phi-2-dpo-test-iter-0 / trainer_state.json
BraylonDash's picture
Model save
9237faa verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.9936,
"eval_steps": 100,
"global_step": 1248,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 4e-08,
"logits/chosen": 0.76749187707901,
"logits/rejected": 1.0001295804977417,
"logps/chosen": -205.27383422851562,
"logps/rejected": -130.56936645507812,
"loss": 0.0001,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.03,
"learning_rate": 4.0000000000000003e-07,
"logits/chosen": 0.7141001224517822,
"logits/rejected": 0.7724499106407166,
"logps/chosen": -190.74786376953125,
"logps/rejected": -138.00537109375,
"loss": 0.0001,
"rewards/accuracies": 0.4027777910232544,
"rewards/chosen": 0.0002450596366543323,
"rewards/margins": 0.0004758307186421007,
"rewards/rejected": -0.0002307710237801075,
"step": 10
},
{
"epoch": 0.06,
"learning_rate": 8.000000000000001e-07,
"logits/chosen": 0.601749062538147,
"logits/rejected": 0.879185676574707,
"logps/chosen": -173.01181030273438,
"logps/rejected": -116.30070495605469,
"loss": 0.0001,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": 4.058447757415706e-06,
"rewards/margins": -0.00013263085565995425,
"rewards/rejected": 0.00013668931205756962,
"step": 20
},
{
"epoch": 0.1,
"learning_rate": 1.2000000000000002e-06,
"logits/chosen": 0.6547126173973083,
"logits/rejected": 0.833116352558136,
"logps/chosen": -187.99520874023438,
"logps/rejected": -127.53055572509766,
"loss": 0.0001,
"rewards/accuracies": 0.45625001192092896,
"rewards/chosen": 2.28220596909523e-06,
"rewards/margins": 0.00018625140364747494,
"rewards/rejected": -0.00018396916857454926,
"step": 30
},
{
"epoch": 0.13,
"learning_rate": 1.6000000000000001e-06,
"logits/chosen": 0.5628782510757446,
"logits/rejected": 0.745602548122406,
"logps/chosen": -183.92758178710938,
"logps/rejected": -134.64688110351562,
"loss": 0.0001,
"rewards/accuracies": 0.45625001192092896,
"rewards/chosen": -0.0005207593785598874,
"rewards/margins": -0.0005336635513231158,
"rewards/rejected": 1.290418458665954e-05,
"step": 40
},
{
"epoch": 0.16,
"learning_rate": 2.0000000000000003e-06,
"logits/chosen": 0.6242247819900513,
"logits/rejected": 0.8314679265022278,
"logps/chosen": -171.67556762695312,
"logps/rejected": -120.4903335571289,
"loss": 0.0001,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.00020891983876936138,
"rewards/margins": 0.0005679951282218099,
"rewards/rejected": -0.00035907537676393986,
"step": 50
},
{
"epoch": 0.19,
"learning_rate": 2.4000000000000003e-06,
"logits/chosen": 0.7160875797271729,
"logits/rejected": 0.8556329607963562,
"logps/chosen": -185.2340087890625,
"logps/rejected": -124.66600036621094,
"loss": 0.0001,
"rewards/accuracies": 0.4312500059604645,
"rewards/chosen": 0.0006271885358728468,
"rewards/margins": 0.0005722854984924197,
"rewards/rejected": 5.4903095588088036e-05,
"step": 60
},
{
"epoch": 0.22,
"learning_rate": 2.8000000000000003e-06,
"logits/chosen": 0.6916844844818115,
"logits/rejected": 0.8196538090705872,
"logps/chosen": -161.62498474121094,
"logps/rejected": -107.52471923828125,
"loss": 0.0001,
"rewards/accuracies": 0.48750001192092896,
"rewards/chosen": 0.0007440428016707301,
"rewards/margins": 0.0006342666456475854,
"rewards/rejected": 0.00010977611964335665,
"step": 70
},
{
"epoch": 0.26,
"learning_rate": 3.2000000000000003e-06,
"logits/chosen": 0.6701158285140991,
"logits/rejected": 0.8293789029121399,
"logps/chosen": -177.7081298828125,
"logps/rejected": -122.1727294921875,
"loss": 0.0001,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": 0.0014677448198199272,
"rewards/margins": 0.000520342611707747,
"rewards/rejected": 0.0009474022081121802,
"step": 80
},
{
"epoch": 0.29,
"learning_rate": 3.6000000000000003e-06,
"logits/chosen": 0.6228800415992737,
"logits/rejected": 0.7212022542953491,
"logps/chosen": -183.03811645507812,
"logps/rejected": -139.0586700439453,
"loss": 0.0001,
"rewards/accuracies": 0.53125,
"rewards/chosen": 0.0022542846854776144,
"rewards/margins": 0.0009365282021462917,
"rewards/rejected": 0.0013177564833313227,
"step": 90
},
{
"epoch": 0.32,
"learning_rate": 4.000000000000001e-06,
"logits/chosen": 0.6344588994979858,
"logits/rejected": 0.8314794301986694,
"logps/chosen": -163.23043823242188,
"logps/rejected": -121.80052185058594,
"loss": 0.0001,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": 0.002225330099463463,
"rewards/margins": 0.0011866830755025148,
"rewards/rejected": 0.0010386471403762698,
"step": 100
},
{
"epoch": 0.32,
"eval_logits/chosen": 0.8262824416160583,
"eval_logits/rejected": 0.8840107321739197,
"eval_logps/chosen": -256.7341003417969,
"eval_logps/rejected": -233.68739318847656,
"eval_loss": 0.00015139963943511248,
"eval_rewards/accuracies": 0.5199999809265137,
"eval_rewards/chosen": -0.00119913334492594,
"eval_rewards/margins": 0.0002922326675616205,
"eval_rewards/rejected": -0.0014913661871105433,
"eval_runtime": 415.0818,
"eval_samples_per_second": 4.818,
"eval_steps_per_second": 1.205,
"step": 100
},
{
"epoch": 0.35,
"learning_rate": 4.4e-06,
"logits/chosen": 0.6864518523216248,
"logits/rejected": 0.9017802476882935,
"logps/chosen": -174.11480712890625,
"logps/rejected": -118.93388366699219,
"loss": 0.0001,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": 0.00283862859942019,
"rewards/margins": 0.0018309386214241385,
"rewards/rejected": 0.001007690210826695,
"step": 110
},
{
"epoch": 0.38,
"learning_rate": 4.800000000000001e-06,
"logits/chosen": 0.5682826638221741,
"logits/rejected": 0.6713820695877075,
"logps/chosen": -189.4928436279297,
"logps/rejected": -135.20498657226562,
"loss": 0.0001,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": 0.0033208150416612625,
"rewards/margins": 0.001955473329871893,
"rewards/rejected": 0.0013653415953740478,
"step": 120
},
{
"epoch": 0.42,
"learning_rate": 4.999755441268144e-06,
"logits/chosen": 0.7162402868270874,
"logits/rejected": 0.8464711904525757,
"logps/chosen": -175.83523559570312,
"logps/rejected": -126.2065200805664,
"loss": 0.0001,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": 0.0035425268579274416,
"rewards/margins": 0.0022287473548203707,
"rewards/rejected": 0.0013137792702764273,
"step": 130
},
{
"epoch": 0.45,
"learning_rate": 4.997799258487003e-06,
"logits/chosen": 0.6711681485176086,
"logits/rejected": 0.8102658987045288,
"logps/chosen": -182.16258239746094,
"logps/rejected": -128.60398864746094,
"loss": 0.0001,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": 0.0029964938294142485,
"rewards/margins": 0.002129464875906706,
"rewards/rejected": 0.0008670290117152035,
"step": 140
},
{
"epoch": 0.48,
"learning_rate": 4.993888423734898e-06,
"logits/chosen": 0.6097667217254639,
"logits/rejected": 0.7144483327865601,
"logps/chosen": -190.95156860351562,
"logps/rejected": -141.19229125976562,
"loss": 0.0001,
"rewards/accuracies": 0.59375,
"rewards/chosen": 0.0037058107554912567,
"rewards/margins": 0.002363733481615782,
"rewards/rejected": 0.001342077157460153,
"step": 150
},
{
"epoch": 0.51,
"learning_rate": 4.988025997434253e-06,
"logits/chosen": 0.6315719485282898,
"logits/rejected": 0.7987428307533264,
"logps/chosen": -181.68386840820312,
"logps/rejected": -128.19888305664062,
"loss": 0.0001,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": 0.002868856769055128,
"rewards/margins": 0.0025774172972887754,
"rewards/rejected": 0.00029143941355869174,
"step": 160
},
{
"epoch": 0.54,
"learning_rate": 4.980216567224801e-06,
"logits/chosen": 0.6035071611404419,
"logits/rejected": 0.8083668947219849,
"logps/chosen": -181.84107971191406,
"logps/rejected": -126.97474670410156,
"loss": 0.0001,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.0037610617000609636,
"rewards/margins": 0.0028130013961344957,
"rewards/rejected": 0.0009480599546805024,
"step": 170
},
{
"epoch": 0.58,
"learning_rate": 4.970466244373527e-06,
"logits/chosen": 0.6453949213027954,
"logits/rejected": 0.797042727470398,
"logps/chosen": -155.79519653320312,
"logps/rejected": -113.65773010253906,
"loss": 0.0001,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": 0.0029109427705407143,
"rewards/margins": 0.0023111128248274326,
"rewards/rejected": 0.0005998298292979598,
"step": 180
},
{
"epoch": 0.61,
"learning_rate": 4.958782658992307e-06,
"logits/chosen": 0.6763242483139038,
"logits/rejected": 0.8502823710441589,
"logps/chosen": -159.13275146484375,
"logps/rejected": -120.99656677246094,
"loss": 0.0001,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": 0.0026291562244296074,
"rewards/margins": 0.0016078405315056443,
"rewards/rejected": 0.001021315692923963,
"step": 190
},
{
"epoch": 0.64,
"learning_rate": 4.945174954066957e-06,
"logits/chosen": 0.7041156888008118,
"logits/rejected": 0.7947753667831421,
"logps/chosen": -177.0174560546875,
"logps/rejected": -127.36385345458984,
"loss": 0.0001,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.003798459889367223,
"rewards/margins": 0.0031563788652420044,
"rewards/rejected": 0.0006420810823328793,
"step": 200
},
{
"epoch": 0.64,
"eval_logits/chosen": 0.8200604319572449,
"eval_logits/rejected": 0.8777603507041931,
"eval_logps/chosen": -256.8277893066406,
"eval_logps/rejected": -233.76914978027344,
"eval_loss": 0.000158556635142304,
"eval_rewards/accuracies": 0.5005000233650208,
"eval_rewards/chosen": -0.00213597621768713,
"eval_rewards/margins": 0.00017285306239500642,
"eval_rewards/rejected": -0.0023088292218744755,
"eval_runtime": 412.468,
"eval_samples_per_second": 4.849,
"eval_steps_per_second": 1.212,
"step": 200
},
{
"epoch": 0.67,
"learning_rate": 4.929653778302397e-06,
"logits/chosen": 0.6099511981010437,
"logits/rejected": 0.811872661113739,
"logps/chosen": -192.26446533203125,
"logps/rejected": -131.25962829589844,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.0029747250955551863,
"rewards/margins": 0.0029428431298583746,
"rewards/rejected": 3.1882118491921574e-05,
"step": 210
},
{
"epoch": 0.7,
"learning_rate": 4.912231277789509e-06,
"logits/chosen": 0.6272088289260864,
"logits/rejected": 0.767717719078064,
"logps/chosen": -193.66490173339844,
"logps/rejected": -143.585205078125,
"loss": 0.0001,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.003646609140560031,
"rewards/margins": 0.0021339866798371077,
"rewards/rejected": 0.0015126224607229233,
"step": 220
},
{
"epoch": 0.74,
"learning_rate": 4.892921086500219e-06,
"logits/chosen": 0.6231340169906616,
"logits/rejected": 0.7231715321540833,
"logps/chosen": -160.1138458251953,
"logps/rejected": -114.69986724853516,
"loss": 0.0001,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.003955821972340345,
"rewards/margins": 0.003026761580258608,
"rewards/rejected": 0.0009290605084970593,
"step": 230
},
{
"epoch": 0.77,
"learning_rate": 4.871738315618236e-06,
"logits/chosen": 0.7013573050498962,
"logits/rejected": 0.8051062822341919,
"logps/chosen": -186.76821899414062,
"logps/rejected": -135.32489013671875,
"loss": 0.0001,
"rewards/accuracies": 0.59375,
"rewards/chosen": 0.003390461904928088,
"rewards/margins": 0.002614812459796667,
"rewards/rejected": 0.0007756495615467429,
"step": 240
},
{
"epoch": 0.8,
"learning_rate": 4.848699541713801e-06,
"logits/chosen": 0.6483660340309143,
"logits/rejected": 0.8247146606445312,
"logps/chosen": -159.59156799316406,
"logps/rejected": -118.78582763671875,
"loss": 0.0001,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": 0.0031985179521143436,
"rewards/margins": 0.0021302015520632267,
"rewards/rejected": 0.0010683165164664388,
"step": 250
},
{
"epoch": 0.83,
"learning_rate": 4.823822793771696e-06,
"logits/chosen": 0.625135064125061,
"logits/rejected": 0.7819581627845764,
"logps/chosen": -169.98837280273438,
"logps/rejected": -119.9397964477539,
"loss": 0.0001,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": 0.00324636441655457,
"rewards/margins": 0.0028731045313179493,
"rewards/rejected": 0.0003732596233021468,
"step": 260
},
{
"epoch": 0.86,
"learning_rate": 4.797127539082669e-06,
"logits/chosen": 0.649788498878479,
"logits/rejected": 0.8432229161262512,
"logps/chosen": -191.59213256835938,
"logps/rejected": -134.23428344726562,
"loss": 0.0001,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": 0.0034480884205549955,
"rewards/margins": 0.002935568569228053,
"rewards/rejected": 0.0005125202005729079,
"step": 270
},
{
"epoch": 0.9,
"learning_rate": 4.7686346680093135e-06,
"logits/chosen": 0.6038953065872192,
"logits/rejected": 0.8061636686325073,
"logps/chosen": -188.86422729492188,
"logps/rejected": -125.95369720458984,
"loss": 0.0001,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.004464815836399794,
"rewards/margins": 0.004056466277688742,
"rewards/rejected": 0.0004083492676727474,
"step": 280
},
{
"epoch": 0.93,
"learning_rate": 4.738366477638319e-06,
"logits/chosen": 0.5706946849822998,
"logits/rejected": 0.7165879011154175,
"logps/chosen": -172.26846313476562,
"logps/rejected": -113.09400939941406,
"loss": 0.0001,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": 0.0033409663010388613,
"rewards/margins": 0.0031195811461657286,
"rewards/rejected": 0.0002213849511463195,
"step": 290
},
{
"epoch": 0.96,
"learning_rate": 4.7063466543318965e-06,
"logits/chosen": 0.6467469334602356,
"logits/rejected": 0.8215667605400085,
"logps/chosen": -171.28201293945312,
"logps/rejected": -121.31685638427734,
"loss": 0.0001,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": 0.00259224371984601,
"rewards/margins": 0.0022495179437100887,
"rewards/rejected": 0.0003427262417972088,
"step": 300
},
{
"epoch": 0.96,
"eval_logits/chosen": 0.8206124901771545,
"eval_logits/rejected": 0.8782824277877808,
"eval_logps/chosen": -256.82720947265625,
"eval_logps/rejected": -233.77804565429688,
"eval_loss": 0.00015650840941816568,
"eval_rewards/accuracies": 0.4984999895095825,
"eval_rewards/chosen": -0.002130264649167657,
"eval_rewards/margins": 0.00026750334654934704,
"eval_rewards/rejected": -0.0023977679666131735,
"eval_runtime": 412.0828,
"eval_samples_per_second": 4.853,
"eval_steps_per_second": 1.213,
"step": 300
},
{
"epoch": 0.99,
"learning_rate": 4.672600255192022e-06,
"logits/chosen": 0.6745213270187378,
"logits/rejected": 0.7582186460494995,
"logps/chosen": -177.1463623046875,
"logps/rejected": -121.45220947265625,
"loss": 0.0001,
"rewards/accuracies": 0.59375,
"rewards/chosen": 0.003095152322202921,
"rewards/margins": 0.002747387159615755,
"rewards/rejected": 0.00034776475513353944,
"step": 310
},
{
"epoch": 1.02,
"learning_rate": 4.6371536884520115e-06,
"logits/chosen": 0.6856507062911987,
"logits/rejected": 0.8083009719848633,
"logps/chosen": -175.43466186523438,
"logps/rejected": -113.60682678222656,
"loss": 0.0001,
"rewards/accuracies": 0.5625,
"rewards/chosen": 0.0029540827963501215,
"rewards/margins": 0.0019870868418365717,
"rewards/rejected": 0.0009669959545135498,
"step": 320
},
{
"epoch": 1.06,
"learning_rate": 4.600034692810764e-06,
"logits/chosen": 0.6629844307899475,
"logits/rejected": 0.831767737865448,
"logps/chosen": -181.02310180664062,
"logps/rejected": -119.25450134277344,
"loss": 0.0001,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.0041047511622309685,
"rewards/margins": 0.003629709128290415,
"rewards/rejected": 0.0004750423540826887,
"step": 330
},
{
"epoch": 1.09,
"learning_rate": 4.561272315725852e-06,
"logits/chosen": 0.6184499263763428,
"logits/rejected": 0.7539141178131104,
"logps/chosen": -186.89682006835938,
"logps/rejected": -139.5793914794922,
"loss": 0.0001,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": 0.0036196750588715076,
"rewards/margins": 0.0029566895682364702,
"rewards/rejected": 0.0006629853160120547,
"step": 340
},
{
"epoch": 1.12,
"learning_rate": 4.520896890682449e-06,
"logits/chosen": 0.6621273159980774,
"logits/rejected": 0.7836776971817017,
"logps/chosen": -179.44253540039062,
"logps/rejected": -126.19744873046875,
"loss": 0.0001,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.003894126508384943,
"rewards/margins": 0.003107634838670492,
"rewards/rejected": 0.0007864916697144508,
"step": 350
},
{
"epoch": 1.15,
"learning_rate": 4.478940013455864e-06,
"logits/chosen": 0.6130011677742004,
"logits/rejected": 0.7723643183708191,
"logps/chosen": -179.69866943359375,
"logps/rejected": -119.17742919921875,
"loss": 0.0001,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": 0.0030810669995844364,
"rewards/margins": 0.0031474686693400145,
"rewards/rejected": -6.640238279942423e-05,
"step": 360
},
{
"epoch": 1.18,
"learning_rate": 4.435434517386281e-06,
"logits/chosen": 0.6930996179580688,
"logits/rejected": 0.8360698819160461,
"logps/chosen": -189.090087890625,
"logps/rejected": -135.1619873046875,
"loss": 0.0001,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.003456123173236847,
"rewards/margins": 0.0029000742360949516,
"rewards/rejected": 0.0005560485878959298,
"step": 370
},
{
"epoch": 1.22,
"learning_rate": 4.39041444768504e-06,
"logits/chosen": 0.6620668172836304,
"logits/rejected": 0.853411853313446,
"logps/chosen": -177.2210235595703,
"logps/rejected": -124.18900299072266,
"loss": 0.0001,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.00355343846604228,
"rewards/margins": 0.003507003653794527,
"rewards/rejected": 4.643518332159147e-05,
"step": 380
},
{
"epoch": 1.25,
"learning_rate": 4.343915034792569e-06,
"logits/chosen": 0.629216730594635,
"logits/rejected": 0.7686316967010498,
"logps/chosen": -184.9043426513672,
"logps/rejected": -130.17727661132812,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.003557362360879779,
"rewards/margins": 0.003719041822478175,
"rewards/rejected": -0.00016167931607924402,
"step": 390
},
{
"epoch": 1.28,
"learning_rate": 4.295972666808811e-06,
"logits/chosen": 0.6529077291488647,
"logits/rejected": 0.8112252354621887,
"logps/chosen": -172.85665893554688,
"logps/rejected": -127.3191146850586,
"loss": 0.0001,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": 0.0037923946511000395,
"rewards/margins": 0.003340603783726692,
"rewards/rejected": 0.00045179054723121226,
"step": 400
},
{
"epoch": 1.28,
"eval_logits/chosen": 0.8192203044891357,
"eval_logits/rejected": 0.8769342303276062,
"eval_logps/chosen": -256.87567138671875,
"eval_logps/rejected": -233.8277130126953,
"eval_loss": 0.00015775366046000272,
"eval_rewards/accuracies": 0.5195000171661377,
"eval_rewards/chosen": -0.002614969853311777,
"eval_rewards/margins": 0.00027956385747529566,
"eval_rewards/rejected": -0.002894533798098564,
"eval_runtime": 412.2856,
"eval_samples_per_second": 4.851,
"eval_steps_per_second": 1.213,
"step": 400
},
{
"epoch": 1.31,
"learning_rate": 4.246624861017732e-06,
"logits/chosen": 0.625112771987915,
"logits/rejected": 0.7611836791038513,
"logps/chosen": -178.12124633789062,
"logps/rejected": -135.6422882080078,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.004022772889584303,
"rewards/margins": 0.003282977268099785,
"rewards/rejected": 0.0007397954468615353,
"step": 410
},
{
"epoch": 1.34,
"learning_rate": 4.195910234528186e-06,
"logits/chosen": 0.6460477709770203,
"logits/rejected": 0.7641812562942505,
"logps/chosen": -168.17930603027344,
"logps/rejected": -124.3600082397461,
"loss": 0.0001,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": 0.0026535852812230587,
"rewards/margins": 0.0026994948275387287,
"rewards/rejected": -4.590977187035605e-05,
"step": 420
},
{
"epoch": 1.38,
"learning_rate": 4.143868474054098e-06,
"logits/chosen": 0.5987478494644165,
"logits/rejected": 0.7457458972930908,
"logps/chosen": -189.61422729492188,
"logps/rejected": -123.73170471191406,
"loss": 0.0001,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.003896042238920927,
"rewards/margins": 0.00355791044421494,
"rewards/rejected": 0.00033813173649832606,
"step": 430
},
{
"epoch": 1.41,
"learning_rate": 4.0905403048576545e-06,
"logits/chosen": 0.5944562554359436,
"logits/rejected": 0.7794451713562012,
"logps/chosen": -173.83494567871094,
"logps/rejected": -125.55583190917969,
"loss": 0.0001,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": 0.003923391457647085,
"rewards/margins": 0.003630859311670065,
"rewards/rejected": 0.00029253208776935935,
"step": 440
},
{
"epoch": 1.44,
"learning_rate": 4.035967458879751e-06,
"logits/chosen": 0.5830662846565247,
"logits/rejected": 0.7792474031448364,
"logps/chosen": -174.3615264892578,
"logps/rejected": -130.67514038085938,
"loss": 0.0001,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.0029520168900489807,
"rewards/margins": 0.0029378398321568966,
"rewards/rejected": 1.4177237972035073e-05,
"step": 450
},
{
"epoch": 1.47,
"learning_rate": 3.980192642082682e-06,
"logits/chosen": 0.6810758709907532,
"logits/rejected": 0.7834008932113647,
"logps/chosen": -168.80197143554688,
"logps/rejected": -118.42387390136719,
"loss": 0.0001,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": 0.0025485253427177668,
"rewards/margins": 0.002175524365156889,
"rewards/rejected": 0.00037300080293789506,
"step": 460
},
{
"epoch": 1.5,
"learning_rate": 3.923259501030604e-06,
"logits/chosen": 0.6081482768058777,
"logits/rejected": 0.8008432388305664,
"logps/chosen": -158.7695770263672,
"logps/rejected": -117.162109375,
"loss": 0.0001,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": 0.003074315609410405,
"rewards/margins": 0.0024462412111461163,
"rewards/rejected": 0.0006280745146796107,
"step": 470
},
{
"epoch": 1.54,
"learning_rate": 3.865212588733927e-06,
"logits/chosen": 0.5983234643936157,
"logits/rejected": 0.8347161412239075,
"logps/chosen": -177.07232666015625,
"logps/rejected": -113.17292785644531,
"loss": 0.0001,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": 0.004156365990638733,
"rewards/margins": 0.0032986297737807035,
"rewards/rejected": 0.0008577358676120639,
"step": 480
},
{
"epoch": 1.57,
"learning_rate": 3.8060973297843773e-06,
"logits/chosen": 0.6949520111083984,
"logits/rejected": 0.8520463109016418,
"logps/chosen": -180.8011016845703,
"logps/rejected": -135.38951110839844,
"loss": 0.0001,
"rewards/accuracies": 0.59375,
"rewards/chosen": 0.0031148726120591164,
"rewards/margins": 0.0025527984835207462,
"rewards/rejected": 0.000562074186746031,
"step": 490
},
{
"epoch": 1.6,
"learning_rate": 3.7459599848079965e-06,
"logits/chosen": 0.6497796773910522,
"logits/rejected": 0.8080886006355286,
"logps/chosen": -187.10287475585938,
"logps/rejected": -133.8563995361328,
"loss": 0.0001,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": 0.004290526732802391,
"rewards/margins": 0.0023678760044276714,
"rewards/rejected": 0.0019226508447900414,
"step": 500
},
{
"epoch": 1.6,
"eval_logits/chosen": 0.8150748610496521,
"eval_logits/rejected": 0.8728834986686707,
"eval_logps/chosen": -256.8869323730469,
"eval_logps/rejected": -233.8388214111328,
"eval_loss": 0.0001666269963607192,
"eval_rewards/accuracies": 0.5170000195503235,
"eval_rewards/chosen": -0.0027276412583887577,
"eval_rewards/margins": 0.00027794469497166574,
"eval_rewards/rejected": -0.0030055860988795757,
"eval_runtime": 414.2028,
"eval_samples_per_second": 4.829,
"eval_steps_per_second": 1.207,
"step": 500
},
{
"epoch": 1.63,
"learning_rate": 3.684847614263898e-06,
"logits/chosen": 0.649914562702179,
"logits/rejected": 0.8345224261283875,
"logps/chosen": -170.88528442382812,
"logps/rejected": -113.5182876586914,
"loss": 0.0001,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": 0.004132578149437904,
"rewards/margins": 0.004135853610932827,
"rewards/rejected": -3.27564202962094e-06,
"step": 510
},
{
"epoch": 1.66,
"learning_rate": 3.622808041617133e-06,
"logits/chosen": 0.6696175336837769,
"logits/rejected": 0.8282491564750671,
"logps/chosen": -165.21356201171875,
"logps/rejected": -120.80168151855469,
"loss": 0.0001,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": 0.003804347710683942,
"rewards/margins": 0.002934789750725031,
"rewards/rejected": 0.0008695581927895546,
"step": 520
},
{
"epoch": 1.7,
"learning_rate": 3.559889815914441e-06,
"logits/chosen": 0.6223723888397217,
"logits/rejected": 0.8060038685798645,
"logps/chosen": -173.92562866210938,
"logps/rejected": -115.7896728515625,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.0046675438061356544,
"rewards/margins": 0.0034386657644063234,
"rewards/rejected": 0.001228878041729331,
"step": 530
},
{
"epoch": 1.73,
"learning_rate": 3.496142173792219e-06,
"logits/chosen": 0.5860522389411926,
"logits/rejected": 0.7538793683052063,
"logps/chosen": -179.20030212402344,
"logps/rejected": -126.5418930053711,
"loss": 0.0001,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.0036121346056461334,
"rewards/margins": 0.003292589681223035,
"rewards/rejected": 0.00031954512814991176,
"step": 540
},
{
"epoch": 1.76,
"learning_rate": 3.4316150009464023e-06,
"logits/chosen": 0.6498968601226807,
"logits/rejected": 0.8035451173782349,
"logps/chosen": -192.52578735351562,
"logps/rejected": -139.58251953125,
"loss": 0.0001,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": 0.003925256431102753,
"rewards/margins": 0.0028843428008258343,
"rewards/rejected": 0.001040913863107562,
"step": 550
},
{
"epoch": 1.79,
"learning_rate": 3.366358793094433e-06,
"logits/chosen": 0.5884669423103333,
"logits/rejected": 0.8008158802986145,
"logps/chosen": -179.25753784179688,
"logps/rejected": -120.347412109375,
"loss": 0.0001,
"rewards/accuracies": 0.59375,
"rewards/chosen": 0.003893566783517599,
"rewards/margins": 0.0027398644015192986,
"rewards/rejected": 0.0011537026148289442,
"step": 560
},
{
"epoch": 1.82,
"learning_rate": 3.3004246164598535e-06,
"logits/chosen": 0.6389291286468506,
"logits/rejected": 0.8485361933708191,
"logps/chosen": -190.9259490966797,
"logps/rejected": -140.17202758789062,
"loss": 0.0001,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": 0.004239690490067005,
"rewards/margins": 0.003860587952658534,
"rewards/rejected": 0.0003791024792008102,
"step": 570
},
{
"epoch": 1.86,
"learning_rate": 3.233864067810446e-06,
"logits/chosen": 0.6482547521591187,
"logits/rejected": 0.8119586110115051,
"logps/chosen": -183.17164611816406,
"logps/rejected": -119.36767578125,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.004655472934246063,
"rewards/margins": 0.0037689208984375,
"rewards/rejected": 0.0008865518611855805,
"step": 580
},
{
"epoch": 1.89,
"learning_rate": 3.1667292340812077e-06,
"logits/chosen": 0.6013578176498413,
"logits/rejected": 0.7613108158111572,
"logps/chosen": -181.44009399414062,
"logps/rejected": -140.5045166015625,
"loss": 0.0001,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.004445691592991352,
"rewards/margins": 0.003334993962198496,
"rewards/rejected": 0.0011106978636234999,
"step": 590
},
{
"epoch": 1.92,
"learning_rate": 3.099072651613728e-06,
"logits/chosen": 0.7029744386672974,
"logits/rejected": 0.8172439336776733,
"logps/chosen": -165.12625122070312,
"logps/rejected": -119.59954833984375,
"loss": 0.0001,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.0032034076284617186,
"rewards/margins": 0.0023894021287560463,
"rewards/rejected": 0.0008140054414980114,
"step": 600
},
{
"epoch": 1.92,
"eval_logits/chosen": 0.817969799041748,
"eval_logits/rejected": 0.8756601810455322,
"eval_logps/chosen": -256.885986328125,
"eval_logps/rejected": -233.84141540527344,
"eval_loss": 0.00016333417443092912,
"eval_rewards/accuracies": 0.5070000290870667,
"eval_rewards/chosen": -0.0027182498015463352,
"eval_rewards/margins": 0.0003131589328404516,
"eval_rewards/rejected": -0.003031408879905939,
"eval_runtime": 412.1696,
"eval_samples_per_second": 4.852,
"eval_steps_per_second": 1.213,
"step": 600
},
{
"epoch": 1.95,
"learning_rate": 3.0309472650438982e-06,
"logits/chosen": 0.7164211869239807,
"logits/rejected": 0.8135250806808472,
"logps/chosen": -167.83938598632812,
"logps/rejected": -123.3685531616211,
"loss": 0.0001,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": 0.004080263432115316,
"rewards/margins": 0.002866895869374275,
"rewards/rejected": 0.0012133677955716848,
"step": 610
},
{
"epoch": 1.98,
"learning_rate": 2.9624063858701006e-06,
"logits/chosen": 0.6123205423355103,
"logits/rejected": 0.8067408800125122,
"logps/chosen": -174.879638671875,
"logps/rejected": -119.88896179199219,
"loss": 0.0001,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.0032961233519017696,
"rewards/margins": 0.0031299355905503035,
"rewards/rejected": 0.00016618790687061846,
"step": 620
},
{
"epoch": 2.02,
"learning_rate": 2.8935036507343185e-06,
"logits/chosen": 0.6171947717666626,
"logits/rejected": 0.7731830477714539,
"logps/chosen": -173.52308654785156,
"logps/rejected": -122.30467224121094,
"loss": 0.0001,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": 0.004019217099994421,
"rewards/margins": 0.003546707332134247,
"rewards/rejected": 0.0004725100880023092,
"step": 630
},
{
"epoch": 2.05,
"learning_rate": 2.8242929794487926e-06,
"logits/chosen": 0.6720027923583984,
"logits/rejected": 0.8209166526794434,
"logps/chosen": -167.04946899414062,
"logps/rejected": -120.01603698730469,
"loss": 0.0001,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.004417883697897196,
"rewards/margins": 0.00379347731359303,
"rewards/rejected": 0.000624406267888844,
"step": 640
},
{
"epoch": 2.08,
"learning_rate": 2.7548285328010984e-06,
"logits/chosen": 0.5649908781051636,
"logits/rejected": 0.7522573471069336,
"logps/chosen": -167.37994384765625,
"logps/rejected": -118.6939697265625,
"loss": 0.0001,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": 0.003759379032999277,
"rewards/margins": 0.0030631672125309706,
"rewards/rejected": 0.0006962117040529847,
"step": 650
},
{
"epoch": 2.11,
"learning_rate": 2.6851646701706306e-06,
"logits/chosen": 0.7023177742958069,
"logits/rejected": 0.8027107119560242,
"logps/chosen": -177.06430053710938,
"logps/rejected": -129.97714233398438,
"loss": 0.0001,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.003546078223735094,
"rewards/margins": 0.0027953279204666615,
"rewards/rejected": 0.0007507502450607717,
"step": 660
},
{
"epoch": 2.14,
"learning_rate": 2.6153559069897007e-06,
"logits/chosen": 0.6293431520462036,
"logits/rejected": 0.7750530242919922,
"logps/chosen": -168.621337890625,
"logps/rejected": -122.5462875366211,
"loss": 0.0001,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": 0.003033037530258298,
"rewards/margins": 0.0028599021025002003,
"rewards/rejected": 0.00017313548596575856,
"step": 670
},
{
"epoch": 2.18,
"learning_rate": 2.5454568720824937e-06,
"logits/chosen": 0.5758141279220581,
"logits/rejected": 0.7897475957870483,
"logps/chosen": -174.54208374023438,
"logps/rejected": -118.59149169921875,
"loss": 0.0001,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.003853294998407364,
"rewards/margins": 0.003840038087219,
"rewards/rejected": 1.3256654710858129e-05,
"step": 680
},
{
"epoch": 2.21,
"learning_rate": 2.4755222649153014e-06,
"logits/chosen": 0.6393508911132812,
"logits/rejected": 0.8246806263923645,
"logps/chosen": -184.72274780273438,
"logps/rejected": -126.2255859375,
"loss": 0.0001,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": 0.0034170313738286495,
"rewards/margins": 0.0032350593246519566,
"rewards/rejected": 0.00018197241297457367,
"step": 690
},
{
"epoch": 2.24,
"learning_rate": 2.4056068127914803e-06,
"logits/chosen": 0.7041198015213013,
"logits/rejected": 0.8579221963882446,
"logps/chosen": -165.37600708007812,
"logps/rejected": -113.6267318725586,
"loss": 0.0001,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": 0.0038474693428725004,
"rewards/margins": 0.004361593164503574,
"rewards/rejected": -0.0005141238798387349,
"step": 700
},
{
"epoch": 2.24,
"eval_logits/chosen": 0.8142222166061401,
"eval_logits/rejected": 0.8718712329864502,
"eval_logps/chosen": -256.9122619628906,
"eval_logps/rejected": -233.85919189453125,
"eval_loss": 0.00016846887592691928,
"eval_rewards/accuracies": 0.5065000057220459,
"eval_rewards/chosen": -0.0029809277039021254,
"eval_rewards/margins": 0.00022821790480520576,
"eval_rewards/rejected": -0.003209145972505212,
"eval_runtime": 412.3465,
"eval_samples_per_second": 4.85,
"eval_steps_per_second": 1.213,
"step": 700
},
{
"epoch": 2.27,
"learning_rate": 2.3357652280246125e-06,
"logits/chosen": 0.6320663690567017,
"logits/rejected": 0.785031259059906,
"logps/chosen": -189.28390502929688,
"logps/rejected": -144.56167602539062,
"loss": 0.0001,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": 0.0025972402654588223,
"rewards/margins": 0.0028369042556732893,
"rewards/rejected": -0.00023966425214894116,
"step": 710
},
{
"epoch": 2.3,
"learning_rate": 2.2660521651234036e-06,
"logits/chosen": 0.6551303267478943,
"logits/rejected": 0.8406831622123718,
"logps/chosen": -186.629638671875,
"logps/rejected": -126.31327819824219,
"loss": 0.0001,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": 0.0038953598123043776,
"rewards/margins": 0.003374651074409485,
"rewards/rejected": 0.0005207090289331973,
"step": 720
},
{
"epoch": 2.34,
"learning_rate": 2.1965221780218173e-06,
"logits/chosen": 0.6312896013259888,
"logits/rejected": 0.7552987337112427,
"logps/chosen": -173.728271484375,
"logps/rejected": -124.27177429199219,
"loss": 0.0001,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.004026113077998161,
"rewards/margins": 0.0039476132951676846,
"rewards/rejected": 7.849968096707016e-05,
"step": 730
},
{
"epoch": 2.37,
"learning_rate": 2.1272296773879107e-06,
"logits/chosen": 0.6133307814598083,
"logits/rejected": 0.8152937889099121,
"logps/chosen": -178.77232360839844,
"logps/rejected": -125.87672424316406,
"loss": 0.0001,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": 0.0026045762933790684,
"rewards/margins": 0.0032864962704479694,
"rewards/rejected": -0.0006819200934842229,
"step": 740
},
{
"epoch": 2.4,
"learning_rate": 2.058228888044788e-06,
"logits/chosen": 0.6092817187309265,
"logits/rejected": 0.7013009190559387,
"logps/chosen": -165.13404846191406,
"logps/rejected": -122.30549621582031,
"loss": 0.0001,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": 0.003001943463459611,
"rewards/margins": 0.0025921487249433994,
"rewards/rejected": 0.0004097948840353638,
"step": 750
},
{
"epoch": 2.43,
"learning_rate": 1.989573806536978e-06,
"logits/chosen": 0.6568277478218079,
"logits/rejected": 0.8574051856994629,
"logps/chosen": -179.80409240722656,
"logps/rejected": -120.96827697753906,
"loss": 0.0001,
"rewards/accuracies": 0.65625,
"rewards/chosen": 0.004101686645299196,
"rewards/margins": 0.00385199673473835,
"rewards/rejected": 0.0002496893866918981,
"step": 760
},
{
"epoch": 2.46,
"learning_rate": 1.921318158875459e-06,
"logits/chosen": 0.6586846113204956,
"logits/rejected": 0.8662702441215515,
"logps/chosen": -174.4876708984375,
"logps/rejected": -128.13922119140625,
"loss": 0.0001,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.004603301174938679,
"rewards/margins": 0.0032628674525767565,
"rewards/rejected": 0.0013404333731159568,
"step": 770
},
{
"epoch": 2.5,
"learning_rate": 1.8535153584943915e-06,
"logits/chosen": 0.6371630430221558,
"logits/rejected": 0.7737770080566406,
"logps/chosen": -170.006591796875,
"logps/rejected": -126.65934753417969,
"loss": 0.0001,
"rewards/accuracies": 0.59375,
"rewards/chosen": 0.004735985770821571,
"rewards/margins": 0.003097447333857417,
"rewards/rejected": 0.001638538553379476,
"step": 780
},
{
"epoch": 2.53,
"learning_rate": 1.7862184644524422e-06,
"logits/chosen": 0.6403388977050781,
"logits/rejected": 0.7904574275016785,
"logps/chosen": -178.15911865234375,
"logps/rejected": -133.64309692382812,
"loss": 0.0001,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": 0.00445596594363451,
"rewards/margins": 0.004045891109853983,
"rewards/rejected": 0.0004100751248188317,
"step": 790
},
{
"epoch": 2.56,
"learning_rate": 1.7194801399114471e-06,
"logits/chosen": 0.6325648427009583,
"logits/rejected": 0.747418999671936,
"logps/chosen": -174.4492950439453,
"logps/rejected": -125.17801666259766,
"loss": 0.0001,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": 0.004687703680247068,
"rewards/margins": 0.004329483024775982,
"rewards/rejected": 0.00035822103382088244,
"step": 800
},
{
"epoch": 2.56,
"eval_logits/chosen": 0.8135467767715454,
"eval_logits/rejected": 0.8712592124938965,
"eval_logps/chosen": -256.8898010253906,
"eval_logps/rejected": -233.84222412109375,
"eval_loss": 0.00016861619951669127,
"eval_rewards/accuracies": 0.5189999938011169,
"eval_rewards/chosen": -0.002756227506324649,
"eval_rewards/margins": 0.00028340137214399874,
"eval_rewards/rejected": -0.003039628965780139,
"eval_runtime": 412.1883,
"eval_samples_per_second": 4.852,
"eval_steps_per_second": 1.213,
"step": 800
},
{
"epoch": 2.59,
"learning_rate": 1.6533526109248632e-06,
"logits/chosen": 0.6175631284713745,
"logits/rejected": 0.74301677942276,
"logps/chosen": -179.27880859375,
"logps/rejected": -123.3892822265625,
"loss": 0.0001,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.003877174574881792,
"rewards/margins": 0.003652264829725027,
"rewards/rejected": 0.00022490958508569747,
"step": 810
},
{
"epoch": 2.62,
"learning_rate": 1.5878876255682951e-06,
"logits/chosen": 0.6676111817359924,
"logits/rejected": 0.7865282893180847,
"logps/chosen": -189.4076385498047,
"logps/rejected": -137.06271362304688,
"loss": 0.0001,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.004411728121340275,
"rewards/margins": 0.0037712506018579006,
"rewards/rejected": 0.0006404774612747133,
"step": 820
},
{
"epoch": 2.66,
"learning_rate": 1.5231364134440485e-06,
"logits/chosen": 0.6061269044876099,
"logits/rejected": 0.7513775825500488,
"logps/chosen": -190.88760375976562,
"logps/rejected": -134.599365234375,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.004417582880705595,
"rewards/margins": 0.0038115177303552628,
"rewards/rejected": 0.000606064626481384,
"step": 830
},
{
"epoch": 2.69,
"learning_rate": 1.4591496455914292e-06,
"logits/chosen": 0.6278064846992493,
"logits/rejected": 0.7927815914154053,
"logps/chosen": -173.77554321289062,
"logps/rejected": -128.66485595703125,
"loss": 0.0001,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": 0.003960388712584972,
"rewards/margins": 0.0037871100939810276,
"rewards/rejected": 0.0001732785312924534,
"step": 840
},
{
"epoch": 2.72,
"learning_rate": 1.395977394834132e-06,
"logits/chosen": 0.6175497174263,
"logits/rejected": 0.8719260096549988,
"logps/chosen": -172.75567626953125,
"logps/rejected": -118.21751403808594,
"loss": 0.0001,
"rewards/accuracies": 0.65625,
"rewards/chosen": 0.004103314597159624,
"rewards/margins": 0.003662864211946726,
"rewards/rejected": 0.0004404502979014069,
"step": 850
},
{
"epoch": 2.75,
"learning_rate": 1.3336690965957733e-06,
"logits/chosen": 0.6757524013519287,
"logits/rejected": 0.8124428987503052,
"logps/chosen": -192.186279296875,
"logps/rejected": -144.07810974121094,
"loss": 0.0001,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": 0.004593437071889639,
"rewards/margins": 0.004751748405396938,
"rewards/rejected": -0.0001583110133651644,
"step": 860
},
{
"epoch": 2.78,
"learning_rate": 1.2722735102142192e-06,
"logits/chosen": 0.5904192328453064,
"logits/rejected": 0.8412498235702515,
"logps/chosen": -178.94638061523438,
"logps/rejected": -116.67149353027344,
"loss": 0.0001,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": 0.0042373063042759895,
"rewards/margins": 0.004430609289556742,
"rewards/rejected": -0.00019330321811139584,
"step": 870
},
{
"epoch": 2.82,
"learning_rate": 1.2118386807849733e-06,
"logits/chosen": 0.6753177642822266,
"logits/rejected": 0.8218367695808411,
"logps/chosen": -169.46713256835938,
"logps/rejected": -124.23944091796875,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.0036155576817691326,
"rewards/margins": 0.0030525142792612314,
"rewards/rejected": 0.0005630434607155621,
"step": 880
},
{
"epoch": 2.85,
"learning_rate": 1.1524119015635116e-06,
"logits/chosen": 0.5247241258621216,
"logits/rejected": 0.7601326704025269,
"logps/chosen": -187.26348876953125,
"logps/rejected": -123.6650390625,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.004491317551583052,
"rewards/margins": 0.003613727632910013,
"rewards/rejected": 0.0008775900350883603,
"step": 890
},
{
"epoch": 2.88,
"learning_rate": 1.0940396769559584e-06,
"logits/chosen": 0.6696980595588684,
"logits/rejected": 0.7714194059371948,
"logps/chosen": -191.7372283935547,
"logps/rejected": -128.82192993164062,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.004666435532271862,
"rewards/margins": 0.0038504847325384617,
"rewards/rejected": 0.0008159511489793658,
"step": 900
},
{
"epoch": 2.88,
"eval_logits/chosen": 0.8135749101638794,
"eval_logits/rejected": 0.8714309930801392,
"eval_logps/chosen": -256.91107177734375,
"eval_logps/rejected": -233.8529052734375,
"eval_loss": 0.00016910216072574258,
"eval_rewards/accuracies": 0.5015000104904175,
"eval_rewards/chosen": -0.0029687141068279743,
"eval_rewards/margins": 0.00017772088176570833,
"eval_rewards/rejected": -0.0031464346684515476,
"eval_runtime": 412.7326,
"eval_samples_per_second": 4.846,
"eval_steps_per_second": 1.211,
"step": 900
},
{
"epoch": 2.91,
"learning_rate": 1.036767686127079e-06,
"logits/chosen": 0.6320682168006897,
"logits/rejected": 0.7783384323120117,
"logps/chosen": -189.1212158203125,
"logps/rejected": -130.10711669921875,
"loss": 0.0001,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.004713307600468397,
"rewards/margins": 0.003414090257138014,
"rewards/rejected": 0.0012992171104997396,
"step": 910
},
{
"epoch": 2.94,
"learning_rate": 9.806407472540644e-07,
"logits/chosen": 0.6682049036026001,
"logits/rejected": 0.8254071474075317,
"logps/chosen": -184.09628295898438,
"logps/rejected": -123.4487075805664,
"loss": 0.0001,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.004300036001950502,
"rewards/margins": 0.003941989503800869,
"rewards/rejected": 0.00035804632352665067,
"step": 920
},
{
"epoch": 2.98,
"learning_rate": 9.257027824540823e-07,
"logits/chosen": 0.6992205381393433,
"logits/rejected": 0.8208533525466919,
"logps/chosen": -164.10159301757812,
"logps/rejected": -110.07139587402344,
"loss": 0.0001,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": 0.003934869542717934,
"rewards/margins": 0.003884183941408992,
"rewards/rejected": 5.068551399745047e-05,
"step": 930
},
{
"epoch": 3.01,
"learning_rate": 8.719967834130385e-07,
"logits/chosen": 0.6116827726364136,
"logits/rejected": 0.8160429000854492,
"logps/chosen": -169.47280883789062,
"logps/rejected": -115.8711166381836,
"loss": 0.0001,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": 0.005224294029176235,
"rewards/margins": 0.004378092475235462,
"rewards/rejected": 0.0008462019031867385,
"step": 940
},
{
"epoch": 3.04,
"learning_rate": 8.195647777424479e-07,
"logits/chosen": 0.5635887384414673,
"logits/rejected": 0.7940059900283813,
"logps/chosen": -177.22665405273438,
"logps/rejected": -120.8065414428711,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.003887483151629567,
"rewards/margins": 0.0035337016452103853,
"rewards/rejected": 0.00035378162283450365,
"step": 950
},
{
"epoch": 3.07,
"learning_rate": 7.684477960907422e-07,
"logits/chosen": 0.5837413668632507,
"logits/rejected": 0.7604612112045288,
"logps/chosen": -175.0729217529297,
"logps/rejected": -118.60954284667969,
"loss": 0.0001,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.004512741696089506,
"rewards/margins": 0.0036391555331647396,
"rewards/rejected": 0.0008735861629247665,
"step": 960
},
{
"epoch": 3.1,
"learning_rate": 7.186858400347455e-07,
"logits/chosen": 0.6265005469322205,
"logits/rejected": 0.7885332703590393,
"logps/chosen": -182.97769165039062,
"logps/rejected": -138.80723571777344,
"loss": 0.0001,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": 0.004079051315784454,
"rewards/margins": 0.004020996857434511,
"rewards/rejected": 5.8054854889633134e-05,
"step": 970
},
{
"epoch": 3.14,
"learning_rate": 6.703178507764618e-07,
"logits/chosen": 0.6634309887886047,
"logits/rejected": 0.7914069890975952,
"logps/chosen": -167.03756713867188,
"logps/rejected": -124.2927474975586,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.005084961652755737,
"rewards/margins": 0.0035723126493394375,
"rewards/rejected": 0.0015126490034162998,
"step": 980
},
{
"epoch": 3.17,
"learning_rate": 6.233816786696414e-07,
"logits/chosen": 0.6042094230651855,
"logits/rejected": 0.7901323437690735,
"logps/chosen": -195.58741760253906,
"logps/rejected": -132.481689453125,
"loss": 0.0001,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": 0.005596310831606388,
"rewards/margins": 0.0045918067917227745,
"rewards/rejected": 0.001004504389129579,
"step": 990
},
{
"epoch": 3.2,
"learning_rate": 5.77914053600005e-07,
"logits/chosen": 0.5928013920783997,
"logits/rejected": 0.8330589532852173,
"logps/chosen": -176.38113403320312,
"logps/rejected": -123.68299865722656,
"loss": 0.0001,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.0048389798030257225,
"rewards/margins": 0.00368002662435174,
"rewards/rejected": 0.001158953527919948,
"step": 1000
},
{
"epoch": 3.2,
"eval_logits/chosen": 0.8156088590621948,
"eval_logits/rejected": 0.8733118176460266,
"eval_logps/chosen": -256.9035949707031,
"eval_logps/rejected": -233.8666229248047,
"eval_loss": 0.00016867661906871945,
"eval_rewards/accuracies": 0.5180000066757202,
"eval_rewards/chosen": -0.002894038800150156,
"eval_rewards/margins": 0.0003897584683727473,
"eval_rewards/rejected": -0.0032837972976267338,
"eval_runtime": 412.3586,
"eval_samples_per_second": 4.85,
"eval_steps_per_second": 1.213,
"step": 1000
},
{
"epoch": 3.23,
"learning_rate": 5.339505562422851e-07,
"logits/chosen": 0.6899782419204712,
"logits/rejected": 0.8187691569328308,
"logps/chosen": -177.58436584472656,
"logps/rejected": -121.7637939453125,
"loss": 0.0001,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": 0.0037464499473571777,
"rewards/margins": 0.0031559974886476994,
"rewards/rejected": 0.0005904529243707657,
"step": 1010
},
{
"epoch": 3.26,
"learning_rate": 4.915255902165734e-07,
"logits/chosen": 0.6754826903343201,
"logits/rejected": 0.8096501231193542,
"logps/chosen": -173.49542236328125,
"logps/rejected": -120.12110900878906,
"loss": 0.0001,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.004369380883872509,
"rewards/margins": 0.0036476694513112307,
"rewards/rejected": 0.000721711665391922,
"step": 1020
},
{
"epoch": 3.3,
"learning_rate": 4.506723551657879e-07,
"logits/chosen": 0.5708236694335938,
"logits/rejected": 0.7320288419723511,
"logps/chosen": -179.21517944335938,
"logps/rejected": -131.3833465576172,
"loss": 0.0001,
"rewards/accuracies": 0.65625,
"rewards/chosen": 0.004431615583598614,
"rewards/margins": 0.003471215022727847,
"rewards/rejected": 0.0009604001534171402,
"step": 1030
},
{
"epoch": 3.33,
"learning_rate": 4.11422820775299e-07,
"logits/chosen": 0.6452796459197998,
"logits/rejected": 0.8163026571273804,
"logps/chosen": -205.68319702148438,
"logps/rejected": -129.33828735351562,
"loss": 0.0001,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": 0.004141085781157017,
"rewards/margins": 0.0046521080657839775,
"rewards/rejected": -0.0005110226338729262,
"step": 1040
},
{
"epoch": 3.36,
"learning_rate": 3.7380770175506397e-07,
"logits/chosen": 0.5789592862129211,
"logits/rejected": 0.7904280424118042,
"logps/chosen": -180.31417846679688,
"logps/rejected": -129.52687072753906,
"loss": 0.0001,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": 0.004802372306585312,
"rewards/margins": 0.004047960974276066,
"rewards/rejected": 0.0007544115069322288,
"step": 1050
},
{
"epoch": 3.39,
"learning_rate": 3.3785643380384063e-07,
"logits/chosen": 0.6884064674377441,
"logits/rejected": 0.8517038226127625,
"logps/chosen": -177.23248291015625,
"logps/rejected": -117.1332778930664,
"loss": 0.0001,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": 0.0034295388031750917,
"rewards/margins": 0.0038840598426759243,
"rewards/rejected": -0.0004545215633697808,
"step": 1060
},
{
"epoch": 3.42,
"learning_rate": 3.0359715057429186e-07,
"logits/chosen": 0.5767031908035278,
"logits/rejected": 0.6466313004493713,
"logps/chosen": -172.95346069335938,
"logps/rejected": -127.76350402832031,
"loss": 0.0001,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": 0.0043509481474757195,
"rewards/margins": 0.0033387758303433657,
"rewards/rejected": 0.0010121725499629974,
"step": 1070
},
{
"epoch": 3.46,
"learning_rate": 2.710566616570048e-07,
"logits/chosen": 0.661910891532898,
"logits/rejected": 0.8002559542655945,
"logps/chosen": -177.2826690673828,
"logps/rejected": -131.19056701660156,
"loss": 0.0001,
"rewards/accuracies": 0.59375,
"rewards/chosen": 0.0037956517189741135,
"rewards/margins": 0.003371240571141243,
"rewards/rejected": 0.0004244106821715832,
"step": 1080
},
{
"epoch": 3.49,
"learning_rate": 2.40260431600654e-07,
"logits/chosen": 0.6113812923431396,
"logits/rejected": 0.8205118179321289,
"logps/chosen": -174.49224853515625,
"logps/rejected": -135.5587158203125,
"loss": 0.0001,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": 0.004168152809143066,
"rewards/margins": 0.0029846313409507275,
"rewards/rejected": 0.0011835219338536263,
"step": 1090
},
{
"epoch": 3.52,
"learning_rate": 2.1123255998472952e-07,
"logits/chosen": 0.6319399476051331,
"logits/rejected": 0.7784051895141602,
"logps/chosen": -160.6739501953125,
"logps/rejected": -114.9722671508789,
"loss": 0.0001,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": 0.004197821486741304,
"rewards/margins": 0.0031375852413475513,
"rewards/rejected": 0.0010602364782243967,
"step": 1100
},
{
"epoch": 3.52,
"eval_logits/chosen": 0.8145170211791992,
"eval_logits/rejected": 0.8724321126937866,
"eval_logps/chosen": -256.90802001953125,
"eval_logps/rejected": -233.87786865234375,
"eval_loss": 0.00016770198999438435,
"eval_rewards/accuracies": 0.5264999866485596,
"eval_rewards/chosen": -0.00293838232755661,
"eval_rewards/margins": 0.00045752059668302536,
"eval_rewards/rejected": -0.003395902691408992,
"eval_runtime": 412.1486,
"eval_samples_per_second": 4.853,
"eval_steps_per_second": 1.213,
"step": 1100
},
{
"epoch": 3.55,
"learning_rate": 1.8399576256041525e-07,
"logits/chosen": 0.6978103518486023,
"logits/rejected": 0.7642599940299988,
"logps/chosen": -188.5946502685547,
"logps/rejected": -142.72915649414062,
"loss": 0.0001,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": 0.004109769128262997,
"rewards/margins": 0.003021980170160532,
"rewards/rejected": 0.0010877888416871428,
"step": 1110
},
{
"epoch": 3.58,
"learning_rate": 1.58571353474391e-07,
"logits/chosen": 0.6541138887405396,
"logits/rejected": 0.7682735323905945,
"logps/chosen": -157.97927856445312,
"logps/rejected": -120.40803527832031,
"loss": 0.0001,
"rewards/accuracies": 0.5625,
"rewards/chosen": 0.0030703709926456213,
"rewards/margins": 0.0024266496766358614,
"rewards/rejected": 0.0006437213160097599,
"step": 1120
},
{
"epoch": 3.62,
"learning_rate": 1.3497922858944857e-07,
"logits/chosen": 0.7424976229667664,
"logits/rejected": 0.8078197240829468,
"logps/chosen": -174.2366943359375,
"logps/rejected": -131.4811553955078,
"loss": 0.0001,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": 0.004558051936328411,
"rewards/margins": 0.003959262277930975,
"rewards/rejected": 0.0005987894837744534,
"step": 1130
},
{
"epoch": 3.65,
"learning_rate": 1.1323784991499471e-07,
"logits/chosen": 0.6016920804977417,
"logits/rejected": 0.7492619752883911,
"logps/chosen": -172.1629638671875,
"logps/rejected": -116.34178161621094,
"loss": 0.0001,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": 0.004433467984199524,
"rewards/margins": 0.004034010227769613,
"rewards/rejected": 0.0003994574653916061,
"step": 1140
},
{
"epoch": 3.68,
"learning_rate": 9.336423115961002e-08,
"logits/chosen": 0.6536157727241516,
"logits/rejected": 0.8383282423019409,
"logps/chosen": -179.27108764648438,
"logps/rejected": -121.70698547363281,
"loss": 0.0001,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.00442055519670248,
"rewards/margins": 0.003978613764047623,
"rewards/rejected": 0.00044194143265485764,
"step": 1150
},
{
"epoch": 3.71,
"learning_rate": 7.537392441697793e-08,
"logits/chosen": 0.6289481520652771,
"logits/rejected": 0.8159587979316711,
"logps/chosen": -178.74453735351562,
"logps/rejected": -126.57807922363281,
"loss": 0.0001,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": 0.004435301758348942,
"rewards/margins": 0.0039503672160208225,
"rewards/rejected": 0.0004849349206779152,
"step": 1160
},
{
"epoch": 3.74,
"learning_rate": 5.928100799559938e-08,
"logits/chosen": 0.6689172387123108,
"logits/rejected": 0.7863209843635559,
"logps/chosen": -178.4945068359375,
"logps/rejected": -124.7654037475586,
"loss": 0.0001,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.005153838545084,
"rewards/margins": 0.003751266747713089,
"rewards/rejected": 0.001402571564540267,
"step": 1170
},
{
"epoch": 3.78,
"learning_rate": 4.5098075401815435e-08,
"logits/chosen": 0.6192042827606201,
"logits/rejected": 0.76641446352005,
"logps/chosen": -171.137939453125,
"logps/rejected": -125.4515151977539,
"loss": 0.0001,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.004533619619905949,
"rewards/margins": 0.003654058324173093,
"rewards/rejected": 0.000879561179317534,
"step": 1180
},
{
"epoch": 3.81,
"learning_rate": 3.283622548476445e-08,
"logits/chosen": 0.6808220148086548,
"logits/rejected": 0.8677403330802917,
"logps/chosen": -178.3741455078125,
"logps/rejected": -125.1767349243164,
"loss": 0.0001,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": 0.005535735748708248,
"rewards/margins": 0.004433406982570887,
"rewards/rejected": 0.0011023286497220397,
"step": 1190
},
{
"epoch": 3.84,
"learning_rate": 2.250505375098161e-08,
"logits/chosen": 0.669333815574646,
"logits/rejected": 0.808275043964386,
"logps/chosen": -176.30514526367188,
"logps/rejected": -125.28890228271484,
"loss": 0.0001,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.004538280423730612,
"rewards/margins": 0.0036300034262239933,
"rewards/rejected": 0.0009082768228836358,
"step": 1200
},
{
"epoch": 3.84,
"eval_logits/chosen": 0.8126665353775024,
"eval_logits/rejected": 0.8705229759216309,
"eval_logps/chosen": -256.92266845703125,
"eval_logps/rejected": -233.87326049804688,
"eval_loss": 0.00017112624482251704,
"eval_rewards/accuracies": 0.5044999718666077,
"eval_rewards/chosen": -0.0030848486348986626,
"eval_rewards/margins": 0.0002651172399055213,
"eval_rewards/rejected": -0.0033499656710773706,
"eval_runtime": 412.2112,
"eval_samples_per_second": 4.852,
"eval_steps_per_second": 1.213,
"step": 1200
},
{
"epoch": 3.87,
"learning_rate": 1.4112644855438228e-08,
"logits/chosen": 0.6185505390167236,
"logits/rejected": 0.786270260810852,
"logps/chosen": -172.150634765625,
"logps/rejected": -123.99739074707031,
"loss": 0.0001,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": 0.005183863919228315,
"rewards/margins": 0.004181054420769215,
"rewards/rejected": 0.0010028090327978134,
"step": 1210
},
{
"epoch": 3.9,
"learning_rate": 7.665566274897007e-09,
"logits/chosen": 0.6319261789321899,
"logits/rejected": 0.8173401951789856,
"logps/chosen": -197.80917358398438,
"logps/rejected": -135.27621459960938,
"loss": 0.0001,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.005665967706590891,
"rewards/margins": 0.005035373382270336,
"rewards/rejected": 0.0006305938586592674,
"step": 1220
},
{
"epoch": 3.94,
"learning_rate": 3.1688631685364292e-09,
"logits/chosen": 0.5452470183372498,
"logits/rejected": 0.7646620869636536,
"logps/chosen": -184.90304565429688,
"logps/rejected": -128.32806396484375,
"loss": 0.0001,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.0048486096784472466,
"rewards/margins": 0.004199582617729902,
"rewards/rejected": 0.0006490271771326661,
"step": 1230
},
{
"epoch": 3.97,
"learning_rate": 6.260544298619664e-10,
"logits/chosen": 0.6011223793029785,
"logits/rejected": 0.7510023713111877,
"logps/chosen": -178.15272521972656,
"logps/rejected": -124.17759704589844,
"loss": 0.0001,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": 0.0051615191623568535,
"rewards/margins": 0.003729583229869604,
"rewards/rejected": 0.0014319362817332149,
"step": 1240
},
{
"epoch": 3.99,
"step": 1248,
"total_flos": 0.0,
"train_loss": 9.35627439654733e-05,
"train_runtime": 14053.8323,
"train_samples_per_second": 1.423,
"train_steps_per_second": 0.089
}
],
"logging_steps": 10,
"max_steps": 1248,
"num_input_tokens_seen": 0,
"num_train_epochs": 4,
"save_steps": 100,
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}