zephyr-7b-ipo-lora / trainer_state.json
Mastane's picture
Model save
3a36520
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.9992254066615027,
"eval_steps": 100,
"global_step": 726,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 6.84931506849315e-09,
"logits/chosen": -2.55204439163208,
"logits/rejected": -2.328748941421509,
"logps/chosen": -243.65313720703125,
"logps/rejected": -203.76687622070312,
"loss": 25.0,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.04,
"learning_rate": 6.84931506849315e-08,
"logits/chosen": -2.6284379959106445,
"logits/rejected": -2.217939853668213,
"logps/chosen": -283.397705078125,
"logps/rejected": -223.02455139160156,
"loss": 25.2356,
"rewards/accuracies": 0.4652777910232544,
"rewards/chosen": 0.000989701016806066,
"rewards/margins": 0.0014238969888538122,
"rewards/rejected": -0.0004341956228017807,
"step": 10
},
{
"epoch": 0.08,
"learning_rate": 1.36986301369863e-07,
"logits/chosen": -2.6449296474456787,
"logits/rejected": -2.2949368953704834,
"logps/chosen": -276.48468017578125,
"logps/rejected": -204.2726593017578,
"loss": 25.4323,
"rewards/accuracies": 0.5093749761581421,
"rewards/chosen": 0.0012504293117672205,
"rewards/margins": 0.0009743297705426812,
"rewards/rejected": 0.0002760997449513525,
"step": 20
},
{
"epoch": 0.12,
"learning_rate": 2.054794520547945e-07,
"logits/chosen": -2.61030912399292,
"logits/rejected": -2.28235125541687,
"logps/chosen": -242.44509887695312,
"logps/rejected": -195.07154846191406,
"loss": 25.3196,
"rewards/accuracies": 0.5093749761581421,
"rewards/chosen": -0.0013605057029053569,
"rewards/margins": 0.0007983079412952065,
"rewards/rejected": -0.0021588136442005634,
"step": 30
},
{
"epoch": 0.17,
"learning_rate": 2.73972602739726e-07,
"logits/chosen": -2.676435947418213,
"logits/rejected": -2.2325174808502197,
"logps/chosen": -276.4580383300781,
"logps/rejected": -195.8528289794922,
"loss": 24.8252,
"rewards/accuracies": 0.546875,
"rewards/chosen": 0.004295675549656153,
"rewards/margins": 0.006117778830230236,
"rewards/rejected": -0.001822102814912796,
"step": 40
},
{
"epoch": 0.21,
"learning_rate": 3.424657534246575e-07,
"logits/chosen": -2.5881507396698,
"logits/rejected": -2.340297222137451,
"logps/chosen": -254.146484375,
"logps/rejected": -227.5242919921875,
"loss": 24.8309,
"rewards/accuracies": 0.484375,
"rewards/chosen": 0.0033471225760877132,
"rewards/margins": -0.0003331190091557801,
"rewards/rejected": 0.003680241061374545,
"step": 50
},
{
"epoch": 0.25,
"learning_rate": 4.10958904109589e-07,
"logits/chosen": -2.6254591941833496,
"logits/rejected": -2.264179229736328,
"logps/chosen": -268.96685791015625,
"logps/rejected": -212.42770385742188,
"loss": 23.9609,
"rewards/accuracies": 0.6156250238418579,
"rewards/chosen": 0.015944072976708412,
"rewards/margins": 0.016598206013441086,
"rewards/rejected": -0.0006541347247548401,
"step": 60
},
{
"epoch": 0.29,
"learning_rate": 4.794520547945205e-07,
"logits/chosen": -2.596461534500122,
"logits/rejected": -2.2788121700286865,
"logps/chosen": -244.193115234375,
"logps/rejected": -198.00149536132812,
"loss": 23.4551,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": 0.01318939495831728,
"rewards/margins": 0.014477437362074852,
"rewards/rejected": -0.0012880431022495031,
"step": 70
},
{
"epoch": 0.33,
"learning_rate": 4.946401225114854e-07,
"logits/chosen": -2.6231789588928223,
"logits/rejected": -2.248992443084717,
"logps/chosen": -256.1118469238281,
"logps/rejected": -212.2821502685547,
"loss": 22.8216,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": 0.021900424733757973,
"rewards/margins": 0.034875936806201935,
"rewards/rejected": -0.012975512072443962,
"step": 80
},
{
"epoch": 0.37,
"learning_rate": 4.869831546707504e-07,
"logits/chosen": -2.6711127758026123,
"logits/rejected": -2.2961134910583496,
"logps/chosen": -265.5664978027344,
"logps/rejected": -222.6741180419922,
"loss": 22.2504,
"rewards/accuracies": 0.621874988079071,
"rewards/chosen": 0.037670500576496124,
"rewards/margins": 0.04158278554677963,
"rewards/rejected": -0.003912287298589945,
"step": 90
},
{
"epoch": 0.41,
"learning_rate": 4.793261868300153e-07,
"logits/chosen": -2.6314404010772705,
"logits/rejected": -2.367823362350464,
"logps/chosen": -272.08331298828125,
"logps/rejected": -240.5725555419922,
"loss": 21.7851,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": 0.04255244508385658,
"rewards/margins": 0.05080147832632065,
"rewards/rejected": -0.00824903603643179,
"step": 100
},
{
"epoch": 0.45,
"learning_rate": 4.7166921898928023e-07,
"logits/chosen": -2.665757417678833,
"logits/rejected": -2.239075183868408,
"logps/chosen": -236.4951934814453,
"logps/rejected": -197.68630981445312,
"loss": 21.284,
"rewards/accuracies": 0.715624988079071,
"rewards/chosen": 0.03496263176202774,
"rewards/margins": 0.06847874075174332,
"rewards/rejected": -0.03351610153913498,
"step": 110
},
{
"epoch": 0.5,
"learning_rate": 4.640122511485451e-07,
"logits/chosen": -2.61603045463562,
"logits/rejected": -2.319265127182007,
"logps/chosen": -263.4698791503906,
"logps/rejected": -221.0123748779297,
"loss": 20.8846,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": 0.03651345893740654,
"rewards/margins": 0.06544128805398941,
"rewards/rejected": -0.02892782725393772,
"step": 120
},
{
"epoch": 0.54,
"learning_rate": 4.563552833078101e-07,
"logits/chosen": -2.642108201980591,
"logits/rejected": -2.2785372734069824,
"logps/chosen": -265.32977294921875,
"logps/rejected": -217.34768676757812,
"loss": 20.4002,
"rewards/accuracies": 0.671875,
"rewards/chosen": 0.039137523621320724,
"rewards/margins": 0.07437606900930405,
"rewards/rejected": -0.035238541662693024,
"step": 130
},
{
"epoch": 0.58,
"learning_rate": 4.4869831546707505e-07,
"logits/chosen": -2.6350350379943848,
"logits/rejected": -2.323685646057129,
"logps/chosen": -266.7952575683594,
"logps/rejected": -226.29617309570312,
"loss": 20.3918,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": 0.04871717840433121,
"rewards/margins": 0.0838700383901596,
"rewards/rejected": -0.0351528637111187,
"step": 140
},
{
"epoch": 0.62,
"learning_rate": 4.4104134762633994e-07,
"logits/chosen": -2.6673989295959473,
"logits/rejected": -2.305692195892334,
"logps/chosen": -267.29180908203125,
"logps/rejected": -207.8380889892578,
"loss": 19.8348,
"rewards/accuracies": 0.703125,
"rewards/chosen": 0.04811176657676697,
"rewards/margins": 0.09522075951099396,
"rewards/rejected": -0.047109004110097885,
"step": 150
},
{
"epoch": 0.66,
"learning_rate": 4.333843797856049e-07,
"logits/chosen": -2.664452314376831,
"logits/rejected": -2.3129963874816895,
"logps/chosen": -264.9499816894531,
"logps/rejected": -236.6409454345703,
"loss": 19.7381,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": 0.045278288424015045,
"rewards/margins": 0.0909709632396698,
"rewards/rejected": -0.04569266736507416,
"step": 160
},
{
"epoch": 0.7,
"learning_rate": 4.257274119448698e-07,
"logits/chosen": -2.581209421157837,
"logits/rejected": -2.3164689540863037,
"logps/chosen": -269.43072509765625,
"logps/rejected": -230.83047485351562,
"loss": 19.8434,
"rewards/accuracies": 0.640625,
"rewards/chosen": 0.03498955816030502,
"rewards/margins": 0.08386311680078506,
"rewards/rejected": -0.048873551189899445,
"step": 170
},
{
"epoch": 0.74,
"learning_rate": 4.180704441041347e-07,
"logits/chosen": -2.648585557937622,
"logits/rejected": -2.3146426677703857,
"logps/chosen": -254.5563507080078,
"logps/rejected": -222.23922729492188,
"loss": 20.3028,
"rewards/accuracies": 0.6875,
"rewards/chosen": 0.03642386570572853,
"rewards/margins": 0.10294415056705475,
"rewards/rejected": -0.06652027368545532,
"step": 180
},
{
"epoch": 0.78,
"learning_rate": 4.1041347626339966e-07,
"logits/chosen": -2.684251308441162,
"logits/rejected": -2.282341241836548,
"logps/chosen": -262.4200744628906,
"logps/rejected": -213.4808807373047,
"loss": 20.0038,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.03100644052028656,
"rewards/margins": 0.10564477741718292,
"rewards/rejected": -0.07463832199573517,
"step": 190
},
{
"epoch": 0.83,
"learning_rate": 4.027565084226646e-07,
"logits/chosen": -2.6774110794067383,
"logits/rejected": -2.31669020652771,
"logps/chosen": -242.54110717773438,
"logps/rejected": -201.0673370361328,
"loss": 18.8904,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": 0.029109233990311623,
"rewards/margins": 0.09885577857494354,
"rewards/rejected": -0.06974655389785767,
"step": 200
},
{
"epoch": 0.87,
"learning_rate": 3.9509954058192954e-07,
"logits/chosen": -2.6493515968322754,
"logits/rejected": -2.2453055381774902,
"logps/chosen": -269.30072021484375,
"logps/rejected": -215.35482788085938,
"loss": 18.7438,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": 0.03713355213403702,
"rewards/margins": 0.12264309078454971,
"rewards/rejected": -0.0855095237493515,
"step": 210
},
{
"epoch": 0.91,
"learning_rate": 3.874425727411945e-07,
"logits/chosen": -2.6454288959503174,
"logits/rejected": -2.2814364433288574,
"logps/chosen": -264.0574035644531,
"logps/rejected": -206.1600341796875,
"loss": 19.3111,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.03974270448088646,
"rewards/margins": 0.12600289285182953,
"rewards/rejected": -0.08626019209623337,
"step": 220
},
{
"epoch": 0.95,
"learning_rate": 3.797856049004594e-07,
"logits/chosen": -2.6513967514038086,
"logits/rejected": -2.437913179397583,
"logps/chosen": -261.82855224609375,
"logps/rejected": -220.0791778564453,
"loss": 19.1461,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": 0.04797480255365372,
"rewards/margins": 0.12008707225322723,
"rewards/rejected": -0.07211227715015411,
"step": 230
},
{
"epoch": 0.99,
"learning_rate": 3.7212863705972436e-07,
"logits/chosen": -2.664517879486084,
"logits/rejected": -2.3380627632141113,
"logps/chosen": -268.38922119140625,
"logps/rejected": -203.86264038085938,
"loss": 19.1303,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.037844784557819366,
"rewards/margins": 0.11463721841573715,
"rewards/rejected": -0.07679243385791779,
"step": 240
},
{
"epoch": 1.0,
"eval_logits/chosen": -2.688377618789673,
"eval_logits/rejected": -2.3029286861419678,
"eval_logps/chosen": -268.4410400390625,
"eval_logps/rejected": -212.60609436035156,
"eval_loss": 18.912391662597656,
"eval_rewards/accuracies": 0.7360000014305115,
"eval_rewards/chosen": 0.03018273413181305,
"eval_rewards/margins": 0.12408097833395004,
"eval_rewards/rejected": -0.09389825165271759,
"eval_runtime": 352.0426,
"eval_samples_per_second": 5.681,
"eval_steps_per_second": 0.71,
"step": 242
},
{
"epoch": 1.03,
"learning_rate": 3.6447166921898925e-07,
"logits/chosen": -2.585019588470459,
"logits/rejected": -2.2706055641174316,
"logps/chosen": -248.5968017578125,
"logps/rejected": -192.33859252929688,
"loss": 19.3292,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": 0.026646222919225693,
"rewards/margins": 0.12298872321844101,
"rewards/rejected": -0.09634250402450562,
"step": 250
},
{
"epoch": 1.07,
"learning_rate": 3.568147013782542e-07,
"logits/chosen": -2.6509532928466797,
"logits/rejected": -2.2828612327575684,
"logps/chosen": -250.4876708984375,
"logps/rejected": -202.44654846191406,
"loss": 18.5341,
"rewards/accuracies": 0.734375,
"rewards/chosen": 0.04957858473062515,
"rewards/margins": 0.13254784047603607,
"rewards/rejected": -0.08296926319599152,
"step": 260
},
{
"epoch": 1.12,
"learning_rate": 3.4915773353751913e-07,
"logits/chosen": -2.6814422607421875,
"logits/rejected": -2.252516269683838,
"logps/chosen": -273.83038330078125,
"logps/rejected": -227.3008270263672,
"loss": 18.3748,
"rewards/accuracies": 0.768750011920929,
"rewards/chosen": 0.038934461772441864,
"rewards/margins": 0.14948731660842896,
"rewards/rejected": -0.11055286228656769,
"step": 270
},
{
"epoch": 1.16,
"learning_rate": 3.41500765696784e-07,
"logits/chosen": -2.6279919147491455,
"logits/rejected": -2.3474836349487305,
"logps/chosen": -230.10598754882812,
"logps/rejected": -215.902587890625,
"loss": 18.5872,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.03518626466393471,
"rewards/margins": 0.108550526201725,
"rewards/rejected": -0.0733642652630806,
"step": 280
},
{
"epoch": 1.2,
"learning_rate": 3.33843797856049e-07,
"logits/chosen": -2.625840663909912,
"logits/rejected": -2.388363838195801,
"logps/chosen": -287.68280029296875,
"logps/rejected": -237.7518768310547,
"loss": 18.1373,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": 0.03980474919080734,
"rewards/margins": 0.125565305352211,
"rewards/rejected": -0.08576056361198425,
"step": 290
},
{
"epoch": 1.24,
"learning_rate": 3.2618683001531396e-07,
"logits/chosen": -2.592059373855591,
"logits/rejected": -2.242305278778076,
"logps/chosen": -239.9512939453125,
"logps/rejected": -212.0597686767578,
"loss": 18.2727,
"rewards/accuracies": 0.746874988079071,
"rewards/chosen": 0.030053243041038513,
"rewards/margins": 0.1290222704410553,
"rewards/rejected": -0.09896902740001678,
"step": 300
},
{
"epoch": 1.28,
"learning_rate": 3.1852986217457885e-07,
"logits/chosen": -2.6135897636413574,
"logits/rejected": -2.2893946170806885,
"logps/chosen": -252.56002807617188,
"logps/rejected": -218.57937622070312,
"loss": 18.9516,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.027646947652101517,
"rewards/margins": 0.11891412734985352,
"rewards/rejected": -0.0912671834230423,
"step": 310
},
{
"epoch": 1.32,
"learning_rate": 3.108728943338438e-07,
"logits/chosen": -2.6720099449157715,
"logits/rejected": -2.3789660930633545,
"logps/chosen": -255.5870361328125,
"logps/rejected": -232.6374969482422,
"loss": 18.993,
"rewards/accuracies": 0.684374988079071,
"rewards/chosen": 0.03462428227066994,
"rewards/margins": 0.11329145729541779,
"rewards/rejected": -0.07866715639829636,
"step": 320
},
{
"epoch": 1.36,
"learning_rate": 3.0321592649310873e-07,
"logits/chosen": -2.636521816253662,
"logits/rejected": -2.260267496109009,
"logps/chosen": -242.1477508544922,
"logps/rejected": -213.030029296875,
"loss": 18.1642,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.04163060337305069,
"rewards/margins": 0.13628455996513367,
"rewards/rejected": -0.09465396404266357,
"step": 330
},
{
"epoch": 1.4,
"learning_rate": 2.955589586523736e-07,
"logits/chosen": -2.6310007572174072,
"logits/rejected": -2.3438334465026855,
"logps/chosen": -268.5300598144531,
"logps/rejected": -229.24057006835938,
"loss": 19.3254,
"rewards/accuracies": 0.734375,
"rewards/chosen": 0.03425261378288269,
"rewards/margins": 0.1407606601715088,
"rewards/rejected": -0.1065080538392067,
"step": 340
},
{
"epoch": 1.45,
"learning_rate": 2.8790199081163856e-07,
"logits/chosen": -2.6794886589050293,
"logits/rejected": -2.297974109649658,
"logps/chosen": -271.8599853515625,
"logps/rejected": -222.39688110351562,
"loss": 19.1924,
"rewards/accuracies": 0.684374988079071,
"rewards/chosen": 0.022753948345780373,
"rewards/margins": 0.10870112478733063,
"rewards/rejected": -0.0859471783041954,
"step": 350
},
{
"epoch": 1.49,
"learning_rate": 2.802450229709035e-07,
"logits/chosen": -2.635558843612671,
"logits/rejected": -2.196741819381714,
"logps/chosen": -250.3131103515625,
"logps/rejected": -201.41946411132812,
"loss": 18.4515,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.018490195274353027,
"rewards/margins": 0.12475041300058365,
"rewards/rejected": -0.10626020282506943,
"step": 360
},
{
"epoch": 1.53,
"learning_rate": 2.725880551301684e-07,
"logits/chosen": -2.6633524894714355,
"logits/rejected": -2.3473284244537354,
"logps/chosen": -271.9929504394531,
"logps/rejected": -228.14157104492188,
"loss": 19.0148,
"rewards/accuracies": 0.684374988079071,
"rewards/chosen": 0.015023022890090942,
"rewards/margins": 0.12562182545661926,
"rewards/rejected": -0.11059880256652832,
"step": 370
},
{
"epoch": 1.57,
"learning_rate": 2.649310872894334e-07,
"logits/chosen": -2.6451079845428467,
"logits/rejected": -2.3189663887023926,
"logps/chosen": -245.8695831298828,
"logps/rejected": -210.0269775390625,
"loss": 18.4524,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.007006034255027771,
"rewards/margins": 0.11740052700042725,
"rewards/rejected": -0.11039450019598007,
"step": 380
},
{
"epoch": 1.61,
"learning_rate": 2.572741194486983e-07,
"logits/chosen": -2.6605451107025146,
"logits/rejected": -2.206817626953125,
"logps/chosen": -275.4052429199219,
"logps/rejected": -203.68002319335938,
"loss": 18.4799,
"rewards/accuracies": 0.746874988079071,
"rewards/chosen": 0.017081182450056076,
"rewards/margins": 0.14246238768100739,
"rewards/rejected": -0.1253812164068222,
"step": 390
},
{
"epoch": 1.65,
"learning_rate": 2.496171516079632e-07,
"logits/chosen": -2.74629282951355,
"logits/rejected": -2.338416337966919,
"logps/chosen": -276.91033935546875,
"logps/rejected": -221.68307495117188,
"loss": 18.5723,
"rewards/accuracies": 0.7281249761581421,
"rewards/chosen": 0.018960032612085342,
"rewards/margins": 0.1522599160671234,
"rewards/rejected": -0.13329990208148956,
"step": 400
},
{
"epoch": 1.69,
"learning_rate": 2.4196018376722816e-07,
"logits/chosen": -2.616495370864868,
"logits/rejected": -2.2858164310455322,
"logps/chosen": -260.0771484375,
"logps/rejected": -216.95919799804688,
"loss": 18.8108,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": 0.022241920232772827,
"rewards/margins": 0.14108815789222717,
"rewards/rejected": -0.11884623765945435,
"step": 410
},
{
"epoch": 1.74,
"learning_rate": 2.343032159264931e-07,
"logits/chosen": -2.651947259902954,
"logits/rejected": -2.266605854034424,
"logps/chosen": -264.1819152832031,
"logps/rejected": -221.7666473388672,
"loss": 18.6108,
"rewards/accuracies": 0.746874988079071,
"rewards/chosen": 0.019768675789237022,
"rewards/margins": 0.14661307632923126,
"rewards/rejected": -0.1268443912267685,
"step": 420
},
{
"epoch": 1.78,
"learning_rate": 2.26646248085758e-07,
"logits/chosen": -2.6224217414855957,
"logits/rejected": -2.2512006759643555,
"logps/chosen": -242.04757690429688,
"logps/rejected": -202.48556518554688,
"loss": 18.291,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": 0.020827816799283028,
"rewards/margins": 0.13378821313381195,
"rewards/rejected": -0.11296038329601288,
"step": 430
},
{
"epoch": 1.82,
"learning_rate": 2.1898928024502298e-07,
"logits/chosen": -2.658205509185791,
"logits/rejected": -2.3186447620391846,
"logps/chosen": -247.4537353515625,
"logps/rejected": -198.62149047851562,
"loss": 18.2582,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": 0.008651083335280418,
"rewards/margins": 0.12013272196054459,
"rewards/rejected": -0.11148162931203842,
"step": 440
},
{
"epoch": 1.86,
"learning_rate": 2.113323124042879e-07,
"logits/chosen": -2.6413986682891846,
"logits/rejected": -2.323753595352173,
"logps/chosen": -265.66241455078125,
"logps/rejected": -233.0228729248047,
"loss": 18.5032,
"rewards/accuracies": 0.734375,
"rewards/chosen": 0.01642756350338459,
"rewards/margins": 0.13231472671031952,
"rewards/rejected": -0.11588716506958008,
"step": 450
},
{
"epoch": 1.9,
"learning_rate": 2.036753445635528e-07,
"logits/chosen": -2.6705307960510254,
"logits/rejected": -2.3020169734954834,
"logps/chosen": -286.9468078613281,
"logps/rejected": -232.2310028076172,
"loss": 18.0061,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.01540865283459425,
"rewards/margins": 0.14157035946846008,
"rewards/rejected": -0.12616169452667236,
"step": 460
},
{
"epoch": 1.94,
"learning_rate": 1.9601837672281775e-07,
"logits/chosen": -2.645066022872925,
"logits/rejected": -2.2868847846984863,
"logps/chosen": -275.22344970703125,
"logps/rejected": -230.9308624267578,
"loss": 18.0886,
"rewards/accuracies": 0.7593749761581421,
"rewards/chosen": 0.025068962946534157,
"rewards/margins": 0.1567041575908661,
"rewards/rejected": -0.13163520395755768,
"step": 470
},
{
"epoch": 1.98,
"learning_rate": 1.883614088820827e-07,
"logits/chosen": -2.6855411529541016,
"logits/rejected": -2.2669596672058105,
"logps/chosen": -257.3984680175781,
"logps/rejected": -210.7415771484375,
"loss": 19.0824,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.01750902086496353,
"rewards/margins": 0.12825700640678406,
"rewards/rejected": -0.11074797064065933,
"step": 480
},
{
"epoch": 2.0,
"eval_logits/chosen": -2.687664747238159,
"eval_logits/rejected": -2.3010897636413574,
"eval_logps/chosen": -268.55670166015625,
"eval_logps/rejected": -212.82968139648438,
"eval_loss": 18.22655487060547,
"eval_rewards/accuracies": 0.7599999904632568,
"eval_rewards/chosen": 0.01861482858657837,
"eval_rewards/margins": 0.13487249612808228,
"eval_rewards/rejected": -0.1162576824426651,
"eval_runtime": 352.6258,
"eval_samples_per_second": 5.672,
"eval_steps_per_second": 0.709,
"step": 484
},
{
"epoch": 2.02,
"learning_rate": 1.807044410413476e-07,
"logits/chosen": -2.63578200340271,
"logits/rejected": -2.335254430770874,
"logps/chosen": -252.0665740966797,
"logps/rejected": -221.12002563476562,
"loss": 18.6586,
"rewards/accuracies": 0.734375,
"rewards/chosen": 0.00841272622346878,
"rewards/margins": 0.12638814747333527,
"rewards/rejected": -0.11797541379928589,
"step": 490
},
{
"epoch": 2.07,
"learning_rate": 1.7304747320061255e-07,
"logits/chosen": -2.576451063156128,
"logits/rejected": -2.343707799911499,
"logps/chosen": -253.4521484375,
"logps/rejected": -228.43115234375,
"loss": 18.2798,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": 0.0265754796564579,
"rewards/margins": 0.1290186494588852,
"rewards/rejected": -0.1024431437253952,
"step": 500
},
{
"epoch": 2.11,
"learning_rate": 1.6539050535987747e-07,
"logits/chosen": -2.672646999359131,
"logits/rejected": -2.302988290786743,
"logps/chosen": -275.57440185546875,
"logps/rejected": -221.45455932617188,
"loss": 18.1,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.02416495978832245,
"rewards/margins": 0.14235563576221466,
"rewards/rejected": -0.11819068342447281,
"step": 510
},
{
"epoch": 2.15,
"learning_rate": 1.5773353751914243e-07,
"logits/chosen": -2.759183406829834,
"logits/rejected": -2.3177244663238525,
"logps/chosen": -282.8927307128906,
"logps/rejected": -217.7075958251953,
"loss": 18.4222,
"rewards/accuracies": 0.7093750238418579,
"rewards/chosen": 0.021920911967754364,
"rewards/margins": 0.13253231346607208,
"rewards/rejected": -0.11061139404773712,
"step": 520
},
{
"epoch": 2.19,
"learning_rate": 1.5007656967840735e-07,
"logits/chosen": -2.6353068351745605,
"logits/rejected": -2.291774272918701,
"logps/chosen": -260.0713806152344,
"logps/rejected": -232.3697052001953,
"loss": 18.084,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": 0.020517101511359215,
"rewards/margins": 0.15263554453849792,
"rewards/rejected": -0.13211843371391296,
"step": 530
},
{
"epoch": 2.23,
"learning_rate": 1.4241960183767226e-07,
"logits/chosen": -2.665095806121826,
"logits/rejected": -2.3143036365509033,
"logps/chosen": -270.5109558105469,
"logps/rejected": -228.31759643554688,
"loss": 18.2089,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.020305803045630455,
"rewards/margins": 0.12145675718784332,
"rewards/rejected": -0.10115096718072891,
"step": 540
},
{
"epoch": 2.27,
"learning_rate": 1.347626339969372e-07,
"logits/chosen": -2.6581876277923584,
"logits/rejected": -2.402988910675049,
"logps/chosen": -276.17437744140625,
"logps/rejected": -239.04080200195312,
"loss": 18.2431,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": 0.04555383324623108,
"rewards/margins": 0.13798871636390686,
"rewards/rejected": -0.09243487566709518,
"step": 550
},
{
"epoch": 2.31,
"learning_rate": 1.2710566615620215e-07,
"logits/chosen": -2.6366963386535645,
"logits/rejected": -2.2572150230407715,
"logps/chosen": -272.67034912109375,
"logps/rejected": -214.80441284179688,
"loss": 17.8127,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.020294126123189926,
"rewards/margins": 0.14420394599437714,
"rewards/rejected": -0.1239098310470581,
"step": 560
},
{
"epoch": 2.35,
"learning_rate": 1.1944869831546706e-07,
"logits/chosen": -2.663076162338257,
"logits/rejected": -2.3145666122436523,
"logps/chosen": -273.4615478515625,
"logps/rejected": -232.04428100585938,
"loss": 18.2306,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": 0.03313525393605232,
"rewards/margins": 0.13954436779022217,
"rewards/rejected": -0.10640911012887955,
"step": 570
},
{
"epoch": 2.4,
"learning_rate": 1.11791730474732e-07,
"logits/chosen": -2.637507915496826,
"logits/rejected": -2.294003963470459,
"logps/chosen": -254.7287139892578,
"logps/rejected": -214.94534301757812,
"loss": 17.9129,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": 0.023206666111946106,
"rewards/margins": 0.14315126836299896,
"rewards/rejected": -0.11994459480047226,
"step": 580
},
{
"epoch": 2.44,
"learning_rate": 1.0413476263399694e-07,
"logits/chosen": -2.5812573432922363,
"logits/rejected": -2.254936695098877,
"logps/chosen": -267.64752197265625,
"logps/rejected": -209.09072875976562,
"loss": 18.3503,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": 0.01215180940926075,
"rewards/margins": 0.13401289284229279,
"rewards/rejected": -0.12186107784509659,
"step": 590
},
{
"epoch": 2.48,
"learning_rate": 9.647779479326186e-08,
"logits/chosen": -2.645982265472412,
"logits/rejected": -2.3950445652008057,
"logps/chosen": -253.10336303710938,
"logps/rejected": -217.19949340820312,
"loss": 19.0115,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": 0.022207045927643776,
"rewards/margins": 0.11882482469081879,
"rewards/rejected": -0.09661778062582016,
"step": 600
},
{
"epoch": 2.52,
"learning_rate": 8.88208269525268e-08,
"logits/chosen": -2.606689929962158,
"logits/rejected": -2.274811029434204,
"logps/chosen": -254.597412109375,
"logps/rejected": -207.0297088623047,
"loss": 18.634,
"rewards/accuracies": 0.721875011920929,
"rewards/chosen": 0.020870240405201912,
"rewards/margins": 0.11707819998264313,
"rewards/rejected": -0.09620794653892517,
"step": 610
},
{
"epoch": 2.56,
"learning_rate": 8.116385911179173e-08,
"logits/chosen": -2.6234843730926514,
"logits/rejected": -2.277554750442505,
"logps/chosen": -265.1994934082031,
"logps/rejected": -217.2080841064453,
"loss": 18.3686,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.03428371995687485,
"rewards/margins": 0.13564249873161316,
"rewards/rejected": -0.1013587936758995,
"step": 620
},
{
"epoch": 2.6,
"learning_rate": 7.350689127105667e-08,
"logits/chosen": -2.681215286254883,
"logits/rejected": -2.2755188941955566,
"logps/chosen": -263.49530029296875,
"logps/rejected": -226.5316619873047,
"loss": 18.1639,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": 0.024900907650589943,
"rewards/margins": 0.14352695643901825,
"rewards/rejected": -0.11862604320049286,
"step": 630
},
{
"epoch": 2.64,
"learning_rate": 6.584992343032159e-08,
"logits/chosen": -2.589116096496582,
"logits/rejected": -2.3119215965270996,
"logps/chosen": -255.6792449951172,
"logps/rejected": -225.32754516601562,
"loss": 18.7407,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": 0.022887857630848885,
"rewards/margins": 0.13587646186351776,
"rewards/rejected": -0.11298861354589462,
"step": 640
},
{
"epoch": 2.69,
"learning_rate": 5.819295558958652e-08,
"logits/chosen": -2.664092779159546,
"logits/rejected": -2.245487928390503,
"logps/chosen": -287.26129150390625,
"logps/rejected": -202.8896942138672,
"loss": 17.9121,
"rewards/accuracies": 0.765625,
"rewards/chosen": 0.03479045629501343,
"rewards/margins": 0.17097266018390656,
"rewards/rejected": -0.13618221879005432,
"step": 650
},
{
"epoch": 2.73,
"learning_rate": 5.0535987748851455e-08,
"logits/chosen": -2.6955533027648926,
"logits/rejected": -2.341909646987915,
"logps/chosen": -276.48583984375,
"logps/rejected": -231.2989959716797,
"loss": 18.0214,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": 0.03698002174496651,
"rewards/margins": 0.14650048315525055,
"rewards/rejected": -0.10952047258615494,
"step": 660
},
{
"epoch": 2.77,
"learning_rate": 4.287901990811638e-08,
"logits/chosen": -2.691222667694092,
"logits/rejected": -2.269242286682129,
"logps/chosen": -257.7073974609375,
"logps/rejected": -232.1117706298828,
"loss": 18.0439,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.030216818675398827,
"rewards/margins": 0.14609763026237488,
"rewards/rejected": -0.1158808022737503,
"step": 670
},
{
"epoch": 2.81,
"learning_rate": 3.522205206738132e-08,
"logits/chosen": -2.6416354179382324,
"logits/rejected": -2.310128688812256,
"logps/chosen": -271.9048767089844,
"logps/rejected": -223.27969360351562,
"loss": 18.4412,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": 0.02033752016723156,
"rewards/margins": 0.12728351354599,
"rewards/rejected": -0.10694599151611328,
"step": 680
},
{
"epoch": 2.85,
"learning_rate": 2.7565084226646246e-08,
"logits/chosen": -2.660130739212036,
"logits/rejected": -2.3302083015441895,
"logps/chosen": -269.9329833984375,
"logps/rejected": -232.0976104736328,
"loss": 18.3673,
"rewards/accuracies": 0.746874988079071,
"rewards/chosen": 0.030508223921060562,
"rewards/margins": 0.14415839314460754,
"rewards/rejected": -0.11365016549825668,
"step": 690
},
{
"epoch": 2.89,
"learning_rate": 1.9908116385911178e-08,
"logits/chosen": -2.6514875888824463,
"logits/rejected": -2.333763599395752,
"logps/chosen": -251.9337158203125,
"logps/rejected": -223.32437133789062,
"loss": 18.373,
"rewards/accuracies": 0.684374988079071,
"rewards/chosen": 0.020953018218278885,
"rewards/margins": 0.114670529961586,
"rewards/rejected": -0.09371750801801682,
"step": 700
},
{
"epoch": 2.93,
"learning_rate": 1.225114854517611e-08,
"logits/chosen": -2.6763319969177246,
"logits/rejected": -2.2669501304626465,
"logps/chosen": -274.1488037109375,
"logps/rejected": -220.2371063232422,
"loss": 18.4056,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.005063548684120178,
"rewards/margins": 0.12724530696868896,
"rewards/rejected": -0.12218175083398819,
"step": 710
},
{
"epoch": 2.97,
"learning_rate": 4.594180704441042e-09,
"logits/chosen": -2.5772223472595215,
"logits/rejected": -2.3834898471832275,
"logps/chosen": -257.3438720703125,
"logps/rejected": -220.6257781982422,
"loss": 18.0366,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.019895819947123528,
"rewards/margins": 0.12051371484994888,
"rewards/rejected": -0.10061788558959961,
"step": 720
},
{
"epoch": 3.0,
"eval_logits/chosen": -2.687349796295166,
"eval_logits/rejected": -2.300260305404663,
"eval_logps/chosen": -268.54150390625,
"eval_logps/rejected": -212.86087036132812,
"eval_loss": 18.088815689086914,
"eval_rewards/accuracies": 0.7319999933242798,
"eval_rewards/chosen": 0.02013408951461315,
"eval_rewards/margins": 0.13951165974140167,
"eval_rewards/rejected": -0.11937756091356277,
"eval_runtime": 350.3279,
"eval_samples_per_second": 5.709,
"eval_steps_per_second": 0.714,
"step": 726
},
{
"epoch": 3.0,
"step": 726,
"total_flos": 0.0,
"train_loss": 19.467516181882747,
"train_runtime": 42806.6761,
"train_samples_per_second": 4.343,
"train_steps_per_second": 0.017
}
],
"logging_steps": 10,
"max_steps": 726,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 0.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}