, + "loss": 0.6683, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.5831518769264221, + "rewards/margins": 0.18081454932689667, + "rewards/rejected": -0.76396644115448, + "step": 457 + }, + { + "epoch": 0.6, + "learning_rate": 4.678081430699594e-05, + "logits/chosen": -1.9277567863464355, + "logits/rejected": -1.9991313219070435, + "logps/chosen": -177.146484375, + "logps/rejected": -180.6768798828125, + "loss": 0.549, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.5843685865402222, + "rewards/margins": 0.5196143388748169, + "rewards/rejected": -1.103982925415039, + "step": 458 + }, + { + "epoch": 0.6, + "learning_rate": 4.676320395254146e-05, + "logits/chosen": -1.8018032312393188, + "logits/rejected": -1.7631927728652954, + "logps/chosen": -193.01077270507812, + "logps/rejected": -193.25103759765625, + "loss": 0.6489, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.5062118172645569, + "rewards/margins": 0.23078583180904388, + "rewards/rejected": -0.736997663974762, + "step": 459 + }, + { + "epoch": 0.6, + "learning_rate": 4.674554889459968e-05, + "logits/chosen": -1.7966090440750122, + "logits/rejected": -1.7850843667984009, + "logps/chosen": -182.4764404296875, + "logps/rejected": -184.0175018310547, + "loss": 0.7334, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.4785413146018982, + "rewards/margins": -0.006460566073656082, + "rewards/rejected": -0.4720807671546936, + "step": 460 + }, + { + "epoch": 0.6, + "learning_rate": 4.672784916943562e-05, + "logits/chosen": -1.6125917434692383, + "logits/rejected": -1.6394853591918945, + "logps/chosen": -182.64862060546875, + "logps/rejected": -196.8616943359375, + "loss": 0.5382, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.8303923010826111, + "rewards/margins": 0.4725598096847534, + "rewards/rejected": -1.3029520511627197, + "step": 461 + }, + { + "epoch": 0.6, + "learning_rate": 4.6710104813406034e-05, + "logits/chosen": -1.7587897777557373, + "logits/rejected": -1.7165967226028442, + "logps/chosen": -178.50250244140625, + "logps/rejected": -159.33375549316406, + "loss": 0.9045, + "rewards/accuracies": 0.25, + "rewards/chosen": -0.5282204151153564, + "rewards/margins": -0.22660110890865326, + "rewards/rejected": -0.30161935091018677, + "step": 462 + }, + { + "epoch": 0.61, + "learning_rate": 4.669231586295934e-05, + "logits/chosen": -1.8907124996185303, + "logits/rejected": -1.9228183031082153, + "logps/chosen": -169.16119384765625, + "logps/rejected": -180.715087890625, + "loss": 0.7763, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.8981459736824036, + "rewards/margins": -0.11685739457607269, + "rewards/rejected": -0.7812885046005249, + "step": 463 + }, + { + "epoch": 0.61, + "learning_rate": 4.667448235463557e-05, + "logits/chosen": -1.2660267353057861, + "logits/rejected": -1.2475149631500244, + "logps/chosen": -183.68353271484375, + "logps/rejected": -182.84422302246094, + "loss": 0.8929, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.7595192790031433, + "rewards/margins": -0.14992079138755798, + "rewards/rejected": -0.6095985174179077, + "step": 464 + }, + { + "epoch": 0.61, + "learning_rate": 4.665660432506629e-05, + "logits/chosen": -1.8095303773880005, + "logits/rejected": -1.8506840467453003, + "logps/chosen": -213.27145385742188, + "logps/rejected": -220.65640258789062, + "loss": 0.7946, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.6156535744667053, + "rewards/margins": -0.057905957102775574, + "rewards/rejected": -0.5577476024627686, + "step": 465 + }, + { + "epoch": 0.61, + "learning_rate": 4.6638681810974496e-05, + "logits/chosen": -1.758918285369873, + "logits/rejected": -1.7417278289794922, + "logps/chosen": -180.2194366455078, + "logps/rejected": -199.62014770507812, + "loss": 0.5957, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.17319843173027039, + "rewards/margins": 0.4147520363330841, + "rewards/rejected": -0.5879504680633545, + "step": 466 + }, + { + "epoch": 0.61, + "learning_rate": 4.6620714849174576e-05, + "logits/chosen": -1.5012279748916626, + "logits/rejected": -1.495218276977539, + "logps/chosen": -227.26577758789062, + "logps/rejected": -223.47470092773438, + "loss": 0.7009, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.7020196318626404, + "rewards/margins": 0.10670151561498642, + "rewards/rejected": -0.808721125125885, + "step": 467 + }, + { + "epoch": 0.61, + "learning_rate": 4.660270347657219e-05, + "logits/chosen": -1.4245662689208984, + "logits/rejected": -1.4722357988357544, + "logps/chosen": -219.63504028320312, + "logps/rejected": -246.0736846923828, + "loss": 0.6091, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.7042097449302673, + "rewards/margins": 0.5171206593513489, + "rewards/rejected": -1.2213302850723267, + "step": 468 + }, + { + "epoch": 0.61, + "learning_rate": 4.658464773016428e-05, + "logits/chosen": -1.7068259716033936, + "logits/rejected": -1.6351027488708496, + "logps/chosen": -199.54136657714844, + "logps/rejected": -181.50997924804688, + "loss": 0.8131, + "rewards/accuracies": 0.5625, + "rewards/chosen": -1.2041518688201904, + "rewards/margins": -0.023167330771684647, + "rewards/rejected": -1.180984616279602, + "step": 469 + }, + { + "epoch": 0.62, + "learning_rate": 4.6566547647038864e-05, + "logits/chosen": -1.7098909616470337, + "logits/rejected": -1.80801522731781, + "logps/chosen": -167.95101928710938, + "logps/rejected": -180.0511016845703, + "loss": 0.5463, + "rewards/accuracies": 0.75, + "rewards/chosen": -0.2591591477394104, + "rewards/margins": 0.41101884841918945, + "rewards/rejected": -0.6701779961585999, + "step": 470 + }, + { + "epoch": 0.62, + "learning_rate": 4.6548403264375074e-05, + "logits/chosen": -2.014415979385376, + "logits/rejected": -2.0018627643585205, + "logps/chosen": -181.22947692871094, + "logps/rejected": -189.5872344970703, + "loss": 0.8377, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.9818540811538696, + "rewards/margins": -0.06248188391327858, + "rewards/rejected": -0.9193722009658813, + "step": 471 + }, + { + "epoch": 0.62, + "learning_rate": 4.6530214619443037e-05, + "logits/chosen": -1.902940034866333, + "logits/rejected": -1.940006971359253, + "logps/chosen": -156.01939392089844, + "logps/rejected": -156.20623779296875, + "loss": 0.8502, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.6196620464324951, + "rewards/margins": -0.18280625343322754, + "rewards/rejected": -0.4368557929992676, + "step": 472 + }, + { + "epoch": 0.62, + "learning_rate": 4.6511981749603775e-05, + "logits/chosen": -1.8872562646865845, + "logits/rejected": -1.9487504959106445, + "logps/chosen": -178.33872985839844, + "logps/rejected": -182.2080535888672, + "loss": 0.7332, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.6403207778930664, + "rewards/margins": 0.12355762720108032, + "rewards/rejected": -0.7638784050941467, + "step": 473 + }, + { + "epoch": 0.62, + "learning_rate": 4.6493704692309175e-05, + "logits/chosen": -1.8873028755187988, + "logits/rejected": -1.8430054187774658, + "logps/chosen": -248.9535675048828, + "logps/rejected": -238.04327392578125, + "loss": 1.062, + "rewards/accuracies": 0.25, + "rewards/chosen": -1.2121449708938599, + "rewards/margins": -0.3689318299293518, + "rewards/rejected": -0.8432131409645081, + "step": 474 + }, + { + "epoch": 0.62, + "learning_rate": 4.647538348510189e-05, + "logits/chosen": -1.8361527919769287, + "logits/rejected": -1.856339454650879, + "logps/chosen": -171.12091064453125, + "logps/rejected": -179.4962158203125, + "loss": 0.6846, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.7720082998275757, + "rewards/margins": 0.10657864063978195, + "rewards/rejected": -0.8785868883132935, + "step": 475 + }, + { + "epoch": 0.62, + "learning_rate": 4.645701816561523e-05, + "logits/chosen": -1.6982722282409668, + "logits/rejected": -1.7370768785476685, + "logps/chosen": -232.54293823242188, + "logps/rejected": -213.66964721679688, + "loss": 0.7178, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.8362730741500854, + "rewards/margins": 0.07781472057104111, + "rewards/rejected": -0.9140878319740295, + "step": 476 + }, + { + "epoch": 0.62, + "learning_rate": 4.643860877157314e-05, + "logits/chosen": -1.7802523374557495, + "logits/rejected": -1.7304799556732178, + "logps/chosen": -168.3419189453125, + "logps/rejected": -205.67333984375, + "loss": 0.8153, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.4274996221065521, + "rewards/margins": -0.046173423528671265, + "rewards/rejected": -0.38132619857788086, + "step": 477 + }, + { + "epoch": 0.63, + "learning_rate": 4.642015534079012e-05, + "logits/chosen": -1.9037768840789795, + "logits/rejected": -1.8988232612609863, + "logps/chosen": -173.9936981201172, + "logps/rejected": -197.27523803710938, + "loss": 0.6135, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.32596248388290405, + "rewards/margins": 0.23917633295059204, + "rewards/rejected": -0.5651388168334961, + "step": 478 + }, + { + "epoch": 0.63, + "learning_rate": 4.640165791117106e-05, + "logits/chosen": -1.9618606567382812, + "logits/rejected": -1.9455369710922241, + "logps/chosen": -190.39830017089844, + "logps/rejected": -175.3238067626953, + "loss": 0.9091, + "rewards/accuracies": 0.5625, + "rewards/chosen": -1.0257893800735474, + "rewards/margins": -0.2109348475933075, + "rewards/rejected": -0.8148545622825623, + "step": 479 + }, + { + "epoch": 0.63, + "learning_rate": 4.63831165207113e-05, + "logits/chosen": -1.864621877670288, + "logits/rejected": -1.889084815979004, + "logps/chosen": -205.9281005859375, + "logps/rejected": -230.0751953125, + "loss": 0.7761, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.6850899457931519, + "rewards/margins": -0.05720193684101105, + "rewards/rejected": -0.6278879046440125, + "step": 480 + }, + { + "epoch": 0.63, + "learning_rate": 4.6364531207496426e-05, + "logits/chosen": -1.737329363822937, + "logits/rejected": -1.745915412902832, + "logps/chosen": -171.1361846923828, + "logps/rejected": -175.88906860351562, + "loss": 0.6837, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.743391215801239, + "rewards/margins": 0.0780097097158432, + "rewards/rejected": -0.8214008808135986, + "step": 481 + }, + { + "epoch": 0.63, + "learning_rate": 4.634590200970227e-05, + "logits/chosen": -1.8280831575393677, + "logits/rejected": -1.8323631286621094, + "logps/chosen": -188.1634521484375, + "logps/rejected": -213.10360717773438, + "loss": 0.7336, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.7827669382095337, + "rewards/margins": 0.3351660966873169, + "rewards/rejected": -1.1179330348968506, + "step": 482 + }, + { + "epoch": 0.63, + "learning_rate": 4.632722896559481e-05, + "logits/chosen": -1.9295848608016968, + "logits/rejected": -1.9276199340820312, + "logps/chosen": -167.31385803222656, + "logps/rejected": -186.1995391845703, + "loss": 0.6013, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.16915923357009888, + "rewards/margins": 0.3335033357143402, + "rewards/rejected": -0.5026625394821167, + "step": 483 + }, + { + "epoch": 0.63, + "learning_rate": 4.630851211353007e-05, + "logits/chosen": -1.6712524890899658, + "logits/rejected": -1.788968801498413, + "logps/chosen": -164.82725524902344, + "logps/rejected": -182.7891387939453, + "loss": 0.7642, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.609403669834137, + "rewards/margins": 0.03477644547820091, + "rewards/rejected": -0.644180178642273, + "step": 484 + }, + { + "epoch": 0.63, + "learning_rate": 4.628975149195407e-05, + "logits/chosen": -1.2296477556228638, + "logits/rejected": -1.2632193565368652, + "logps/chosen": -204.8614044189453, + "logps/rejected": -224.30543518066406, + "loss": 0.7234, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.7619196176528931, + "rewards/margins": 0.12807466089725494, + "rewards/rejected": -0.8899943232536316, + "step": 485 + }, + { + "epoch": 0.64, + "learning_rate": 4.6270947139402744e-05, + "logits/chosen": -2.047361373901367, + "logits/rejected": -2.1051080226898193, + "logps/chosen": -169.24703979492188, + "logps/rejected": -184.35586547851562, + "loss": 0.66, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.6050511002540588, + "rewards/margins": 0.13156384229660034, + "rewards/rejected": -0.7366149425506592, + "step": 486 + }, + { + "epoch": 0.64, + "learning_rate": 4.6252099094501834e-05, + "logits/chosen": -1.9163178205490112, + "logits/rejected": -1.8832037448883057, + "logps/chosen": -202.50064086914062, + "logps/rejected": -206.1787109375, + "loss": 0.8078, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.8862631320953369, + "rewards/margins": -0.06633087992668152, + "rewards/rejected": -0.819932222366333, + "step": 487 + }, + { + "epoch": 0.64, + "learning_rate": 4.623320739596685e-05, + "logits/chosen": -1.943336009979248, + "logits/rejected": -1.9594800472259521, + "logps/chosen": -184.20272827148438, + "logps/rejected": -185.5780029296875, + "loss": 0.948, + "rewards/accuracies": 0.1875, + "rewards/chosen": -0.8648966550827026, + "rewards/margins": -0.32853201031684875, + "rewards/rejected": -0.5363646745681763, + "step": 488 + }, + { + "epoch": 0.64, + "learning_rate": 4.621427208260296e-05, + "logits/chosen": -2.0543949604034424, + "logits/rejected": -2.09141206741333, + "logps/chosen": -186.11021423339844, + "logps/rejected": -197.07164001464844, + "loss": 0.6593, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.6588226556777954, + "rewards/margins": 0.23558923602104187, + "rewards/rejected": -0.8944119811058044, + "step": 489 + }, + { + "epoch": 0.64, + "learning_rate": 4.6195293193304915e-05, + "logits/chosen": -2.2013731002807617, + "logits/rejected": -2.209264039993286, + "logps/chosen": -192.5195770263672, + "logps/rejected": -188.9171600341797, + "loss": 0.8303, + "rewards/accuracies": 0.375, + "rewards/chosen": -0.8534584641456604, + "rewards/margins": -0.10596348345279694, + "rewards/rejected": -0.7474948763847351, + "step": 490 + }, + { + "epoch": 0.64, + "learning_rate": 4.6176270767056976e-05, + "logits/chosen": -1.8635625839233398, + "logits/rejected": -1.8899545669555664, + "logps/chosen": -193.61715698242188, + "logps/rejected": -196.3071746826172, + "loss": 0.5859, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.620331346988678, + "rewards/margins": 0.35552215576171875, + "rewards/rejected": -0.9758535027503967, + "step": 491 + }, + { + "epoch": 0.64, + "learning_rate": 4.615720484293286e-05, + "logits/chosen": -2.0970966815948486, + "logits/rejected": -2.0922045707702637, + "logps/chosen": -171.4237060546875, + "logps/rejected": -173.91969299316406, + "loss": 0.7777, + "rewards/accuracies": 0.5, + "rewards/chosen": -0.7394740581512451, + "rewards/margins": 0.09215141832828522, + "rewards/rejected": -0.8316254615783691, + "step": 492 + }, + { + "epoch": 0.65, + "learning_rate": 4.613809546009558e-05, + "logits/chosen": -1.923639178276062, + "logits/rejected": -1.9087320566177368, + "logps/chosen": -210.64447021484375, + "logps/rejected": -202.98309326171875, + "loss": 0.7005, + "rewards/accuracies": 0.6875, + "rewards/chosen": -0.7617734670639038, + "rewards/margins": 0.28139957785606384, + "rewards/rejected": -1.04317307472229, + "step": 493 + }, + { + "epoch": 0.65, + "learning_rate": 4.611894265779748e-05, + "logits/chosen": -1.7692898511886597, + "logits/rejected": -1.8441616296768188, + "logps/chosen": -181.05316162109375, + "logps/rejected": -190.46311950683594, + "loss": 0.8492, + "rewards/accuracies": 0.3125, + "rewards/chosen": -0.9923998117446899, + "rewards/margins": -0.16852112114429474, + "rewards/rejected": -0.8238787651062012, + "step": 494 + }, + { + "epoch": 0.65, + "learning_rate": 4.609974647538003e-05, + "logits/chosen": -2.242365837097168, + "logits/rejected": -2.2200753688812256, + "logps/chosen": -192.88491821289062, + "logps/rejected": -209.79190063476562, + "loss": 0.7824, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.7258377075195312, + "rewards/margins": 0.06173846498131752, + "rewards/rejected": -0.7875760793685913, + "step": 495 + }, + { + "epoch": 0.65, + "learning_rate": 4.608050695227385e-05, + "logits/chosen": -2.0390326976776123, + "logits/rejected": -2.0602505207061768, + "logps/chosen": -159.614013671875, + "logps/rejected": -159.90530395507812, + "loss": 0.6495, + "rewards/accuracies": 0.625, + "rewards/chosen": -0.6496396660804749, + "rewards/margins": 0.15987172722816467, + "rewards/rejected": -0.8095113635063171, + "step": 496 + }, + { + "epoch": 0.65, + "learning_rate": 4.606122412799857e-05, + "logits/chosen": -1.8621050119400024, + "logits/rejected": -1.843872308731079, + "logps/chosen": -191.3387451171875, + "logps/rejected": -212.04867553710938, + "loss": 0.838, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.9131224751472473, + "rewards/margins": -0.06589814275503159, + "rewards/rejected": -0.847224235534668, + "step": 497 + }, + { + "epoch": 0.65, + "learning_rate": 4.6041898042162764e-05, + "logits/chosen": -1.9165095090866089, + "logits/rejected": -1.9768743515014648, + "logps/chosen": -179.1850128173828, + "logps/rejected": -197.76953125, + "loss": 0.7127, + "rewards/accuracies": 0.5625, + "rewards/chosen": -0.8291003704071045, + "rewards/margins": 0.08120033144950867, + "rewards/rejected": -0.9103007316589355, + "step": 498 + }, + { + "epoch": 0.65, + "learning_rate": 4.602252873446386e-05, + "logits/chosen": -1.71052086353302, + "logits/rejected": -1.7267752885818481, + "logps/chosen": -233.17083740234375, + "logps/rejected": -238.11651611328125, + "loss": 0.7183, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.6033438444137573, + "rewards/margins": 0.14365322887897491, + "rewards/rejected": -0.7469971179962158, + "step": 499 + }, + { + "epoch": 0.65, + "learning_rate": 4.60031162446881e-05, + "logits/chosen": -1.685623049736023, + "logits/rejected": -1.759178876876831, + "logps/chosen": -179.87600708007812, + "logps/rejected": -183.2005615234375, + "loss": 0.7049, + "rewards/accuracies": 0.4375, + "rewards/chosen": -0.7123013734817505, + "rewards/margins": 0.11958488076925278, + "rewards/rejected": -0.8318862915039062, + "step": 500 + } + ], + "logging_steps": 1, + "max_steps": 2292, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 500, + "total_flos": 0.0, + "train_batch_size": 2, + "trial_name": null, + "trial_params": null +}