| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9997382884061764, | |
| "eval_steps": 500, | |
| "global_step": 1910, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0005234231876472127, | |
| "grad_norm": 14.640187733663568, | |
| "learning_rate": 2.617801047120419e-09, | |
| "logits/chosen": -2.959083080291748, | |
| "logits/rejected": -3.0469329357147217, | |
| "logps/chosen": -134.99343872070312, | |
| "logps/rejected": -161.30029296875, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": 0.0001932144077727571, | |
| "rewards/margins": 0.0005058574606664479, | |
| "rewards/rejected": -0.00031264303834177554, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.005234231876472127, | |
| "grad_norm": 14.254383428129056, | |
| "learning_rate": 2.6178010471204188e-08, | |
| "logits/chosen": -2.9816136360168457, | |
| "logits/rejected": -2.893190860748291, | |
| "logps/chosen": -186.0762176513672, | |
| "logps/rejected": -186.07012939453125, | |
| "loss": 0.6933, | |
| "rewards/accuracies": 0.4583333432674408, | |
| "rewards/chosen": 0.0004193554923404008, | |
| "rewards/margins": -0.00021269405260682106, | |
| "rewards/rejected": 0.0006320495158433914, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.010468463752944255, | |
| "grad_norm": 12.949573017221987, | |
| "learning_rate": 5.2356020942408376e-08, | |
| "logits/chosen": -2.829813003540039, | |
| "logits/rejected": -2.8478147983551025, | |
| "logps/chosen": -174.7886199951172, | |
| "logps/rejected": -193.36080932617188, | |
| "loss": 0.6932, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": -0.0009858906269073486, | |
| "rewards/margins": -0.0011966052697971463, | |
| "rewards/rejected": 0.00021071455557830632, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.015702695629416383, | |
| "grad_norm": 15.371844314146115, | |
| "learning_rate": 7.853403141361257e-08, | |
| "logits/chosen": -3.01656174659729, | |
| "logits/rejected": -2.9938042163848877, | |
| "logps/chosen": -214.79541015625, | |
| "logps/rejected": -201.54867553710938, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": -0.000996897229924798, | |
| "rewards/margins": 0.00021716937771998346, | |
| "rewards/rejected": -0.00121406652033329, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02093692750588851, | |
| "grad_norm": 14.22366978101997, | |
| "learning_rate": 1.0471204188481675e-07, | |
| "logits/chosen": -2.874187469482422, | |
| "logits/rejected": -2.886640787124634, | |
| "logps/chosen": -172.05311584472656, | |
| "logps/rejected": -216.9876251220703, | |
| "loss": 0.6925, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.002353773918002844, | |
| "rewards/margins": 0.0017049995949491858, | |
| "rewards/rejected": -0.004058773163706064, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.02617115938236064, | |
| "grad_norm": 13.356014407693477, | |
| "learning_rate": 1.3089005235602092e-07, | |
| "logits/chosen": -2.9540908336639404, | |
| "logits/rejected": -2.9672648906707764, | |
| "logps/chosen": -196.1494598388672, | |
| "logps/rejected": -180.84803771972656, | |
| "loss": 0.6919, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.00872573908418417, | |
| "rewards/margins": 0.002879455918446183, | |
| "rewards/rejected": -0.01160519476979971, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.031405391258832765, | |
| "grad_norm": 14.564330446769505, | |
| "learning_rate": 1.5706806282722514e-07, | |
| "logits/chosen": -2.9785428047180176, | |
| "logits/rejected": -2.9725818634033203, | |
| "logps/chosen": -183.6147918701172, | |
| "logps/rejected": -203.66824340820312, | |
| "loss": 0.6902, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.01978887803852558, | |
| "rewards/margins": 0.005196017678827047, | |
| "rewards/rejected": -0.024984892457723618, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.036639623135304895, | |
| "grad_norm": 16.449869342928505, | |
| "learning_rate": 1.8324607329842932e-07, | |
| "logits/chosen": -2.995837688446045, | |
| "logits/rejected": -2.8539786338806152, | |
| "logps/chosen": -197.07168579101562, | |
| "logps/rejected": -206.65469360351562, | |
| "loss": 0.6886, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.02455015294253826, | |
| "rewards/margins": 0.011808360926806927, | |
| "rewards/rejected": -0.03635851666331291, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04187385501177702, | |
| "grad_norm": 14.79926563160046, | |
| "learning_rate": 2.094240837696335e-07, | |
| "logits/chosen": -2.9914395809173584, | |
| "logits/rejected": -2.942915439605713, | |
| "logps/chosen": -205.7972869873047, | |
| "logps/rejected": -192.72232055664062, | |
| "loss": 0.6881, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.04919552057981491, | |
| "rewards/margins": 0.009290899150073528, | |
| "rewards/rejected": -0.058486420661211014, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04710808688824915, | |
| "grad_norm": 14.05004280238572, | |
| "learning_rate": 2.356020942408377e-07, | |
| "logits/chosen": -2.888996124267578, | |
| "logits/rejected": -2.804115056991577, | |
| "logps/chosen": -184.47732543945312, | |
| "logps/rejected": -172.6226348876953, | |
| "loss": 0.6844, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.07440824806690216, | |
| "rewards/margins": 0.013659747317433357, | |
| "rewards/rejected": -0.08806798607110977, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05234231876472128, | |
| "grad_norm": 13.559496262792273, | |
| "learning_rate": 2.6178010471204185e-07, | |
| "logits/chosen": -2.9032299518585205, | |
| "logits/rejected": -2.857269525527954, | |
| "logps/chosen": -206.85140991210938, | |
| "logps/rejected": -219.37026977539062, | |
| "loss": 0.6773, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.0838988870382309, | |
| "rewards/margins": 0.028635507449507713, | |
| "rewards/rejected": -0.11253438144922256, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05757655064119341, | |
| "grad_norm": 13.4996248822944, | |
| "learning_rate": 2.879581151832461e-07, | |
| "logits/chosen": -2.940826892852783, | |
| "logits/rejected": -2.8834738731384277, | |
| "logps/chosen": -212.55807495117188, | |
| "logps/rejected": -201.2547607421875, | |
| "loss": 0.675, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.10653187334537506, | |
| "rewards/margins": 0.05816306918859482, | |
| "rewards/rejected": -0.16469493508338928, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06281078251766553, | |
| "grad_norm": 12.884354544368191, | |
| "learning_rate": 3.1413612565445027e-07, | |
| "logits/chosen": -2.927412986755371, | |
| "logits/rejected": -2.985417604446411, | |
| "logps/chosen": -228.93350219726562, | |
| "logps/rejected": -238.04269409179688, | |
| "loss": 0.6747, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.10475633293390274, | |
| "rewards/margins": 0.046889565885066986, | |
| "rewards/rejected": -0.15164589881896973, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06804501439413765, | |
| "grad_norm": 14.312004568183767, | |
| "learning_rate": 3.4031413612565446e-07, | |
| "logits/chosen": -2.8365414142608643, | |
| "logits/rejected": -2.876112461090088, | |
| "logps/chosen": -200.61264038085938, | |
| "logps/rejected": -213.1061248779297, | |
| "loss": 0.6622, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.13382001221179962, | |
| "rewards/margins": 0.06009231135249138, | |
| "rewards/rejected": -0.1939123272895813, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07327924627060979, | |
| "grad_norm": 15.573435236138954, | |
| "learning_rate": 3.6649214659685864e-07, | |
| "logits/chosen": -2.8712472915649414, | |
| "logits/rejected": -2.8083159923553467, | |
| "logps/chosen": -212.3485870361328, | |
| "logps/rejected": -216.5910186767578, | |
| "loss": 0.659, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.2070785015821457, | |
| "rewards/margins": 0.06957153975963593, | |
| "rewards/rejected": -0.27665001153945923, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07851347814708191, | |
| "grad_norm": 17.962582071011855, | |
| "learning_rate": 3.926701570680628e-07, | |
| "logits/chosen": -2.8894333839416504, | |
| "logits/rejected": -2.783268928527832, | |
| "logps/chosen": -211.5115509033203, | |
| "logps/rejected": -224.9191131591797, | |
| "loss": 0.6492, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.206833153963089, | |
| "rewards/margins": 0.1691361963748932, | |
| "rewards/rejected": -0.3759693503379822, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08374771002355404, | |
| "grad_norm": 17.592745804055944, | |
| "learning_rate": 4.18848167539267e-07, | |
| "logits/chosen": -2.866983413696289, | |
| "logits/rejected": -2.766951084136963, | |
| "logps/chosen": -217.34207153320312, | |
| "logps/rejected": -228.534912109375, | |
| "loss": 0.6384, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.25625622272491455, | |
| "rewards/margins": 0.15949755907058716, | |
| "rewards/rejected": -0.4157537817955017, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.08898194190002617, | |
| "grad_norm": 17.652878273699034, | |
| "learning_rate": 4.450261780104712e-07, | |
| "logits/chosen": -2.8391916751861572, | |
| "logits/rejected": -2.832740306854248, | |
| "logps/chosen": -199.69781494140625, | |
| "logps/rejected": -227.84652709960938, | |
| "loss": 0.6301, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.13288532197475433, | |
| "rewards/margins": 0.19410356879234314, | |
| "rewards/rejected": -0.32698890566825867, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0942161737764983, | |
| "grad_norm": 17.777586322420262, | |
| "learning_rate": 4.712041884816754e-07, | |
| "logits/chosen": -2.715904951095581, | |
| "logits/rejected": -2.5992846488952637, | |
| "logps/chosen": -210.23941040039062, | |
| "logps/rejected": -229.0234832763672, | |
| "loss": 0.6307, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.21204690635204315, | |
| "rewards/margins": 0.17581574618816376, | |
| "rewards/rejected": -0.3878627121448517, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.09945040565297043, | |
| "grad_norm": 28.072972397122147, | |
| "learning_rate": 4.973821989528796e-07, | |
| "logits/chosen": -2.6811397075653076, | |
| "logits/rejected": -2.6616740226745605, | |
| "logps/chosen": -216.45114135742188, | |
| "logps/rejected": -258.03021240234375, | |
| "loss": 0.6089, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.33117493987083435, | |
| "rewards/margins": 0.33262476325035095, | |
| "rewards/rejected": -0.6637996435165405, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.10468463752944256, | |
| "grad_norm": 29.910313896117426, | |
| "learning_rate": 4.999661831436498e-07, | |
| "logits/chosen": -2.5953736305236816, | |
| "logits/rejected": -2.6287741661071777, | |
| "logps/chosen": -201.4033966064453, | |
| "logps/rejected": -260.02728271484375, | |
| "loss": 0.6132, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.2563866972923279, | |
| "rewards/margins": 0.3042460083961487, | |
| "rewards/rejected": -0.5606327652931213, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.10991886940591468, | |
| "grad_norm": 35.27450975604578, | |
| "learning_rate": 4.998492971140339e-07, | |
| "logits/chosen": -2.543915271759033, | |
| "logits/rejected": -2.3116703033447266, | |
| "logps/chosen": -247.22891235351562, | |
| "logps/rejected": -270.3451843261719, | |
| "loss": 0.6417, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.595511794090271, | |
| "rewards/margins": 0.2611444592475891, | |
| "rewards/rejected": -0.8566562533378601, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.11515310128238682, | |
| "grad_norm": 48.58365341922755, | |
| "learning_rate": 4.996489634487865e-07, | |
| "logits/chosen": -2.545841932296753, | |
| "logits/rejected": -2.2179739475250244, | |
| "logps/chosen": -234.15786743164062, | |
| "logps/rejected": -287.24920654296875, | |
| "loss": 0.585, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.533110499382019, | |
| "rewards/margins": 0.5109399557113647, | |
| "rewards/rejected": -1.0440504550933838, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.12038733315885894, | |
| "grad_norm": 75.83009735431122, | |
| "learning_rate": 4.993652490577246e-07, | |
| "logits/chosen": -2.5557217597961426, | |
| "logits/rejected": -2.3648390769958496, | |
| "logps/chosen": -249.6676483154297, | |
| "logps/rejected": -277.823486328125, | |
| "loss": 0.6025, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.5915195941925049, | |
| "rewards/margins": 0.36318284273147583, | |
| "rewards/rejected": -0.9547024965286255, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.12562156503533106, | |
| "grad_norm": 33.52383186890237, | |
| "learning_rate": 4.9899824869915e-07, | |
| "logits/chosen": -2.333035945892334, | |
| "logits/rejected": -2.079817295074463, | |
| "logps/chosen": -255.44595336914062, | |
| "logps/rejected": -313.8713684082031, | |
| "loss": 0.5681, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.8247841000556946, | |
| "rewards/margins": 0.47075214982032776, | |
| "rewards/rejected": -1.2955362796783447, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.13085579691180318, | |
| "grad_norm": 35.13466621709411, | |
| "learning_rate": 4.985480849482012e-07, | |
| "logits/chosen": -2.3787760734558105, | |
| "logits/rejected": -2.364150047302246, | |
| "logps/chosen": -304.33441162109375, | |
| "logps/rejected": -327.4510498046875, | |
| "loss": 0.5645, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -1.0099525451660156, | |
| "rewards/margins": 0.29796892404556274, | |
| "rewards/rejected": -1.3079214096069336, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1360900287882753, | |
| "grad_norm": 33.359619674605376, | |
| "learning_rate": 4.980149081559142e-07, | |
| "logits/chosen": -2.2392945289611816, | |
| "logits/rejected": -2.0072715282440186, | |
| "logps/chosen": -299.57879638671875, | |
| "logps/rejected": -395.2071838378906, | |
| "loss": 0.5998, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.3633458614349365, | |
| "rewards/margins": 0.728509247303009, | |
| "rewards/rejected": -2.09185528755188, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.14132426066474746, | |
| "grad_norm": 47.681686539836825, | |
| "learning_rate": 4.973988963990065e-07, | |
| "logits/chosen": -2.056598424911499, | |
| "logits/rejected": -2.0986738204956055, | |
| "logps/chosen": -318.61956787109375, | |
| "logps/rejected": -412.1875915527344, | |
| "loss": 0.548, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.372604489326477, | |
| "rewards/margins": 0.7224138975143433, | |
| "rewards/rejected": -2.095018148422241, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.14655849254121958, | |
| "grad_norm": 30.114115223768707, | |
| "learning_rate": 4.967002554204008e-07, | |
| "logits/chosen": -2.0974724292755127, | |
| "logits/rejected": -2.0098748207092285, | |
| "logps/chosen": -350.09600830078125, | |
| "logps/rejected": -403.1016540527344, | |
| "loss": 0.5879, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -1.6619269847869873, | |
| "rewards/margins": 0.39111390709877014, | |
| "rewards/rejected": -2.0530409812927246, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.1517927244176917, | |
| "grad_norm": 31.769810828425065, | |
| "learning_rate": 4.959192185605087e-07, | |
| "logits/chosen": -2.51796817779541, | |
| "logits/rejected": -2.2334158420562744, | |
| "logps/chosen": -324.0982360839844, | |
| "logps/rejected": -340.4048767089844, | |
| "loss": 0.5945, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -1.1041514873504639, | |
| "rewards/margins": 0.38266053795814514, | |
| "rewards/rejected": -1.486811876296997, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.15702695629416383, | |
| "grad_norm": 48.774597859411024, | |
| "learning_rate": 4.950560466792969e-07, | |
| "logits/chosen": -2.362421751022339, | |
| "logits/rejected": -2.193986654281616, | |
| "logps/chosen": -305.4976806640625, | |
| "logps/rejected": -367.10467529296875, | |
| "loss": 0.5823, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.2003720998764038, | |
| "rewards/margins": 0.572384774684906, | |
| "rewards/rejected": -1.7727569341659546, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.16226118817063595, | |
| "grad_norm": 39.19160116710573, | |
| "learning_rate": 4.941110280691619e-07, | |
| "logits/chosen": -2.353766441345215, | |
| "logits/rejected": -2.0096635818481445, | |
| "logps/chosen": -309.58905029296875, | |
| "logps/rejected": -353.7257080078125, | |
| "loss": 0.5836, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -1.201560378074646, | |
| "rewards/margins": 0.48785895109176636, | |
| "rewards/rejected": -1.6894193887710571, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.16749542004710807, | |
| "grad_norm": 37.587189222436905, | |
| "learning_rate": 4.930844783586424e-07, | |
| "logits/chosen": -2.396035671234131, | |
| "logits/rejected": -2.165098190307617, | |
| "logps/chosen": -288.3567810058594, | |
| "logps/rejected": -332.8038635253906, | |
| "loss": 0.5415, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.062434196472168, | |
| "rewards/margins": 0.4423864781856537, | |
| "rewards/rejected": -1.5048205852508545, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.17272965192358022, | |
| "grad_norm": 50.8199153591931, | |
| "learning_rate": 4.919767404070033e-07, | |
| "logits/chosen": -1.8470354080200195, | |
| "logits/rejected": -1.600237250328064, | |
| "logps/chosen": -307.25360107421875, | |
| "logps/rejected": -398.39910888671875, | |
| "loss": 0.5536, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.3850140571594238, | |
| "rewards/margins": 0.7685943245887756, | |
| "rewards/rejected": -2.153608560562134, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.17796388380005235, | |
| "grad_norm": 39.2860888155997, | |
| "learning_rate": 4.907881841897216e-07, | |
| "logits/chosen": -2.0841307640075684, | |
| "logits/rejected": -1.8823187351226807, | |
| "logps/chosen": -306.69305419921875, | |
| "logps/rejected": -375.80352783203125, | |
| "loss": 0.5377, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.2261284589767456, | |
| "rewards/margins": 0.7133340835571289, | |
| "rewards/rejected": -1.939462423324585, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.18319811567652447, | |
| "grad_norm": 44.441009515641795, | |
| "learning_rate": 4.895192066749189e-07, | |
| "logits/chosen": -2.40903902053833, | |
| "logits/rejected": -2.2892374992370605, | |
| "logps/chosen": -299.1591796875, | |
| "logps/rejected": -401.603271484375, | |
| "loss": 0.5481, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.1807595491409302, | |
| "rewards/margins": 0.7055211663246155, | |
| "rewards/rejected": -1.8862807750701904, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.1884323475529966, | |
| "grad_norm": 33.989270442301404, | |
| "learning_rate": 4.881702316907768e-07, | |
| "logits/chosen": -2.2026100158691406, | |
| "logits/rejected": -2.008274793624878, | |
| "logps/chosen": -290.2915344238281, | |
| "logps/rejected": -346.9622497558594, | |
| "loss": 0.5596, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -1.1741074323654175, | |
| "rewards/margins": 0.5104838609695435, | |
| "rewards/rejected": -1.68459153175354, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.19366657942946872, | |
| "grad_norm": 39.74252231849878, | |
| "learning_rate": 4.86741709783982e-07, | |
| "logits/chosen": -2.0070841312408447, | |
| "logits/rejected": -1.5279141664505005, | |
| "logps/chosen": -293.36578369140625, | |
| "logps/rejected": -345.587646484375, | |
| "loss": 0.5468, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -1.2320070266723633, | |
| "rewards/margins": 0.6169141530990601, | |
| "rewards/rejected": -1.8489214181900024, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.19890081130594087, | |
| "grad_norm": 41.66915169536549, | |
| "learning_rate": 4.85234118069247e-07, | |
| "logits/chosen": -1.4260574579238892, | |
| "logits/rejected": -1.1182783842086792, | |
| "logps/chosen": -323.9841003417969, | |
| "logps/rejected": -404.05517578125, | |
| "loss": 0.5282, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.595336675643921, | |
| "rewards/margins": 0.6996889114379883, | |
| "rewards/rejected": -2.295025587081909, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.204135043182413, | |
| "grad_norm": 38.40542199265962, | |
| "learning_rate": 4.836479600699578e-07, | |
| "logits/chosen": -1.7142446041107178, | |
| "logits/rejected": -1.7196638584136963, | |
| "logps/chosen": -361.3770446777344, | |
| "logps/rejected": -439.6658630371094, | |
| "loss": 0.5487, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.594753623008728, | |
| "rewards/margins": 0.683542788028717, | |
| "rewards/rejected": -2.27829647064209, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.2093692750588851, | |
| "grad_norm": 39.63693507119565, | |
| "learning_rate": 4.819837655500013e-07, | |
| "logits/chosen": -1.911871314048767, | |
| "logits/rejected": -1.7180120944976807, | |
| "logps/chosen": -339.29705810546875, | |
| "logps/rejected": -399.7021179199219, | |
| "loss": 0.54, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.528294563293457, | |
| "rewards/margins": 0.6349833011627197, | |
| "rewards/rejected": -2.163278102874756, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.21460350693535724, | |
| "grad_norm": 39.95185592857185, | |
| "learning_rate": 4.802420903368285e-07, | |
| "logits/chosen": -1.7220875024795532, | |
| "logits/rejected": -1.3575185537338257, | |
| "logps/chosen": -365.69964599609375, | |
| "logps/rejected": -418.9811096191406, | |
| "loss": 0.5823, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.7411553859710693, | |
| "rewards/margins": 0.48729467391967773, | |
| "rewards/rejected": -2.228450059890747, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.21983773881182936, | |
| "grad_norm": 42.46558644572104, | |
| "learning_rate": 4.784235161358123e-07, | |
| "logits/chosen": -1.8128083944320679, | |
| "logits/rejected": -1.8185402154922485, | |
| "logps/chosen": -309.54937744140625, | |
| "logps/rejected": -385.9242248535156, | |
| "loss": 0.5788, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.3886371850967407, | |
| "rewards/margins": 0.5779408812522888, | |
| "rewards/rejected": -1.9665781259536743, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.22507197068830148, | |
| "grad_norm": 40.391730044431704, | |
| "learning_rate": 4.7652865033596314e-07, | |
| "logits/chosen": -2.0464603900909424, | |
| "logits/rejected": -1.5721585750579834, | |
| "logps/chosen": -317.18560791015625, | |
| "logps/rejected": -354.8387756347656, | |
| "loss": 0.5339, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.305790662765503, | |
| "rewards/margins": 0.5792006254196167, | |
| "rewards/rejected": -1.8849912881851196, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.23030620256477363, | |
| "grad_norm": 44.15647862665091, | |
| "learning_rate": 4.7455812580706534e-07, | |
| "logits/chosen": -2.0872130393981934, | |
| "logits/rejected": -1.5626317262649536, | |
| "logps/chosen": -345.29278564453125, | |
| "logps/rejected": -409.45989990234375, | |
| "loss": 0.549, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.5832579135894775, | |
| "rewards/margins": 0.6660935282707214, | |
| "rewards/rejected": -2.2493512630462646, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.23554043444124576, | |
| "grad_norm": 42.04086790745509, | |
| "learning_rate": 4.725126006883046e-07, | |
| "logits/chosen": -1.773993730545044, | |
| "logits/rejected": -1.5137646198272705, | |
| "logps/chosen": -326.9532165527344, | |
| "logps/rejected": -401.57867431640625, | |
| "loss": 0.577, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.5454232692718506, | |
| "rewards/margins": 0.6505107879638672, | |
| "rewards/rejected": -2.1959340572357178, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.24077466631771788, | |
| "grad_norm": 36.31044897637686, | |
| "learning_rate": 4.703927581684539e-07, | |
| "logits/chosen": -2.1727023124694824, | |
| "logits/rejected": -1.8378498554229736, | |
| "logps/chosen": -350.1428527832031, | |
| "logps/rejected": -373.5480041503906, | |
| "loss": 0.5687, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.4619617462158203, | |
| "rewards/margins": 0.49050265550613403, | |
| "rewards/rejected": -1.9524643421173096, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.24600889819419, | |
| "grad_norm": 38.97906947550607, | |
| "learning_rate": 4.68199306257695e-07, | |
| "logits/chosen": -1.960343599319458, | |
| "logits/rejected": -1.3587312698364258, | |
| "logps/chosen": -331.9888916015625, | |
| "logps/rejected": -399.2967834472656, | |
| "loss": 0.4892, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.4141899347305298, | |
| "rewards/margins": 0.7412781715393066, | |
| "rewards/rejected": -2.155468225479126, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.2512431300706621, | |
| "grad_norm": 60.96840256666392, | |
| "learning_rate": 4.6593297755114776e-07, | |
| "logits/chosen": -1.4715737104415894, | |
| "logits/rejected": -0.8570035099983215, | |
| "logps/chosen": -363.36376953125, | |
| "logps/rejected": -419.7437438964844, | |
| "loss": 0.5466, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.9223047494888306, | |
| "rewards/margins": 0.5888181328773499, | |
| "rewards/rejected": -2.511122941970825, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.2564773619471343, | |
| "grad_norm": 56.85832650681642, | |
| "learning_rate": 4.635945289841902e-07, | |
| "logits/chosen": -1.736699104309082, | |
| "logits/rejected": -1.6159862279891968, | |
| "logps/chosen": -331.16339111328125, | |
| "logps/rejected": -429.3055114746094, | |
| "loss": 0.5324, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.511039137840271, | |
| "rewards/margins": 0.6905651092529297, | |
| "rewards/rejected": -2.2016043663024902, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.26171159382360637, | |
| "grad_norm": 45.91788216290567, | |
| "learning_rate": 4.611847415796476e-07, | |
| "logits/chosen": -1.6824716329574585, | |
| "logits/rejected": -1.536476731300354, | |
| "logps/chosen": -365.19488525390625, | |
| "logps/rejected": -471.90985107421875, | |
| "loss": 0.5294, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.727085828781128, | |
| "rewards/margins": 0.8007200956344604, | |
| "rewards/rejected": -2.527806043624878, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.26171159382360637, | |
| "eval_logits/chosen": -0.863903820514679, | |
| "eval_logits/rejected": -0.4723878502845764, | |
| "eval_logps/chosen": -358.6011047363281, | |
| "eval_logps/rejected": -428.4360656738281, | |
| "eval_loss": 0.5469957590103149, | |
| "eval_rewards/accuracies": 0.6980000138282776, | |
| "eval_rewards/chosen": -1.7358005046844482, | |
| "eval_rewards/margins": 0.6566510796546936, | |
| "eval_rewards/rejected": -2.392451763153076, | |
| "eval_runtime": 316.8135, | |
| "eval_samples_per_second": 6.313, | |
| "eval_steps_per_second": 0.789, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.2669458257000785, | |
| "grad_norm": 37.615935219623054, | |
| "learning_rate": 4.5870442018693773e-07, | |
| "logits/chosen": -1.8111470937728882, | |
| "logits/rejected": -1.5349056720733643, | |
| "logps/chosen": -327.2267761230469, | |
| "logps/rejected": -381.962890625, | |
| "loss": 0.5606, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -1.547119140625, | |
| "rewards/margins": 0.5296986699104309, | |
| "rewards/rejected": -2.076817750930786, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.2721800575765506, | |
| "grad_norm": 35.85517663702724, | |
| "learning_rate": 4.5615439321325735e-07, | |
| "logits/chosen": -2.0327141284942627, | |
| "logits/rejected": -1.460468053817749, | |
| "logps/chosen": -312.06671142578125, | |
| "logps/rejected": -395.5126647949219, | |
| "loss": 0.5241, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -1.3687623739242554, | |
| "rewards/margins": 0.6768960952758789, | |
| "rewards/rejected": -2.045658588409424, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.27741428945302277, | |
| "grad_norm": 50.37074560251419, | |
| "learning_rate": 4.535355123469008e-07, | |
| "logits/chosen": -1.4753310680389404, | |
| "logits/rejected": -1.471920132637024, | |
| "logps/chosen": -338.68060302734375, | |
| "logps/rejected": -424.1499938964844, | |
| "loss": 0.545, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -1.7495324611663818, | |
| "rewards/margins": 0.5848132371902466, | |
| "rewards/rejected": -2.334345579147339, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.2826485213294949, | |
| "grad_norm": 55.36008562841294, | |
| "learning_rate": 4.5084865227280366e-07, | |
| "logits/chosen": -1.087989091873169, | |
| "logits/rejected": -1.2594499588012695, | |
| "logps/chosen": -424.70941162109375, | |
| "logps/rejected": -481.74688720703125, | |
| "loss": 0.5205, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -2.1392130851745605, | |
| "rewards/margins": 0.6303030252456665, | |
| "rewards/rejected": -2.7695162296295166, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.287882753205967, | |
| "grad_norm": 39.75849189232047, | |
| "learning_rate": 4.4809471038040437e-07, | |
| "logits/chosen": -1.642165184020996, | |
| "logits/rejected": -0.9477740526199341, | |
| "logps/chosen": -399.1435546875, | |
| "logps/rejected": -444.47882080078125, | |
| "loss": 0.5659, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.8148092031478882, | |
| "rewards/margins": 0.7540385127067566, | |
| "rewards/rejected": -2.568847894668579, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.29311698508243916, | |
| "grad_norm": 33.49141951717591, | |
| "learning_rate": 4.4527460646392386e-07, | |
| "logits/chosen": -2.0694565773010254, | |
| "logits/rejected": -1.7511459589004517, | |
| "logps/chosen": -310.82574462890625, | |
| "logps/rejected": -369.46539306640625, | |
| "loss": 0.5329, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -1.1495193243026733, | |
| "rewards/margins": 0.6521326303482056, | |
| "rewards/rejected": -1.801652193069458, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.29835121695891126, | |
| "grad_norm": 56.88589774049656, | |
| "learning_rate": 4.4238928241516163e-07, | |
| "logits/chosen": -1.8898957967758179, | |
| "logits/rejected": -1.2891989946365356, | |
| "logps/chosen": -350.3455810546875, | |
| "logps/rejected": -402.1921081542969, | |
| "loss": 0.5373, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.413482427597046, | |
| "rewards/margins": 0.6170186996459961, | |
| "rewards/rejected": -2.030501127243042, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.3035854488353834, | |
| "grad_norm": 42.10520204001843, | |
| "learning_rate": 4.394397019089116e-07, | |
| "logits/chosen": -1.6958551406860352, | |
| "logits/rejected": -0.6804537773132324, | |
| "logps/chosen": -387.3909606933594, | |
| "logps/rejected": -451.7494201660156, | |
| "loss": 0.544, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.97348952293396, | |
| "rewards/margins": 0.7865778207778931, | |
| "rewards/rejected": -2.7600669860839844, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.30881968071185556, | |
| "grad_norm": 35.07476918232387, | |
| "learning_rate": 4.3642685008110246e-07, | |
| "logits/chosen": -1.4044368267059326, | |
| "logits/rejected": -0.7555860280990601, | |
| "logps/chosen": -428.0267639160156, | |
| "logps/rejected": -496.0921936035156, | |
| "loss": 0.5421, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -2.2372324466705322, | |
| "rewards/margins": 0.8896536827087402, | |
| "rewards/rejected": -3.1268858909606934, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.31405391258832765, | |
| "grad_norm": 56.64712231153677, | |
| "learning_rate": 4.333517331997704e-07, | |
| "logits/chosen": -1.425999641418457, | |
| "logits/rejected": -0.7339752912521362, | |
| "logps/chosen": -357.58050537109375, | |
| "logps/rejected": -459.1241149902344, | |
| "loss": 0.5675, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.811273217201233, | |
| "rewards/margins": 0.9072386622428894, | |
| "rewards/rejected": -2.7185120582580566, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.3192881444647998, | |
| "grad_norm": 53.88796332939181, | |
| "learning_rate": 4.302153783289736e-07, | |
| "logits/chosen": -1.348676085472107, | |
| "logits/rejected": -0.9413366317749023, | |
| "logps/chosen": -354.7875061035156, | |
| "logps/rejected": -408.4991149902344, | |
| "loss": 0.5262, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.5594650506973267, | |
| "rewards/margins": 0.6488285064697266, | |
| "rewards/rejected": -2.2082934379577637, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.3245223763412719, | |
| "grad_norm": 52.50744960152232, | |
| "learning_rate": 4.2701883298576124e-07, | |
| "logits/chosen": -1.720445990562439, | |
| "logits/rejected": -1.290917992591858, | |
| "logps/chosen": -375.7829284667969, | |
| "logps/rejected": -434.217529296875, | |
| "loss": 0.5917, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.6508047580718994, | |
| "rewards/margins": 0.5466897487640381, | |
| "rewards/rejected": -2.1974947452545166, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.32975660821774405, | |
| "grad_norm": 44.222624438240096, | |
| "learning_rate": 4.237631647903115e-07, | |
| "logits/chosen": -1.4673423767089844, | |
| "logits/rejected": -0.8496716618537903, | |
| "logps/chosen": -377.01470947265625, | |
| "logps/rejected": -444.0126953125, | |
| "loss": 0.5623, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -1.9182418584823608, | |
| "rewards/margins": 0.6217330694198608, | |
| "rewards/rejected": -2.539975166320801, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.33499084009421615, | |
| "grad_norm": 56.74613947557357, | |
| "learning_rate": 4.204494611093548e-07, | |
| "logits/chosen": -1.0175368785858154, | |
| "logits/rejected": -1.0307013988494873, | |
| "logps/chosen": -352.26666259765625, | |
| "logps/rejected": -444.2943420410156, | |
| "loss": 0.5419, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -2.003497362136841, | |
| "rewards/margins": 0.6112686395645142, | |
| "rewards/rejected": -2.6147658824920654, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.3402250719706883, | |
| "grad_norm": 56.35474765839724, | |
| "learning_rate": 4.1707882869300235e-07, | |
| "logits/chosen": -1.4247758388519287, | |
| "logits/rejected": -0.8727043271064758, | |
| "logps/chosen": -357.51800537109375, | |
| "logps/rejected": -452.87677001953125, | |
| "loss": 0.526, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -1.8759056329727173, | |
| "rewards/margins": 0.9082478284835815, | |
| "rewards/rejected": -2.784153699874878, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.34545930384716045, | |
| "grad_norm": 51.29897875137078, | |
| "learning_rate": 4.136523933051005e-07, | |
| "logits/chosen": -1.3449084758758545, | |
| "logits/rejected": -1.0442228317260742, | |
| "logps/chosen": -372.7139587402344, | |
| "logps/rejected": -517.5125732421875, | |
| "loss": 0.5216, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -2.0670166015625, | |
| "rewards/margins": 1.0036545991897583, | |
| "rewards/rejected": -3.0706706047058105, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.35069353572363254, | |
| "grad_norm": 53.649661826724454, | |
| "learning_rate": 4.101712993472348e-07, | |
| "logits/chosen": -1.49883234500885, | |
| "logits/rejected": -1.0886845588684082, | |
| "logps/chosen": -370.93817138671875, | |
| "logps/rejected": -424.97064208984375, | |
| "loss": 0.5716, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.9939110279083252, | |
| "rewards/margins": 0.5497623085975647, | |
| "rewards/rejected": -2.543673276901245, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.3559277676001047, | |
| "grad_norm": 44.18426608864023, | |
| "learning_rate": 4.066367094765091e-07, | |
| "logits/chosen": -1.3670942783355713, | |
| "logits/rejected": -1.2187774181365967, | |
| "logps/chosen": -374.8455810546875, | |
| "logps/rejected": -434.95867919921875, | |
| "loss": 0.535, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.9276262521743774, | |
| "rewards/margins": 0.5358133912086487, | |
| "rewards/rejected": -2.463439464569092, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.3611619994765768, | |
| "grad_norm": 52.799712890225976, | |
| "learning_rate": 4.0304980421722766e-07, | |
| "logits/chosen": -1.5443974733352661, | |
| "logits/rejected": -0.7440096139907837, | |
| "logps/chosen": -365.8990173339844, | |
| "logps/rejected": -443.71368408203125, | |
| "loss": 0.5022, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.7589327096939087, | |
| "rewards/margins": 0.8254102468490601, | |
| "rewards/rejected": -2.5843429565429688, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.36639623135304894, | |
| "grad_norm": 51.1426156745725, | |
| "learning_rate": 3.994117815666095e-07, | |
| "logits/chosen": -1.7846711874008179, | |
| "logits/rejected": -0.7315829992294312, | |
| "logps/chosen": -387.86376953125, | |
| "logps/rejected": -462.34503173828125, | |
| "loss": 0.5186, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.9715665578842163, | |
| "rewards/margins": 0.752856969833374, | |
| "rewards/rejected": -2.724423885345459, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.3716304632295211, | |
| "grad_norm": 62.36681697005891, | |
| "learning_rate": 3.957238565946671e-07, | |
| "logits/chosen": -1.2870978116989136, | |
| "logits/rejected": -0.7070925235748291, | |
| "logps/chosen": -403.86444091796875, | |
| "logps/rejected": -515.5906982421875, | |
| "loss": 0.483, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -2.2011804580688477, | |
| "rewards/margins": 1.022341012954712, | |
| "rewards/rejected": -3.2235217094421387, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.3768646951059932, | |
| "grad_norm": 57.328832948777865, | |
| "learning_rate": 3.9198726103838306e-07, | |
| "logits/chosen": -1.2161122560501099, | |
| "logits/rejected": -0.788101315498352, | |
| "logps/chosen": -387.68768310546875, | |
| "logps/rejected": -519.7088012695312, | |
| "loss": 0.5069, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -1.9085991382598877, | |
| "rewards/margins": 1.1666193008422852, | |
| "rewards/rejected": -3.0752182006835938, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.38209892698246534, | |
| "grad_norm": 68.04530741097648, | |
| "learning_rate": 3.8820324289031946e-07, | |
| "logits/chosen": -1.356323003768921, | |
| "logits/rejected": -1.176628828048706, | |
| "logps/chosen": -346.1966247558594, | |
| "logps/rejected": -447.4593811035156, | |
| "loss": 0.5455, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.8683630228042603, | |
| "rewards/margins": 0.8649228811264038, | |
| "rewards/rejected": -2.733285903930664, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.38733315885893743, | |
| "grad_norm": 39.4868442275961, | |
| "learning_rate": 3.84373065981799e-07, | |
| "logits/chosen": -1.5190155506134033, | |
| "logits/rejected": -1.4792649745941162, | |
| "logps/chosen": -346.9692077636719, | |
| "logps/rejected": -456.06634521484375, | |
| "loss": 0.4934, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -1.7277275323867798, | |
| "rewards/margins": 0.8931981921195984, | |
| "rewards/rejected": -2.6209259033203125, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.3925673907354096, | |
| "grad_norm": 63.695440464895746, | |
| "learning_rate": 3.8049800956079545e-07, | |
| "logits/chosen": -1.434788703918457, | |
| "logits/rejected": -1.0202864408493042, | |
| "logps/chosen": -409.87701416015625, | |
| "logps/rejected": -506.40179443359375, | |
| "loss": 0.492, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -2.0980966091156006, | |
| "rewards/margins": 0.8423215746879578, | |
| "rewards/rejected": -2.940418243408203, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.39780162261188173, | |
| "grad_norm": 74.4694029881303, | |
| "learning_rate": 3.7657936786467525e-07, | |
| "logits/chosen": -1.2294615507125854, | |
| "logits/rejected": -0.37707972526550293, | |
| "logps/chosen": -439.51025390625, | |
| "logps/rejected": -541.850830078125, | |
| "loss": 0.5554, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -2.648507595062256, | |
| "rewards/margins": 1.04387366771698, | |
| "rewards/rejected": -3.6923813819885254, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.40303585448835383, | |
| "grad_norm": 71.73793938729527, | |
| "learning_rate": 3.7261844968793226e-07, | |
| "logits/chosen": -1.8120813369750977, | |
| "logits/rejected": -1.2725474834442139, | |
| "logps/chosen": -390.5841979980469, | |
| "logps/rejected": -455.8675231933594, | |
| "loss": 0.5758, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.8539800643920898, | |
| "rewards/margins": 0.7287551164627075, | |
| "rewards/rejected": -2.582735300064087, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.408270086364826, | |
| "grad_norm": 39.07736477725613, | |
| "learning_rate": 3.6861657794506187e-07, | |
| "logits/chosen": -1.9753271341323853, | |
| "logits/rejected": -1.181422233581543, | |
| "logps/chosen": -334.88177490234375, | |
| "logps/rejected": -404.42864990234375, | |
| "loss": 0.5208, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.412209153175354, | |
| "rewards/margins": 0.8599991798400879, | |
| "rewards/rejected": -2.2722084522247314, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.4135043182412981, | |
| "grad_norm": 43.71312206302924, | |
| "learning_rate": 3.6457508922871777e-07, | |
| "logits/chosen": -1.889682412147522, | |
| "logits/rejected": -1.6067005395889282, | |
| "logps/chosen": -302.65997314453125, | |
| "logps/rejected": -356.685302734375, | |
| "loss": 0.5254, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.3029696941375732, | |
| "rewards/margins": 0.5454732775688171, | |
| "rewards/rejected": -1.8484432697296143, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.4187385501177702, | |
| "grad_norm": 55.95387505858846, | |
| "learning_rate": 3.6049533336330084e-07, | |
| "logits/chosen": -1.7215135097503662, | |
| "logits/rejected": -1.557579755783081, | |
| "logps/chosen": -316.13299560546875, | |
| "logps/rejected": -418.8580017089844, | |
| "loss": 0.574, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.4686353206634521, | |
| "rewards/margins": 0.7653969526290894, | |
| "rewards/rejected": -2.234032154083252, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.4239727819942423, | |
| "grad_norm": 55.856275598760135, | |
| "learning_rate": 3.56378672954129e-07, | |
| "logits/chosen": -1.6764100790023804, | |
| "logits/rejected": -1.513831377029419, | |
| "logps/chosen": -329.6163635253906, | |
| "logps/rejected": -439.32666015625, | |
| "loss": 0.5538, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.5339328050613403, | |
| "rewards/margins": 0.7833306789398193, | |
| "rewards/rejected": -2.317263126373291, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.42920701387071447, | |
| "grad_norm": 30.497969544668866, | |
| "learning_rate": 3.5222648293233803e-07, | |
| "logits/chosen": -1.4442280530929565, | |
| "logits/rejected": -1.609626054763794, | |
| "logps/chosen": -373.2001037597656, | |
| "logps/rejected": -476.081787109375, | |
| "loss": 0.5508, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.8375556468963623, | |
| "rewards/margins": 0.7855282425880432, | |
| "rewards/rejected": -2.6230838298797607, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.4344412457471866, | |
| "grad_norm": 61.68802213789782, | |
| "learning_rate": 3.480401500956657e-07, | |
| "logits/chosen": -1.523938536643982, | |
| "logits/rejected": -1.2494620084762573, | |
| "logps/chosen": -351.20379638671875, | |
| "logps/rejected": -404.7106018066406, | |
| "loss": 0.566, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.7532094717025757, | |
| "rewards/margins": 0.5370227098464966, | |
| "rewards/rejected": -2.2902321815490723, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.4396754776236587, | |
| "grad_norm": 60.810924331570874, | |
| "learning_rate": 3.438210726452724e-07, | |
| "logits/chosen": -1.4685875177383423, | |
| "logits/rejected": -0.2091219425201416, | |
| "logps/chosen": -392.4021911621094, | |
| "logps/rejected": -455.44482421875, | |
| "loss": 0.5227, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.634267807006836, | |
| "rewards/margins": 0.8823213577270508, | |
| "rewards/rejected": -2.5165891647338867, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.44490970950013087, | |
| "grad_norm": 96.20779655004723, | |
| "learning_rate": 3.395706597187538e-07, | |
| "logits/chosen": -1.657881498336792, | |
| "logits/rejected": -1.0053375959396362, | |
| "logps/chosen": -368.91693115234375, | |
| "logps/rejected": -448.57928466796875, | |
| "loss": 0.5498, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.6729650497436523, | |
| "rewards/margins": 0.8312269449234009, | |
| "rewards/rejected": -2.5041918754577637, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.45014394137660296, | |
| "grad_norm": 42.3904090971244, | |
| "learning_rate": 3.3529033091949986e-07, | |
| "logits/chosen": -1.2889825105667114, | |
| "logits/rejected": -1.143614411354065, | |
| "logps/chosen": -324.1905212402344, | |
| "logps/rejected": -454.4544372558594, | |
| "loss": 0.5259, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.6198065280914307, | |
| "rewards/margins": 0.8226202130317688, | |
| "rewards/rejected": -2.4424266815185547, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.4553781732530751, | |
| "grad_norm": 58.49911093913357, | |
| "learning_rate": 3.309815158425591e-07, | |
| "logits/chosen": -1.6366761922836304, | |
| "logits/rejected": -1.365851640701294, | |
| "logps/chosen": -323.0370178222656, | |
| "logps/rejected": -400.08746337890625, | |
| "loss": 0.5314, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.317899465560913, | |
| "rewards/margins": 0.7919186353683472, | |
| "rewards/rejected": -2.1098179817199707, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.46061240512954726, | |
| "grad_norm": 51.25947355514214, | |
| "learning_rate": 3.2664565359716536e-07, | |
| "logits/chosen": -1.6139739751815796, | |
| "logits/rejected": -1.3585468530654907, | |
| "logps/chosen": -331.5315856933594, | |
| "logps/rejected": -412.1536560058594, | |
| "loss": 0.5143, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.580944299697876, | |
| "rewards/margins": 0.688396155834198, | |
| "rewards/rejected": -2.269340753555298, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.46584663700601936, | |
| "grad_norm": 46.47446781450997, | |
| "learning_rate": 3.222841923260869e-07, | |
| "logits/chosen": -1.56208074092865, | |
| "logits/rejected": -1.448430061340332, | |
| "logps/chosen": -320.86212158203125, | |
| "logps/rejected": -417.16864013671875, | |
| "loss": 0.5148, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.4867175817489624, | |
| "rewards/margins": 0.8568013310432434, | |
| "rewards/rejected": -2.3435187339782715, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.4710808688824915, | |
| "grad_norm": 48.96651105399355, | |
| "learning_rate": 3.1789858872195887e-07, | |
| "logits/chosen": -1.9190874099731445, | |
| "logits/rejected": -1.5447766780853271, | |
| "logps/chosen": -371.22479248046875, | |
| "logps/rejected": -460.0589904785156, | |
| "loss": 0.5126, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.576744794845581, | |
| "rewards/margins": 0.9420223236083984, | |
| "rewards/rejected": -2.5187671184539795, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.4763151007589636, | |
| "grad_norm": 52.14197578469426, | |
| "learning_rate": 3.1349030754075937e-07, | |
| "logits/chosen": -1.5406408309936523, | |
| "logits/rejected": -1.2619167566299438, | |
| "logps/chosen": -352.7300109863281, | |
| "logps/rejected": -432.47027587890625, | |
| "loss": 0.5743, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.5983822345733643, | |
| "rewards/margins": 0.7195128202438354, | |
| "rewards/rejected": -2.3178951740264893, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.48154933263543576, | |
| "grad_norm": 40.54722236321293, | |
| "learning_rate": 3.090608211125931e-07, | |
| "logits/chosen": -1.6173185110092163, | |
| "logits/rejected": -1.3097602128982544, | |
| "logps/chosen": -330.4248046875, | |
| "logps/rejected": -428.1268005371094, | |
| "loss": 0.5167, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -1.4730908870697021, | |
| "rewards/margins": 0.8812786340713501, | |
| "rewards/rejected": -2.354369640350342, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.48678356451190785, | |
| "grad_norm": 72.50233844212927, | |
| "learning_rate": 3.0461160884994487e-07, | |
| "logits/chosen": -1.5237468481063843, | |
| "logits/rejected": -1.1552776098251343, | |
| "logps/chosen": -362.2909240722656, | |
| "logps/rejected": -435.55718994140625, | |
| "loss": 0.5268, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.6922056674957275, | |
| "rewards/margins": 0.7138403654098511, | |
| "rewards/rejected": -2.4060463905334473, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.49201779638838, | |
| "grad_norm": 43.01313334263546, | |
| "learning_rate": 3.001441567535681e-07, | |
| "logits/chosen": -1.4215238094329834, | |
| "logits/rejected": -0.5556894540786743, | |
| "logps/chosen": -325.53814697265625, | |
| "logps/rejected": -438.48602294921875, | |
| "loss": 0.4948, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -1.556456208229065, | |
| "rewards/margins": 1.1343437433242798, | |
| "rewards/rejected": -2.6907999515533447, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.49725202826485215, | |
| "grad_norm": 53.39321127260436, | |
| "learning_rate": 2.956599569161724e-07, | |
| "logits/chosen": -1.5301493406295776, | |
| "logits/rejected": -0.6811811327934265, | |
| "logps/chosen": -359.93988037109375, | |
| "logps/rejected": -444.7823181152344, | |
| "loss": 0.504, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.7924388647079468, | |
| "rewards/margins": 0.9484294652938843, | |
| "rewards/rejected": -2.740868330001831, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.5024862601413242, | |
| "grad_norm": 54.86988246933394, | |
| "learning_rate": 2.91160507024077e-07, | |
| "logits/chosen": -1.3366544246673584, | |
| "logits/rejected": -0.8686436414718628, | |
| "logps/chosen": -387.9107360839844, | |
| "logps/rejected": -483.66546630859375, | |
| "loss": 0.5133, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -2.0935609340667725, | |
| "rewards/margins": 0.9254472851753235, | |
| "rewards/rejected": -3.019008159637451, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.5077204920177963, | |
| "grad_norm": 40.90699971350834, | |
| "learning_rate": 2.866473098569953e-07, | |
| "logits/chosen": -1.3852179050445557, | |
| "logits/rejected": -1.096529483795166, | |
| "logps/chosen": -406.48736572265625, | |
| "logps/rejected": -503.5655822753906, | |
| "loss": 0.5081, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -2.1852097511291504, | |
| "rewards/margins": 0.9200779795646667, | |
| "rewards/rejected": -3.105287551879883, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.5129547238942685, | |
| "grad_norm": 58.095451640836856, | |
| "learning_rate": 2.8212187278611905e-07, | |
| "logits/chosen": -1.689866065979004, | |
| "logits/rejected": -1.0209802389144897, | |
| "logps/chosen": -391.45513916015625, | |
| "logps/rejected": -496.5826110839844, | |
| "loss": 0.5035, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.8062623739242554, | |
| "rewards/margins": 0.9972459077835083, | |
| "rewards/rejected": -2.8035082817077637, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.5181889557707406, | |
| "grad_norm": 59.318676556528416, | |
| "learning_rate": 2.775857072706684e-07, | |
| "logits/chosen": -1.3584972620010376, | |
| "logits/rejected": -1.3198730945587158, | |
| "logps/chosen": -378.9326171875, | |
| "logps/rejected": -501.0203552246094, | |
| "loss": 0.5252, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -2.007432460784912, | |
| "rewards/margins": 0.9733474850654602, | |
| "rewards/rejected": -2.9807801246643066, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.5234231876472127, | |
| "grad_norm": 71.87530393074267, | |
| "learning_rate": 2.7304032835307667e-07, | |
| "logits/chosen": -1.4289054870605469, | |
| "logits/rejected": -1.1081018447875977, | |
| "logps/chosen": -383.5538024902344, | |
| "logps/rejected": -441.691650390625, | |
| "loss": 0.5232, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -2.0912938117980957, | |
| "rewards/margins": 0.5263177156448364, | |
| "rewards/rejected": -2.6176114082336426, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5234231876472127, | |
| "eval_logits/chosen": -0.6161556839942932, | |
| "eval_logits/rejected": -0.05726008489727974, | |
| "eval_logps/chosen": -376.86456298828125, | |
| "eval_logps/rejected": -464.8497009277344, | |
| "eval_loss": 0.5098932385444641, | |
| "eval_rewards/accuracies": 0.7160000205039978, | |
| "eval_rewards/chosen": -1.9184350967407227, | |
| "eval_rewards/margins": 0.838153064250946, | |
| "eval_rewards/rejected": -2.7565884590148926, | |
| "eval_runtime": 316.7389, | |
| "eval_samples_per_second": 6.314, | |
| "eval_steps_per_second": 0.789, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.528657419523685, | |
| "grad_norm": 34.830235400374406, | |
| "learning_rate": 2.6848725415297884e-07, | |
| "logits/chosen": -1.6549265384674072, | |
| "logits/rejected": -1.5296745300292969, | |
| "logps/chosen": -367.93243408203125, | |
| "logps/rejected": -463.02581787109375, | |
| "loss": 0.5217, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.8079150915145874, | |
| "rewards/margins": 0.8082486391067505, | |
| "rewards/rejected": -2.616163730621338, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.533891651400157, | |
| "grad_norm": 46.15714344837178, | |
| "learning_rate": 2.6392800536017183e-07, | |
| "logits/chosen": -1.3246982097625732, | |
| "logits/rejected": -1.457916498184204, | |
| "logps/chosen": -347.6453552246094, | |
| "logps/rejected": -456.356689453125, | |
| "loss": 0.5123, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.6062390804290771, | |
| "rewards/margins": 0.8783868551254272, | |
| "rewards/rejected": -2.484626054763794, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.5391258832766291, | |
| "grad_norm": 40.29007295543721, | |
| "learning_rate": 2.59364104726716e-07, | |
| "logits/chosen": -1.495118498802185, | |
| "logits/rejected": -1.118583083152771, | |
| "logps/chosen": -361.8853454589844, | |
| "logps/rejected": -440.2359924316406, | |
| "loss": 0.4939, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.8378098011016846, | |
| "rewards/margins": 0.7821410894393921, | |
| "rewards/rejected": -2.619950771331787, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.5443601151531012, | |
| "grad_norm": 41.4595455203303, | |
| "learning_rate": 2.547970765583491e-07, | |
| "logits/chosen": -1.4750334024429321, | |
| "logits/rejected": -1.2206333875656128, | |
| "logps/chosen": -345.68243408203125, | |
| "logps/rejected": -457.96417236328125, | |
| "loss": 0.5369, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.8482234477996826, | |
| "rewards/margins": 0.8760908246040344, | |
| "rewards/rejected": -2.7243142127990723, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.5495943470295734, | |
| "grad_norm": 69.76278032549452, | |
| "learning_rate": 2.502284462053799e-07, | |
| "logits/chosen": -1.6379356384277344, | |
| "logits/rejected": -1.1788249015808105, | |
| "logps/chosen": -364.0078125, | |
| "logps/rejected": -483.25750732421875, | |
| "loss": 0.5505, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.7867047786712646, | |
| "rewards/margins": 0.9141599535942078, | |
| "rewards/rejected": -2.700864553451538, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.5548285789060455, | |
| "grad_norm": 43.473418630985485, | |
| "learning_rate": 2.4565973955323374e-07, | |
| "logits/chosen": -1.9050061702728271, | |
| "logits/rejected": -0.7371460795402527, | |
| "logps/chosen": -347.5364990234375, | |
| "logps/rejected": -419.16436767578125, | |
| "loss": 0.5246, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.645029067993164, | |
| "rewards/margins": 0.8056646585464478, | |
| "rewards/rejected": -2.4506936073303223, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.5600628107825176, | |
| "grad_norm": 44.18467471903625, | |
| "learning_rate": 2.410924825128195e-07, | |
| "logits/chosen": -1.677795648574829, | |
| "logits/rejected": -1.0934385061264038, | |
| "logps/chosen": -336.7928161621094, | |
| "logps/rejected": -414.9443359375, | |
| "loss": 0.5456, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.7231500148773193, | |
| "rewards/margins": 0.6935104727745056, | |
| "rewards/rejected": -2.416660785675049, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.5652970426589898, | |
| "grad_norm": 43.21905358731484, | |
| "learning_rate": 2.365282005108875e-07, | |
| "logits/chosen": -1.9169113636016846, | |
| "logits/rejected": -1.1039310693740845, | |
| "logps/chosen": -326.3893127441406, | |
| "logps/rejected": -395.2596435546875, | |
| "loss": 0.5318, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.3917843103408813, | |
| "rewards/margins": 0.8932709693908691, | |
| "rewards/rejected": -2.28505539894104, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.5705312745354619, | |
| "grad_norm": 57.74038702140702, | |
| "learning_rate": 2.319684179805491e-07, | |
| "logits/chosen": -1.7448060512542725, | |
| "logits/rejected": -0.4081326425075531, | |
| "logps/chosen": -359.01214599609375, | |
| "logps/rejected": -431.38531494140625, | |
| "loss": 0.5478, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.73810613155365, | |
| "rewards/margins": 0.8466591835021973, | |
| "rewards/rejected": -2.5847649574279785, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.575765506411934, | |
| "grad_norm": 47.91534163404264, | |
| "learning_rate": 2.2741465785212902e-07, | |
| "logits/chosen": -1.360697627067566, | |
| "logits/rejected": -1.1678688526153564, | |
| "logps/chosen": -341.6931457519531, | |
| "logps/rejected": -434.61749267578125, | |
| "loss": 0.5126, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -1.7018409967422485, | |
| "rewards/margins": 0.6925634741783142, | |
| "rewards/rejected": -2.394404649734497, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.5809997382884062, | |
| "grad_norm": 45.37428786887181, | |
| "learning_rate": 2.2286844104451843e-07, | |
| "logits/chosen": -1.6492805480957031, | |
| "logits/rejected": -1.0351827144622803, | |
| "logps/chosen": -381.6548156738281, | |
| "logps/rejected": -453.56561279296875, | |
| "loss": 0.5463, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -1.7667547464370728, | |
| "rewards/margins": 0.7139405012130737, | |
| "rewards/rejected": -2.4806952476501465, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.5862339701648783, | |
| "grad_norm": 55.038838111114046, | |
| "learning_rate": 2.183312859572008e-07, | |
| "logits/chosen": -1.5666944980621338, | |
| "logits/rejected": -1.518080711364746, | |
| "logps/chosen": -349.3604431152344, | |
| "logps/rejected": -461.66033935546875, | |
| "loss": 0.5142, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.716058373451233, | |
| "rewards/margins": 0.9028706550598145, | |
| "rewards/rejected": -2.618929147720337, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.5914682020413504, | |
| "grad_norm": 51.86256999226559, | |
| "learning_rate": 2.138047079631184e-07, | |
| "logits/chosen": -1.71830153465271, | |
| "logits/rejected": -1.3731257915496826, | |
| "logps/chosen": -348.7271423339844, | |
| "logps/rejected": -429.04730224609375, | |
| "loss": 0.5304, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.4229063987731934, | |
| "rewards/margins": 0.8486738204956055, | |
| "rewards/rejected": -2.271580219268799, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.5967024339178225, | |
| "grad_norm": 61.89587232832013, | |
| "learning_rate": 2.0929021890255068e-07, | |
| "logits/chosen": -1.7627557516098022, | |
| "logits/rejected": -0.9542855024337769, | |
| "logps/chosen": -336.31402587890625, | |
| "logps/rejected": -434.7019958496094, | |
| "loss": 0.4759, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -1.5431474447250366, | |
| "rewards/margins": 1.0138742923736572, | |
| "rewards/rejected": -2.5570216178894043, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.6019366657942947, | |
| "grad_norm": 40.16605369465375, | |
| "learning_rate": 2.0478932657817102e-07, | |
| "logits/chosen": -1.7182365655899048, | |
| "logits/rejected": -1.2019648551940918, | |
| "logps/chosen": -383.36444091796875, | |
| "logps/rejected": -474.964599609375, | |
| "loss": 0.5167, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.7786900997161865, | |
| "rewards/margins": 0.8999784588813782, | |
| "rewards/rejected": -2.67866849899292, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.6071708976707668, | |
| "grad_norm": 65.12157906582378, | |
| "learning_rate": 2.0030353425145374e-07, | |
| "logits/chosen": -1.5357849597930908, | |
| "logits/rejected": -0.887830913066864, | |
| "logps/chosen": -351.23516845703125, | |
| "logps/rejected": -427.96795654296875, | |
| "loss": 0.5519, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.722428560256958, | |
| "rewards/margins": 0.9013321995735168, | |
| "rewards/rejected": -2.62376070022583, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.6124051295472389, | |
| "grad_norm": 56.641784845018215, | |
| "learning_rate": 1.9583434014059635e-07, | |
| "logits/chosen": -1.632693886756897, | |
| "logits/rejected": -1.4601401090621948, | |
| "logps/chosen": -347.542236328125, | |
| "logps/rejected": -451.15875244140625, | |
| "loss": 0.5421, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.6873490810394287, | |
| "rewards/margins": 0.8980420827865601, | |
| "rewards/rejected": -2.5853912830352783, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.6176393614237111, | |
| "grad_norm": 45.373706821931066, | |
| "learning_rate": 1.9138323692012733e-07, | |
| "logits/chosen": -1.8676795959472656, | |
| "logits/rejected": -1.3623721599578857, | |
| "logps/chosen": -318.84417724609375, | |
| "logps/rejected": -431.2315979003906, | |
| "loss": 0.4796, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.421907901763916, | |
| "rewards/margins": 0.911504864692688, | |
| "rewards/rejected": -2.3334126472473145, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.6228735933001832, | |
| "grad_norm": 33.99709051781791, | |
| "learning_rate": 1.8695171122236442e-07, | |
| "logits/chosen": -1.6899328231811523, | |
| "logits/rejected": -1.231277585029602, | |
| "logps/chosen": -341.48797607421875, | |
| "logps/rejected": -474.08343505859375, | |
| "loss": 0.4651, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -1.5275802612304688, | |
| "rewards/margins": 1.2658531665802002, | |
| "rewards/rejected": -2.793433427810669, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.6281078251766553, | |
| "grad_norm": 59.87117531330853, | |
| "learning_rate": 1.8254124314089223e-07, | |
| "logits/chosen": -1.715743064880371, | |
| "logits/rejected": -1.2481369972229004, | |
| "logps/chosen": -357.3684997558594, | |
| "logps/rejected": -465.179443359375, | |
| "loss": 0.4937, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -1.7480823993682861, | |
| "rewards/margins": 0.9352186322212219, | |
| "rewards/rejected": -2.6833012104034424, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.6333420570531274, | |
| "grad_norm": 47.15786058024146, | |
| "learning_rate": 1.7815330573622205e-07, | |
| "logits/chosen": -1.7806997299194336, | |
| "logits/rejected": -1.0503987073898315, | |
| "logps/chosen": -366.83319091796875, | |
| "logps/rejected": -447.2417907714844, | |
| "loss": 0.5129, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -1.8660345077514648, | |
| "rewards/margins": 0.938534140586853, | |
| "rewards/rejected": -2.8045687675476074, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.6385762889295996, | |
| "grad_norm": 59.66066585989714, | |
| "learning_rate": 1.7378936454380274e-07, | |
| "logits/chosen": -1.665154218673706, | |
| "logits/rejected": -1.2627506256103516, | |
| "logps/chosen": -375.3360290527344, | |
| "logps/rejected": -451.8492736816406, | |
| "loss": 0.5141, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.9925143718719482, | |
| "rewards/margins": 0.8351390957832336, | |
| "rewards/rejected": -2.827653408050537, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.6438105208060717, | |
| "grad_norm": 41.626090508354544, | |
| "learning_rate": 1.694508770845427e-07, | |
| "logits/chosen": -1.8555021286010742, | |
| "logits/rejected": -1.1900012493133545, | |
| "logps/chosen": -381.927490234375, | |
| "logps/rejected": -469.22900390625, | |
| "loss": 0.481, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.8637516498565674, | |
| "rewards/margins": 0.8393726348876953, | |
| "rewards/rejected": -2.7031240463256836, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.6490447526825438, | |
| "grad_norm": 46.575209941519326, | |
| "learning_rate": 1.651392923780105e-07, | |
| "logits/chosen": -1.7130018472671509, | |
| "logits/rejected": -1.2077395915985107, | |
| "logps/chosen": -395.5782775878906, | |
| "logps/rejected": -510.3377380371094, | |
| "loss": 0.5015, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -2.0861849784851074, | |
| "rewards/margins": 1.0165139436721802, | |
| "rewards/rejected": -3.102698802947998, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.654278984559016, | |
| "grad_norm": 37.25325816311792, | |
| "learning_rate": 1.6085605045847367e-07, | |
| "logits/chosen": -1.3418257236480713, | |
| "logits/rejected": -0.8439585566520691, | |
| "logps/chosen": -367.1421813964844, | |
| "logps/rejected": -509.10919189453125, | |
| "loss": 0.4888, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -2.0315098762512207, | |
| "rewards/margins": 1.2717547416687012, | |
| "rewards/rejected": -3.303264617919922, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.6595132164354881, | |
| "grad_norm": 56.54421548734596, | |
| "learning_rate": 1.5660258189393944e-07, | |
| "logits/chosen": -1.6082178354263306, | |
| "logits/rejected": -0.8709881901741028, | |
| "logps/chosen": -418.84100341796875, | |
| "logps/rejected": -491.2918395996094, | |
| "loss": 0.4837, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -2.1445069313049316, | |
| "rewards/margins": 0.9897298812866211, | |
| "rewards/rejected": -3.1342368125915527, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.6647474483119602, | |
| "grad_norm": 58.872726779228444, | |
| "learning_rate": 1.5238030730835577e-07, | |
| "logits/chosen": -1.4633188247680664, | |
| "logits/rejected": -1.0070345401763916, | |
| "logps/chosen": -417.1031188964844, | |
| "logps/rejected": -512.1041870117188, | |
| "loss": 0.4794, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -2.1558361053466797, | |
| "rewards/margins": 1.0906795263290405, | |
| "rewards/rejected": -3.246515989303589, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.6699816801884323, | |
| "grad_norm": 81.68062237565607, | |
| "learning_rate": 1.4819063690713564e-07, | |
| "logits/chosen": -1.230272650718689, | |
| "logits/rejected": -1.1420851945877075, | |
| "logps/chosen": -400.37384033203125, | |
| "logps/rejected": -501.90203857421875, | |
| "loss": 0.4836, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.241870403289795, | |
| "rewards/margins": 0.9584578275680542, | |
| "rewards/rejected": -3.2003283500671387, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.6752159120649045, | |
| "grad_norm": 43.4045980317099, | |
| "learning_rate": 1.4403497000615883e-07, | |
| "logits/chosen": -1.5301976203918457, | |
| "logits/rejected": -1.0589085817337036, | |
| "logps/chosen": -406.1276550292969, | |
| "logps/rejected": -488.4261779785156, | |
| "loss": 0.5306, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -2.061319589614868, | |
| "rewards/margins": 0.8899547457695007, | |
| "rewards/rejected": -2.9512743949890137, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.6804501439413766, | |
| "grad_norm": 89.57498346480389, | |
| "learning_rate": 1.3991469456441272e-07, | |
| "logits/chosen": -1.6822435855865479, | |
| "logits/rejected": -1.314344882965088, | |
| "logps/chosen": -366.78118896484375, | |
| "logps/rejected": -476.983642578125, | |
| "loss": 0.4653, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.9514825344085693, | |
| "rewards/margins": 0.9372948408126831, | |
| "rewards/rejected": -2.888777256011963, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.6856843758178487, | |
| "grad_norm": 79.22800819366738, | |
| "learning_rate": 1.358311867204244e-07, | |
| "logits/chosen": -1.3850901126861572, | |
| "logits/rejected": -1.0397329330444336, | |
| "logps/chosen": -410.082763671875, | |
| "logps/rejected": -514.4329833984375, | |
| "loss": 0.5204, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -2.1403348445892334, | |
| "rewards/margins": 0.8943174481391907, | |
| "rewards/rejected": -3.0346522331237793, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.6909186076943209, | |
| "grad_norm": 38.40836601776419, | |
| "learning_rate": 1.3178581033264216e-07, | |
| "logits/chosen": -1.7234236001968384, | |
| "logits/rejected": -1.0658862590789795, | |
| "logps/chosen": -384.4350280761719, | |
| "logps/rejected": -510.3180236816406, | |
| "loss": 0.4734, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.981745719909668, | |
| "rewards/margins": 1.077263355255127, | |
| "rewards/rejected": -3.059009075164795, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.696152839570793, | |
| "grad_norm": 64.64336518776386, | |
| "learning_rate": 1.2777991652391757e-07, | |
| "logits/chosen": -1.686440110206604, | |
| "logits/rejected": -1.139702558517456, | |
| "logps/chosen": -364.24102783203125, | |
| "logps/rejected": -508.68310546875, | |
| "loss": 0.4863, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -1.8615505695343018, | |
| "rewards/margins": 1.3014369010925293, | |
| "rewards/rejected": -3.162987232208252, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.7013870714472651, | |
| "grad_norm": 54.064436812936954, | |
| "learning_rate": 1.2381484323024178e-07, | |
| "logits/chosen": -1.5776959657669067, | |
| "logits/rejected": -1.3716309070587158, | |
| "logps/chosen": -381.18890380859375, | |
| "logps/rejected": -485.0956115722656, | |
| "loss": 0.4748, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -2.0563852787017822, | |
| "rewards/margins": 0.8801882863044739, | |
| "rewards/rejected": -2.9365735054016113, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.7066213033237373, | |
| "grad_norm": 71.26268968336478, | |
| "learning_rate": 1.1989191475388516e-07, | |
| "logits/chosen": -1.5418208837509155, | |
| "logits/rejected": -1.3163645267486572, | |
| "logps/chosen": -389.85479736328125, | |
| "logps/rejected": -469.4613342285156, | |
| "loss": 0.5114, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.9432039260864258, | |
| "rewards/margins": 0.8496206402778625, | |
| "rewards/rejected": -2.7928245067596436, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.7118555352002094, | |
| "grad_norm": 47.16713356681667, | |
| "learning_rate": 1.1601244132109179e-07, | |
| "logits/chosen": -1.547049641609192, | |
| "logits/rejected": -1.1506636142730713, | |
| "logps/chosen": -378.78546142578125, | |
| "logps/rejected": -475.17340087890625, | |
| "loss": 0.5175, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.9611685276031494, | |
| "rewards/margins": 0.8856902122497559, | |
| "rewards/rejected": -2.8468587398529053, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.7170897670766815, | |
| "grad_norm": 57.49724175574713, | |
| "learning_rate": 1.1217771864447395e-07, | |
| "logits/chosen": -1.5181536674499512, | |
| "logits/rejected": -1.0610802173614502, | |
| "logps/chosen": -382.1415710449219, | |
| "logps/rejected": -458.43310546875, | |
| "loss": 0.446, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.9159305095672607, | |
| "rewards/margins": 0.8874963521957397, | |
| "rewards/rejected": -2.803427219390869, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.7223239989531536, | |
| "grad_norm": 48.43604137330991, | |
| "learning_rate": 1.0838902749025499e-07, | |
| "logits/chosen": -1.5803115367889404, | |
| "logits/rejected": -0.9147381782531738, | |
| "logps/chosen": -364.0269470214844, | |
| "logps/rejected": -485.8170471191406, | |
| "loss": 0.4886, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -1.7391821146011353, | |
| "rewards/margins": 1.2611137628555298, | |
| "rewards/rejected": -3.000295639038086, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.7275582308296258, | |
| "grad_norm": 46.02896607743019, | |
| "learning_rate": 1.0464763325050358e-07, | |
| "logits/chosen": -1.389105200767517, | |
| "logits/rejected": -1.2149088382720947, | |
| "logps/chosen": -363.1921081542969, | |
| "logps/rejected": -481.0618591308594, | |
| "loss": 0.4901, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.8936526775360107, | |
| "rewards/margins": 1.019033670425415, | |
| "rewards/rejected": -2.912686586380005, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.7327924627060979, | |
| "grad_norm": 47.49065837586764, | |
| "learning_rate": 1.0095478552050346e-07, | |
| "logits/chosen": -1.6921743154525757, | |
| "logits/rejected": -1.1181484460830688, | |
| "logps/chosen": -369.4226989746094, | |
| "logps/rejected": -469.757568359375, | |
| "loss": 0.5117, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.803931474685669, | |
| "rewards/margins": 0.8779823184013367, | |
| "rewards/rejected": -2.6819140911102295, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.73802669458257, | |
| "grad_norm": 51.70967904748882, | |
| "learning_rate": 9.731171768139806e-08, | |
| "logits/chosen": -1.5017741918563843, | |
| "logits/rejected": -0.8476959466934204, | |
| "logps/chosen": -366.8438415527344, | |
| "logps/rejected": -477.4512634277344, | |
| "loss": 0.5513, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.8564783334732056, | |
| "rewards/margins": 1.0491094589233398, | |
| "rewards/rejected": -2.905587673187256, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.7432609264590422, | |
| "grad_norm": 38.12827677711878, | |
| "learning_rate": 9.37196464882522e-08, | |
| "logits/chosen": -1.7628695964813232, | |
| "logits/rejected": -0.9748938679695129, | |
| "logps/chosen": -358.43377685546875, | |
| "logps/rejected": -426.41839599609375, | |
| "loss": 0.5369, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.729318380355835, | |
| "rewards/margins": 0.7932037711143494, | |
| "rewards/rejected": -2.522522211074829, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.7484951583355143, | |
| "grad_norm": 54.47585201435841, | |
| "learning_rate": 9.017977166366444e-08, | |
| "logits/chosen": -1.7095372676849365, | |
| "logits/rejected": -1.3382246494293213, | |
| "logps/chosen": -386.80419921875, | |
| "logps/rejected": -485.66217041015625, | |
| "loss": 0.4759, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.8498786687850952, | |
| "rewards/margins": 0.956847071647644, | |
| "rewards/rejected": -2.8067257404327393, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.7537293902119864, | |
| "grad_norm": 55.994908017980144, | |
| "learning_rate": 8.669327549707095e-08, | |
| "logits/chosen": -1.8079534769058228, | |
| "logits/rejected": -1.6061460971832275, | |
| "logps/chosen": -372.0968017578125, | |
| "logps/rejected": -449.3272399902344, | |
| "loss": 0.4883, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.8120940923690796, | |
| "rewards/margins": 0.7753244042396545, | |
| "rewards/rejected": -2.587418794631958, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.7589636220884585, | |
| "grad_norm": 44.71580454221332, | |
| "learning_rate": 8.326132244986931e-08, | |
| "logits/chosen": -1.501529335975647, | |
| "logits/rejected": -1.167120337486267, | |
| "logps/chosen": -371.7989807128906, | |
| "logps/rejected": -466.7962951660156, | |
| "loss": 0.4955, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.9070402383804321, | |
| "rewards/margins": 0.8023300170898438, | |
| "rewards/rejected": -2.7093701362609863, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.7641978539649307, | |
| "grad_norm": 47.918962412956276, | |
| "learning_rate": 7.988505876649862e-08, | |
| "logits/chosen": -1.6484168767929077, | |
| "logits/rejected": -1.049037218093872, | |
| "logps/chosen": -387.1507263183594, | |
| "logps/rejected": -471.23455810546875, | |
| "loss": 0.5227, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.9400913715362549, | |
| "rewards/margins": 0.9767180681228638, | |
| "rewards/rejected": -2.916809320449829, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.7694320858414028, | |
| "grad_norm": 44.681178148336286, | |
| "learning_rate": 7.656561209160248e-08, | |
| "logits/chosen": -1.5934867858886719, | |
| "logits/rejected": -0.9533841013908386, | |
| "logps/chosen": -415.989013671875, | |
| "logps/rejected": -503.83038330078125, | |
| "loss": 0.5124, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.1351356506347656, | |
| "rewards/margins": 1.072280764579773, | |
| "rewards/rejected": -3.2074170112609863, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.7746663177178749, | |
| "grad_norm": 42.957667070339845, | |
| "learning_rate": 7.330409109340562e-08, | |
| "logits/chosen": -1.6417922973632812, | |
| "logits/rejected": -0.9782828092575073, | |
| "logps/chosen": -368.1243591308594, | |
| "logps/rejected": -457.6202697753906, | |
| "loss": 0.53, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.979144811630249, | |
| "rewards/margins": 0.9068864583969116, | |
| "rewards/rejected": -2.88603138923645, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.7799005495943471, | |
| "grad_norm": 50.32703939295202, | |
| "learning_rate": 7.010158509342681e-08, | |
| "logits/chosen": -1.492157220840454, | |
| "logits/rejected": -1.0186296701431274, | |
| "logps/chosen": -405.69769287109375, | |
| "logps/rejected": -512.1472778320312, | |
| "loss": 0.4922, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -2.2098875045776367, | |
| "rewards/margins": 0.9893035888671875, | |
| "rewards/rejected": -3.199190855026245, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.7851347814708192, | |
| "grad_norm": 46.95925974979752, | |
| "learning_rate": 6.695916370265527e-08, | |
| "logits/chosen": -1.5276159048080444, | |
| "logits/rejected": -1.1085699796676636, | |
| "logps/chosen": -377.8671875, | |
| "logps/rejected": -545.2662353515625, | |
| "loss": 0.4707, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -2.0651357173919678, | |
| "rewards/margins": 1.2801060676574707, | |
| "rewards/rejected": -3.3452422618865967, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.7851347814708192, | |
| "eval_logits/chosen": -0.6702331304550171, | |
| "eval_logits/rejected": -0.05479540303349495, | |
| "eval_logps/chosen": -403.77130126953125, | |
| "eval_logps/rejected": -503.5488586425781, | |
| "eval_loss": 0.500029981136322, | |
| "eval_rewards/accuracies": 0.7319999933242798, | |
| "eval_rewards/chosen": -2.1875030994415283, | |
| "eval_rewards/margins": 0.9560768604278564, | |
| "eval_rewards/rejected": -3.1435797214508057, | |
| "eval_runtime": 316.8429, | |
| "eval_samples_per_second": 6.312, | |
| "eval_steps_per_second": 0.789, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.7903690133472913, | |
| "grad_norm": 54.6800615563418, | |
| "learning_rate": 6.387787646430853e-08, | |
| "logits/chosen": -1.6765468120574951, | |
| "logits/rejected": -0.902447521686554, | |
| "logps/chosen": -411.84197998046875, | |
| "logps/rejected": -520.2639770507812, | |
| "loss": 0.5386, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -2.209589719772339, | |
| "rewards/margins": 1.2369428873062134, | |
| "rewards/rejected": -3.4465324878692627, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.7956032452237635, | |
| "grad_norm": 39.76345106053052, | |
| "learning_rate": 6.0858752503294e-08, | |
| "logits/chosen": -1.9017261266708374, | |
| "logits/rejected": -1.027804970741272, | |
| "logps/chosen": -375.6845397949219, | |
| "logps/rejected": -452.0986328125, | |
| "loss": 0.524, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.9550809860229492, | |
| "rewards/margins": 0.9305828213691711, | |
| "rewards/rejected": -2.8856637477874756, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.8008374771002356, | |
| "grad_norm": 58.54818603530255, | |
| "learning_rate": 5.7902800182489385e-08, | |
| "logits/chosen": -1.340591311454773, | |
| "logits/rejected": -0.8900511860847473, | |
| "logps/chosen": -382.7643737792969, | |
| "logps/rejected": -484.4600524902344, | |
| "loss": 0.4566, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.102140188217163, | |
| "rewards/margins": 0.9162777066230774, | |
| "rewards/rejected": -3.018418073654175, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.8060717089767077, | |
| "grad_norm": 31.245998478258013, | |
| "learning_rate": 5.5011006765957604e-08, | |
| "logits/chosen": -1.618656873703003, | |
| "logits/rejected": -1.1801456212997437, | |
| "logps/chosen": -419.6280822753906, | |
| "logps/rejected": -499.01751708984375, | |
| "loss": 0.4916, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -2.230376720428467, | |
| "rewards/margins": 0.7931150197982788, | |
| "rewards/rejected": -3.023491621017456, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.8113059408531798, | |
| "grad_norm": 51.33396122501131, | |
| "learning_rate": 5.218433808920883e-08, | |
| "logits/chosen": -1.60371994972229, | |
| "logits/rejected": -1.1156429052352905, | |
| "logps/chosen": -398.92974853515625, | |
| "logps/rejected": -472.16973876953125, | |
| "loss": 0.5013, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -2.029679775238037, | |
| "rewards/margins": 0.9714105725288391, | |
| "rewards/rejected": -3.0010905265808105, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.816540172729652, | |
| "grad_norm": 39.552994793370026, | |
| "learning_rate": 4.942373823661927e-08, | |
| "logits/chosen": -1.6282466650009155, | |
| "logits/rejected": -0.91362464427948, | |
| "logps/chosen": -435.72216796875, | |
| "logps/rejected": -547.359619140625, | |
| "loss": 0.5372, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -2.306370496749878, | |
| "rewards/margins": 0.9918804168701172, | |
| "rewards/rejected": -3.298250913619995, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.821774404606124, | |
| "grad_norm": 60.08870167062165, | |
| "learning_rate": 4.6730129226114354e-08, | |
| "logits/chosen": -1.6371328830718994, | |
| "logits/rejected": -0.6333679556846619, | |
| "logps/chosen": -426.95703125, | |
| "logps/rejected": -538.3595581054688, | |
| "loss": 0.5232, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.2964425086975098, | |
| "rewards/margins": 1.1182937622070312, | |
| "rewards/rejected": -3.414735794067383, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.8270086364825961, | |
| "grad_norm": 46.54218001272997, | |
| "learning_rate": 4.41044107012227e-08, | |
| "logits/chosen": -1.46670401096344, | |
| "logits/rejected": -1.1195374727249146, | |
| "logps/chosen": -394.80078125, | |
| "logps/rejected": -524.2930297851562, | |
| "loss": 0.5174, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.0618937015533447, | |
| "rewards/margins": 1.137542486190796, | |
| "rewards/rejected": -3.1994361877441406, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.8322428683590684, | |
| "grad_norm": 61.54813763490487, | |
| "learning_rate": 4.1547459630601966e-08, | |
| "logits/chosen": -1.5916258096694946, | |
| "logits/rejected": -0.7485260963439941, | |
| "logps/chosen": -393.9290771484375, | |
| "logps/rejected": -469.08026123046875, | |
| "loss": 0.488, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -2.228323459625244, | |
| "rewards/margins": 0.8031214475631714, | |
| "rewards/rejected": -3.031445026397705, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.8374771002355405, | |
| "grad_norm": 47.454976444510464, | |
| "learning_rate": 3.9060130015138857e-08, | |
| "logits/chosen": -1.6673399209976196, | |
| "logits/rejected": -0.9725383520126343, | |
| "logps/chosen": -391.20428466796875, | |
| "logps/rejected": -521.3951416015625, | |
| "loss": 0.5182, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -2.1082262992858887, | |
| "rewards/margins": 1.1958465576171875, | |
| "rewards/rejected": -3.304072856903076, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.8427113321120125, | |
| "grad_norm": 49.49623357781679, | |
| "learning_rate": 3.664325260271953e-08, | |
| "logits/chosen": -1.785679578781128, | |
| "logits/rejected": -1.3486297130584717, | |
| "logps/chosen": -425.0037536621094, | |
| "logps/rejected": -499.3848571777344, | |
| "loss": 0.5221, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.9705803394317627, | |
| "rewards/margins": 1.0100176334381104, | |
| "rewards/rejected": -2.980597972869873, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.8479455639884846, | |
| "grad_norm": 46.41346156870081, | |
| "learning_rate": 3.429763461076676e-08, | |
| "logits/chosen": -1.5286333560943604, | |
| "logits/rejected": -1.219725251197815, | |
| "logps/chosen": -410.5511169433594, | |
| "logps/rejected": -490.86669921875, | |
| "loss": 0.499, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -2.219921112060547, | |
| "rewards/margins": 0.7538778185844421, | |
| "rewards/rejected": -2.973799228668213, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.8531797958649568, | |
| "grad_norm": 48.04705054674233, | |
| "learning_rate": 3.202405945663555e-08, | |
| "logits/chosen": -1.2301688194274902, | |
| "logits/rejected": -1.1187329292297363, | |
| "logps/chosen": -398.3143615722656, | |
| "logps/rejected": -471.55010986328125, | |
| "loss": 0.5216, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -2.1741843223571777, | |
| "rewards/margins": 0.7370268702507019, | |
| "rewards/rejected": -2.9112114906311035, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.8584140277414289, | |
| "grad_norm": 55.340731067768736, | |
| "learning_rate": 2.9823286495958556e-08, | |
| "logits/chosen": -1.6326100826263428, | |
| "logits/rejected": -1.1603213548660278, | |
| "logps/chosen": -397.24005126953125, | |
| "logps/rejected": -491.3506774902344, | |
| "loss": 0.5211, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -1.9936778545379639, | |
| "rewards/margins": 1.0326869487762451, | |
| "rewards/rejected": -3.026364803314209, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.863648259617901, | |
| "grad_norm": 50.97109608180561, | |
| "learning_rate": 2.769605076902695e-08, | |
| "logits/chosen": -1.5886093378067017, | |
| "logits/rejected": -1.497037649154663, | |
| "logps/chosen": -364.107666015625, | |
| "logps/rejected": -491.516357421875, | |
| "loss": 0.5201, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -2.014535903930664, | |
| "rewards/margins": 0.9813510179519653, | |
| "rewards/rejected": -2.9958865642547607, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.8688824914943732, | |
| "grad_norm": 79.38139421221125, | |
| "learning_rate": 2.5643062755293403e-08, | |
| "logits/chosen": -1.504066824913025, | |
| "logits/rejected": -0.8931864500045776, | |
| "logps/chosen": -391.62713623046875, | |
| "logps/rejected": -493.92425537109375, | |
| "loss": 0.5236, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.1034412384033203, | |
| "rewards/margins": 0.948072075843811, | |
| "rewards/rejected": -3.051513195037842, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.8741167233708453, | |
| "grad_norm": 51.56700786640102, | |
| "learning_rate": 2.366500813607733e-08, | |
| "logits/chosen": -1.5938122272491455, | |
| "logits/rejected": -1.2632464170455933, | |
| "logps/chosen": -385.2617492675781, | |
| "logps/rejected": -489.09228515625, | |
| "loss": 0.4905, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.0015676021575928, | |
| "rewards/margins": 1.0208126306533813, | |
| "rewards/rejected": -3.0223803520202637, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.8793509552473174, | |
| "grad_norm": 57.01702831334661, | |
| "learning_rate": 2.176254756555329e-08, | |
| "logits/chosen": -1.5943069458007812, | |
| "logits/rejected": -1.2856522798538208, | |
| "logps/chosen": -391.60736083984375, | |
| "logps/rejected": -528.451171875, | |
| "loss": 0.4412, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.9151815176010132, | |
| "rewards/margins": 1.0721461772918701, | |
| "rewards/rejected": -2.987327814102173, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.8845851871237895, | |
| "grad_norm": 52.04207661692601, | |
| "learning_rate": 1.9936316450097468e-08, | |
| "logits/chosen": -1.6583569049835205, | |
| "logits/rejected": -1.2720321416854858, | |
| "logps/chosen": -375.07989501953125, | |
| "logps/rejected": -463.97015380859375, | |
| "loss": 0.48, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.0725209712982178, | |
| "rewards/margins": 0.813088595867157, | |
| "rewards/rejected": -2.8856096267700195, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.8898194190002617, | |
| "grad_norm": 56.72817214823504, | |
| "learning_rate": 1.8186924736067477e-08, | |
| "logits/chosen": -1.6529877185821533, | |
| "logits/rejected": -1.2598158121109009, | |
| "logps/chosen": -395.845703125, | |
| "logps/rejected": -497.45574951171875, | |
| "loss": 0.4753, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.0222179889678955, | |
| "rewards/margins": 1.0427659749984741, | |
| "rewards/rejected": -3.06498384475708, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.8950536508767338, | |
| "grad_norm": 60.40350237801338, | |
| "learning_rate": 1.651495670608488e-08, | |
| "logits/chosen": -1.6280845403671265, | |
| "logits/rejected": -1.0049400329589844, | |
| "logps/chosen": -408.4595642089844, | |
| "logps/rejected": -528.3615112304688, | |
| "loss": 0.4851, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.1130669116973877, | |
| "rewards/margins": 1.1445512771606445, | |
| "rewards/rejected": -3.2576184272766113, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.9002878827532059, | |
| "grad_norm": 42.82234513435798, | |
| "learning_rate": 1.4920970783889737e-08, | |
| "logits/chosen": -1.784874677658081, | |
| "logits/rejected": -1.022477388381958, | |
| "logps/chosen": -403.08294677734375, | |
| "logps/rejected": -465.8329162597656, | |
| "loss": 0.4906, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -2.028446912765503, | |
| "rewards/margins": 0.7884195446968079, | |
| "rewards/rejected": -2.816866636276245, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.9055221146296781, | |
| "grad_norm": 67.83919792587483, | |
| "learning_rate": 1.340549934783164e-08, | |
| "logits/chosen": -1.6942017078399658, | |
| "logits/rejected": -1.3202614784240723, | |
| "logps/chosen": -391.9894104003906, | |
| "logps/rejected": -499.2957458496094, | |
| "loss": 0.5082, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -2.1304984092712402, | |
| "rewards/margins": 0.9605563879013062, | |
| "rewards/rejected": -3.091054916381836, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.9107563465061502, | |
| "grad_norm": 38.38762358526929, | |
| "learning_rate": 1.1969048553059608e-08, | |
| "logits/chosen": -1.5242860317230225, | |
| "logits/rejected": -1.0694557428359985, | |
| "logps/chosen": -385.64447021484375, | |
| "logps/rejected": -532.7508544921875, | |
| "loss": 0.5175, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -2.149869441986084, | |
| "rewards/margins": 1.1629979610443115, | |
| "rewards/rejected": -3.3128669261932373, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.9159905783826223, | |
| "grad_norm": 47.57788484474261, | |
| "learning_rate": 1.06120981624703e-08, | |
| "logits/chosen": -1.8843555450439453, | |
| "logits/rejected": -0.6692565083503723, | |
| "logps/chosen": -416.1067810058594, | |
| "logps/rejected": -449.87103271484375, | |
| "loss": 0.524, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -2.03684401512146, | |
| "rewards/margins": 0.8309102058410645, | |
| "rewards/rejected": -2.8677542209625244, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.9212248102590945, | |
| "grad_norm": 51.79734307723889, | |
| "learning_rate": 9.335101386471284e-09, | |
| "logits/chosen": -1.769464135169983, | |
| "logits/rejected": -1.5171881914138794, | |
| "logps/chosen": -366.7747497558594, | |
| "logps/rejected": -517.2105712890625, | |
| "loss": 0.4596, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -1.8451915979385376, | |
| "rewards/margins": 1.2242134809494019, | |
| "rewards/rejected": -3.0694050788879395, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.9264590421355666, | |
| "grad_norm": 59.243332805566055, | |
| "learning_rate": 8.138484731612273e-09, | |
| "logits/chosen": -1.610945701599121, | |
| "logits/rejected": -0.8597782850265503, | |
| "logps/chosen": -412.3548278808594, | |
| "logps/rejected": -507.70050048828125, | |
| "loss": 0.4844, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -2.1251347064971924, | |
| "rewards/margins": 1.0798755884170532, | |
| "rewards/rejected": -3.205010175704956, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.9316932740120387, | |
| "grad_norm": 35.23329301018955, | |
| "learning_rate": 7.0226478581355e-09, | |
| "logits/chosen": -1.5735182762145996, | |
| "logits/rejected": -1.0823094844818115, | |
| "logps/chosen": -382.31378173828125, | |
| "logps/rejected": -459.026611328125, | |
| "loss": 0.522, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -2.1322665214538574, | |
| "rewards/margins": 0.7317706346511841, | |
| "rewards/rejected": -2.86403751373291, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.9369275058885108, | |
| "grad_norm": 39.12208203761793, | |
| "learning_rate": 5.987963446492383e-09, | |
| "logits/chosen": -1.6163190603256226, | |
| "logits/rejected": -1.3726081848144531, | |
| "logps/chosen": -368.1934509277344, | |
| "logps/rejected": -484.42620849609375, | |
| "loss": 0.4783, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.939570426940918, | |
| "rewards/margins": 0.9698532819747925, | |
| "rewards/rejected": -2.909423828125, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.942161737764983, | |
| "grad_norm": 45.32578662406385, | |
| "learning_rate": 5.0347770728713935e-09, | |
| "logits/chosen": -1.5549806356430054, | |
| "logits/rejected": -0.9020807147026062, | |
| "logps/chosen": -382.2792053222656, | |
| "logps/rejected": -480.25714111328125, | |
| "loss": 0.4825, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -2.1894912719726562, | |
| "rewards/margins": 0.8568480610847473, | |
| "rewards/rejected": -3.046339273452759, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.9473959696414551, | |
| "grad_norm": 51.50392990407485, | |
| "learning_rate": 4.1634070937782424e-09, | |
| "logits/chosen": -1.563454270362854, | |
| "logits/rejected": -1.4620710611343384, | |
| "logps/chosen": -374.35662841796875, | |
| "logps/rejected": -459.2840881347656, | |
| "loss": 0.4949, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -2.0908713340759277, | |
| "rewards/margins": 0.8073171377182007, | |
| "rewards/rejected": -2.898188829421997, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.9526302015179272, | |
| "grad_norm": 52.57515906506144, | |
| "learning_rate": 3.3741445397075797e-09, | |
| "logits/chosen": -1.5654033422470093, | |
| "logits/rejected": -0.9612579345703125, | |
| "logps/chosen": -403.11767578125, | |
| "logps/rejected": -518.7571411132812, | |
| "loss": 0.499, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -2.0279488563537598, | |
| "rewards/margins": 1.2528560161590576, | |
| "rewards/rejected": -3.2808051109313965, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.9578644333943994, | |
| "grad_norm": 80.78348799446286, | |
| "learning_rate": 2.667253017941018e-09, | |
| "logits/chosen": -1.5542982816696167, | |
| "logits/rejected": -1.2357369661331177, | |
| "logps/chosen": -414.04571533203125, | |
| "logps/rejected": -515.9486083984375, | |
| "loss": 0.5405, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -2.095529079437256, | |
| "rewards/margins": 0.8272086977958679, | |
| "rewards/rejected": -2.9227375984191895, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.9630986652708715, | |
| "grad_norm": 60.481486866817654, | |
| "learning_rate": 2.0429686245045097e-09, | |
| "logits/chosen": -1.7054321765899658, | |
| "logits/rejected": -0.9457400441169739, | |
| "logps/chosen": -383.46612548828125, | |
| "logps/rejected": -498.7964782714844, | |
| "loss": 0.4679, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -2.065452814102173, | |
| "rewards/margins": 1.179323434829712, | |
| "rewards/rejected": -3.244776487350464, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.9683328971473436, | |
| "grad_norm": 42.47290413747811, | |
| "learning_rate": 1.5014998653141708e-09, | |
| "logits/chosen": -1.4126255512237549, | |
| "logits/rejected": -1.1500160694122314, | |
| "logps/chosen": -424.77337646484375, | |
| "logps/rejected": -545.006591796875, | |
| "loss": 0.4857, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -2.2447969913482666, | |
| "rewards/margins": 1.013723611831665, | |
| "rewards/rejected": -3.2585208415985107, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.9735671290238157, | |
| "grad_norm": 55.24460892129673, | |
| "learning_rate": 1.0430275865371263e-09, | |
| "logits/chosen": -1.460265874862671, | |
| "logits/rejected": -0.5893803238868713, | |
| "logps/chosen": -364.1796569824219, | |
| "logps/rejected": -458.278564453125, | |
| "loss": 0.4837, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -2.0408921241760254, | |
| "rewards/margins": 1.031553864479065, | |
| "rewards/rejected": -3.0724456310272217, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.9788013609002879, | |
| "grad_norm": 78.57983765704049, | |
| "learning_rate": 6.677049141901314e-10, | |
| "logits/chosen": -1.6051515340805054, | |
| "logits/rejected": -0.9829438924789429, | |
| "logps/chosen": -398.05267333984375, | |
| "logps/rejected": -477.8648376464844, | |
| "loss": 0.4942, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -2.0092108249664307, | |
| "rewards/margins": 0.8877508044242859, | |
| "rewards/rejected": -2.8969614505767822, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.98403559277676, | |
| "grad_norm": 37.16308130632785, | |
| "learning_rate": 3.7565720299687077e-10, | |
| "logits/chosen": -1.8912632465362549, | |
| "logits/rejected": -0.9422454833984375, | |
| "logps/chosen": -400.77197265625, | |
| "logps/rejected": -511.6893005371094, | |
| "loss": 0.4815, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.160900115966797, | |
| "rewards/margins": 1.268087387084961, | |
| "rewards/rejected": -3.428987503051758, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.9892698246532321, | |
| "grad_norm": 54.378639302860876, | |
| "learning_rate": 1.6698199452053197e-10, | |
| "logits/chosen": -1.7411140203475952, | |
| "logits/rejected": -1.1444064378738403, | |
| "logps/chosen": -405.96466064453125, | |
| "logps/rejected": -516.4884643554688, | |
| "loss": 0.5036, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -2.084994077682495, | |
| "rewards/margins": 0.9552907943725586, | |
| "rewards/rejected": -3.0402846336364746, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.9945040565297043, | |
| "grad_norm": 55.60143338165864, | |
| "learning_rate": 4.174898458556009e-11, | |
| "logits/chosen": -1.7632405757904053, | |
| "logits/rejected": -1.2880979776382446, | |
| "logps/chosen": -400.0120544433594, | |
| "logps/rejected": -445.19036865234375, | |
| "loss": 0.516, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -2.067795515060425, | |
| "rewards/margins": 0.6980506777763367, | |
| "rewards/rejected": -2.7658464908599854, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.9997382884061764, | |
| "grad_norm": 50.31539314937899, | |
| "learning_rate": 0.0, | |
| "logits/chosen": -1.4952467679977417, | |
| "logits/rejected": -0.9173043966293335, | |
| "logps/chosen": -421.25, | |
| "logps/rejected": -543.969482421875, | |
| "loss": 0.4891, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -2.100468635559082, | |
| "rewards/margins": 1.1773890256881714, | |
| "rewards/rejected": -3.277857542037964, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.9997382884061764, | |
| "step": 1910, | |
| "total_flos": 0.0, | |
| "train_loss": 0.5374546366836388, | |
| "train_runtime": 38043.1849, | |
| "train_samples_per_second": 1.607, | |
| "train_steps_per_second": 0.05 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1910, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000, | |
| "total_flos": 0.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |