martimfasantos's picture
Model save
5ceb57d verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.99582225598177,
"eval_steps": 800,
"global_step": 1479,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.002025572857323712,
"grad_norm": 31.013870239257812,
"learning_rate": 6.756756756756757e-10,
"logits/chosen": -2.5177597999572754,
"logits/rejected": -2.4276583194732666,
"logps/chosen": -79.6932373046875,
"logps/rejected": -86.58649444580078,
"loss": 0.6929,
"rewards/accuracies": 0.03125,
"rewards/chosen": -0.0008372783777303994,
"rewards/margins": 0.00045527220936492085,
"rewards/rejected": -0.0012925505870953202,
"step": 1
},
{
"epoch": 0.020255728573237118,
"grad_norm": 28.71278190612793,
"learning_rate": 6.756756756756757e-09,
"logits/chosen": -2.587923526763916,
"logits/rejected": -2.421647787094116,
"logps/chosen": -72.02790069580078,
"logps/rejected": -68.7666015625,
"loss": 0.6928,
"rewards/accuracies": 0.4479166567325592,
"rewards/chosen": -0.000452535372460261,
"rewards/margins": 0.0008923111017793417,
"rewards/rejected": -0.001344846561551094,
"step": 10
},
{
"epoch": 0.040511457146474236,
"grad_norm": 24.486814498901367,
"learning_rate": 1.3513513513513514e-08,
"logits/chosen": -2.5588934421539307,
"logits/rejected": -2.3707621097564697,
"logps/chosen": -77.4730453491211,
"logps/rejected": -71.17650604248047,
"loss": 0.693,
"rewards/accuracies": 0.49687498807907104,
"rewards/chosen": 0.0015422820579260588,
"rewards/margins": 0.00048240157775580883,
"rewards/rejected": 0.0010598807130008936,
"step": 20
},
{
"epoch": 0.060767185719711354,
"grad_norm": 26.952857971191406,
"learning_rate": 2.027027027027027e-08,
"logits/chosen": -2.5552210807800293,
"logits/rejected": -2.3964200019836426,
"logps/chosen": -75.58769226074219,
"logps/rejected": -74.38423156738281,
"loss": 0.6941,
"rewards/accuracies": 0.4625000059604645,
"rewards/chosen": -0.00154799222946167,
"rewards/margins": -0.0016709610354155302,
"rewards/rejected": 0.0001229687622981146,
"step": 30
},
{
"epoch": 0.08102291429294847,
"grad_norm": 30.491893768310547,
"learning_rate": 2.7027027027027028e-08,
"logits/chosen": -2.538985013961792,
"logits/rejected": -2.3956074714660645,
"logps/chosen": -84.64269256591797,
"logps/rejected": -82.15937042236328,
"loss": 0.6927,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": 0.0005608886131085455,
"rewards/margins": 0.0012397856917232275,
"rewards/rejected": -0.000678897020407021,
"step": 40
},
{
"epoch": 0.1012786428661856,
"grad_norm": 27.152774810791016,
"learning_rate": 3.378378378378378e-08,
"logits/chosen": -2.515413522720337,
"logits/rejected": -2.358457565307617,
"logps/chosen": -81.1507568359375,
"logps/rejected": -78.68826293945312,
"loss": 0.6921,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.0013986782869324088,
"rewards/margins": 0.002325823763385415,
"rewards/rejected": -0.0037245028652250767,
"step": 50
},
{
"epoch": 0.12153437143942271,
"grad_norm": 27.534177780151367,
"learning_rate": 4.054054054054054e-08,
"logits/chosen": -2.520850419998169,
"logits/rejected": -2.3658010959625244,
"logps/chosen": -78.13814544677734,
"logps/rejected": -75.04551696777344,
"loss": 0.6938,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.0025924122892320156,
"rewards/margins": -0.0010297519620507956,
"rewards/rejected": -0.00156266032718122,
"step": 60
},
{
"epoch": 0.14179010001265982,
"grad_norm": 30.60668182373047,
"learning_rate": 4.72972972972973e-08,
"logits/chosen": -2.5382590293884277,
"logits/rejected": -2.37661075592041,
"logps/chosen": -83.97273254394531,
"logps/rejected": -80.8182373046875,
"loss": 0.6926,
"rewards/accuracies": 0.515625,
"rewards/chosen": -0.003294873284175992,
"rewards/margins": 0.0013795426348224282,
"rewards/rejected": -0.004674416035413742,
"step": 70
},
{
"epoch": 0.16204582858589694,
"grad_norm": 27.501964569091797,
"learning_rate": 5.4054054054054056e-08,
"logits/chosen": -2.4654035568237305,
"logits/rejected": -2.3297314643859863,
"logps/chosen": -75.83648681640625,
"logps/rejected": -76.66287994384766,
"loss": 0.6902,
"rewards/accuracies": 0.578125,
"rewards/chosen": -0.0013893753057345748,
"rewards/margins": 0.006102095358073711,
"rewards/rejected": -0.007491470314562321,
"step": 80
},
{
"epoch": 0.18230155715913407,
"grad_norm": 24.944175720214844,
"learning_rate": 6.081081081081081e-08,
"logits/chosen": -2.5287299156188965,
"logits/rejected": -2.379664659500122,
"logps/chosen": -86.45475769042969,
"logps/rejected": -79.61102294921875,
"loss": 0.6904,
"rewards/accuracies": 0.578125,
"rewards/chosen": -0.0014329934492707253,
"rewards/margins": 0.005873243790119886,
"rewards/rejected": -0.007306237705051899,
"step": 90
},
{
"epoch": 0.2025572857323712,
"grad_norm": 29.129093170166016,
"learning_rate": 6.756756756756756e-08,
"logits/chosen": -2.5153121948242188,
"logits/rejected": -2.361551523208618,
"logps/chosen": -84.0345230102539,
"logps/rejected": -78.61013793945312,
"loss": 0.6881,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.0028929836116731167,
"rewards/margins": 0.010393200442194939,
"rewards/rejected": -0.013286183588206768,
"step": 100
},
{
"epoch": 0.2228130143056083,
"grad_norm": 28.691940307617188,
"learning_rate": 7.432432432432432e-08,
"logits/chosen": -2.546154260635376,
"logits/rejected": -2.389882802963257,
"logps/chosen": -74.24641418457031,
"logps/rejected": -72.99244689941406,
"loss": 0.6872,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.003837780561298132,
"rewards/margins": 0.012213540263473988,
"rewards/rejected": -0.016051320359110832,
"step": 110
},
{
"epoch": 0.24306874287884542,
"grad_norm": 26.047407150268555,
"learning_rate": 8.108108108108108e-08,
"logits/chosen": -2.530447006225586,
"logits/rejected": -2.3604226112365723,
"logps/chosen": -79.45042419433594,
"logps/rejected": -75.46896362304688,
"loss": 0.6834,
"rewards/accuracies": 0.6781250238418579,
"rewards/chosen": -0.004467087332159281,
"rewards/margins": 0.020118705928325653,
"rewards/rejected": -0.024585790932178497,
"step": 120
},
{
"epoch": 0.26332447145208254,
"grad_norm": 30.345191955566406,
"learning_rate": 8.783783783783784e-08,
"logits/chosen": -2.4959208965301514,
"logits/rejected": -2.344454526901245,
"logps/chosen": -86.02290344238281,
"logps/rejected": -81.23602294921875,
"loss": 0.6824,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.006714115384966135,
"rewards/margins": 0.02207200787961483,
"rewards/rejected": -0.028786126524209976,
"step": 130
},
{
"epoch": 0.28358020002531964,
"grad_norm": 23.314868927001953,
"learning_rate": 9.45945945945946e-08,
"logits/chosen": -2.5607352256774902,
"logits/rejected": -2.394366502761841,
"logps/chosen": -72.60206604003906,
"logps/rejected": -67.85148620605469,
"loss": 0.6799,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.008559314534068108,
"rewards/margins": 0.027249369770288467,
"rewards/rejected": -0.035808682441711426,
"step": 140
},
{
"epoch": 0.3038359285985568,
"grad_norm": 27.51814079284668,
"learning_rate": 9.999944288759615e-08,
"logits/chosen": -2.5597286224365234,
"logits/rejected": -2.4156954288482666,
"logps/chosen": -74.42972564697266,
"logps/rejected": -70.92676544189453,
"loss": 0.6768,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.00905610155314207,
"rewards/margins": 0.03394917771220207,
"rewards/rejected": -0.043005283921957016,
"step": 150
},
{
"epoch": 0.3240916571717939,
"grad_norm": 30.097389221191406,
"learning_rate": 9.99799452570021e-08,
"logits/chosen": -2.508636236190796,
"logits/rejected": -2.3848562240600586,
"logps/chosen": -77.739013671875,
"logps/rejected": -73.92839813232422,
"loss": 0.6753,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.017720907926559448,
"rewards/margins": 0.037171002477407455,
"rewards/rejected": -0.054891906678676605,
"step": 160
},
{
"epoch": 0.34434738574503104,
"grad_norm": 32.0848503112793,
"learning_rate": 9.993260441994115e-08,
"logits/chosen": -2.5097594261169434,
"logits/rejected": -2.3447771072387695,
"logps/chosen": -81.94526672363281,
"logps/rejected": -78.39651489257812,
"loss": 0.6634,
"rewards/accuracies": 0.7406250238418579,
"rewards/chosen": -0.013811466284096241,
"rewards/margins": 0.06231771036982536,
"rewards/rejected": -0.07612917572259903,
"step": 170
},
{
"epoch": 0.36460311431826814,
"grad_norm": 29.288728713989258,
"learning_rate": 9.985744674940535e-08,
"logits/chosen": -2.5364279747009277,
"logits/rejected": -2.354965925216675,
"logps/chosen": -80.46150207519531,
"logps/rejected": -75.10428619384766,
"loss": 0.6577,
"rewards/accuracies": 0.7406250238418579,
"rewards/chosen": -0.0209208894520998,
"rewards/margins": 0.07542888820171356,
"rewards/rejected": -0.09634977579116821,
"step": 180
},
{
"epoch": 0.38485884289150524,
"grad_norm": 27.00816535949707,
"learning_rate": 9.975451411479911e-08,
"logits/chosen": -2.499474048614502,
"logits/rejected": -2.337350606918335,
"logps/chosen": -78.61238098144531,
"logps/rejected": -78.64549255371094,
"loss": 0.657,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.02487110160291195,
"rewards/margins": 0.07822562754154205,
"rewards/rejected": -0.10309673845767975,
"step": 190
},
{
"epoch": 0.4051145714647424,
"grad_norm": 27.34921646118164,
"learning_rate": 9.962386385861412e-08,
"logits/chosen": -2.50087308883667,
"logits/rejected": -2.360152006149292,
"logps/chosen": -76.67208862304688,
"logps/rejected": -78.57847595214844,
"loss": 0.652,
"rewards/accuracies": 0.746874988079071,
"rewards/chosen": -0.038105227053165436,
"rewards/margins": 0.08875634521245956,
"rewards/rejected": -0.126861572265625,
"step": 200
},
{
"epoch": 0.4253703000379795,
"grad_norm": 25.707185745239258,
"learning_rate": 9.946556876448468e-08,
"logits/chosen": -2.4654879570007324,
"logits/rejected": -2.312530994415283,
"logps/chosen": -78.15449523925781,
"logps/rejected": -77.42134857177734,
"loss": 0.6439,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.03906805440783501,
"rewards/margins": 0.10793592780828476,
"rewards/rejected": -0.14700399339199066,
"step": 210
},
{
"epoch": 0.4456260286112166,
"grad_norm": 27.729816436767578,
"learning_rate": 9.927971701664084e-08,
"logits/chosen": -2.4674429893493652,
"logits/rejected": -2.3009190559387207,
"logps/chosen": -75.07694244384766,
"logps/rejected": -75.41253662109375,
"loss": 0.6417,
"rewards/accuracies": 0.7593749761581421,
"rewards/chosen": -0.0388757549226284,
"rewards/margins": 0.1117323786020279,
"rewards/rejected": -0.1506081372499466,
"step": 220
},
{
"epoch": 0.46588175718445374,
"grad_norm": 24.861696243286133,
"learning_rate": 9.906641215078196e-08,
"logits/chosen": -2.462665557861328,
"logits/rejected": -2.309985876083374,
"logps/chosen": -77.72923278808594,
"logps/rejected": -75.81291961669922,
"loss": 0.6384,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.052964676171541214,
"rewards/margins": 0.12125100940465927,
"rewards/rejected": -0.1742156744003296,
"step": 230
},
{
"epoch": 0.48613748575769083,
"grad_norm": 30.613037109375,
"learning_rate": 9.882577299639835e-08,
"logits/chosen": -2.4711391925811768,
"logits/rejected": -2.329216957092285,
"logps/chosen": -80.07206726074219,
"logps/rejected": -80.30625915527344,
"loss": 0.6333,
"rewards/accuracies": 0.7406250238418579,
"rewards/chosen": -0.0626656636595726,
"rewards/margins": 0.13491353392601013,
"rewards/rejected": -0.19757920503616333,
"step": 240
},
{
"epoch": 0.506393214330928,
"grad_norm": 26.08587074279785,
"learning_rate": 9.85579336105728e-08,
"logits/chosen": -2.443732738494873,
"logits/rejected": -2.2919225692749023,
"logps/chosen": -81.61358642578125,
"logps/rejected": -77.46446990966797,
"loss": 0.6342,
"rewards/accuracies": 0.7281249761581421,
"rewards/chosen": -0.05705754831433296,
"rewards/margins": 0.13192656636238098,
"rewards/rejected": -0.18898411095142365,
"step": 250
},
{
"epoch": 0.5266489429041651,
"grad_norm": 24.169845581054688,
"learning_rate": 9.826304320329907e-08,
"logits/chosen": -2.478874683380127,
"logits/rejected": -2.297999382019043,
"logps/chosen": -83.48451232910156,
"logps/rejected": -78.53952026367188,
"loss": 0.6226,
"rewards/accuracies": 0.7093750238418579,
"rewards/chosen": -0.07330447435379028,
"rewards/margins": 0.16096489131450653,
"rewards/rejected": -0.23426935076713562,
"step": 260
},
{
"epoch": 0.5469046714774022,
"grad_norm": 26.509349822998047,
"learning_rate": 9.794126605435884e-08,
"logits/chosen": -2.452291488647461,
"logits/rejected": -2.2766990661621094,
"logps/chosen": -83.7742691040039,
"logps/rejected": -80.86663818359375,
"loss": 0.6085,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -0.08826017379760742,
"rewards/margins": 0.20091946423053741,
"rewards/rejected": -0.28917962312698364,
"step": 270
},
{
"epoch": 0.5671604000506393,
"grad_norm": 23.097434997558594,
"learning_rate": 9.759278142180347e-08,
"logits/chosen": -2.4537911415100098,
"logits/rejected": -2.291194438934326,
"logps/chosen": -77.87368774414062,
"logps/rejected": -77.5306625366211,
"loss": 0.6047,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.08215166628360748,
"rewards/margins": 0.20681920647621155,
"rewards/rejected": -0.28897085785865784,
"step": 280
},
{
"epoch": 0.5874161286238765,
"grad_norm": 25.596263885498047,
"learning_rate": 9.72177834420916e-08,
"logits/chosen": -2.4155325889587402,
"logits/rejected": -2.2689411640167236,
"logps/chosen": -84.03662109375,
"logps/rejected": -83.03952026367188,
"loss": 0.6053,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.12551462650299072,
"rewards/margins": 0.20985326170921326,
"rewards/rejected": -0.33536791801452637,
"step": 290
},
{
"epoch": 0.6076718571971136,
"grad_norm": 32.68680191040039,
"learning_rate": 9.68164810219381e-08,
"logits/chosen": -2.4283127784729004,
"logits/rejected": -2.3022093772888184,
"logps/chosen": -74.84422302246094,
"logps/rejected": -76.46062469482422,
"loss": 0.6136,
"rewards/accuracies": 0.7093750238418579,
"rewards/chosen": -0.13987191021442413,
"rewards/margins": 0.19149479269981384,
"rewards/rejected": -0.33136671781539917,
"step": 300
},
{
"epoch": 0.6279275857703507,
"grad_norm": 26.36361312866211,
"learning_rate": 9.638909772193478e-08,
"logits/chosen": -2.3842902183532715,
"logits/rejected": -2.242034435272217,
"logps/chosen": -85.19241333007812,
"logps/rejected": -81.81166076660156,
"loss": 0.6018,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": -0.12717826664447784,
"rewards/margins": 0.21827277541160583,
"rewards/rejected": -0.3454510569572449,
"step": 310
},
{
"epoch": 0.6481833143435878,
"grad_norm": 26.719867706298828,
"learning_rate": 9.593587163200753e-08,
"logits/chosen": -2.4053542613983154,
"logits/rejected": -2.277993679046631,
"logps/chosen": -82.21893310546875,
"logps/rejected": -82.72879791259766,
"loss": 0.5946,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.14859510958194733,
"rewards/margins": 0.2397138774394989,
"rewards/rejected": -0.38830894231796265,
"step": 320
},
{
"epoch": 0.6684390429168249,
"grad_norm": 29.379661560058594,
"learning_rate": 9.545705523877943e-08,
"logits/chosen": -2.39337420463562,
"logits/rejected": -2.243393659591675,
"logps/chosen": -89.27748107910156,
"logps/rejected": -88.36190795898438,
"loss": 0.583,
"rewards/accuracies": 0.7593749761581421,
"rewards/chosen": -0.15845103561878204,
"rewards/margins": 0.27531546354293823,
"rewards/rejected": -0.43376651406288147,
"step": 330
},
{
"epoch": 0.6886947714900621,
"grad_norm": 29.51100730895996,
"learning_rate": 9.495291528491348e-08,
"logits/chosen": -2.4061636924743652,
"logits/rejected": -2.2737927436828613,
"logps/chosen": -77.5340805053711,
"logps/rejected": -78.92658996582031,
"loss": 0.6076,
"rewards/accuracies": 0.715624988079071,
"rewards/chosen": -0.18812718987464905,
"rewards/margins": 0.22010421752929688,
"rewards/rejected": -0.4082314372062683,
"step": 340
},
{
"epoch": 0.7089505000632992,
"grad_norm": 28.85425567626953,
"learning_rate": 9.442373262051371e-08,
"logits/chosen": -2.3706448078155518,
"logits/rejected": -2.207597017288208,
"logps/chosen": -84.70035552978516,
"logps/rejected": -79.90083312988281,
"loss": 0.596,
"rewards/accuracies": 0.7093750238418579,
"rewards/chosen": -0.1977623850107193,
"rewards/margins": 0.24996185302734375,
"rewards/rejected": -0.44772419333457947,
"step": 350
},
{
"epoch": 0.7292062286365363,
"grad_norm": 28.540573120117188,
"learning_rate": 9.386980204666698e-08,
"logits/chosen": -2.369175910949707,
"logits/rejected": -2.214489459991455,
"logps/chosen": -80.21549224853516,
"logps/rejected": -79.48722839355469,
"loss": 0.5941,
"rewards/accuracies": 0.7281249761581421,
"rewards/chosen": -0.21811044216156006,
"rewards/margins": 0.2636975646018982,
"rewards/rejected": -0.48180800676345825,
"step": 360
},
{
"epoch": 0.7494619572097734,
"grad_norm": 32.249847412109375,
"learning_rate": 9.3291432151213e-08,
"logits/chosen": -2.3587095737457275,
"logits/rejected": -2.218735456466675,
"logps/chosen": -85.94621276855469,
"logps/rejected": -85.6765365600586,
"loss": 0.5842,
"rewards/accuracies": 0.746874988079071,
"rewards/chosen": -0.21563701331615448,
"rewards/margins": 0.27373427152633667,
"rewards/rejected": -0.48937129974365234,
"step": 370
},
{
"epoch": 0.7697176857830105,
"grad_norm": 26.788440704345703,
"learning_rate": 9.26889451368339e-08,
"logits/chosen": -2.368450164794922,
"logits/rejected": -2.2382750511169434,
"logps/chosen": -84.80735778808594,
"logps/rejected": -83.58937072753906,
"loss": 0.5836,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.22893838584423065,
"rewards/margins": 0.27746590971946716,
"rewards/rejected": -0.5064042806625366,
"step": 380
},
{
"epoch": 0.7899734143562476,
"grad_norm": 26.557083129882812,
"learning_rate": 9.206267664155906e-08,
"logits/chosen": -2.389660120010376,
"logits/rejected": -2.2198188304901123,
"logps/chosen": -80.32451629638672,
"logps/rejected": -81.00599670410156,
"loss": 0.5576,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": -0.22655515372753143,
"rewards/margins": 0.3575701117515564,
"rewards/rejected": -0.584125280380249,
"step": 390
},
{
"epoch": 0.8102291429294848,
"grad_norm": 22.922771453857422,
"learning_rate": 9.141297555178535e-08,
"logits/chosen": -2.4124135971069336,
"logits/rejected": -2.25138521194458,
"logps/chosen": -73.63737487792969,
"logps/rejected": -74.43919372558594,
"loss": 0.5756,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.2471962869167328,
"rewards/margins": 0.31536445021629333,
"rewards/rejected": -0.5625607371330261,
"step": 400
},
{
"epoch": 0.8304848715027219,
"grad_norm": 27.076038360595703,
"learning_rate": 9.074020380791693e-08,
"logits/chosen": -2.387418270111084,
"logits/rejected": -2.233450412750244,
"logps/chosen": -75.89783477783203,
"logps/rejected": -77.44602966308594,
"loss": 0.5727,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.2607758641242981,
"rewards/margins": 0.3421139121055603,
"rewards/rejected": -0.6028897762298584,
"step": 410
},
{
"epoch": 0.850740600075959,
"grad_norm": 28.83265495300293,
"learning_rate": 9.004473620273263e-08,
"logits/chosen": -2.3343796730041504,
"logits/rejected": -2.207730293273926,
"logps/chosen": -80.99537658691406,
"logps/rejected": -83.44149017333984,
"loss": 0.5767,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.2705189883708954,
"rewards/margins": 0.32222992181777954,
"rewards/rejected": -0.5927489399909973,
"step": 420
},
{
"epoch": 0.8709963286491961,
"grad_norm": 29.984909057617188,
"learning_rate": 8.932696017259361e-08,
"logits/chosen": -2.3199007511138916,
"logits/rejected": -2.1576333045959473,
"logps/chosen": -85.59019470214844,
"logps/rejected": -84.45293426513672,
"loss": 0.5712,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.25998255610466003,
"rewards/margins": 0.3359551429748535,
"rewards/rejected": -0.5959377288818359,
"step": 430
},
{
"epoch": 0.8912520572224332,
"grad_norm": 25.058645248413086,
"learning_rate": 8.858727558160743e-08,
"logits/chosen": -2.3427436351776123,
"logits/rejected": -2.1888678073883057,
"logps/chosen": -82.66050720214844,
"logps/rejected": -82.61741638183594,
"loss": 0.5589,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.26411646604537964,
"rewards/margins": 0.3582358956336975,
"rewards/rejected": -0.6223524212837219,
"step": 440
},
{
"epoch": 0.9115077857956704,
"grad_norm": 27.469467163085938,
"learning_rate": 8.782609449886861e-08,
"logits/chosen": -2.325899839401245,
"logits/rejected": -2.1819796562194824,
"logps/chosen": -82.68738555908203,
"logps/rejected": -82.69558715820312,
"loss": 0.5571,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": -0.27801764011383057,
"rewards/margins": 0.37557533383369446,
"rewards/rejected": -0.6535929441452026,
"step": 450
},
{
"epoch": 0.9317635143689075,
"grad_norm": 25.38697052001953,
"learning_rate": 8.704384096890013e-08,
"logits/chosen": -2.3276028633117676,
"logits/rejected": -2.1806609630584717,
"logps/chosen": -84.02021789550781,
"logps/rejected": -83.25233459472656,
"loss": 0.5403,
"rewards/accuracies": 0.784375011920929,
"rewards/chosen": -0.2819129526615143,
"rewards/margins": 0.41642332077026367,
"rewards/rejected": -0.6983363032341003,
"step": 460
},
{
"epoch": 0.9520192429421446,
"grad_norm": 32.92365264892578,
"learning_rate": 8.62409507754235e-08,
"logits/chosen": -2.2575485706329346,
"logits/rejected": -2.1456449031829834,
"logps/chosen": -87.51969909667969,
"logps/rejected": -87.74186706542969,
"loss": 0.5594,
"rewards/accuracies": 0.7406250238418579,
"rewards/chosen": -0.3425014913082123,
"rewards/margins": 0.37962430715560913,
"rewards/rejected": -0.7221258878707886,
"step": 470
},
{
"epoch": 0.9722749715153817,
"grad_norm": 29.529056549072266,
"learning_rate": 8.541787119858902e-08,
"logits/chosen": -2.302694082260132,
"logits/rejected": -2.162090539932251,
"logps/chosen": -79.02600860595703,
"logps/rejected": -78.34095764160156,
"loss": 0.5721,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.35388362407684326,
"rewards/margins": 0.3509567975997925,
"rewards/rejected": -0.7048403024673462,
"step": 480
},
{
"epoch": 0.9925307000886188,
"grad_norm": 23.92174530029297,
"learning_rate": 8.457506076580162e-08,
"logits/chosen": -2.3030953407287598,
"logits/rejected": -2.158973217010498,
"logps/chosen": -81.10468292236328,
"logps/rejected": -83.77845764160156,
"loss": 0.547,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": -0.3333788514137268,
"rewards/margins": 0.4156631529331207,
"rewards/rejected": -0.7490419745445251,
"step": 490
},
{
"epoch": 1.012786428661856,
"grad_norm": 30.66814613342285,
"learning_rate": 8.371298899628089e-08,
"logits/chosen": -2.2549185752868652,
"logits/rejected": -2.122537612915039,
"logps/chosen": -83.01513671875,
"logps/rejected": -86.776123046875,
"loss": 0.5316,
"rewards/accuracies": 0.778124988079071,
"rewards/chosen": -0.380868136882782,
"rewards/margins": 0.471442848443985,
"rewards/rejected": -0.8523109555244446,
"step": 500
},
{
"epoch": 1.033042157235093,
"grad_norm": 26.109542846679688,
"learning_rate": 8.28321361394978e-08,
"logits/chosen": -2.2775070667266846,
"logits/rejected": -2.1255202293395996,
"logps/chosen": -81.81797790527344,
"logps/rejected": -84.06095123291016,
"loss": 0.5349,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -0.3651012182235718,
"rewards/margins": 0.4572354853153229,
"rewards/rejected": -0.8223366737365723,
"step": 510
},
{
"epoch": 1.0532978858083302,
"grad_norm": 28.901084899902344,
"learning_rate": 8.193299290763362e-08,
"logits/chosen": -2.2764482498168945,
"logits/rejected": -2.128359317779541,
"logps/chosen": -83.87080383300781,
"logps/rejected": -83.10397338867188,
"loss": 0.5413,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.38391047716140747,
"rewards/margins": 0.43836045265197754,
"rewards/rejected": -0.8222709894180298,
"step": 520
},
{
"epoch": 1.0735536143815674,
"grad_norm": 26.954652786254883,
"learning_rate": 8.101606020221038e-08,
"logits/chosen": -2.26556134223938,
"logits/rejected": -2.12338924407959,
"logps/chosen": -85.02649688720703,
"logps/rejected": -85.88096618652344,
"loss": 0.5519,
"rewards/accuracies": 0.7593749761581421,
"rewards/chosen": -0.4198201298713684,
"rewards/margins": 0.4140090048313141,
"rewards/rejected": -0.8338291049003601,
"step": 530
},
{
"epoch": 1.0938093429548044,
"grad_norm": 27.471698760986328,
"learning_rate": 8.008184883504472e-08,
"logits/chosen": -2.285780668258667,
"logits/rejected": -2.15956449508667,
"logps/chosen": -91.95730590820312,
"logps/rejected": -89.56153869628906,
"loss": 0.5388,
"rewards/accuracies": 0.778124988079071,
"rewards/chosen": -0.37540143728256226,
"rewards/margins": 0.4454117715358734,
"rewards/rejected": -0.8208131790161133,
"step": 540
},
{
"epoch": 1.1140650715280416,
"grad_norm": 27.839784622192383,
"learning_rate": 7.913087924368102e-08,
"logits/chosen": -2.272618055343628,
"logits/rejected": -2.146136522293091,
"logps/chosen": -82.60453033447266,
"logps/rejected": -84.6351547241211,
"loss": 0.5466,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": -0.3854660391807556,
"rewards/margins": 0.4508994221687317,
"rewards/rejected": -0.8363655209541321,
"step": 550
},
{
"epoch": 1.1343208001012786,
"grad_norm": 29.220504760742188,
"learning_rate": 7.816368120146224e-08,
"logits/chosen": -2.2264904975891113,
"logits/rejected": -2.1163620948791504,
"logps/chosen": -80.90209197998047,
"logps/rejected": -84.96456909179688,
"loss": 0.5371,
"rewards/accuracies": 0.778124988079071,
"rewards/chosen": -0.3850155472755432,
"rewards/margins": 0.47849899530410767,
"rewards/rejected": -0.8635146021842957,
"step": 560
},
{
"epoch": 1.1545765286745158,
"grad_norm": 24.049087524414062,
"learning_rate": 7.718079352239955e-08,
"logits/chosen": -2.2715275287628174,
"logits/rejected": -2.095773220062256,
"logps/chosen": -83.15594482421875,
"logps/rejected": -83.11366271972656,
"loss": 0.5244,
"rewards/accuracies": 0.778124988079071,
"rewards/chosen": -0.38344329595565796,
"rewards/margins": 0.4991793632507324,
"rewards/rejected": -0.8826227188110352,
"step": 570
},
{
"epoch": 1.174832257247753,
"grad_norm": 32.831993103027344,
"learning_rate": 7.618276376100587e-08,
"logits/chosen": -2.2670161724090576,
"logits/rejected": -2.106199264526367,
"logps/chosen": -81.29884338378906,
"logps/rejected": -82.74469757080078,
"loss": 0.5303,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.4227599501609802,
"rewards/margins": 0.4853205680847168,
"rewards/rejected": -0.9080804586410522,
"step": 580
},
{
"epoch": 1.19508798582099,
"grad_norm": 26.217998504638672,
"learning_rate": 7.517014790726011e-08,
"logits/chosen": -2.2205467224121094,
"logits/rejected": -2.0953052043914795,
"logps/chosen": -83.64337921142578,
"logps/rejected": -86.88542938232422,
"loss": 0.5513,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.43212181329727173,
"rewards/margins": 0.44605493545532227,
"rewards/rejected": -0.8781768083572388,
"step": 590
},
{
"epoch": 1.2153437143942272,
"grad_norm": 22.538532257080078,
"learning_rate": 7.414351007687187e-08,
"logits/chosen": -2.205540418624878,
"logits/rejected": -2.0615344047546387,
"logps/chosen": -88.0082015991211,
"logps/rejected": -85.7721176147461,
"loss": 0.5173,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -0.4331473708152771,
"rewards/margins": 0.5372229814529419,
"rewards/rejected": -0.9703702926635742,
"step": 600
},
{
"epoch": 1.2355994429674642,
"grad_norm": 29.271059036254883,
"learning_rate": 7.310342219701981e-08,
"logits/chosen": -2.2468390464782715,
"logits/rejected": -2.107861280441284,
"logps/chosen": -82.70673370361328,
"logps/rejected": -83.1991958618164,
"loss": 0.5279,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.4486440122127533,
"rewards/margins": 0.5201537013053894,
"rewards/rejected": -0.9687976837158203,
"step": 610
},
{
"epoch": 1.2558551715407014,
"grad_norm": 26.889476776123047,
"learning_rate": 7.205046368773794e-08,
"logits/chosen": -2.1630682945251465,
"logits/rejected": -2.018644332885742,
"logps/chosen": -86.0914535522461,
"logps/rejected": -87.1644515991211,
"loss": 0.5364,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -0.48391470313072205,
"rewards/margins": 0.474750280380249,
"rewards/rejected": -0.9586650133132935,
"step": 620
},
{
"epoch": 1.2761109001139386,
"grad_norm": 32.43236541748047,
"learning_rate": 7.098522113912808e-08,
"logits/chosen": -2.2398154735565186,
"logits/rejected": -2.0971333980560303,
"logps/chosen": -81.19099426269531,
"logps/rejected": -82.07807922363281,
"loss": 0.5372,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": -0.441434383392334,
"rewards/margins": 0.5117905735969543,
"rewards/rejected": -0.9532249569892883,
"step": 630
},
{
"epoch": 1.2963666286871756,
"grad_norm": 26.83734130859375,
"learning_rate": 6.990828798457764e-08,
"logits/chosen": -2.2109534740448,
"logits/rejected": -2.083139657974243,
"logps/chosen": -80.3914566040039,
"logps/rejected": -87.99849700927734,
"loss": 0.5314,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.4658758044242859,
"rewards/margins": 0.5169892311096191,
"rewards/rejected": -0.9828651547431946,
"step": 640
},
{
"epoch": 1.3166223572604128,
"grad_norm": 25.39097785949707,
"learning_rate": 6.882026417016541e-08,
"logits/chosen": -2.230027437210083,
"logits/rejected": -2.100419759750366,
"logps/chosen": -82.01457214355469,
"logps/rejected": -83.42109680175781,
"loss": 0.5346,
"rewards/accuracies": 0.746874988079071,
"rewards/chosen": -0.4833486080169678,
"rewards/margins": 0.4825879633426666,
"rewards/rejected": -0.9659366607666016,
"step": 650
},
{
"epoch": 1.3368780858336498,
"grad_norm": 30.31522560119629,
"learning_rate": 6.772175582043889e-08,
"logits/chosen": -2.1776843070983887,
"logits/rejected": -2.039802074432373,
"logps/chosen": -83.42335510253906,
"logps/rejected": -88.11833953857422,
"loss": 0.4909,
"rewards/accuracies": 0.828125,
"rewards/chosen": -0.45507392287254333,
"rewards/margins": 0.6273630857467651,
"rewards/rejected": -1.0824369192123413,
"step": 660
},
{
"epoch": 1.357133814406887,
"grad_norm": 34.97041702270508,
"learning_rate": 6.661337490075003e-08,
"logits/chosen": -2.2355475425720215,
"logits/rejected": -2.095804452896118,
"logps/chosen": -83.29669952392578,
"logps/rejected": -84.55702209472656,
"loss": 0.5307,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": -0.5031043291091919,
"rewards/margins": 0.5249863862991333,
"rewards/rejected": -1.0280907154083252,
"step": 670
},
{
"epoch": 1.377389542980124,
"grad_norm": 26.33791160583496,
"learning_rate": 6.549573887633676e-08,
"logits/chosen": -2.1734795570373535,
"logits/rejected": -2.028352737426758,
"logps/chosen": -83.61205291748047,
"logps/rejected": -84.9587173461914,
"loss": 0.4813,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.46062904596328735,
"rewards/margins": 0.6703583002090454,
"rewards/rejected": -1.1309874057769775,
"step": 680
},
{
"epoch": 1.3976452715533612,
"grad_norm": 35.68913269042969,
"learning_rate": 6.436947036834086e-08,
"logits/chosen": -2.1777005195617676,
"logits/rejected": -2.054405689239502,
"logps/chosen": -83.69397735595703,
"logps/rejected": -86.34680938720703,
"loss": 0.5313,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.518213152885437,
"rewards/margins": 0.5348538756370544,
"rewards/rejected": -1.0530669689178467,
"step": 690
},
{
"epoch": 1.4179010001265984,
"grad_norm": 30.295581817626953,
"learning_rate": 6.323519680695349e-08,
"logits/chosen": -2.1419105529785156,
"logits/rejected": -1.9936020374298096,
"logps/chosen": -90.16146087646484,
"logps/rejected": -89.37017822265625,
"loss": 0.522,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.5005131363868713,
"rewards/margins": 0.5668459534645081,
"rewards/rejected": -1.067359209060669,
"step": 700
},
{
"epoch": 1.4381567286998354,
"grad_norm": 26.36173439025879,
"learning_rate": 6.209355008188152e-08,
"logits/chosen": -2.1437783241271973,
"logits/rejected": -2.0539603233337402,
"logps/chosen": -89.0562973022461,
"logps/rejected": -92.04231262207031,
"loss": 0.5409,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.5396249294281006,
"rewards/margins": 0.5132231116294861,
"rewards/rejected": -1.0528481006622314,
"step": 710
},
{
"epoch": 1.4584124572730726,
"grad_norm": 27.820674896240234,
"learning_rate": 6.094516619032975e-08,
"logits/chosen": -2.1499810218811035,
"logits/rejected": -2.025269031524658,
"logps/chosen": -83.47040557861328,
"logps/rejected": -86.60690307617188,
"loss": 0.5196,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -0.49484682083129883,
"rewards/margins": 0.5879716277122498,
"rewards/rejected": -1.0828183889389038,
"step": 720
},
{
"epoch": 1.4786681858463098,
"grad_norm": 24.657302856445312,
"learning_rate": 5.979068488269468e-08,
"logits/chosen": -2.1996073722839355,
"logits/rejected": -2.0537047386169434,
"logps/chosen": -86.91001892089844,
"logps/rejected": -89.12824249267578,
"loss": 0.4914,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": -0.542976975440979,
"rewards/margins": 0.6267004013061523,
"rewards/rejected": -1.169677495956421,
"step": 730
},
{
"epoch": 1.4989239144195468,
"grad_norm": 28.313621520996094,
"learning_rate": 5.8630749306167556e-08,
"logits/chosen": -2.1813175678253174,
"logits/rejected": -2.0757999420166016,
"logps/chosen": -83.51826477050781,
"logps/rejected": -89.63159942626953,
"loss": 0.5115,
"rewards/accuracies": 0.7406250238418579,
"rewards/chosen": -0.515524685382843,
"rewards/margins": 0.594115674495697,
"rewards/rejected": -1.1096404790878296,
"step": 740
},
{
"epoch": 1.5191796429927837,
"grad_norm": 29.172090530395508,
"learning_rate": 5.7466005646445095e-08,
"logits/chosen": -2.1559250354766846,
"logits/rejected": -2.0057528018951416,
"logps/chosen": -83.4648208618164,
"logps/rejected": -85.3985366821289,
"loss": 0.4947,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.5010935068130493,
"rewards/margins": 0.6447556018829346,
"rewards/rejected": -1.1458488702774048,
"step": 750
},
{
"epoch": 1.5394353715660212,
"grad_norm": 30.766334533691406,
"learning_rate": 5.6297102767747325e-08,
"logits/chosen": -2.1724162101745605,
"logits/rejected": -2.0574355125427246,
"logps/chosen": -90.57199096679688,
"logps/rejected": -92.65487670898438,
"loss": 0.5309,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.5361162424087524,
"rewards/margins": 0.5525388121604919,
"rewards/rejected": -1.0886551141738892,
"step": 760
},
{
"epoch": 1.5596911001392582,
"grad_norm": 32.11360549926758,
"learning_rate": 5.512469185134354e-08,
"logits/chosen": -2.1918747425079346,
"logits/rejected": -2.054835796356201,
"logps/chosen": -84.68449401855469,
"logps/rejected": -87.59661865234375,
"loss": 0.4972,
"rewards/accuracies": 0.778124988079071,
"rewards/chosen": -0.5611705780029297,
"rewards/margins": 0.6253485083580017,
"rewards/rejected": -1.1865190267562866,
"step": 770
},
{
"epoch": 1.5799468287124951,
"grad_norm": 28.335988998413086,
"learning_rate": 5.394942603278726e-08,
"logits/chosen": -2.1388983726501465,
"logits/rejected": -2.0282931327819824,
"logps/chosen": -89.77628326416016,
"logps/rejected": -90.62747955322266,
"loss": 0.5258,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.524773895740509,
"rewards/margins": 0.5710369944572449,
"rewards/rejected": -1.095810890197754,
"step": 780
},
{
"epoch": 1.6002025572857324,
"grad_norm": 25.966876983642578,
"learning_rate": 5.277196003806249e-08,
"logits/chosen": -2.2010576725006104,
"logits/rejected": -2.0596330165863037,
"logps/chosen": -79.80914306640625,
"logps/rejected": -82.98551940917969,
"loss": 0.4784,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -0.4947393536567688,
"rewards/margins": 0.6838704347610474,
"rewards/rejected": -1.1786099672317505,
"step": 790
},
{
"epoch": 1.6204582858589696,
"grad_norm": 29.524734497070312,
"learning_rate": 5.1592949818844046e-08,
"logits/chosen": -2.1284611225128174,
"logits/rejected": -2.0030832290649414,
"logps/chosen": -86.19153594970703,
"logps/rejected": -87.28998565673828,
"loss": 0.5126,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": -0.6002845764160156,
"rewards/margins": 0.6240141987800598,
"rewards/rejected": -1.2242988348007202,
"step": 800
},
{
"epoch": 1.6407140144322065,
"grad_norm": 25.266782760620117,
"learning_rate": 5.0413052187075054e-08,
"logits/chosen": -2.168487787246704,
"logits/rejected": -2.0182714462280273,
"logps/chosen": -79.49079132080078,
"logps/rejected": -81.5199203491211,
"loss": 0.5011,
"rewards/accuracies": 0.778124988079071,
"rewards/chosen": -0.48622050881385803,
"rewards/margins": 0.6199517846107483,
"rewards/rejected": -1.1061723232269287,
"step": 810
},
{
"epoch": 1.6609697430054438,
"grad_norm": 26.315034866333008,
"learning_rate": 4.9232924449065095e-08,
"logits/chosen": -2.1593496799468994,
"logits/rejected": -2.030149459838867,
"logps/chosen": -81.82089233398438,
"logps/rejected": -89.67662811279297,
"loss": 0.5042,
"rewards/accuracies": 0.778124988079071,
"rewards/chosen": -0.5738162994384766,
"rewards/margins": 0.6879658102989197,
"rewards/rejected": -1.261782169342041,
"step": 820
},
{
"epoch": 1.681225471578681,
"grad_norm": 33.470664978027344,
"learning_rate": 4.8053224039313114e-08,
"logits/chosen": -2.1270673274993896,
"logits/rejected": -2.012338638305664,
"logps/chosen": -86.01063537597656,
"logps/rejected": -85.80992889404297,
"loss": 0.5459,
"rewards/accuracies": 0.715624988079071,
"rewards/chosen": -0.535088837146759,
"rewards/margins": 0.49604400992393494,
"rewards/rejected": -1.0311328172683716,
"step": 830
},
{
"epoch": 1.701481200151918,
"grad_norm": 20.869911193847656,
"learning_rate": 4.687460815425878e-08,
"logits/chosen": -2.157341480255127,
"logits/rejected": -2.007072925567627,
"logps/chosen": -80.98677825927734,
"logps/rejected": -83.40042114257812,
"loss": 0.4965,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -0.568179190158844,
"rewards/margins": 0.660231351852417,
"rewards/rejected": -1.2284104824066162,
"step": 840
},
{
"epoch": 1.721736928725155,
"grad_norm": 23.50938606262207,
"learning_rate": 4.5697733386166524e-08,
"logits/chosen": -2.1210384368896484,
"logits/rejected": -1.9905798435211182,
"logps/chosen": -87.40711975097656,
"logps/rejected": -86.93902587890625,
"loss": 0.5181,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.5617082715034485,
"rewards/margins": 0.592042088508606,
"rewards/rejected": -1.1537501811981201,
"step": 850
},
{
"epoch": 1.7419926572983924,
"grad_norm": 27.28333282470703,
"learning_rate": 4.4523255357346187e-08,
"logits/chosen": -2.1478943824768066,
"logits/rejected": -2.024747371673584,
"logps/chosen": -84.65662384033203,
"logps/rejected": -88.84449005126953,
"loss": 0.4993,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.5495315790176392,
"rewards/margins": 0.6314437985420227,
"rewards/rejected": -1.1809751987457275,
"step": 860
},
{
"epoch": 1.7622483858716294,
"grad_norm": 25.31683349609375,
"learning_rate": 4.335182835491387e-08,
"logits/chosen": -2.1592297554016113,
"logits/rejected": -2.031510829925537,
"logps/chosen": -83.72755432128906,
"logps/rejected": -90.34858703613281,
"loss": 0.5023,
"rewards/accuracies": 0.7718750238418579,
"rewards/chosen": -0.6199240684509277,
"rewards/margins": 0.634971022605896,
"rewards/rejected": -1.2548949718475342,
"step": 870
},
{
"epoch": 1.7825041144448663,
"grad_norm": 29.31256103515625,
"learning_rate": 4.218410496629684e-08,
"logits/chosen": -2.1241517066955566,
"logits/rejected": -1.9871305227279663,
"logps/chosen": -76.77335357666016,
"logps/rejected": -80.37085723876953,
"loss": 0.4854,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.5658852458000183,
"rewards/margins": 0.6777737736701965,
"rewards/rejected": -1.2436590194702148,
"step": 880
},
{
"epoch": 1.8027598430181035,
"grad_norm": 29.485347747802734,
"learning_rate": 4.102073571568516e-08,
"logits/chosen": -2.1224985122680664,
"logits/rejected": -1.993857741355896,
"logps/chosen": -86.9496078491211,
"logps/rejected": -88.30699157714844,
"loss": 0.506,
"rewards/accuracies": 0.7718750238418579,
"rewards/chosen": -0.5687106847763062,
"rewards/margins": 0.6147977709770203,
"rewards/rejected": -1.1835086345672607,
"step": 890
},
{
"epoch": 1.8230155715913408,
"grad_norm": 27.672901153564453,
"learning_rate": 3.986236870163262e-08,
"logits/chosen": -2.1112308502197266,
"logits/rejected": -1.99734628200531,
"logps/chosen": -85.15098571777344,
"logps/rejected": -90.8910140991211,
"loss": 0.5078,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.5743888020515442,
"rewards/margins": 0.6254476308822632,
"rewards/rejected": -1.1998364925384521,
"step": 900
},
{
"epoch": 1.8432713001645777,
"grad_norm": 28.670635223388672,
"learning_rate": 3.870964923600923e-08,
"logits/chosen": -2.088013172149658,
"logits/rejected": -1.9703779220581055,
"logps/chosen": -83.16795349121094,
"logps/rejected": -86.55892181396484,
"loss": 0.496,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.5139526724815369,
"rewards/margins": 0.6616466045379639,
"rewards/rejected": -1.1755993366241455,
"step": 910
},
{
"epoch": 1.863527028737815,
"grad_norm": 30.791370391845703,
"learning_rate": 3.756321948450599e-08,
"logits/chosen": -2.120954751968384,
"logits/rejected": -1.9975354671478271,
"logps/chosen": -84.26751708984375,
"logps/rejected": -86.91252899169922,
"loss": 0.5453,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.6508383750915527,
"rewards/margins": 0.5812313556671143,
"rewards/rejected": -1.232069730758667,
"step": 920
},
{
"epoch": 1.8837827573110522,
"grad_norm": 33.708187103271484,
"learning_rate": 3.642371810889222e-08,
"logits/chosen": -2.092048168182373,
"logits/rejected": -1.9911048412322998,
"logps/chosen": -84.46561431884766,
"logps/rejected": -88.94342041015625,
"loss": 0.5404,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.5762837529182434,
"rewards/margins": 0.5255244970321655,
"rewards/rejected": -1.1018081903457642,
"step": 930
},
{
"epoch": 1.9040384858842891,
"grad_norm": 25.534833908081055,
"learning_rate": 3.529177991122518e-08,
"logits/chosen": -2.066344738006592,
"logits/rejected": -1.9486182928085327,
"logps/chosen": -91.85676574707031,
"logps/rejected": -94.47185516357422,
"loss": 0.493,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.5671579837799072,
"rewards/margins": 0.6500786542892456,
"rewards/rejected": -1.2172366380691528,
"step": 940
},
{
"epoch": 1.9242942144575261,
"grad_norm": 26.457944869995117,
"learning_rate": 3.416803548020969e-08,
"logits/chosen": -2.115591049194336,
"logits/rejected": -1.9885094165802002,
"logps/chosen": -89.46985626220703,
"logps/rejected": -93.4120864868164,
"loss": 0.5111,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": -0.5443202257156372,
"rewards/margins": 0.6258162260055542,
"rewards/rejected": -1.1701364517211914,
"step": 950
},
{
"epoch": 1.9445499430307633,
"grad_norm": 27.294607162475586,
"learning_rate": 3.305311083990496e-08,
"logits/chosen": -2.1644487380981445,
"logits/rejected": -2.040801525115967,
"logps/chosen": -76.92610931396484,
"logps/rejected": -82.45623016357422,
"loss": 0.5196,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.6151694655418396,
"rewards/margins": 0.5893052220344543,
"rewards/rejected": -1.204474687576294,
"step": 960
},
{
"epoch": 1.9648056716040005,
"grad_norm": 27.27229881286621,
"learning_rate": 3.194762710097436e-08,
"logits/chosen": -2.1350479125976562,
"logits/rejected": -2.030987501144409,
"logps/chosen": -80.64533996582031,
"logps/rejected": -85.28868103027344,
"loss": 0.5371,
"rewards/accuracies": 0.7281249761581421,
"rewards/chosen": -0.5596300363540649,
"rewards/margins": 0.5202323794364929,
"rewards/rejected": -1.0798624753952026,
"step": 970
},
{
"epoch": 1.9850614001772375,
"grad_norm": 24.55060386657715,
"learning_rate": 3.0852200114672453e-08,
"logits/chosen": -2.127375602722168,
"logits/rejected": -1.991199254989624,
"logps/chosen": -85.77812957763672,
"logps/rejected": -89.46642303466797,
"loss": 0.5081,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.5709745287895203,
"rewards/margins": 0.6586155891418457,
"rewards/rejected": -1.2295901775360107,
"step": 980
},
{
"epoch": 2.005317128750475,
"grad_norm": 25.645376205444336,
"learning_rate": 2.976744012976189e-08,
"logits/chosen": -2.1159276962280273,
"logits/rejected": -1.9866254329681396,
"logps/chosen": -87.69465637207031,
"logps/rejected": -88.93052673339844,
"loss": 0.4894,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.5695109367370605,
"rewards/margins": 0.6490300893783569,
"rewards/rejected": -1.218540906906128,
"step": 990
},
{
"epoch": 2.025572857323712,
"grad_norm": 27.878236770629883,
"learning_rate": 2.8693951452551307e-08,
"logits/chosen": -2.0782949924468994,
"logits/rejected": -1.9823192358016968,
"logps/chosen": -79.88198852539062,
"logps/rejected": -86.08316802978516,
"loss": 0.5315,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.5980249643325806,
"rewards/margins": 0.5769001245498657,
"rewards/rejected": -1.1749250888824463,
"step": 1000
},
{
"epoch": 2.045828585896949,
"grad_norm": 28.799379348754883,
"learning_rate": 2.7632332110243967e-08,
"logits/chosen": -2.0895416736602783,
"logits/rejected": -1.975619912147522,
"logps/chosen": -86.46625518798828,
"logps/rejected": -90.41603088378906,
"loss": 0.4972,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -0.5257114171981812,
"rewards/margins": 0.6659296751022339,
"rewards/rejected": -1.1916412115097046,
"step": 1010
},
{
"epoch": 2.066084314470186,
"grad_norm": 48.88608169555664,
"learning_rate": 2.658317351778412e-08,
"logits/chosen": -2.099612236022949,
"logits/rejected": -1.9862468242645264,
"logps/chosen": -86.78905487060547,
"logps/rejected": -92.29573059082031,
"loss": 0.4904,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.564150869846344,
"rewards/margins": 0.7153151631355286,
"rewards/rejected": -1.279465913772583,
"step": 1020
},
{
"epoch": 2.0863400430434234,
"grad_norm": 24.239578247070312,
"learning_rate": 2.554706014838705e-08,
"logits/chosen": -2.1574556827545166,
"logits/rejected": -2.014895439147949,
"logps/chosen": -84.7563247680664,
"logps/rejected": -85.77194213867188,
"loss": 0.4855,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.5780550837516785,
"rewards/margins": 0.6501516103744507,
"rewards/rejected": -1.2282066345214844,
"step": 1030
},
{
"epoch": 2.1065957716166603,
"grad_norm": 28.262638092041016,
"learning_rate": 2.4524569207936445e-08,
"logits/chosen": -2.0934982299804688,
"logits/rejected": -1.962937355041504,
"logps/chosen": -86.09654235839844,
"logps/rejected": -91.18133544921875,
"loss": 0.4598,
"rewards/accuracies": 0.815625011920929,
"rewards/chosen": -0.567767322063446,
"rewards/margins": 0.8064430356025696,
"rewards/rejected": -1.374210238456726,
"step": 1040
},
{
"epoch": 2.1268515001898973,
"grad_norm": 30.161561965942383,
"learning_rate": 2.351627031343008e-08,
"logits/chosen": -2.134225368499756,
"logits/rejected": -1.9962198734283447,
"logps/chosen": -87.08121490478516,
"logps/rejected": -90.91963958740234,
"loss": 0.5068,
"rewards/accuracies": 0.7718750238418579,
"rewards/chosen": -0.5947022438049316,
"rewards/margins": 0.6418746113777161,
"rewards/rejected": -1.236576795578003,
"step": 1050
},
{
"epoch": 2.1471072287631348,
"grad_norm": 29.3469295501709,
"learning_rate": 2.2522725175653233e-08,
"logits/chosen": -2.0764639377593994,
"logits/rejected": -1.9766466617584229,
"logps/chosen": -87.24481964111328,
"logps/rejected": -93.39856719970703,
"loss": 0.533,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -0.6166614294052124,
"rewards/margins": 0.5650585293769836,
"rewards/rejected": -1.1817197799682617,
"step": 1060
},
{
"epoch": 2.1673629573363717,
"grad_norm": 23.277862548828125,
"learning_rate": 2.154448728625668e-08,
"logits/chosen": -2.1141998767852783,
"logits/rejected": -1.9909133911132812,
"logps/chosen": -84.21327209472656,
"logps/rejected": -86.62323760986328,
"loss": 0.4652,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.5331219434738159,
"rewards/margins": 0.7241859436035156,
"rewards/rejected": -1.257307767868042,
"step": 1070
},
{
"epoch": 2.1876186859096087,
"grad_norm": 26.303924560546875,
"learning_rate": 2.0582101609413333e-08,
"logits/chosen": -2.0846378803253174,
"logits/rejected": -1.9430017471313477,
"logps/chosen": -90.30846405029297,
"logps/rejected": -91.7738265991211,
"loss": 0.483,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.5784596800804138,
"rewards/margins": 0.7019392251968384,
"rewards/rejected": -1.2803988456726074,
"step": 1080
},
{
"epoch": 2.2078744144828457,
"grad_norm": 29.66044044494629,
"learning_rate": 1.9636104278225413e-08,
"logits/chosen": -2.113520860671997,
"logits/rejected": -2.006913661956787,
"logps/chosen": -87.39739990234375,
"logps/rejected": -91.94351959228516,
"loss": 0.4885,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -0.6303068995475769,
"rewards/margins": 0.7132034301757812,
"rewards/rejected": -1.343510389328003,
"step": 1090
},
{
"epoch": 2.228130143056083,
"grad_norm": 31.82991600036621,
"learning_rate": 1.8707022296051462e-08,
"logits/chosen": -2.1318724155426025,
"logits/rejected": -1.997719407081604,
"logps/chosen": -83.96778869628906,
"logps/rejected": -91.7020492553711,
"loss": 0.4837,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.5827969312667847,
"rewards/margins": 0.7294033765792847,
"rewards/rejected": -1.3122001886367798,
"step": 1100
},
{
"epoch": 2.24838587162932,
"grad_norm": 30.810279846191406,
"learning_rate": 1.779537324291926e-08,
"logits/chosen": -2.087120771408081,
"logits/rejected": -1.974585771560669,
"logps/chosen": -85.29585266113281,
"logps/rejected": -90.92556762695312,
"loss": 0.5036,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.6058118939399719,
"rewards/margins": 0.6656503081321716,
"rewards/rejected": -1.2714622020721436,
"step": 1110
},
{
"epoch": 2.268641600202557,
"grad_norm": 24.00498390197754,
"learning_rate": 1.6901664987188425e-08,
"logits/chosen": -2.0903851985931396,
"logits/rejected": -1.9751678705215454,
"logps/chosen": -83.34523010253906,
"logps/rejected": -86.21971130371094,
"loss": 0.5044,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.5954999923706055,
"rewards/margins": 0.6783249378204346,
"rewards/rejected": -1.27382493019104,
"step": 1120
},
{
"epoch": 2.2888973287757945,
"grad_norm": 39.50699234008789,
"learning_rate": 1.6026395402623272e-08,
"logits/chosen": -2.0663511753082275,
"logits/rejected": -1.9365609884262085,
"logps/chosen": -94.30004119873047,
"logps/rejected": -97.46401977539062,
"loss": 0.4973,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -0.6117586493492126,
"rewards/margins": 0.6866195797920227,
"rewards/rejected": -1.2983782291412354,
"step": 1130
},
{
"epoch": 2.3091530573490315,
"grad_norm": 26.588275909423828,
"learning_rate": 1.5170052091033552e-08,
"logits/chosen": -2.1126387119293213,
"logits/rejected": -1.959896445274353,
"logps/chosen": -80.80674743652344,
"logps/rejected": -83.63319396972656,
"loss": 0.4726,
"rewards/accuracies": 0.8218749761581421,
"rewards/chosen": -0.616263747215271,
"rewards/margins": 0.7580442428588867,
"rewards/rejected": -1.3743079900741577,
"step": 1140
},
{
"epoch": 2.3294087859222685,
"grad_norm": 28.325511932373047,
"learning_rate": 1.4333112110637453e-08,
"logits/chosen": -2.064669609069824,
"logits/rejected": -1.9410665035247803,
"logps/chosen": -84.78388977050781,
"logps/rejected": -87.63703155517578,
"loss": 0.493,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.5784262418746948,
"rewards/margins": 0.691449761390686,
"rewards/rejected": -1.2698760032653809,
"step": 1150
},
{
"epoch": 2.349664514495506,
"grad_norm": 27.597017288208008,
"learning_rate": 1.3516041710298498e-08,
"logits/chosen": -2.1402578353881836,
"logits/rejected": -2.004826068878174,
"logps/chosen": -87.74010467529297,
"logps/rejected": -89.4506607055664,
"loss": 0.5047,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.5920398831367493,
"rewards/margins": 0.6377407908439636,
"rewards/rejected": -1.2297805547714233,
"step": 1160
},
{
"epoch": 2.369920243068743,
"grad_norm": 33.10106658935547,
"learning_rate": 1.2719296069784063e-08,
"logits/chosen": -2.062407970428467,
"logits/rejected": -1.9447336196899414,
"logps/chosen": -89.19010925292969,
"logps/rejected": -95.0318374633789,
"loss": 0.4953,
"rewards/accuracies": 0.8187500238418579,
"rewards/chosen": -0.5857541561126709,
"rewards/margins": 0.7076044082641602,
"rewards/rejected": -1.293358564376831,
"step": 1170
},
{
"epoch": 2.39017597164198,
"grad_norm": 33.07532501220703,
"learning_rate": 1.1943319046190332e-08,
"logits/chosen": -2.074035167694092,
"logits/rejected": -1.965685486793518,
"logps/chosen": -80.5416030883789,
"logps/rejected": -84.71125030517578,
"loss": 0.4871,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": -0.5754114389419556,
"rewards/margins": 0.7324446439743042,
"rewards/rejected": -1.3078559637069702,
"step": 1180
},
{
"epoch": 2.4104317002152174,
"grad_norm": 26.195051193237305,
"learning_rate": 1.1188542926675104e-08,
"logits/chosen": -2.117806911468506,
"logits/rejected": -1.9781955480575562,
"logps/chosen": -86.0428466796875,
"logps/rejected": -90.29086303710938,
"loss": 0.4505,
"rewards/accuracies": 0.831250011920929,
"rewards/chosen": -0.5670899152755737,
"rewards/margins": 0.8041001558303833,
"rewards/rejected": -1.371190071105957,
"step": 1190
},
{
"epoch": 2.4306874287884543,
"grad_norm": 23.817218780517578,
"learning_rate": 1.0455388187635933e-08,
"logits/chosen": -2.1228573322296143,
"logits/rejected": -1.9943969249725342,
"logps/chosen": -81.49883270263672,
"logps/rejected": -83.27964782714844,
"loss": 0.4844,
"rewards/accuracies": 0.7718750238418579,
"rewards/chosen": -0.6024131774902344,
"rewards/margins": 0.6800934076309204,
"rewards/rejected": -1.2825065851211548,
"step": 1200
},
{
"epoch": 2.4509431573616913,
"grad_norm": 30.316747665405273,
"learning_rate": 9.744263260468005e-09,
"logits/chosen": -2.059378147125244,
"logits/rejected": -1.9458458423614502,
"logps/chosen": -92.07587432861328,
"logps/rejected": -95.30807495117188,
"loss": 0.4914,
"rewards/accuracies": 0.7906249761581421,
"rewards/chosen": -0.5844647884368896,
"rewards/margins": 0.7100616097450256,
"rewards/rejected": -1.2945263385772705,
"step": 1210
},
{
"epoch": 2.4711988859349283,
"grad_norm": 28.162200927734375,
"learning_rate": 9.055564304031981e-09,
"logits/chosen": -2.082139730453491,
"logits/rejected": -1.9701576232910156,
"logps/chosen": -86.1491928100586,
"logps/rejected": -92.10936737060547,
"loss": 0.4954,
"rewards/accuracies": 0.7718750238418579,
"rewards/chosen": -0.5959383845329285,
"rewards/margins": 0.7093779444694519,
"rewards/rejected": -1.3053163290023804,
"step": 1220
},
{
"epoch": 2.4914546145081657,
"grad_norm": 29.766904830932617,
"learning_rate": 8.38967498395895e-09,
"logits/chosen": -2.1094155311584473,
"logits/rejected": -1.9789727926254272,
"logps/chosen": -80.73603820800781,
"logps/rejected": -84.64231872558594,
"loss": 0.4904,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.5383921265602112,
"rewards/margins": 0.6760575175285339,
"rewards/rejected": -1.2144496440887451,
"step": 1230
},
{
"epoch": 2.5117103430814027,
"grad_norm": 35.819358825683594,
"learning_rate": 7.746966258914988e-09,
"logits/chosen": -2.1154112815856934,
"logits/rejected": -1.9815971851348877,
"logps/chosen": -86.95155334472656,
"logps/rejected": -85.64454650878906,
"loss": 0.5197,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.6145626306533813,
"rewards/margins": 0.631356418132782,
"rewards/rejected": -1.2459189891815186,
"step": 1240
},
{
"epoch": 2.5319660716546397,
"grad_norm": 29.06343650817871,
"learning_rate": 7.127796173944695e-09,
"logits/chosen": -2.085669994354248,
"logits/rejected": -1.9675817489624023,
"logps/chosen": -89.59977722167969,
"logps/rejected": -91.9139404296875,
"loss": 0.4879,
"rewards/accuracies": 0.796875,
"rewards/chosen": -0.5644815564155579,
"rewards/margins": 0.7162417769432068,
"rewards/rejected": -1.2807233333587646,
"step": 1250
},
{
"epoch": 2.552221800227877,
"grad_norm": 32.731590270996094,
"learning_rate": 6.532509661008789e-09,
"logits/chosen": -2.1157116889953613,
"logits/rejected": -1.9942877292633057,
"logps/chosen": -81.25712585449219,
"logps/rejected": -85.86751556396484,
"loss": 0.4855,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.5965838432312012,
"rewards/margins": 0.7261613011360168,
"rewards/rejected": -1.3227452039718628,
"step": 1260
},
{
"epoch": 2.572477528801114,
"grad_norm": 21.976213455200195,
"learning_rate": 5.9614383468267916e-09,
"logits/chosen": -2.064387321472168,
"logits/rejected": -1.939223289489746,
"logps/chosen": -87.50946807861328,
"logps/rejected": -90.81806945800781,
"loss": 0.4677,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.5381378531455994,
"rewards/margins": 0.7613744735717773,
"rewards/rejected": -1.2995123863220215,
"step": 1270
},
{
"epoch": 2.592733257374351,
"grad_norm": 26.645793914794922,
"learning_rate": 5.4149003681318525e-09,
"logits/chosen": -2.0983309745788574,
"logits/rejected": -1.9681360721588135,
"logps/chosen": -86.5027847290039,
"logps/rejected": -88.28022003173828,
"loss": 0.4846,
"rewards/accuracies": 0.784375011920929,
"rewards/chosen": -0.6178566217422485,
"rewards/margins": 0.7156898975372314,
"rewards/rejected": -1.3335466384887695,
"step": 1280
},
{
"epoch": 2.612988985947588,
"grad_norm": 34.658424377441406,
"learning_rate": 4.8932001944408e-09,
"logits/chosen": -2.114567995071411,
"logits/rejected": -2.003492832183838,
"logps/chosen": -86.31463623046875,
"logps/rejected": -87.3377456665039,
"loss": 0.5289,
"rewards/accuracies": 0.746874988079071,
"rewards/chosen": -0.5726319551467896,
"rewards/margins": 0.5831801891326904,
"rewards/rejected": -1.1558120250701904,
"step": 1290
},
{
"epoch": 2.6332447145208255,
"grad_norm": 29.894866943359375,
"learning_rate": 4.396628458437912e-09,
"logits/chosen": -2.057438373565674,
"logits/rejected": -1.9271215200424194,
"logps/chosen": -86.70679473876953,
"logps/rejected": -89.4132308959961,
"loss": 0.4833,
"rewards/accuracies": 0.828125,
"rewards/chosen": -0.5987198948860168,
"rewards/margins": 0.7157739996910095,
"rewards/rejected": -1.3144938945770264,
"step": 1300
},
{
"epoch": 2.6535004430940625,
"grad_norm": 28.439125061035156,
"learning_rate": 3.9254617940670474e-09,
"logits/chosen": -2.0954787731170654,
"logits/rejected": -1.9630225896835327,
"logps/chosen": -84.08492279052734,
"logps/rejected": -87.75090026855469,
"loss": 0.4635,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.523126482963562,
"rewards/margins": 0.7484750151634216,
"rewards/rejected": -1.2716015577316284,
"step": 1310
},
{
"epoch": 2.6737561716672995,
"grad_norm": 25.333951950073242,
"learning_rate": 3.479962682422366e-09,
"logits/chosen": -2.1200668811798096,
"logits/rejected": -1.9594342708587646,
"logps/chosen": -83.06964111328125,
"logps/rejected": -84.25973510742188,
"loss": 0.4752,
"rewards/accuracies": 0.8062499761581421,
"rewards/chosen": -0.6142527461051941,
"rewards/margins": 0.7399193048477173,
"rewards/rejected": -1.3541719913482666,
"step": 1320
},
{
"epoch": 2.694011900240537,
"grad_norm": 38.79462814331055,
"learning_rate": 3.0603793055233194e-09,
"logits/chosen": -2.078015089035034,
"logits/rejected": -1.9544031620025635,
"logps/chosen": -86.95954895019531,
"logps/rejected": -87.41011810302734,
"loss": 0.5153,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -0.6789627075195312,
"rewards/margins": 0.6111471652984619,
"rewards/rejected": -1.2901098728179932,
"step": 1330
},
{
"epoch": 2.714267628813774,
"grad_norm": 25.715852737426758,
"learning_rate": 2.6669454080555707e-09,
"logits/chosen": -2.081672191619873,
"logits/rejected": -1.9599504470825195,
"logps/chosen": -81.85166931152344,
"logps/rejected": -84.5459213256836,
"loss": 0.4941,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.5518099069595337,
"rewards/margins": 0.6579364538192749,
"rewards/rejected": -1.2097463607788086,
"step": 1340
},
{
"epoch": 2.734523357387011,
"grad_norm": 31.674190521240234,
"learning_rate": 2.299880167154694e-09,
"logits/chosen": -2.0664610862731934,
"logits/rejected": -1.9412147998809814,
"logps/chosen": -86.22771453857422,
"logps/rejected": -90.61781311035156,
"loss": 0.4909,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.5787885785102844,
"rewards/margins": 0.6425621509552002,
"rewards/rejected": -1.2213506698608398,
"step": 1350
},
{
"epoch": 2.754779085960248,
"grad_norm": 21.180307388305664,
"learning_rate": 1.959388070305368e-09,
"logits/chosen": -2.1191658973693848,
"logits/rejected": -1.9780559539794922,
"logps/chosen": -84.12895202636719,
"logps/rejected": -87.06913757324219,
"loss": 0.4683,
"rewards/accuracies": 0.784375011920929,
"rewards/chosen": -0.598002016544342,
"rewards/margins": 0.7502027750015259,
"rewards/rejected": -1.3482048511505127,
"step": 1360
},
{
"epoch": 2.7750348145334853,
"grad_norm": 31.467397689819336,
"learning_rate": 1.6456588014238826e-09,
"logits/chosen": -2.0679941177368164,
"logits/rejected": -1.984758973121643,
"logps/chosen": -82.20966339111328,
"logps/rejected": -89.56126403808594,
"loss": 0.5158,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.6504064798355103,
"rewards/margins": 0.6637855768203735,
"rewards/rejected": -1.3141919374465942,
"step": 1370
},
{
"epoch": 2.7952905431067223,
"grad_norm": 25.813554763793945,
"learning_rate": 1.3588671351876358e-09,
"logits/chosen": -2.088512897491455,
"logits/rejected": -1.9785858392715454,
"logps/chosen": -86.8198013305664,
"logps/rejected": -90.5335922241211,
"loss": 0.4835,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.6225306987762451,
"rewards/margins": 0.7479228973388672,
"rewards/rejected": -1.3704535961151123,
"step": 1380
},
{
"epoch": 2.8155462716799597,
"grad_norm": 27.169641494750977,
"learning_rate": 1.099172839670298e-09,
"logits/chosen": -2.0676891803741455,
"logits/rejected": -1.9737918376922607,
"logps/chosen": -78.63626861572266,
"logps/rejected": -83.41529083251953,
"loss": 0.5417,
"rewards/accuracies": 0.715624988079071,
"rewards/chosen": -0.6757029294967651,
"rewards/margins": 0.6025967597961426,
"rewards/rejected": -1.2782996892929077,
"step": 1390
},
{
"epoch": 2.8358020002531967,
"grad_norm": 27.115802764892578,
"learning_rate": 8.66720587337011e-10,
"logits/chosen": -2.065960645675659,
"logits/rejected": -1.9609451293945312,
"logps/chosen": -87.2630615234375,
"logps/rejected": -90.57234191894531,
"loss": 0.5015,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.619225263595581,
"rewards/margins": 0.6657966375350952,
"rewards/rejected": -1.2850219011306763,
"step": 1400
},
{
"epoch": 2.8560577288264337,
"grad_norm": 24.163036346435547,
"learning_rate": 6.616398744491825e-10,
"logits/chosen": -2.079883337020874,
"logits/rejected": -1.9338324069976807,
"logps/chosen": -88.01893615722656,
"logps/rejected": -89.63729095458984,
"loss": 0.4572,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.5343765020370483,
"rewards/margins": 0.7622194886207581,
"rewards/rejected": -1.2965959310531616,
"step": 1410
},
{
"epoch": 2.8763134573996707,
"grad_norm": 33.94268798828125,
"learning_rate": 4.840449489236786e-10,
"logits/chosen": -2.0775258541107178,
"logits/rejected": -1.9470727443695068,
"logps/chosen": -83.77029418945312,
"logps/rejected": -87.10340881347656,
"loss": 0.478,
"rewards/accuracies": 0.809374988079071,
"rewards/chosen": -0.578041672706604,
"rewards/margins": 0.695804238319397,
"rewards/rejected": -1.273845911026001,
"step": 1420
},
{
"epoch": 2.8965691859729077,
"grad_norm": 31.864702224731445,
"learning_rate": 3.3403474668677324e-10,
"logits/chosen": -2.063178062438965,
"logits/rejected": -1.9387401342391968,
"logps/chosen": -88.2574234008789,
"logps/rejected": -90.79651641845703,
"loss": 0.5111,
"rewards/accuracies": 0.7875000238418579,
"rewards/chosen": -0.6772388219833374,
"rewards/margins": 0.6761503219604492,
"rewards/rejected": -1.3533891439437866,
"step": 1430
},
{
"epoch": 2.916824914546145,
"grad_norm": 23.90968894958496,
"learning_rate": 2.1169283655815274e-10,
"logits/chosen": -2.082099199295044,
"logits/rejected": -1.949532151222229,
"logps/chosen": -87.7041244506836,
"logps/rejected": -91.3929214477539,
"loss": 0.4748,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.6174831390380859,
"rewards/margins": 0.7841922044754028,
"rewards/rejected": -1.4016753435134888,
"step": 1440
},
{
"epoch": 2.937080643119382,
"grad_norm": 25.46677589416504,
"learning_rate": 1.1708737369576228e-10,
"logits/chosen": -2.081848621368408,
"logits/rejected": -1.9594366550445557,
"logps/chosen": -78.64682006835938,
"logps/rejected": -87.83316040039062,
"loss": 0.4826,
"rewards/accuracies": 0.8125,
"rewards/chosen": -0.5947951078414917,
"rewards/margins": 0.7158970832824707,
"rewards/rejected": -1.3106920719146729,
"step": 1450
},
{
"epoch": 2.9573363716926195,
"grad_norm": 34.160301208496094,
"learning_rate": 5.0271061627427115e-11,
"logits/chosen": -2.1119987964630127,
"logits/rejected": -1.9852664470672607,
"logps/chosen": -80.1048355102539,
"logps/rejected": -88.52069091796875,
"loss": 0.5029,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -0.6696670651435852,
"rewards/margins": 0.718749463558197,
"rewards/rejected": -1.3884165287017822,
"step": 1460
},
{
"epoch": 2.9775921002658565,
"grad_norm": 26.002410888671875,
"learning_rate": 1.1281122890355322e-11,
"logits/chosen": -2.076430082321167,
"logits/rejected": -1.9372243881225586,
"logps/chosen": -83.05271911621094,
"logps/rejected": -85.09577941894531,
"loss": 0.5069,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -0.542018711566925,
"rewards/margins": 0.6729723215103149,
"rewards/rejected": -1.2149909734725952,
"step": 1470
},
{
"epoch": 2.99582225598177,
"step": 1479,
"total_flos": 0.0,
"train_loss": 0.546259533964032,
"train_runtime": 14357.5178,
"train_samples_per_second": 3.301,
"train_steps_per_second": 0.103
}
],
"logging_steps": 10,
"max_steps": 1479,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 100,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}