10174 lines
337 KiB
JSON
10174 lines
337 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 2.9666818804198996,
|
|
"eval_steps": 100,
|
|
"global_step": 6500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.559270516717325e-10,
|
|
"logits/chosen": -0.9643518328666687,
|
|
"logits/rejected": -0.9552459120750427,
|
|
"logps/chosen": -76.82135772705078,
|
|
"logps/rejected": -59.52644348144531,
|
|
"loss": 0.6931,
|
|
"rewards/accuracies": 0.0,
|
|
"rewards/chosen": 0.0,
|
|
"rewards/margins": 0.0,
|
|
"rewards/rejected": 0.0,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.559270516717325e-09,
|
|
"logits/chosen": -0.9517099857330322,
|
|
"logits/rejected": -0.9616715908050537,
|
|
"logps/chosen": -93.39585876464844,
|
|
"logps/rejected": -69.06417846679688,
|
|
"loss": 0.6969,
|
|
"rewards/accuracies": 0.3055555522441864,
|
|
"rewards/chosen": -0.04679955542087555,
|
|
"rewards/margins": -0.035246770828962326,
|
|
"rewards/rejected": -0.011552784591913223,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 9.11854103343465e-09,
|
|
"logits/chosen": -0.9816751480102539,
|
|
"logits/rejected": -1.003169298171997,
|
|
"logps/chosen": -90.08583068847656,
|
|
"logps/rejected": -69.84603118896484,
|
|
"loss": 0.6993,
|
|
"rewards/accuracies": 0.4749999940395355,
|
|
"rewards/chosen": -0.00020122528076171875,
|
|
"rewards/margins": -0.014502143487334251,
|
|
"rewards/rejected": 0.014300918206572533,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.3677811550151975e-08,
|
|
"logits/chosen": -0.9795465469360352,
|
|
"logits/rejected": -0.9944146275520325,
|
|
"logps/chosen": -90.3318099975586,
|
|
"logps/rejected": -66.77433013916016,
|
|
"loss": 0.6898,
|
|
"rewards/accuracies": 0.48750001192092896,
|
|
"rewards/chosen": 0.007479047868400812,
|
|
"rewards/margins": 0.011048078536987305,
|
|
"rewards/rejected": -0.0035690306685864925,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 1.82370820668693e-08,
|
|
"logits/chosen": -0.9575172662734985,
|
|
"logits/rejected": -0.974955677986145,
|
|
"logps/chosen": -94.9634017944336,
|
|
"logps/rejected": -64.13943481445312,
|
|
"loss": 0.6901,
|
|
"rewards/accuracies": 0.512499988079071,
|
|
"rewards/chosen": -0.020472276955842972,
|
|
"rewards/margins": -0.0019337296253070235,
|
|
"rewards/rejected": -0.018538545817136765,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.2796352583586623e-08,
|
|
"logits/chosen": -0.9589303731918335,
|
|
"logits/rejected": -0.9642618894577026,
|
|
"logps/chosen": -96.45872497558594,
|
|
"logps/rejected": -69.0987319946289,
|
|
"loss": 0.679,
|
|
"rewards/accuracies": 0.5,
|
|
"rewards/chosen": -0.0022797822020947933,
|
|
"rewards/margins": 0.03321406990289688,
|
|
"rewards/rejected": -0.03549385070800781,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 2.735562310030395e-08,
|
|
"logits/chosen": -0.9560245275497437,
|
|
"logits/rejected": -0.9745736122131348,
|
|
"logps/chosen": -90.70116424560547,
|
|
"logps/rejected": -70.5436782836914,
|
|
"loss": 0.6589,
|
|
"rewards/accuracies": 0.637499988079071,
|
|
"rewards/chosen": 0.014189362525939941,
|
|
"rewards/margins": 0.053305577486753464,
|
|
"rewards/rejected": -0.03911621496081352,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.191489361702128e-08,
|
|
"logits/chosen": -1.006732702255249,
|
|
"logits/rejected": -1.0067856311798096,
|
|
"logps/chosen": -86.19349670410156,
|
|
"logps/rejected": -66.15892028808594,
|
|
"loss": 0.6445,
|
|
"rewards/accuracies": 0.7749999761581421,
|
|
"rewards/chosen": 0.066792331635952,
|
|
"rewards/margins": 0.12021535634994507,
|
|
"rewards/rejected": -0.05342302471399307,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.64741641337386e-08,
|
|
"logits/chosen": -1.0003823041915894,
|
|
"logits/rejected": -1.0047554969787598,
|
|
"logps/chosen": -96.32222747802734,
|
|
"logps/rejected": -68.91173553466797,
|
|
"loss": 0.6131,
|
|
"rewards/accuracies": 0.8374999761581421,
|
|
"rewards/chosen": 0.09268401563167572,
|
|
"rewards/margins": 0.19902931153774261,
|
|
"rewards/rejected": -0.1063452959060669,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 4.1033434650455923e-08,
|
|
"logits/chosen": -0.9865853190422058,
|
|
"logits/rejected": -0.9936298131942749,
|
|
"logps/chosen": -89.23389434814453,
|
|
"logps/rejected": -64.02348327636719,
|
|
"loss": 0.5757,
|
|
"rewards/accuracies": 0.925000011920929,
|
|
"rewards/chosen": 0.13552097976207733,
|
|
"rewards/margins": 0.2616090774536133,
|
|
"rewards/rejected": -0.12608811259269714,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 4.559270516717325e-08,
|
|
"logits/chosen": -0.9565617442131042,
|
|
"logits/rejected": -0.9617575407028198,
|
|
"logps/chosen": -91.22322845458984,
|
|
"logps/rejected": -68.244140625,
|
|
"loss": 0.5412,
|
|
"rewards/accuracies": 0.9375,
|
|
"rewards/chosen": 0.14585547149181366,
|
|
"rewards/margins": 0.3651939332485199,
|
|
"rewards/rejected": -0.21933846175670624,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"eval_logits/chosen": -0.9840940237045288,
|
|
"eval_logits/rejected": -0.9992539286613464,
|
|
"eval_logps/chosen": -89.1026611328125,
|
|
"eval_logps/rejected": -65.7355728149414,
|
|
"eval_loss": 0.5342935919761658,
|
|
"eval_rewards/accuracies": 0.9441340565681458,
|
|
"eval_rewards/chosen": 0.16487935185432434,
|
|
"eval_rewards/margins": 0.3618040084838867,
|
|
"eval_rewards/rejected": -0.19692467153072357,
|
|
"eval_runtime": 71.4873,
|
|
"eval_samples_per_second": 40.035,
|
|
"eval_steps_per_second": 2.504,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 5.015197568389058e-08,
|
|
"logits/chosen": -0.9964181780815125,
|
|
"logits/rejected": -1.000222086906433,
|
|
"logps/chosen": -89.18244934082031,
|
|
"logps/rejected": -67.94490814208984,
|
|
"loss": 0.5304,
|
|
"rewards/accuracies": 0.9125000238418579,
|
|
"rewards/chosen": 0.12820395827293396,
|
|
"rewards/margins": 0.37578168511390686,
|
|
"rewards/rejected": -0.2475777566432953,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 5.47112462006079e-08,
|
|
"logits/chosen": -0.9542981386184692,
|
|
"logits/rejected": -0.9645611643791199,
|
|
"logps/chosen": -85.86537170410156,
|
|
"logps/rejected": -67.51045989990234,
|
|
"loss": 0.4784,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.18605713546276093,
|
|
"rewards/margins": 0.5225586891174316,
|
|
"rewards/rejected": -0.3365015983581543,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 5.9270516717325223e-08,
|
|
"logits/chosen": -0.9908889532089233,
|
|
"logits/rejected": -0.9953571557998657,
|
|
"logps/chosen": -92.44700622558594,
|
|
"logps/rejected": -69.91798400878906,
|
|
"loss": 0.3954,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.338260680437088,
|
|
"rewards/margins": 0.743990957736969,
|
|
"rewards/rejected": -0.4057301878929138,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 6.382978723404255e-08,
|
|
"logits/chosen": -0.954239010810852,
|
|
"logits/rejected": -0.968449592590332,
|
|
"logps/chosen": -91.10675811767578,
|
|
"logps/rejected": -67.44385528564453,
|
|
"loss": 0.3584,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.36729928851127625,
|
|
"rewards/margins": 0.9051514863967896,
|
|
"rewards/rejected": -0.5378521680831909,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 6.838905775075987e-08,
|
|
"logits/chosen": -0.9643675088882446,
|
|
"logits/rejected": -0.9654477834701538,
|
|
"logps/chosen": -93.19139099121094,
|
|
"logps/rejected": -69.21319580078125,
|
|
"loss": 0.3223,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.491039901971817,
|
|
"rewards/margins": 1.1034448146820068,
|
|
"rewards/rejected": -0.6124049425125122,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 7.29483282674772e-08,
|
|
"logits/chosen": -0.9584270715713501,
|
|
"logits/rejected": -0.9729134440422058,
|
|
"logps/chosen": -93.25248718261719,
|
|
"logps/rejected": -67.92393493652344,
|
|
"loss": 0.3027,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.5195748805999756,
|
|
"rewards/margins": 1.1697877645492554,
|
|
"rewards/rejected": -0.6502128839492798,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 7.750759878419453e-08,
|
|
"logits/chosen": -0.9704807996749878,
|
|
"logits/rejected": -0.9716874361038208,
|
|
"logps/chosen": -83.62016296386719,
|
|
"logps/rejected": -66.22118377685547,
|
|
"loss": 0.3037,
|
|
"rewards/accuracies": 0.9624999761581421,
|
|
"rewards/chosen": 0.4028417468070984,
|
|
"rewards/margins": 1.2314786911010742,
|
|
"rewards/rejected": -0.8286369442939758,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 8.206686930091185e-08,
|
|
"logits/chosen": -0.9470335245132446,
|
|
"logits/rejected": -0.963607668876648,
|
|
"logps/chosen": -87.67333984375,
|
|
"logps/rejected": -67.34207916259766,
|
|
"loss": 0.2843,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.5842958688735962,
|
|
"rewards/margins": 1.32895028591156,
|
|
"rewards/rejected": -0.7446545362472534,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 8.662613981762918e-08,
|
|
"logits/chosen": -0.965084433555603,
|
|
"logits/rejected": -0.9808289408683777,
|
|
"logps/chosen": -89.90495300292969,
|
|
"logps/rejected": -65.31876373291016,
|
|
"loss": 0.2496,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.6354175806045532,
|
|
"rewards/margins": 1.4908440113067627,
|
|
"rewards/rejected": -0.8554266095161438,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 9.11854103343465e-08,
|
|
"logits/chosen": -0.9802848100662231,
|
|
"logits/rejected": -0.9860008358955383,
|
|
"logps/chosen": -83.04852294921875,
|
|
"logps/rejected": -66.60576629638672,
|
|
"loss": 0.2248,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 0.6749335527420044,
|
|
"rewards/margins": 1.5874103307724,
|
|
"rewards/rejected": -0.9124768972396851,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"eval_logits/chosen": -0.9838407635688782,
|
|
"eval_logits/rejected": -0.996880292892456,
|
|
"eval_logps/chosen": -88.08007049560547,
|
|
"eval_logps/rejected": -67.23590850830078,
|
|
"eval_loss": 0.22697897255420685,
|
|
"eval_rewards/accuracies": 0.9832402467727661,
|
|
"eval_rewards/chosen": 0.6761797666549683,
|
|
"eval_rewards/margins": 1.6232739686965942,
|
|
"eval_rewards/rejected": -0.9470942616462708,
|
|
"eval_runtime": 105.1739,
|
|
"eval_samples_per_second": 27.212,
|
|
"eval_steps_per_second": 1.702,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 9.574468085106384e-08,
|
|
"logits/chosen": -0.9623929262161255,
|
|
"logits/rejected": -0.9724255800247192,
|
|
"logps/chosen": -82.79269409179688,
|
|
"logps/rejected": -68.85140228271484,
|
|
"loss": 0.1957,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6824924349784851,
|
|
"rewards/margins": 1.8234453201293945,
|
|
"rewards/rejected": -1.1409530639648438,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 1.0030395136778115e-07,
|
|
"logits/chosen": -0.9426813125610352,
|
|
"logits/rejected": -0.95277339220047,
|
|
"logps/chosen": -90.78047943115234,
|
|
"logps/rejected": -70.69515991210938,
|
|
"loss": 0.1768,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.7748227715492249,
|
|
"rewards/margins": 2.0056302547454834,
|
|
"rewards/rejected": -1.2308075428009033,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 1.0486322188449848e-07,
|
|
"logits/chosen": -0.9746893048286438,
|
|
"logits/rejected": -0.9837571978569031,
|
|
"logps/chosen": -91.14712524414062,
|
|
"logps/rejected": -69.38472747802734,
|
|
"loss": 0.1403,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9424777030944824,
|
|
"rewards/margins": 2.5599138736724854,
|
|
"rewards/rejected": -1.6174360513687134,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 1.094224924012158e-07,
|
|
"logits/chosen": -0.9637104868888855,
|
|
"logits/rejected": -0.9687842130661011,
|
|
"logps/chosen": -88.99737548828125,
|
|
"logps/rejected": -69.65807342529297,
|
|
"loss": 0.1228,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.0617095232009888,
|
|
"rewards/margins": 2.6283512115478516,
|
|
"rewards/rejected": -1.5666416883468628,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 1.1398176291793313e-07,
|
|
"logits/chosen": -0.9593694806098938,
|
|
"logits/rejected": -0.9832932353019714,
|
|
"logps/chosen": -89.79000091552734,
|
|
"logps/rejected": -71.10289001464844,
|
|
"loss": 0.116,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.0695412158966064,
|
|
"rewards/margins": 2.6741912364959717,
|
|
"rewards/rejected": -1.6046499013900757,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 1.1854103343465045e-07,
|
|
"logits/chosen": -0.9898384809494019,
|
|
"logits/rejected": -0.9927492141723633,
|
|
"logps/chosen": -84.07673645019531,
|
|
"logps/rejected": -67.94017028808594,
|
|
"loss": 0.1081,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1412315368652344,
|
|
"rewards/margins": 2.985485315322876,
|
|
"rewards/rejected": -1.8442538976669312,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 1.2310030395136776e-07,
|
|
"logits/chosen": -0.9738420248031616,
|
|
"logits/rejected": -0.9792343974113464,
|
|
"logps/chosen": -88.48084259033203,
|
|
"logps/rejected": -75.20460510253906,
|
|
"loss": 0.1053,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.2394745349884033,
|
|
"rewards/margins": 3.0349550247192383,
|
|
"rewards/rejected": -1.7954803705215454,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 1.276595744680851e-07,
|
|
"logits/chosen": -0.973858654499054,
|
|
"logits/rejected": -0.977423369884491,
|
|
"logps/chosen": -88.2381820678711,
|
|
"logps/rejected": -69.44757080078125,
|
|
"loss": 0.0878,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0679655075073242,
|
|
"rewards/margins": 3.237596035003662,
|
|
"rewards/rejected": -2.169631004333496,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 1.3221884498480242e-07,
|
|
"logits/chosen": -0.9805269241333008,
|
|
"logits/rejected": -0.9859040975570679,
|
|
"logps/chosen": -86.40400695800781,
|
|
"logps/rejected": -73.2696533203125,
|
|
"loss": 0.0876,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1778849363327026,
|
|
"rewards/margins": 3.2878527641296387,
|
|
"rewards/rejected": -2.1099677085876465,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 1.3677811550151974e-07,
|
|
"logits/chosen": -0.9615720510482788,
|
|
"logits/rejected": -0.9615543484687805,
|
|
"logps/chosen": -85.02987670898438,
|
|
"logps/rejected": -69.7044906616211,
|
|
"loss": 0.0827,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.123947262763977,
|
|
"rewards/margins": 3.40171480178833,
|
|
"rewards/rejected": -2.2777674198150635,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"eval_logits/chosen": -0.9878929853439331,
|
|
"eval_logits/rejected": -0.9959621429443359,
|
|
"eval_logps/chosen": -86.76927947998047,
|
|
"eval_logps/rejected": -69.99213409423828,
|
|
"eval_loss": 0.08308280259370804,
|
|
"eval_rewards/accuracies": 0.9888268113136292,
|
|
"eval_rewards/chosen": 1.3315751552581787,
|
|
"eval_rewards/margins": 3.6567811965942383,
|
|
"eval_rewards/rejected": -2.3252058029174805,
|
|
"eval_runtime": 61.2711,
|
|
"eval_samples_per_second": 46.71,
|
|
"eval_steps_per_second": 2.921,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 1.4133738601823708e-07,
|
|
"logits/chosen": -0.9472341537475586,
|
|
"logits/rejected": -0.9612518548965454,
|
|
"logps/chosen": -89.58518981933594,
|
|
"logps/rejected": -70.13312530517578,
|
|
"loss": 0.0787,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.4316556453704834,
|
|
"rewards/margins": 3.8420116901397705,
|
|
"rewards/rejected": -2.410356044769287,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 1.458966565349544e-07,
|
|
"logits/chosen": -0.9657572507858276,
|
|
"logits/rejected": -0.9684290885925293,
|
|
"logps/chosen": -86.15750122070312,
|
|
"logps/rejected": -69.44698333740234,
|
|
"loss": 0.0743,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.158479928970337,
|
|
"rewards/margins": 3.4451236724853516,
|
|
"rewards/rejected": -2.2866437435150146,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 1.5045592705167174e-07,
|
|
"logits/chosen": -0.9523347020149231,
|
|
"logits/rejected": -0.9598930478096008,
|
|
"logps/chosen": -91.19930267333984,
|
|
"logps/rejected": -71.31523895263672,
|
|
"loss": 0.0718,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.529072642326355,
|
|
"rewards/margins": 3.7886452674865723,
|
|
"rewards/rejected": -2.2595720291137695,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 1.5501519756838906e-07,
|
|
"logits/chosen": -0.9740470051765442,
|
|
"logits/rejected": -0.9752241373062134,
|
|
"logps/chosen": -82.92659759521484,
|
|
"logps/rejected": -69.77666473388672,
|
|
"loss": 0.0723,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3220869302749634,
|
|
"rewards/margins": 4.073119163513184,
|
|
"rewards/rejected": -2.7510321140289307,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 1.5957446808510638e-07,
|
|
"logits/chosen": -0.979371190071106,
|
|
"logits/rejected": -0.9843432307243347,
|
|
"logps/chosen": -88.71659851074219,
|
|
"logps/rejected": -76.50203704833984,
|
|
"loss": 0.0682,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.327655553817749,
|
|
"rewards/margins": 4.103621006011963,
|
|
"rewards/rejected": -2.7759652137756348,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 1.641337386018237e-07,
|
|
"logits/chosen": -0.9631199836730957,
|
|
"logits/rejected": -0.9769188165664673,
|
|
"logps/chosen": -88.78105163574219,
|
|
"logps/rejected": -71.40655517578125,
|
|
"loss": 0.0576,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.553839087486267,
|
|
"rewards/margins": 4.10164213180542,
|
|
"rewards/rejected": -2.547802448272705,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 1.6869300911854104e-07,
|
|
"logits/chosen": -0.9622189402580261,
|
|
"logits/rejected": -0.9701143503189087,
|
|
"logps/chosen": -89.26042938232422,
|
|
"logps/rejected": -75.21412658691406,
|
|
"loss": 0.0687,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.384950041770935,
|
|
"rewards/margins": 4.296316623687744,
|
|
"rewards/rejected": -2.9113662242889404,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 1.7325227963525835e-07,
|
|
"logits/chosen": -0.9843562245368958,
|
|
"logits/rejected": -0.9918986558914185,
|
|
"logps/chosen": -86.35823059082031,
|
|
"logps/rejected": -72.2359848022461,
|
|
"loss": 0.0488,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5267091989517212,
|
|
"rewards/margins": 4.862682819366455,
|
|
"rewards/rejected": -3.3359732627868652,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 1.7781155015197567e-07,
|
|
"logits/chosen": -0.9975897073745728,
|
|
"logits/rejected": -0.9955156445503235,
|
|
"logps/chosen": -91.64771270751953,
|
|
"logps/rejected": -74.9527816772461,
|
|
"loss": 0.0481,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.4639146327972412,
|
|
"rewards/margins": 5.018456935882568,
|
|
"rewards/rejected": -3.5545425415039062,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 1.82370820668693e-07,
|
|
"logits/chosen": -0.9872757792472839,
|
|
"logits/rejected": -0.9929397702217102,
|
|
"logps/chosen": -86.04756164550781,
|
|
"logps/rejected": -75.50514221191406,
|
|
"loss": 0.0351,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6687867641448975,
|
|
"rewards/margins": 5.520654201507568,
|
|
"rewards/rejected": -3.851867198944092,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"eval_logits/chosen": -1.001560091972351,
|
|
"eval_logits/rejected": -0.9991575479507446,
|
|
"eval_logps/chosen": -86.00261688232422,
|
|
"eval_logps/rejected": -72.45541381835938,
|
|
"eval_loss": 0.04544173553586006,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.7149040699005127,
|
|
"eval_rewards/margins": 5.271754264831543,
|
|
"eval_rewards/rejected": -3.556849956512451,
|
|
"eval_runtime": 62.4842,
|
|
"eval_samples_per_second": 45.804,
|
|
"eval_steps_per_second": 2.865,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 1.869300911854103e-07,
|
|
"logits/chosen": -1.0112515687942505,
|
|
"logits/rejected": -1.0055806636810303,
|
|
"logps/chosen": -86.33168029785156,
|
|
"logps/rejected": -79.20642852783203,
|
|
"loss": 0.0381,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5995696783065796,
|
|
"rewards/margins": 5.682262420654297,
|
|
"rewards/rejected": -4.082693099975586,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 1.9148936170212767e-07,
|
|
"logits/chosen": -0.9795786142349243,
|
|
"logits/rejected": -0.9767486453056335,
|
|
"logps/chosen": -89.1741714477539,
|
|
"logps/rejected": -75.95233917236328,
|
|
"loss": 0.0414,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.1586321592330933,
|
|
"rewards/margins": 5.276698112487793,
|
|
"rewards/rejected": -4.118066310882568,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 1.96048632218845e-07,
|
|
"logits/chosen": -1.0036985874176025,
|
|
"logits/rejected": -0.9971052408218384,
|
|
"logps/chosen": -86.45294189453125,
|
|
"logps/rejected": -79.10576629638672,
|
|
"loss": 0.0317,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.8366873264312744,
|
|
"rewards/margins": 6.2948222160339355,
|
|
"rewards/rejected": -4.458134651184082,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 2.006079027355623e-07,
|
|
"logits/chosen": -1.0170124769210815,
|
|
"logits/rejected": -1.0096765756607056,
|
|
"logps/chosen": -90.57597351074219,
|
|
"logps/rejected": -72.65457916259766,
|
|
"loss": 0.0339,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.073103427886963,
|
|
"rewards/margins": 6.505836486816406,
|
|
"rewards/rejected": -4.432732582092285,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 2.0516717325227962e-07,
|
|
"logits/chosen": -1.0112982988357544,
|
|
"logits/rejected": -1.0087357759475708,
|
|
"logps/chosen": -87.71390533447266,
|
|
"logps/rejected": -78.8847427368164,
|
|
"loss": 0.037,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.966575264930725,
|
|
"rewards/margins": 7.236502170562744,
|
|
"rewards/rejected": -5.26992654800415,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 2.0972644376899697e-07,
|
|
"logits/chosen": -0.9955867528915405,
|
|
"logits/rejected": -0.9789519309997559,
|
|
"logps/chosen": -93.02728271484375,
|
|
"logps/rejected": -79.93946838378906,
|
|
"loss": 0.0318,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.8666505813598633,
|
|
"rewards/margins": 7.276576042175293,
|
|
"rewards/rejected": -5.409926414489746,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 2.1428571428571428e-07,
|
|
"logits/chosen": -1.0107711553573608,
|
|
"logits/rejected": -1.015987753868103,
|
|
"logps/chosen": -91.70677185058594,
|
|
"logps/rejected": -78.36326599121094,
|
|
"loss": 0.0279,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.9241142272949219,
|
|
"rewards/margins": 7.251917839050293,
|
|
"rewards/rejected": -5.327803611755371,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 2.188449848024316e-07,
|
|
"logits/chosen": -1.0210905075073242,
|
|
"logits/rejected": -1.0056220293045044,
|
|
"logps/chosen": -90.60785675048828,
|
|
"logps/rejected": -77.8875961303711,
|
|
"loss": 0.0252,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.8818817138671875,
|
|
"rewards/margins": 7.938240051269531,
|
|
"rewards/rejected": -6.056358814239502,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 2.2340425531914892e-07,
|
|
"logits/chosen": -1.0591073036193848,
|
|
"logits/rejected": -1.0334078073501587,
|
|
"logps/chosen": -86.05088806152344,
|
|
"logps/rejected": -80.09144592285156,
|
|
"loss": 0.018,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.9079704284667969,
|
|
"rewards/margins": 7.765076637268066,
|
|
"rewards/rejected": -5.857105255126953,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 2.2796352583586626e-07,
|
|
"logits/chosen": -1.044002652168274,
|
|
"logits/rejected": -1.0229895114898682,
|
|
"logps/chosen": -81.89948272705078,
|
|
"logps/rejected": -77.97576141357422,
|
|
"loss": 0.0217,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.5751844644546509,
|
|
"rewards/margins": 7.957855224609375,
|
|
"rewards/rejected": -6.382671356201172,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"eval_logits/chosen": -1.0691347122192383,
|
|
"eval_logits/rejected": -1.0533946752548218,
|
|
"eval_logps/chosen": -85.60762023925781,
|
|
"eval_logps/rejected": -77.9649887084961,
|
|
"eval_loss": 0.021840358152985573,
|
|
"eval_rewards/accuracies": 0.9888268113136292,
|
|
"eval_rewards/chosen": 1.912401795387268,
|
|
"eval_rewards/margins": 8.22403335571289,
|
|
"eval_rewards/rejected": -6.3116326332092285,
|
|
"eval_runtime": 60.6624,
|
|
"eval_samples_per_second": 47.179,
|
|
"eval_steps_per_second": 2.951,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 2.3252279635258358e-07,
|
|
"logits/chosen": -1.0656869411468506,
|
|
"logits/rejected": -1.0443495512008667,
|
|
"logps/chosen": -80.74382781982422,
|
|
"logps/rejected": -80.70405578613281,
|
|
"loss": 0.0211,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.984548807144165,
|
|
"rewards/margins": 8.708304405212402,
|
|
"rewards/rejected": -6.7237548828125,
|
|
"step": 510
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 2.370820668693009e-07,
|
|
"logits/chosen": -1.0407330989837646,
|
|
"logits/rejected": -1.019323706626892,
|
|
"logps/chosen": -94.31208801269531,
|
|
"logps/rejected": -84.92703247070312,
|
|
"loss": 0.0192,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.9942207336425781,
|
|
"rewards/margins": 8.950647354125977,
|
|
"rewards/rejected": -6.956425666809082,
|
|
"step": 520
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 2.4164133738601824e-07,
|
|
"logits/chosen": -1.0353573560714722,
|
|
"logits/rejected": -1.0317944288253784,
|
|
"logps/chosen": -84.43415832519531,
|
|
"logps/rejected": -80.83734130859375,
|
|
"loss": 0.0188,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.38482928276062,
|
|
"rewards/margins": 9.657920837402344,
|
|
"rewards/rejected": -7.273091793060303,
|
|
"step": 530
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 2.4620060790273553e-07,
|
|
"logits/chosen": -1.0673249959945679,
|
|
"logits/rejected": -1.0390758514404297,
|
|
"logps/chosen": -86.52366638183594,
|
|
"logps/rejected": -82.38668823242188,
|
|
"loss": 0.0125,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5419979095458984,
|
|
"rewards/margins": 9.403617858886719,
|
|
"rewards/rejected": -7.861618995666504,
|
|
"step": 540
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 2.5075987841945287e-07,
|
|
"logits/chosen": -1.0771315097808838,
|
|
"logits/rejected": -1.0471489429473877,
|
|
"logps/chosen": -94.00312805175781,
|
|
"logps/rejected": -81.71351623535156,
|
|
"loss": 0.0146,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.8300154209136963,
|
|
"rewards/margins": 9.741002082824707,
|
|
"rewards/rejected": -7.910986423492432,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 2.553191489361702e-07,
|
|
"logits/chosen": -1.0417684316635132,
|
|
"logits/rejected": -1.0242894887924194,
|
|
"logps/chosen": -85.45355987548828,
|
|
"logps/rejected": -84.82173156738281,
|
|
"loss": 0.0132,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.176011323928833,
|
|
"rewards/margins": 10.29346752166748,
|
|
"rewards/rejected": -8.117456436157227,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 2.598784194528875e-07,
|
|
"logits/chosen": -1.0521628856658936,
|
|
"logits/rejected": -1.0290045738220215,
|
|
"logps/chosen": -86.50548553466797,
|
|
"logps/rejected": -84.81304931640625,
|
|
"loss": 0.0163,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.396401286125183,
|
|
"rewards/margins": 9.808305740356445,
|
|
"rewards/rejected": -8.411903381347656,
|
|
"step": 570
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 2.6443768996960485e-07,
|
|
"logits/chosen": -1.0446751117706299,
|
|
"logits/rejected": -1.0253543853759766,
|
|
"logps/chosen": -83.4941635131836,
|
|
"logps/rejected": -81.4127197265625,
|
|
"loss": 0.0107,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.4424480199813843,
|
|
"rewards/margins": 10.39416217803955,
|
|
"rewards/rejected": -8.951713562011719,
|
|
"step": 580
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 2.689969604863222e-07,
|
|
"logits/chosen": -1.0527292490005493,
|
|
"logits/rejected": -1.028416633605957,
|
|
"logps/chosen": -89.69185638427734,
|
|
"logps/rejected": -84.93494415283203,
|
|
"loss": 0.016,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.6588973999023438,
|
|
"rewards/margins": 10.561676025390625,
|
|
"rewards/rejected": -8.902778625488281,
|
|
"step": 590
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 2.735562310030395e-07,
|
|
"logits/chosen": -1.0560492277145386,
|
|
"logits/rejected": -1.031136155128479,
|
|
"logps/chosen": -88.6587142944336,
|
|
"logps/rejected": -89.13833618164062,
|
|
"loss": 0.0112,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.94255793094635,
|
|
"rewards/margins": 11.583595275878906,
|
|
"rewards/rejected": -9.641037940979004,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"eval_logits/chosen": -1.0850204229354858,
|
|
"eval_logits/rejected": -1.0588685274124146,
|
|
"eval_logps/chosen": -85.50906372070312,
|
|
"eval_logps/rejected": -83.90418243408203,
|
|
"eval_loss": 0.013939271681010723,
|
|
"eval_rewards/accuracies": 0.9888268113136292,
|
|
"eval_rewards/chosen": 1.961681842803955,
|
|
"eval_rewards/margins": 11.242914199829102,
|
|
"eval_rewards/rejected": -9.281231880187988,
|
|
"eval_runtime": 67.4874,
|
|
"eval_samples_per_second": 42.408,
|
|
"eval_steps_per_second": 2.652,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 2.781155015197568e-07,
|
|
"logits/chosen": -1.0646154880523682,
|
|
"logits/rejected": -1.0403302907943726,
|
|
"logps/chosen": -87.16560363769531,
|
|
"logps/rejected": -84.03047943115234,
|
|
"loss": 0.0133,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.1723438501358032,
|
|
"rewards/margins": 10.494967460632324,
|
|
"rewards/rejected": -9.322624206542969,
|
|
"step": 610
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 2.8267477203647417e-07,
|
|
"logits/chosen": -1.040969729423523,
|
|
"logits/rejected": -1.025119423866272,
|
|
"logps/chosen": -86.0352783203125,
|
|
"logps/rejected": -84.01143646240234,
|
|
"loss": 0.0171,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.131622314453125,
|
|
"rewards/margins": 11.987198829650879,
|
|
"rewards/rejected": -9.855576515197754,
|
|
"step": 620
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 2.872340425531915e-07,
|
|
"logits/chosen": -1.0877559185028076,
|
|
"logits/rejected": -1.057796835899353,
|
|
"logps/chosen": -88.63595581054688,
|
|
"logps/rejected": -93.89146423339844,
|
|
"loss": 0.0093,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.9702014923095703,
|
|
"rewards/margins": 12.47899055480957,
|
|
"rewards/rejected": -10.508790016174316,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 2.917933130699088e-07,
|
|
"logits/chosen": -1.0618011951446533,
|
|
"logits/rejected": -1.0364282131195068,
|
|
"logps/chosen": -83.58055114746094,
|
|
"logps/rejected": -87.19285583496094,
|
|
"loss": 0.0172,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.3319017887115479,
|
|
"rewards/margins": 12.28033447265625,
|
|
"rewards/rejected": -10.948432922363281,
|
|
"step": 640
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 2.9635258358662614e-07,
|
|
"logits/chosen": -1.0487759113311768,
|
|
"logits/rejected": -1.0237376689910889,
|
|
"logps/chosen": -97.61241149902344,
|
|
"logps/rejected": -89.84696197509766,
|
|
"loss": 0.0056,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.0295894145965576,
|
|
"rewards/margins": 13.6948823928833,
|
|
"rewards/rejected": -11.665290832519531,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 2.9989856297548603e-07,
|
|
"logits/chosen": -1.0819432735443115,
|
|
"logits/rejected": -1.0479309558868408,
|
|
"logps/chosen": -88.46907043457031,
|
|
"logps/rejected": -87.63093566894531,
|
|
"loss": 0.0062,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6469261646270752,
|
|
"rewards/margins": 12.705039978027344,
|
|
"rewards/rejected": -11.058113098144531,
|
|
"step": 660
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 2.9939137785291633e-07,
|
|
"logits/chosen": -1.0461461544036865,
|
|
"logits/rejected": -1.0262444019317627,
|
|
"logps/chosen": -88.99845886230469,
|
|
"logps/rejected": -88.98668670654297,
|
|
"loss": 0.0069,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7188533544540405,
|
|
"rewards/margins": 12.911355972290039,
|
|
"rewards/rejected": -11.192502975463867,
|
|
"step": 670
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 2.9888419273034654e-07,
|
|
"logits/chosen": -1.0983555316925049,
|
|
"logits/rejected": -1.058870792388916,
|
|
"logps/chosen": -92.72132873535156,
|
|
"logps/rejected": -91.1702880859375,
|
|
"loss": 0.0139,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.1177186965942383,
|
|
"rewards/margins": 12.790249824523926,
|
|
"rewards/rejected": -10.672532081604004,
|
|
"step": 680
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 2.9837700760777684e-07,
|
|
"logits/chosen": -1.0601097345352173,
|
|
"logits/rejected": -1.0348825454711914,
|
|
"logps/chosen": -86.94996643066406,
|
|
"logps/rejected": -90.38238525390625,
|
|
"loss": 0.0086,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.6195377111434937,
|
|
"rewards/margins": 13.726669311523438,
|
|
"rewards/rejected": -12.107131958007812,
|
|
"step": 690
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 2.978698224852071e-07,
|
|
"logits/chosen": -1.0934697389602661,
|
|
"logits/rejected": -1.0594290494918823,
|
|
"logps/chosen": -89.61146545410156,
|
|
"logps/rejected": -92.48619842529297,
|
|
"loss": 0.0048,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.0675747394561768,
|
|
"rewards/margins": 13.449430465698242,
|
|
"rewards/rejected": -11.381856918334961,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"eval_logits/chosen": -1.1140062808990479,
|
|
"eval_logits/rejected": -1.0765314102172852,
|
|
"eval_logps/chosen": -85.63339233398438,
|
|
"eval_logps/rejected": -89.27959442138672,
|
|
"eval_loss": 0.010984507389366627,
|
|
"eval_rewards/accuracies": 0.9888268113136292,
|
|
"eval_rewards/chosen": 1.8995180130004883,
|
|
"eval_rewards/margins": 13.868452072143555,
|
|
"eval_rewards/rejected": -11.968934059143066,
|
|
"eval_runtime": 63.0243,
|
|
"eval_samples_per_second": 45.411,
|
|
"eval_steps_per_second": 2.84,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 2.9736263736263735e-07,
|
|
"logits/chosen": -1.1225736141204834,
|
|
"logits/rejected": -1.079655408859253,
|
|
"logps/chosen": -84.65995788574219,
|
|
"logps/rejected": -93.70130920410156,
|
|
"loss": 0.0063,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7059738636016846,
|
|
"rewards/margins": 13.980459213256836,
|
|
"rewards/rejected": -12.27448558807373,
|
|
"step": 710
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 2.968554522400676e-07,
|
|
"logits/chosen": -1.0669975280761719,
|
|
"logits/rejected": -1.0285903215408325,
|
|
"logps/chosen": -89.87354278564453,
|
|
"logps/rejected": -91.52484893798828,
|
|
"loss": 0.0063,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.0469908714294434,
|
|
"rewards/margins": 14.076245307922363,
|
|
"rewards/rejected": -12.029253005981445,
|
|
"step": 720
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 2.9634826711749786e-07,
|
|
"logits/chosen": -1.0944750308990479,
|
|
"logits/rejected": -1.0600286722183228,
|
|
"logps/chosen": -90.2714614868164,
|
|
"logps/rejected": -86.80066680908203,
|
|
"loss": 0.003,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.6248250007629395,
|
|
"rewards/margins": 14.128560066223145,
|
|
"rewards/rejected": -11.503734588623047,
|
|
"step": 730
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 2.958410819949281e-07,
|
|
"logits/chosen": -1.0793449878692627,
|
|
"logits/rejected": -1.0554149150848389,
|
|
"logps/chosen": -78.48091888427734,
|
|
"logps/rejected": -89.2685317993164,
|
|
"loss": 0.0078,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.959774374961853,
|
|
"rewards/margins": 14.275179862976074,
|
|
"rewards/rejected": -12.315404891967773,
|
|
"step": 740
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 2.953338968723584e-07,
|
|
"logits/chosen": -1.0978403091430664,
|
|
"logits/rejected": -1.0679986476898193,
|
|
"logps/chosen": -79.7038345336914,
|
|
"logps/rejected": -89.61735534667969,
|
|
"loss": 0.0117,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.5487966537475586,
|
|
"rewards/margins": 15.569620132446289,
|
|
"rewards/rejected": -13.020822525024414,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 2.948267117497887e-07,
|
|
"logits/chosen": -1.1022833585739136,
|
|
"logits/rejected": -1.0697553157806396,
|
|
"logps/chosen": -87.8471450805664,
|
|
"logps/rejected": -95.08956146240234,
|
|
"loss": 0.0078,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.898350715637207,
|
|
"rewards/margins": 14.861310005187988,
|
|
"rewards/rejected": -11.962959289550781,
|
|
"step": 760
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 2.9431952662721893e-07,
|
|
"logits/chosen": -1.0957014560699463,
|
|
"logits/rejected": -1.0692951679229736,
|
|
"logps/chosen": -85.73948669433594,
|
|
"logps/rejected": -89.58802795410156,
|
|
"loss": 0.0072,
|
|
"rewards/accuracies": 0.9624999761581421,
|
|
"rewards/chosen": 1.3441336154937744,
|
|
"rewards/margins": 13.35987377166748,
|
|
"rewards/rejected": -12.015739440917969,
|
|
"step": 770
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 2.938123415046492e-07,
|
|
"logits/chosen": -1.0931169986724854,
|
|
"logits/rejected": -1.0750610828399658,
|
|
"logps/chosen": -83.94388580322266,
|
|
"logps/rejected": -90.04351043701172,
|
|
"loss": 0.0064,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7953174114227295,
|
|
"rewards/margins": 14.375581741333008,
|
|
"rewards/rejected": -12.5802640914917,
|
|
"step": 780
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 2.9330515638207944e-07,
|
|
"logits/chosen": -1.1000292301177979,
|
|
"logits/rejected": -1.0617127418518066,
|
|
"logps/chosen": -85.10647583007812,
|
|
"logps/rejected": -95.74652099609375,
|
|
"loss": 0.0055,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.9242041110992432,
|
|
"rewards/margins": 15.295672416687012,
|
|
"rewards/rejected": -13.371467590332031,
|
|
"step": 790
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 2.927979712595097e-07,
|
|
"logits/chosen": -1.074942708015442,
|
|
"logits/rejected": -1.0459177494049072,
|
|
"logps/chosen": -87.57856750488281,
|
|
"logps/rejected": -92.28514862060547,
|
|
"loss": 0.0085,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.659388780593872,
|
|
"rewards/margins": 13.915122985839844,
|
|
"rewards/rejected": -12.25573444366455,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"eval_logits/chosen": -1.1495593786239624,
|
|
"eval_logits/rejected": -1.104751467704773,
|
|
"eval_logps/chosen": -85.99646759033203,
|
|
"eval_logps/rejected": -93.16914367675781,
|
|
"eval_loss": 0.01038471981883049,
|
|
"eval_rewards/accuracies": 0.9888268113136292,
|
|
"eval_rewards/chosen": 1.7179815769195557,
|
|
"eval_rewards/margins": 15.631699562072754,
|
|
"eval_rewards/rejected": -13.913717269897461,
|
|
"eval_runtime": 65.9739,
|
|
"eval_samples_per_second": 43.381,
|
|
"eval_steps_per_second": 2.713,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 2.9229078613694e-07,
|
|
"logits/chosen": -1.1248949766159058,
|
|
"logits/rejected": -1.086717128753662,
|
|
"logps/chosen": -86.31236267089844,
|
|
"logps/rejected": -93.25736236572266,
|
|
"loss": 0.0126,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.9837150573730469,
|
|
"rewards/margins": 15.558454513549805,
|
|
"rewards/rejected": -13.574739456176758,
|
|
"step": 810
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 2.917836010143702e-07,
|
|
"logits/chosen": -1.0978708267211914,
|
|
"logits/rejected": -1.057366132736206,
|
|
"logps/chosen": -81.67508697509766,
|
|
"logps/rejected": -91.41726684570312,
|
|
"loss": 0.0056,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.00944185256958,
|
|
"rewards/margins": 15.302815437316895,
|
|
"rewards/rejected": -13.293373107910156,
|
|
"step": 820
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 2.912764158918005e-07,
|
|
"logits/chosen": -1.1044955253601074,
|
|
"logits/rejected": -1.0758087635040283,
|
|
"logps/chosen": -82.3414306640625,
|
|
"logps/rejected": -97.22112274169922,
|
|
"loss": 0.0106,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.6446453332901,
|
|
"rewards/margins": 15.301877975463867,
|
|
"rewards/rejected": -13.657232284545898,
|
|
"step": 830
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 2.9076923076923076e-07,
|
|
"logits/chosen": -1.124459981918335,
|
|
"logits/rejected": -1.0775994062423706,
|
|
"logps/chosen": -90.77351379394531,
|
|
"logps/rejected": -92.74385833740234,
|
|
"loss": 0.0042,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.0721054077148438,
|
|
"rewards/margins": 16.051000595092773,
|
|
"rewards/rejected": -13.97889518737793,
|
|
"step": 840
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 2.90262045646661e-07,
|
|
"logits/chosen": -1.1489288806915283,
|
|
"logits/rejected": -1.1091543436050415,
|
|
"logps/chosen": -83.20817565917969,
|
|
"logps/rejected": -99.59390258789062,
|
|
"loss": 0.0079,
|
|
"rewards/accuracies": 0.9624999761581421,
|
|
"rewards/chosen": 1.3499611616134644,
|
|
"rewards/margins": 16.251693725585938,
|
|
"rewards/rejected": -14.9017333984375,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 2.8975486052409127e-07,
|
|
"logits/chosen": -1.1202361583709717,
|
|
"logits/rejected": -1.0809228420257568,
|
|
"logps/chosen": -93.94721984863281,
|
|
"logps/rejected": -97.80915069580078,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6343969106674194,
|
|
"rewards/margins": 16.120519638061523,
|
|
"rewards/rejected": -14.486124038696289,
|
|
"step": 860
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 2.892476754015215e-07,
|
|
"logits/chosen": -1.1433136463165283,
|
|
"logits/rejected": -1.0998663902282715,
|
|
"logps/chosen": -88.42134094238281,
|
|
"logps/rejected": -97.21299743652344,
|
|
"loss": 0.009,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.4050545692443848,
|
|
"rewards/margins": 17.111103057861328,
|
|
"rewards/rejected": -14.706045150756836,
|
|
"step": 870
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 2.887404902789518e-07,
|
|
"logits/chosen": -1.1467622518539429,
|
|
"logits/rejected": -1.1034185886383057,
|
|
"logps/chosen": -93.99958801269531,
|
|
"logps/rejected": -99.74283599853516,
|
|
"loss": 0.0038,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.3127968311309814,
|
|
"rewards/margins": 16.9749698638916,
|
|
"rewards/rejected": -14.6621732711792,
|
|
"step": 880
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 2.882333051563821e-07,
|
|
"logits/chosen": -1.1312350034713745,
|
|
"logits/rejected": -1.0877126455307007,
|
|
"logps/chosen": -85.80730438232422,
|
|
"logps/rejected": -98.7739028930664,
|
|
"loss": 0.0046,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6915788650512695,
|
|
"rewards/margins": 17.439577102661133,
|
|
"rewards/rejected": -15.747998237609863,
|
|
"step": 890
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 2.8772612003381234e-07,
|
|
"logits/chosen": -1.1451927423477173,
|
|
"logits/rejected": -1.1060543060302734,
|
|
"logps/chosen": -86.88992309570312,
|
|
"logps/rejected": -99.46288299560547,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.0115227699279785,
|
|
"rewards/margins": 17.152305603027344,
|
|
"rewards/rejected": -15.140782356262207,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"eval_logits/chosen": -1.1781193017959595,
|
|
"eval_logits/rejected": -1.1281659603118896,
|
|
"eval_logps/chosen": -86.0967788696289,
|
|
"eval_logps/rejected": -96.05574798583984,
|
|
"eval_loss": 0.009535559453070164,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.6678271293640137,
|
|
"eval_rewards/margins": 17.024843215942383,
|
|
"eval_rewards/rejected": -15.357016563415527,
|
|
"eval_runtime": 61.9434,
|
|
"eval_samples_per_second": 46.203,
|
|
"eval_steps_per_second": 2.89,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 2.872189349112426e-07,
|
|
"logits/chosen": -1.138933539390564,
|
|
"logits/rejected": -1.0921571254730225,
|
|
"logps/chosen": -89.17816162109375,
|
|
"logps/rejected": -98.90169525146484,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6821825504302979,
|
|
"rewards/margins": 17.2696533203125,
|
|
"rewards/rejected": -15.587472915649414,
|
|
"step": 910
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 2.8671174978867285e-07,
|
|
"logits/chosen": -1.132676362991333,
|
|
"logits/rejected": -1.0800034999847412,
|
|
"logps/chosen": -86.4345932006836,
|
|
"logps/rejected": -97.37138366699219,
|
|
"loss": 0.0076,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.264104127883911,
|
|
"rewards/margins": 17.31032371520996,
|
|
"rewards/rejected": -15.046220779418945,
|
|
"step": 920
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 2.862045646661031e-07,
|
|
"logits/chosen": -1.1458117961883545,
|
|
"logits/rejected": -1.0923669338226318,
|
|
"logps/chosen": -92.89430236816406,
|
|
"logps/rejected": -95.99363708496094,
|
|
"loss": 0.0073,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.16265869140625,
|
|
"rewards/margins": 18.03095054626465,
|
|
"rewards/rejected": -15.868295669555664,
|
|
"step": 930
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 2.8569737954353336e-07,
|
|
"logits/chosen": -1.1349055767059326,
|
|
"logits/rejected": -1.083634614944458,
|
|
"logps/chosen": -93.05912780761719,
|
|
"logps/rejected": -103.2133560180664,
|
|
"loss": 0.0057,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1463124752044678,
|
|
"rewards/margins": 17.069040298461914,
|
|
"rewards/rejected": -15.922727584838867,
|
|
"step": 940
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 2.8519019442096367e-07,
|
|
"logits/chosen": -1.14430832862854,
|
|
"logits/rejected": -1.103515386581421,
|
|
"logps/chosen": -86.96769714355469,
|
|
"logps/rejected": -103.5058364868164,
|
|
"loss": 0.0103,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.6363202333450317,
|
|
"rewards/margins": 18.16317367553711,
|
|
"rewards/rejected": -17.52685546875,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 2.846830092983939e-07,
|
|
"logits/chosen": -1.1029218435287476,
|
|
"logits/rejected": -1.0647412538528442,
|
|
"logps/chosen": -89.83350372314453,
|
|
"logps/rejected": -97.29312133789062,
|
|
"loss": 0.0056,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2907755374908447,
|
|
"rewards/margins": 16.094762802124023,
|
|
"rewards/rejected": -14.803988456726074,
|
|
"step": 960
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 2.841758241758242e-07,
|
|
"logits/chosen": -1.1452381610870361,
|
|
"logits/rejected": -1.1065289974212646,
|
|
"logps/chosen": -86.73941802978516,
|
|
"logps/rejected": -92.15290069580078,
|
|
"loss": 0.0094,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6819941997528076,
|
|
"rewards/margins": 15.786001205444336,
|
|
"rewards/rejected": -14.10400676727295,
|
|
"step": 970
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 2.8366863905325443e-07,
|
|
"logits/chosen": -1.1149189472198486,
|
|
"logits/rejected": -1.0778940916061401,
|
|
"logps/chosen": -87.29264068603516,
|
|
"logps/rejected": -98.55472564697266,
|
|
"loss": 0.0038,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.247238278388977,
|
|
"rewards/margins": 16.33363914489746,
|
|
"rewards/rejected": -15.086400032043457,
|
|
"step": 980
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 2.831614539306847e-07,
|
|
"logits/chosen": -1.104259967803955,
|
|
"logits/rejected": -1.0659714937210083,
|
|
"logps/chosen": -83.01313018798828,
|
|
"logps/rejected": -93.28031921386719,
|
|
"loss": 0.0114,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.3180642127990723,
|
|
"rewards/margins": 17.55698585510254,
|
|
"rewards/rejected": -15.238920211791992,
|
|
"step": 990
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 2.8265426880811494e-07,
|
|
"logits/chosen": -1.1296952962875366,
|
|
"logits/rejected": -1.080012559890747,
|
|
"logps/chosen": -88.91751098632812,
|
|
"logps/rejected": -97.22592163085938,
|
|
"loss": 0.0053,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.736224889755249,
|
|
"rewards/margins": 17.38802719116211,
|
|
"rewards/rejected": -15.651802062988281,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"eval_logits/chosen": -1.1499282121658325,
|
|
"eval_logits/rejected": -1.1065919399261475,
|
|
"eval_logps/chosen": -85.68370056152344,
|
|
"eval_logps/rejected": -95.3034439086914,
|
|
"eval_loss": 0.008165295235812664,
|
|
"eval_rewards/accuracies": 0.9888268113136292,
|
|
"eval_rewards/chosen": 1.8743646144866943,
|
|
"eval_rewards/margins": 16.85523223876953,
|
|
"eval_rewards/rejected": -14.980865478515625,
|
|
"eval_runtime": 64.5918,
|
|
"eval_samples_per_second": 44.309,
|
|
"eval_steps_per_second": 2.771,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 2.821470836855452e-07,
|
|
"logits/chosen": -1.105207085609436,
|
|
"logits/rejected": -1.0650384426116943,
|
|
"logps/chosen": -86.5456314086914,
|
|
"logps/rejected": -99.33902740478516,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.171086072921753,
|
|
"rewards/margins": 16.355039596557617,
|
|
"rewards/rejected": -15.183954238891602,
|
|
"step": 1010
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 2.8163989856297545e-07,
|
|
"logits/chosen": -1.093670129776001,
|
|
"logits/rejected": -1.0636560916900635,
|
|
"logps/chosen": -84.88511657714844,
|
|
"logps/rejected": -100.3253173828125,
|
|
"loss": 0.0056,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.8782209157943726,
|
|
"rewards/margins": 17.67445945739746,
|
|
"rewards/rejected": -15.796239852905273,
|
|
"step": 1020
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 2.8113271344040575e-07,
|
|
"logits/chosen": -1.1384307146072388,
|
|
"logits/rejected": -1.0990992784500122,
|
|
"logps/chosen": -82.6616439819336,
|
|
"logps/rejected": -98.50576782226562,
|
|
"loss": 0.0067,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.3514288663864136,
|
|
"rewards/margins": 17.497568130493164,
|
|
"rewards/rejected": -16.14613914489746,
|
|
"step": 1030
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 2.80625528317836e-07,
|
|
"logits/chosen": -1.099442481994629,
|
|
"logits/rejected": -1.072345495223999,
|
|
"logps/chosen": -94.7537612915039,
|
|
"logps/rejected": -99.29774475097656,
|
|
"loss": 0.0079,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.425427198410034,
|
|
"rewards/margins": 17.845937728881836,
|
|
"rewards/rejected": -15.420511245727539,
|
|
"step": 1040
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 2.8011834319526626e-07,
|
|
"logits/chosen": -1.1023457050323486,
|
|
"logits/rejected": -1.0619539022445679,
|
|
"logps/chosen": -89.6052474975586,
|
|
"logps/rejected": -101.50379943847656,
|
|
"loss": 0.0051,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.8860256671905518,
|
|
"rewards/margins": 18.8059024810791,
|
|
"rewards/rejected": -16.919872283935547,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 2.796111580726965e-07,
|
|
"logits/chosen": -1.1182763576507568,
|
|
"logits/rejected": -1.083601474761963,
|
|
"logps/chosen": -86.69522857666016,
|
|
"logps/rejected": -98.14034271240234,
|
|
"loss": 0.0078,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.7847849130630493,
|
|
"rewards/margins": 16.757022857666016,
|
|
"rewards/rejected": -15.972234725952148,
|
|
"step": 1060
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 2.7910397295012677e-07,
|
|
"logits/chosen": -1.0921047925949097,
|
|
"logits/rejected": -1.058253526687622,
|
|
"logps/chosen": -88.3211898803711,
|
|
"logps/rejected": -100.35057067871094,
|
|
"loss": 0.0091,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.095982074737549,
|
|
"rewards/margins": 18.4968318939209,
|
|
"rewards/rejected": -16.40085220336914,
|
|
"step": 1070
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 2.78596787827557e-07,
|
|
"logits/chosen": -1.1086832284927368,
|
|
"logits/rejected": -1.074639916419983,
|
|
"logps/chosen": -88.3951416015625,
|
|
"logps/rejected": -101.92322540283203,
|
|
"loss": 0.0071,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.5710289478302,
|
|
"rewards/margins": 18.696849822998047,
|
|
"rewards/rejected": -16.12582015991211,
|
|
"step": 1080
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 2.7808960270498733e-07,
|
|
"logits/chosen": -1.1255228519439697,
|
|
"logits/rejected": -1.0840017795562744,
|
|
"logps/chosen": -84.0284194946289,
|
|
"logps/rejected": -103.49464416503906,
|
|
"loss": 0.0048,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7648484706878662,
|
|
"rewards/margins": 18.157766342163086,
|
|
"rewards/rejected": -16.39291763305664,
|
|
"step": 1090
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 2.775824175824176e-07,
|
|
"logits/chosen": -1.1340689659118652,
|
|
"logits/rejected": -1.095685362815857,
|
|
"logps/chosen": -86.91542053222656,
|
|
"logps/rejected": -100.10322570800781,
|
|
"loss": 0.0105,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.1016123294830322,
|
|
"rewards/margins": 17.42715835571289,
|
|
"rewards/rejected": -16.325546264648438,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"eval_logits/chosen": -1.1586607694625854,
|
|
"eval_logits/rejected": -1.1157867908477783,
|
|
"eval_logps/chosen": -85.93265533447266,
|
|
"eval_logps/rejected": -96.73734283447266,
|
|
"eval_loss": 0.007867630571126938,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.7498830556869507,
|
|
"eval_rewards/margins": 17.447694778442383,
|
|
"eval_rewards/rejected": -15.6978120803833,
|
|
"eval_runtime": 64.3858,
|
|
"eval_samples_per_second": 44.451,
|
|
"eval_steps_per_second": 2.78,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 2.7707523245984784e-07,
|
|
"logits/chosen": -1.1197665929794312,
|
|
"logits/rejected": -1.078808307647705,
|
|
"logps/chosen": -92.86619567871094,
|
|
"logps/rejected": -102.52835845947266,
|
|
"loss": 0.0051,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.6940488815307617,
|
|
"rewards/margins": 17.81356430053711,
|
|
"rewards/rejected": -16.11951446533203,
|
|
"step": 1110
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 2.765680473372781e-07,
|
|
"logits/chosen": -1.1259820461273193,
|
|
"logits/rejected": -1.0869123935699463,
|
|
"logps/chosen": -93.55479431152344,
|
|
"logps/rejected": -95.98357391357422,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.1045427322387695,
|
|
"rewards/margins": 17.43987464904785,
|
|
"rewards/rejected": -15.335331916809082,
|
|
"step": 1120
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 2.7606086221470835e-07,
|
|
"logits/chosen": -1.137892246246338,
|
|
"logits/rejected": -1.0994970798492432,
|
|
"logps/chosen": -90.780517578125,
|
|
"logps/rejected": -102.3265380859375,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.290585517883301,
|
|
"rewards/margins": 18.35919761657715,
|
|
"rewards/rejected": -16.068613052368164,
|
|
"step": 1130
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 2.755536770921386e-07,
|
|
"logits/chosen": -1.1444514989852905,
|
|
"logits/rejected": -1.1056411266326904,
|
|
"logps/chosen": -86.2841796875,
|
|
"logps/rejected": -100.90594482421875,
|
|
"loss": 0.0068,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.109790325164795,
|
|
"rewards/margins": 17.33005142211914,
|
|
"rewards/rejected": -16.220258712768555,
|
|
"step": 1140
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 2.750464919695689e-07,
|
|
"logits/chosen": -1.1471726894378662,
|
|
"logits/rejected": -1.1120365858078003,
|
|
"logps/chosen": -83.0753402709961,
|
|
"logps/rejected": -98.07260131835938,
|
|
"loss": 0.0027,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3137190341949463,
|
|
"rewards/margins": 17.202226638793945,
|
|
"rewards/rejected": -15.888509750366211,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 2.745393068469991e-07,
|
|
"logits/chosen": -1.131012201309204,
|
|
"logits/rejected": -1.0842430591583252,
|
|
"logps/chosen": -97.24982452392578,
|
|
"logps/rejected": -103.58902740478516,
|
|
"loss": 0.0125,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.803784728050232,
|
|
"rewards/margins": 19.04859161376953,
|
|
"rewards/rejected": -17.244808197021484,
|
|
"step": 1160
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 2.740321217244294e-07,
|
|
"logits/chosen": -1.1448577642440796,
|
|
"logits/rejected": -1.0983836650848389,
|
|
"logps/chosen": -88.2474136352539,
|
|
"logps/rejected": -101.29617309570312,
|
|
"loss": 0.0095,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.321990489959717,
|
|
"rewards/margins": 19.179903030395508,
|
|
"rewards/rejected": -16.857913970947266,
|
|
"step": 1170
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 2.7352493660185967e-07,
|
|
"logits/chosen": -1.15381920337677,
|
|
"logits/rejected": -1.1216586828231812,
|
|
"logps/chosen": -79.86784362792969,
|
|
"logps/rejected": -101.373046875,
|
|
"loss": 0.0065,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6416583061218262,
|
|
"rewards/margins": 18.733495712280273,
|
|
"rewards/rejected": -18.091835021972656,
|
|
"step": 1180
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 2.7301775147928993e-07,
|
|
"logits/chosen": -1.1581026315689087,
|
|
"logits/rejected": -1.1127700805664062,
|
|
"logps/chosen": -95.01976013183594,
|
|
"logps/rejected": -102.31490325927734,
|
|
"loss": 0.0073,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.2217804193496704,
|
|
"rewards/margins": 18.29319953918457,
|
|
"rewards/rejected": -17.07141876220703,
|
|
"step": 1190
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 2.725105663567202e-07,
|
|
"logits/chosen": -1.1556333303451538,
|
|
"logits/rejected": -1.1143467426300049,
|
|
"logps/chosen": -85.31793975830078,
|
|
"logps/rejected": -104.33609771728516,
|
|
"loss": 0.0154,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.109412670135498,
|
|
"rewards/margins": 19.068283081054688,
|
|
"rewards/rejected": -16.958871841430664,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"eval_logits/chosen": -1.2309268712997437,
|
|
"eval_logits/rejected": -1.177735447883606,
|
|
"eval_logps/chosen": -86.51016998291016,
|
|
"eval_logps/rejected": -97.42555236816406,
|
|
"eval_loss": 0.008077413775026798,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.4611291885375977,
|
|
"eval_rewards/margins": 17.5030517578125,
|
|
"eval_rewards/rejected": -16.041919708251953,
|
|
"eval_runtime": 61.5469,
|
|
"eval_samples_per_second": 46.501,
|
|
"eval_steps_per_second": 2.908,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 2.7200338123415044e-07,
|
|
"logits/chosen": -1.1782100200653076,
|
|
"logits/rejected": -1.1402640342712402,
|
|
"logps/chosen": -92.4530029296875,
|
|
"logps/rejected": -104.04048919677734,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6389929056167603,
|
|
"rewards/margins": 18.719791412353516,
|
|
"rewards/rejected": -17.080801010131836,
|
|
"step": 1210
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 2.714961961115807e-07,
|
|
"logits/chosen": -1.1656419038772583,
|
|
"logits/rejected": -1.1300376653671265,
|
|
"logps/chosen": -83.96723175048828,
|
|
"logps/rejected": -97.54676055908203,
|
|
"loss": 0.009,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.120603084564209,
|
|
"rewards/margins": 17.0549259185791,
|
|
"rewards/rejected": -15.934321403503418,
|
|
"step": 1220
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 2.70989010989011e-07,
|
|
"logits/chosen": -1.181616187095642,
|
|
"logits/rejected": -1.140226125717163,
|
|
"logps/chosen": -86.73959350585938,
|
|
"logps/rejected": -100.60263061523438,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.8291637897491455,
|
|
"rewards/margins": 17.071361541748047,
|
|
"rewards/rejected": -15.242197036743164,
|
|
"step": 1230
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 2.7048182586644125e-07,
|
|
"logits/chosen": -1.1932631731033325,
|
|
"logits/rejected": -1.1404846906661987,
|
|
"logps/chosen": -89.65509796142578,
|
|
"logps/rejected": -100.72111511230469,
|
|
"loss": 0.006,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.12876558303833,
|
|
"rewards/margins": 18.558353424072266,
|
|
"rewards/rejected": -16.429588317871094,
|
|
"step": 1240
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 2.699746407438715e-07,
|
|
"logits/chosen": -1.1967064142227173,
|
|
"logits/rejected": -1.1478387117385864,
|
|
"logps/chosen": -86.44863891601562,
|
|
"logps/rejected": -100.36338806152344,
|
|
"loss": 0.0028,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.105329990386963,
|
|
"rewards/margins": 18.038101196289062,
|
|
"rewards/rejected": -16.932769775390625,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 2.6946745562130176e-07,
|
|
"logits/chosen": -1.1940025091171265,
|
|
"logits/rejected": -1.1532320976257324,
|
|
"logps/chosen": -89.0478744506836,
|
|
"logps/rejected": -102.28709411621094,
|
|
"loss": 0.0037,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.9386533498764038,
|
|
"rewards/margins": 18.23946762084961,
|
|
"rewards/rejected": -16.30081558227539,
|
|
"step": 1260
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 2.68960270498732e-07,
|
|
"logits/chosen": -1.1994531154632568,
|
|
"logits/rejected": -1.1430509090423584,
|
|
"logps/chosen": -89.60600280761719,
|
|
"logps/rejected": -107.12632751464844,
|
|
"loss": 0.0077,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.8430372476577759,
|
|
"rewards/margins": 19.31827163696289,
|
|
"rewards/rejected": -18.475234985351562,
|
|
"step": 1270
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 2.6845308537616227e-07,
|
|
"logits/chosen": -1.1602545976638794,
|
|
"logits/rejected": -1.1090123653411865,
|
|
"logps/chosen": -87.08576202392578,
|
|
"logps/rejected": -95.82630920410156,
|
|
"loss": 0.011,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.725023627281189,
|
|
"rewards/margins": 18.602710723876953,
|
|
"rewards/rejected": -16.877687454223633,
|
|
"step": 1280
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 2.679459002535926e-07,
|
|
"logits/chosen": -1.2227842807769775,
|
|
"logits/rejected": -1.1634962558746338,
|
|
"logps/chosen": -88.16740417480469,
|
|
"logps/rejected": -97.47923278808594,
|
|
"loss": 0.0026,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8518360257148743,
|
|
"rewards/margins": 17.11991310119629,
|
|
"rewards/rejected": -16.268077850341797,
|
|
"step": 1290
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 2.674387151310228e-07,
|
|
"logits/chosen": -1.1858384609222412,
|
|
"logits/rejected": -1.1380422115325928,
|
|
"logps/chosen": -91.48844909667969,
|
|
"logps/rejected": -104.26606750488281,
|
|
"loss": 0.0041,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.4121227264404297,
|
|
"rewards/margins": 19.12758445739746,
|
|
"rewards/rejected": -16.71546173095703,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"eval_logits/chosen": -1.2335172891616821,
|
|
"eval_logits/rejected": -1.1754339933395386,
|
|
"eval_logps/chosen": -86.43448638916016,
|
|
"eval_logps/rejected": -99.26506042480469,
|
|
"eval_loss": 0.00815549585968256,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.4989758729934692,
|
|
"eval_rewards/margins": 18.460649490356445,
|
|
"eval_rewards/rejected": -16.961671829223633,
|
|
"eval_runtime": 59.4219,
|
|
"eval_samples_per_second": 48.164,
|
|
"eval_steps_per_second": 3.012,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 2.669315300084531e-07,
|
|
"logits/chosen": -1.2054587602615356,
|
|
"logits/rejected": -1.1616264581680298,
|
|
"logps/chosen": -86.14714050292969,
|
|
"logps/rejected": -103.92906188964844,
|
|
"loss": 0.0053,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1422611474990845,
|
|
"rewards/margins": 18.06179428100586,
|
|
"rewards/rejected": -16.919536590576172,
|
|
"step": 1310
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 2.6642434488588334e-07,
|
|
"logits/chosen": -1.164485216140747,
|
|
"logits/rejected": -1.1202260255813599,
|
|
"logps/chosen": -83.26927947998047,
|
|
"logps/rejected": -103.29952239990234,
|
|
"loss": 0.0101,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.3391577005386353,
|
|
"rewards/margins": 19.02712631225586,
|
|
"rewards/rejected": -17.68796730041504,
|
|
"step": 1320
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 2.659171597633136e-07,
|
|
"logits/chosen": -1.1936196088790894,
|
|
"logits/rejected": -1.1440293788909912,
|
|
"logps/chosen": -84.86544036865234,
|
|
"logps/rejected": -104.18849182128906,
|
|
"loss": 0.0051,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3678613901138306,
|
|
"rewards/margins": 19.402454376220703,
|
|
"rewards/rejected": -18.03459358215332,
|
|
"step": 1330
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 2.6540997464074385e-07,
|
|
"logits/chosen": -1.1995398998260498,
|
|
"logits/rejected": -1.1528923511505127,
|
|
"logps/chosen": -83.30887603759766,
|
|
"logps/rejected": -100.70631408691406,
|
|
"loss": 0.0063,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.5668017864227295,
|
|
"rewards/margins": 20.389301300048828,
|
|
"rewards/rejected": -17.822498321533203,
|
|
"step": 1340
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 2.649027895181741e-07,
|
|
"logits/chosen": -1.2100002765655518,
|
|
"logits/rejected": -1.1548488140106201,
|
|
"logps/chosen": -82.71726989746094,
|
|
"logps/rejected": -100.33686828613281,
|
|
"loss": 0.004,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1490261554718018,
|
|
"rewards/margins": 18.165735244750977,
|
|
"rewards/rejected": -17.016712188720703,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 2.6439560439560436e-07,
|
|
"logits/chosen": -1.2042474746704102,
|
|
"logits/rejected": -1.1507916450500488,
|
|
"logps/chosen": -89.17523956298828,
|
|
"logps/rejected": -103.0042724609375,
|
|
"loss": 0.002,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0882587432861328,
|
|
"rewards/margins": 19.461795806884766,
|
|
"rewards/rejected": -18.37353515625,
|
|
"step": 1360
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 2.6388841927303466e-07,
|
|
"logits/chosen": -1.19635808467865,
|
|
"logits/rejected": -1.1526683568954468,
|
|
"logps/chosen": -84.29436492919922,
|
|
"logps/rejected": -103.22540283203125,
|
|
"loss": 0.0066,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6204719543457031,
|
|
"rewards/margins": 19.50061798095703,
|
|
"rewards/rejected": -17.880146026611328,
|
|
"step": 1370
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 2.633812341504649e-07,
|
|
"logits/chosen": -1.1552878618240356,
|
|
"logits/rejected": -1.1150346994400024,
|
|
"logps/chosen": -84.24638366699219,
|
|
"logps/rejected": -101.41566467285156,
|
|
"loss": 0.0058,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.426559329032898,
|
|
"rewards/margins": 18.867931365966797,
|
|
"rewards/rejected": -17.44137191772461,
|
|
"step": 1380
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 2.6287404902789517e-07,
|
|
"logits/chosen": -1.1951394081115723,
|
|
"logits/rejected": -1.1533236503601074,
|
|
"logps/chosen": -89.08964538574219,
|
|
"logps/rejected": -107.95294189453125,
|
|
"loss": 0.0015,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.272940754890442,
|
|
"rewards/margins": 19.47437858581543,
|
|
"rewards/rejected": -18.201440811157227,
|
|
"step": 1390
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 2.623668639053254e-07,
|
|
"logits/chosen": -1.1739484071731567,
|
|
"logits/rejected": -1.1234138011932373,
|
|
"logps/chosen": -93.20767211914062,
|
|
"logps/rejected": -102.9843521118164,
|
|
"loss": 0.0058,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4396759271621704,
|
|
"rewards/margins": 19.409513473510742,
|
|
"rewards/rejected": -17.969837188720703,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"eval_logits/chosen": -1.2195051908493042,
|
|
"eval_logits/rejected": -1.1641638278961182,
|
|
"eval_logps/chosen": -86.76068115234375,
|
|
"eval_logps/rejected": -102.28946685791016,
|
|
"eval_loss": 0.008062196895480156,
|
|
"eval_rewards/accuracies": 0.9888268113136292,
|
|
"eval_rewards/chosen": 1.3358712196350098,
|
|
"eval_rewards/margins": 19.80974578857422,
|
|
"eval_rewards/rejected": -18.473875045776367,
|
|
"eval_runtime": 69.0484,
|
|
"eval_samples_per_second": 41.449,
|
|
"eval_steps_per_second": 2.592,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 2.618596787827557e-07,
|
|
"logits/chosen": -1.157814621925354,
|
|
"logits/rejected": -1.119381308555603,
|
|
"logps/chosen": -92.300048828125,
|
|
"logps/rejected": -103.40312194824219,
|
|
"loss": 0.0058,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.9399839639663696,
|
|
"rewards/margins": 19.74970054626465,
|
|
"rewards/rejected": -17.80971908569336,
|
|
"step": 1410
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 2.6135249366018593e-07,
|
|
"logits/chosen": -1.1667792797088623,
|
|
"logits/rejected": -1.119706392288208,
|
|
"logps/chosen": -85.43004608154297,
|
|
"logps/rejected": -100.96659851074219,
|
|
"loss": 0.0035,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.285947561264038,
|
|
"rewards/margins": 19.657182693481445,
|
|
"rewards/rejected": -17.371234893798828,
|
|
"step": 1420
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 2.6084530853761624e-07,
|
|
"logits/chosen": -1.1735081672668457,
|
|
"logits/rejected": -1.1159073114395142,
|
|
"logps/chosen": -89.15231323242188,
|
|
"logps/rejected": -104.07325744628906,
|
|
"loss": 0.0017,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.379710853099823,
|
|
"rewards/margins": 18.454669952392578,
|
|
"rewards/rejected": -18.07495880126953,
|
|
"step": 1430
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 2.603381234150465e-07,
|
|
"logits/chosen": -1.177208423614502,
|
|
"logits/rejected": -1.1227295398712158,
|
|
"logps/chosen": -87.95960235595703,
|
|
"logps/rejected": -100.62553405761719,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4888887405395508,
|
|
"rewards/margins": 18.826932907104492,
|
|
"rewards/rejected": -17.338045120239258,
|
|
"step": 1440
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 2.5983093829247675e-07,
|
|
"logits/chosen": -1.135980248451233,
|
|
"logits/rejected": -1.097318410873413,
|
|
"logps/chosen": -90.05815887451172,
|
|
"logps/rejected": -97.01565551757812,
|
|
"loss": 0.0178,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6748930215835571,
|
|
"rewards/margins": 16.284954071044922,
|
|
"rewards/rejected": -14.61005973815918,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 2.59323753169907e-07,
|
|
"logits/chosen": -1.1009352207183838,
|
|
"logits/rejected": -1.0712867975234985,
|
|
"logps/chosen": -90.10000610351562,
|
|
"logps/rejected": -104.41805267333984,
|
|
"loss": 0.0042,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.7969961762428284,
|
|
"rewards/margins": 18.065677642822266,
|
|
"rewards/rejected": -17.268680572509766,
|
|
"step": 1460
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 2.5881656804733726e-07,
|
|
"logits/chosen": -1.1424516439437866,
|
|
"logits/rejected": -1.104478120803833,
|
|
"logps/chosen": -85.87166595458984,
|
|
"logps/rejected": -96.37669372558594,
|
|
"loss": 0.0071,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2775685787200928,
|
|
"rewards/margins": 17.6835880279541,
|
|
"rewards/rejected": -16.40601921081543,
|
|
"step": 1470
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 2.583093829247675e-07,
|
|
"logits/chosen": -1.148421049118042,
|
|
"logits/rejected": -1.1051350831985474,
|
|
"logps/chosen": -84.38792419433594,
|
|
"logps/rejected": -95.69989776611328,
|
|
"loss": 0.0025,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.0020928382873535,
|
|
"rewards/margins": 18.655029296875,
|
|
"rewards/rejected": -16.652935028076172,
|
|
"step": 1480
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 2.5780219780219777e-07,
|
|
"logits/chosen": -1.137450933456421,
|
|
"logits/rejected": -1.1082595586776733,
|
|
"logps/chosen": -88.18030548095703,
|
|
"logps/rejected": -98.36376190185547,
|
|
"loss": 0.0041,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.1630921363830566,
|
|
"rewards/margins": 18.910762786865234,
|
|
"rewards/rejected": -16.747669219970703,
|
|
"step": 1490
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 2.57295012679628e-07,
|
|
"logits/chosen": -1.144364595413208,
|
|
"logits/rejected": -1.1114692687988281,
|
|
"logps/chosen": -87.55670928955078,
|
|
"logps/rejected": -104.33062744140625,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.494736909866333,
|
|
"rewards/margins": 19.5846004486084,
|
|
"rewards/rejected": -18.08986473083496,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"eval_logits/chosen": -1.2191152572631836,
|
|
"eval_logits/rejected": -1.1659319400787354,
|
|
"eval_logps/chosen": -87.01031494140625,
|
|
"eval_logps/rejected": -101.05850982666016,
|
|
"eval_loss": 0.00720271747559309,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.211055040359497,
|
|
"eval_rewards/margins": 19.069448471069336,
|
|
"eval_rewards/rejected": -17.85839080810547,
|
|
"eval_runtime": 77.7452,
|
|
"eval_samples_per_second": 36.813,
|
|
"eval_steps_per_second": 2.302,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 2.5678782755705833e-07,
|
|
"logits/chosen": -1.1631323099136353,
|
|
"logits/rejected": -1.1225086450576782,
|
|
"logps/chosen": -82.05781555175781,
|
|
"logps/rejected": -102.92134094238281,
|
|
"loss": 0.0024,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1484121084213257,
|
|
"rewards/margins": 19.43674087524414,
|
|
"rewards/rejected": -18.288328170776367,
|
|
"step": 1510
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 2.562806424344886e-07,
|
|
"logits/chosen": -1.1722062826156616,
|
|
"logits/rejected": -1.1262253522872925,
|
|
"logps/chosen": -92.10721588134766,
|
|
"logps/rejected": -107.66226959228516,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.5253825187683105,
|
|
"rewards/margins": 21.867202758789062,
|
|
"rewards/rejected": -19.341819763183594,
|
|
"step": 1520
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 2.5577345731191884e-07,
|
|
"logits/chosen": -1.1755383014678955,
|
|
"logits/rejected": -1.1288068294525146,
|
|
"logps/chosen": -86.13807678222656,
|
|
"logps/rejected": -107.95841217041016,
|
|
"loss": 0.0064,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1721524000167847,
|
|
"rewards/margins": 20.330608367919922,
|
|
"rewards/rejected": -19.158458709716797,
|
|
"step": 1530
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 2.552662721893491e-07,
|
|
"logits/chosen": -1.1406983137130737,
|
|
"logits/rejected": -1.1065037250518799,
|
|
"logps/chosen": -88.02510070800781,
|
|
"logps/rejected": -102.81829833984375,
|
|
"loss": 0.0039,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6564058661460876,
|
|
"rewards/margins": 18.354061126708984,
|
|
"rewards/rejected": -17.697656631469727,
|
|
"step": 1540
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 2.5475908706677935e-07,
|
|
"logits/chosen": -1.1616684198379517,
|
|
"logits/rejected": -1.1152303218841553,
|
|
"logps/chosen": -85.58765411376953,
|
|
"logps/rejected": -105.52689361572266,
|
|
"loss": 0.0038,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9486013650894165,
|
|
"rewards/margins": 20.196699142456055,
|
|
"rewards/rejected": -19.24810028076172,
|
|
"step": 1550
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 2.542519019442096e-07,
|
|
"logits/chosen": -1.2028621435165405,
|
|
"logits/rejected": -1.1485238075256348,
|
|
"logps/chosen": -78.52555847167969,
|
|
"logps/rejected": -99.6640853881836,
|
|
"loss": 0.0063,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.5557008981704712,
|
|
"rewards/margins": 19.518081665039062,
|
|
"rewards/rejected": -17.962379455566406,
|
|
"step": 1560
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 2.537447168216399e-07,
|
|
"logits/chosen": -1.176762342453003,
|
|
"logits/rejected": -1.1274961233139038,
|
|
"logps/chosen": -93.12080383300781,
|
|
"logps/rejected": -105.16368103027344,
|
|
"loss": 0.0031,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.659658432006836,
|
|
"rewards/margins": 19.342235565185547,
|
|
"rewards/rejected": -17.682575225830078,
|
|
"step": 1570
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 2.5323753169907016e-07,
|
|
"logits/chosen": -1.2171311378479004,
|
|
"logits/rejected": -1.172149419784546,
|
|
"logps/chosen": -86.0450210571289,
|
|
"logps/rejected": -107.85368347167969,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6793642044067383,
|
|
"rewards/margins": 21.292896270751953,
|
|
"rewards/rejected": -19.613529205322266,
|
|
"step": 1580
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 2.527303465765004e-07,
|
|
"logits/chosen": -1.2056636810302734,
|
|
"logits/rejected": -1.1607674360275269,
|
|
"logps/chosen": -93.74681854248047,
|
|
"logps/rejected": -113.48725891113281,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.6828597187995911,
|
|
"rewards/margins": 20.181041717529297,
|
|
"rewards/rejected": -19.49818229675293,
|
|
"step": 1590
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 2.5222316145393067e-07,
|
|
"logits/chosen": -1.1712194681167603,
|
|
"logits/rejected": -1.124768853187561,
|
|
"logps/chosen": -96.73299407958984,
|
|
"logps/rejected": -108.51261901855469,
|
|
"loss": 0.0035,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9574599266052246,
|
|
"rewards/margins": 20.438793182373047,
|
|
"rewards/rejected": -19.481334686279297,
|
|
"step": 1600
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"eval_logits/chosen": -1.2923858165740967,
|
|
"eval_logits/rejected": -1.227616786956787,
|
|
"eval_logps/chosen": -87.92854309082031,
|
|
"eval_logps/rejected": -104.88460540771484,
|
|
"eval_loss": 0.007888087071478367,
|
|
"eval_rewards/accuracies": 0.9888268113136292,
|
|
"eval_rewards/chosen": 0.7519445419311523,
|
|
"eval_rewards/margins": 20.523387908935547,
|
|
"eval_rewards/rejected": -19.77144432067871,
|
|
"eval_runtime": 58.4414,
|
|
"eval_samples_per_second": 48.972,
|
|
"eval_steps_per_second": 3.063,
|
|
"step": 1600
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 2.517159763313609e-07,
|
|
"logits/chosen": -1.2481939792633057,
|
|
"logits/rejected": -1.1921546459197998,
|
|
"logps/chosen": -90.30425262451172,
|
|
"logps/rejected": -106.55949401855469,
|
|
"loss": 0.0021,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6353106498718262,
|
|
"rewards/margins": 21.251611709594727,
|
|
"rewards/rejected": -20.616300582885742,
|
|
"step": 1610
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 2.512087912087912e-07,
|
|
"logits/chosen": -1.2304586172103882,
|
|
"logits/rejected": -1.173517107963562,
|
|
"logps/chosen": -92.87260437011719,
|
|
"logps/rejected": -110.1197280883789,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5774765610694885,
|
|
"rewards/margins": 20.693300247192383,
|
|
"rewards/rejected": -20.115821838378906,
|
|
"step": 1620
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 2.507016060862215e-07,
|
|
"logits/chosen": -1.2224457263946533,
|
|
"logits/rejected": -1.1627413034439087,
|
|
"logps/chosen": -89.59088134765625,
|
|
"logps/rejected": -107.96586608886719,
|
|
"loss": 0.0049,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 0.6187242865562439,
|
|
"rewards/margins": 21.216934204101562,
|
|
"rewards/rejected": -20.598209381103516,
|
|
"step": 1630
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 2.501944209636517e-07,
|
|
"logits/chosen": -1.2341878414154053,
|
|
"logits/rejected": -1.1646558046340942,
|
|
"logps/chosen": -99.49878692626953,
|
|
"logps/rejected": -107.7359390258789,
|
|
"loss": 0.0032,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.47276169061660767,
|
|
"rewards/margins": 21.252347946166992,
|
|
"rewards/rejected": -20.779584884643555,
|
|
"step": 1640
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 2.49687235841082e-07,
|
|
"logits/chosen": -1.2416927814483643,
|
|
"logits/rejected": -1.1858526468276978,
|
|
"logps/chosen": -92.1099624633789,
|
|
"logps/rejected": -109.5634765625,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.229796051979065,
|
|
"rewards/margins": 20.57327651977539,
|
|
"rewards/rejected": -19.34347915649414,
|
|
"step": 1650
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 2.4918005071851225e-07,
|
|
"logits/chosen": -1.2330695390701294,
|
|
"logits/rejected": -1.1908389329910278,
|
|
"logps/chosen": -87.75404357910156,
|
|
"logps/rejected": -106.69844055175781,
|
|
"loss": 0.0027,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9370654821395874,
|
|
"rewards/margins": 20.381633758544922,
|
|
"rewards/rejected": -19.444568634033203,
|
|
"step": 1660
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 2.486728655959425e-07,
|
|
"logits/chosen": -1.2374944686889648,
|
|
"logits/rejected": -1.178486943244934,
|
|
"logps/chosen": -88.52249145507812,
|
|
"logps/rejected": -102.99385070800781,
|
|
"loss": 0.0105,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.8609209060668945,
|
|
"rewards/margins": 20.464916229248047,
|
|
"rewards/rejected": -19.6039981842041,
|
|
"step": 1670
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 2.481656804733728e-07,
|
|
"logits/chosen": -1.228576421737671,
|
|
"logits/rejected": -1.1720635890960693,
|
|
"logps/chosen": -89.48531341552734,
|
|
"logps/rejected": -108.29301452636719,
|
|
"loss": 0.0027,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3690917491912842,
|
|
"rewards/margins": 20.013029098510742,
|
|
"rewards/rejected": -18.643938064575195,
|
|
"step": 1680
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 2.47658495350803e-07,
|
|
"logits/chosen": -1.2124004364013672,
|
|
"logits/rejected": -1.1609828472137451,
|
|
"logps/chosen": -84.43907928466797,
|
|
"logps/rejected": -104.23795318603516,
|
|
"loss": 0.0029,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7291128635406494,
|
|
"rewards/margins": 20.71861457824707,
|
|
"rewards/rejected": -18.989501953125,
|
|
"step": 1690
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 2.4715131022823327e-07,
|
|
"logits/chosen": -1.2294721603393555,
|
|
"logits/rejected": -1.1744955778121948,
|
|
"logps/chosen": -92.34876251220703,
|
|
"logps/rejected": -111.11873626708984,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.766455888748169,
|
|
"rewards/margins": 21.175350189208984,
|
|
"rewards/rejected": -20.408893585205078,
|
|
"step": 1700
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"eval_logits/chosen": -1.2879317998886108,
|
|
"eval_logits/rejected": -1.226698875427246,
|
|
"eval_logps/chosen": -87.46847534179688,
|
|
"eval_logps/rejected": -104.55460357666016,
|
|
"eval_loss": 0.007508194539695978,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.9819736480712891,
|
|
"eval_rewards/margins": 20.588415145874023,
|
|
"eval_rewards/rejected": -19.606443405151367,
|
|
"eval_runtime": 66.3677,
|
|
"eval_samples_per_second": 43.123,
|
|
"eval_steps_per_second": 2.697,
|
|
"step": 1700
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 2.4664412510566357e-07,
|
|
"logits/chosen": -1.2202476263046265,
|
|
"logits/rejected": -1.158785104751587,
|
|
"logps/chosen": -89.9117202758789,
|
|
"logps/rejected": -105.97224426269531,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9460952877998352,
|
|
"rewards/margins": 20.959388732910156,
|
|
"rewards/rejected": -20.013294219970703,
|
|
"step": 1710
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 2.4613693998309383e-07,
|
|
"logits/chosen": -1.2056442499160767,
|
|
"logits/rejected": -1.1586743593215942,
|
|
"logps/chosen": -84.9280776977539,
|
|
"logps/rejected": -102.05850982666016,
|
|
"loss": 0.0051,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.1483945846557617,
|
|
"rewards/margins": 20.313642501831055,
|
|
"rewards/rejected": -18.16524887084961,
|
|
"step": 1720
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 2.456297548605241e-07,
|
|
"logits/chosen": -1.1853172779083252,
|
|
"logits/rejected": -1.146316409111023,
|
|
"logps/chosen": -85.19999694824219,
|
|
"logps/rejected": -104.33415222167969,
|
|
"loss": 0.0056,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6200752258300781,
|
|
"rewards/margins": 19.86764907836914,
|
|
"rewards/rejected": -18.247573852539062,
|
|
"step": 1730
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 2.4512256973795434e-07,
|
|
"logits/chosen": -1.1810106039047241,
|
|
"logits/rejected": -1.1348565816879272,
|
|
"logps/chosen": -88.91767120361328,
|
|
"logps/rejected": -105.30973815917969,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.908353567123413,
|
|
"rewards/margins": 20.9567928314209,
|
|
"rewards/rejected": -19.04844093322754,
|
|
"step": 1740
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 2.446153846153846e-07,
|
|
"logits/chosen": -1.1730239391326904,
|
|
"logits/rejected": -1.1376618146896362,
|
|
"logps/chosen": -87.39840698242188,
|
|
"logps/rejected": -104.39476013183594,
|
|
"loss": 0.005,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.5575692653656006,
|
|
"rewards/margins": 19.33660125732422,
|
|
"rewards/rejected": -17.779033660888672,
|
|
"step": 1750
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 2.4410819949281484e-07,
|
|
"logits/chosen": -1.1946265697479248,
|
|
"logits/rejected": -1.1434904336929321,
|
|
"logps/chosen": -89.12718963623047,
|
|
"logps/rejected": -110.394287109375,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4408177435398102,
|
|
"rewards/margins": 19.224130630493164,
|
|
"rewards/rejected": -18.78331184387207,
|
|
"step": 1760
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 2.4360101437024515e-07,
|
|
"logits/chosen": -1.1789695024490356,
|
|
"logits/rejected": -1.136823296546936,
|
|
"logps/chosen": -83.54109954833984,
|
|
"logps/rejected": -104.07505798339844,
|
|
"loss": 0.0009,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6552903652191162,
|
|
"rewards/margins": 20.878345489501953,
|
|
"rewards/rejected": -19.223054885864258,
|
|
"step": 1770
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 2.4309382924767535e-07,
|
|
"logits/chosen": -1.1888688802719116,
|
|
"logits/rejected": -1.1330959796905518,
|
|
"logps/chosen": -91.33379364013672,
|
|
"logps/rejected": -110.06937408447266,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4772006869316101,
|
|
"rewards/margins": 20.467754364013672,
|
|
"rewards/rejected": -19.990554809570312,
|
|
"step": 1780
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 2.4258664412510566e-07,
|
|
"logits/chosen": -1.1968967914581299,
|
|
"logits/rejected": -1.1466515064239502,
|
|
"logps/chosen": -90.5532455444336,
|
|
"logps/rejected": -102.20117950439453,
|
|
"loss": 0.0105,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.0483832359313965,
|
|
"rewards/margins": 20.55087661743164,
|
|
"rewards/rejected": -18.502490997314453,
|
|
"step": 1790
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 2.420794590025359e-07,
|
|
"logits/chosen": -1.1845229864120483,
|
|
"logits/rejected": -1.137880802154541,
|
|
"logps/chosen": -90.67330932617188,
|
|
"logps/rejected": -107.00489807128906,
|
|
"loss": 0.005,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5007214546203613,
|
|
"rewards/margins": 20.610965728759766,
|
|
"rewards/rejected": -19.110244750976562,
|
|
"step": 1800
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"eval_logits/chosen": -1.2606803178787231,
|
|
"eval_logits/rejected": -1.1984455585479736,
|
|
"eval_logps/chosen": -86.78810119628906,
|
|
"eval_logps/rejected": -103.85710906982422,
|
|
"eval_loss": 0.008221164345741272,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.322161316871643,
|
|
"eval_rewards/margins": 20.57985496520996,
|
|
"eval_rewards/rejected": -19.257694244384766,
|
|
"eval_runtime": 70.2981,
|
|
"eval_samples_per_second": 40.712,
|
|
"eval_steps_per_second": 2.546,
|
|
"step": 1800
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 2.4157227387996617e-07,
|
|
"logits/chosen": -1.2245190143585205,
|
|
"logits/rejected": -1.1681041717529297,
|
|
"logps/chosen": -90.532470703125,
|
|
"logps/rejected": -104.7442398071289,
|
|
"loss": 0.0009,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.7430979609489441,
|
|
"rewards/margins": 20.217361450195312,
|
|
"rewards/rejected": -19.47426414489746,
|
|
"step": 1810
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 2.410650887573965e-07,
|
|
"logits/chosen": -1.2046802043914795,
|
|
"logits/rejected": -1.153531789779663,
|
|
"logps/chosen": -89.37370300292969,
|
|
"logps/rejected": -104.24407958984375,
|
|
"loss": 0.0066,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3622597455978394,
|
|
"rewards/margins": 19.505048751831055,
|
|
"rewards/rejected": -18.14278793334961,
|
|
"step": 1820
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 2.405579036348267e-07,
|
|
"logits/chosen": -1.1867458820343018,
|
|
"logits/rejected": -1.1448475122451782,
|
|
"logps/chosen": -80.08723449707031,
|
|
"logps/rejected": -102.7812728881836,
|
|
"loss": 0.0024,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2673819065093994,
|
|
"rewards/margins": 19.793813705444336,
|
|
"rewards/rejected": -18.526432037353516,
|
|
"step": 1830
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 2.4005071851225693e-07,
|
|
"logits/chosen": -1.1724661588668823,
|
|
"logits/rejected": -1.1277903318405151,
|
|
"logps/chosen": -90.30950927734375,
|
|
"logps/rejected": -113.6275405883789,
|
|
"loss": 0.0028,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5377770662307739,
|
|
"rewards/margins": 20.727577209472656,
|
|
"rewards/rejected": -20.189800262451172,
|
|
"step": 1840
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 2.3954353338968724e-07,
|
|
"logits/chosen": -1.1979376077651978,
|
|
"logits/rejected": -1.155174970626831,
|
|
"logps/chosen": -83.19313049316406,
|
|
"logps/rejected": -104.564453125,
|
|
"loss": 0.0042,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9587980508804321,
|
|
"rewards/margins": 20.022220611572266,
|
|
"rewards/rejected": -19.063425064086914,
|
|
"step": 1850
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 2.390363482671175e-07,
|
|
"logits/chosen": -1.170663595199585,
|
|
"logits/rejected": -1.11752450466156,
|
|
"logps/chosen": -90.89318084716797,
|
|
"logps/rejected": -109.8968734741211,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.412500262260437,
|
|
"rewards/margins": 21.69009017944336,
|
|
"rewards/rejected": -20.277587890625,
|
|
"step": 1860
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 2.3852916314454775e-07,
|
|
"logits/chosen": -1.1926937103271484,
|
|
"logits/rejected": -1.1473562717437744,
|
|
"logps/chosen": -93.6591796875,
|
|
"logps/rejected": -106.84416198730469,
|
|
"loss": 0.0039,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.8632802963256836,
|
|
"rewards/margins": 19.899490356445312,
|
|
"rewards/rejected": -18.036211013793945,
|
|
"step": 1870
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 2.38021978021978e-07,
|
|
"logits/chosen": -1.1921271085739136,
|
|
"logits/rejected": -1.1413754224777222,
|
|
"logps/chosen": -85.18832397460938,
|
|
"logps/rejected": -105.3952407836914,
|
|
"loss": 0.0038,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8996086120605469,
|
|
"rewards/margins": 20.485759735107422,
|
|
"rewards/rejected": -19.586151123046875,
|
|
"step": 1880
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 2.3751479289940826e-07,
|
|
"logits/chosen": -1.2049638032913208,
|
|
"logits/rejected": -1.1474329233169556,
|
|
"logps/chosen": -92.87411499023438,
|
|
"logps/rejected": -106.48674011230469,
|
|
"loss": 0.0039,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.1706862449645996,
|
|
"rewards/margins": 22.162261962890625,
|
|
"rewards/rejected": -19.991575241088867,
|
|
"step": 1890
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 2.3700760777683854e-07,
|
|
"logits/chosen": -1.1745494604110718,
|
|
"logits/rejected": -1.1166932582855225,
|
|
"logps/chosen": -90.28245544433594,
|
|
"logps/rejected": -111.68940734863281,
|
|
"loss": 0.003,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.081857204437256,
|
|
"rewards/margins": 21.467239379882812,
|
|
"rewards/rejected": -19.385379791259766,
|
|
"step": 1900
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"eval_logits/chosen": -1.2352814674377441,
|
|
"eval_logits/rejected": -1.1811531782150269,
|
|
"eval_logps/chosen": -86.63172149658203,
|
|
"eval_logps/rejected": -104.2937240600586,
|
|
"eval_loss": 0.007481275591999292,
|
|
"eval_rewards/accuracies": 0.994413435459137,
|
|
"eval_rewards/chosen": 1.4003478288650513,
|
|
"eval_rewards/margins": 20.87635612487793,
|
|
"eval_rewards/rejected": -19.47600746154785,
|
|
"eval_runtime": 64.2566,
|
|
"eval_samples_per_second": 44.54,
|
|
"eval_steps_per_second": 2.786,
|
|
"step": 1900
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 2.3650042265426882e-07,
|
|
"logits/chosen": -1.2035939693450928,
|
|
"logits/rejected": -1.1619645357131958,
|
|
"logps/chosen": -87.07538604736328,
|
|
"logps/rejected": -105.97283935546875,
|
|
"loss": 0.0035,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.9201610088348389,
|
|
"rewards/margins": 21.46194839477539,
|
|
"rewards/rejected": -19.54178810119629,
|
|
"step": 1910
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 2.3599323753169907e-07,
|
|
"logits/chosen": -1.1762607097625732,
|
|
"logits/rejected": -1.1266227960586548,
|
|
"logps/chosen": -84.24024200439453,
|
|
"logps/rejected": -105.28382873535156,
|
|
"loss": 0.0088,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.7776741981506348,
|
|
"rewards/margins": 21.163555145263672,
|
|
"rewards/rejected": -20.38587760925293,
|
|
"step": 1920
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 2.354860524091293e-07,
|
|
"logits/chosen": -1.196942925453186,
|
|
"logits/rejected": -1.1432682275772095,
|
|
"logps/chosen": -95.1038818359375,
|
|
"logps/rejected": -106.1786117553711,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3159481287002563,
|
|
"rewards/margins": 21.558012008666992,
|
|
"rewards/rejected": -20.2420654296875,
|
|
"step": 1930
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 2.3497886728655958e-07,
|
|
"logits/chosen": -1.1665886640548706,
|
|
"logits/rejected": -1.1258021593093872,
|
|
"logps/chosen": -83.95616149902344,
|
|
"logps/rejected": -100.2303237915039,
|
|
"loss": 0.0038,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.901753306388855,
|
|
"rewards/margins": 19.35708999633789,
|
|
"rewards/rejected": -18.455339431762695,
|
|
"step": 1940
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 2.3447168216398983e-07,
|
|
"logits/chosen": -1.18831205368042,
|
|
"logits/rejected": -1.1461542844772339,
|
|
"logps/chosen": -89.49617767333984,
|
|
"logps/rejected": -105.2798843383789,
|
|
"loss": 0.007,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2909132242202759,
|
|
"rewards/margins": 20.118661880493164,
|
|
"rewards/rejected": -18.827747344970703,
|
|
"step": 1950
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 2.3396449704142012e-07,
|
|
"logits/chosen": -1.1922476291656494,
|
|
"logits/rejected": -1.1407110691070557,
|
|
"logps/chosen": -94.23783874511719,
|
|
"logps/rejected": -105.92415618896484,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5083615183830261,
|
|
"rewards/margins": 19.7877254486084,
|
|
"rewards/rejected": -19.279361724853516,
|
|
"step": 1960
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 2.334573119188504e-07,
|
|
"logits/chosen": -1.1829791069030762,
|
|
"logits/rejected": -1.1273462772369385,
|
|
"logps/chosen": -88.82598876953125,
|
|
"logps/rejected": -101.88629150390625,
|
|
"loss": 0.0049,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.235897183418274,
|
|
"rewards/margins": 21.914575576782227,
|
|
"rewards/rejected": -20.67867660522461,
|
|
"step": 1970
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 2.3295012679628062e-07,
|
|
"logits/chosen": -1.2115113735198975,
|
|
"logits/rejected": -1.1657658815383911,
|
|
"logps/chosen": -86.12489318847656,
|
|
"logps/rejected": -107.81489562988281,
|
|
"loss": 0.0048,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.653774082660675,
|
|
"rewards/margins": 20.462425231933594,
|
|
"rewards/rejected": -19.808650970458984,
|
|
"step": 1980
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 2.3244294167371088e-07,
|
|
"logits/chosen": -1.2142908573150635,
|
|
"logits/rejected": -1.1777855157852173,
|
|
"logps/chosen": -84.75012969970703,
|
|
"logps/rejected": -106.76557922363281,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0658010244369507,
|
|
"rewards/margins": 20.134159088134766,
|
|
"rewards/rejected": -19.068355560302734,
|
|
"step": 1990
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 2.3193575655114116e-07,
|
|
"logits/chosen": -1.2295953035354614,
|
|
"logits/rejected": -1.174081802368164,
|
|
"logps/chosen": -88.81068420410156,
|
|
"logps/rejected": -106.81929779052734,
|
|
"loss": 0.0078,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5790882110595703,
|
|
"rewards/margins": 20.7359619140625,
|
|
"rewards/rejected": -19.156875610351562,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"eval_logits/chosen": -1.2681105136871338,
|
|
"eval_logits/rejected": -1.2061457633972168,
|
|
"eval_logps/chosen": -86.66743469238281,
|
|
"eval_logps/rejected": -104.74527740478516,
|
|
"eval_loss": 0.007192350458353758,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.3824937343597412,
|
|
"eval_rewards/margins": 21.084270477294922,
|
|
"eval_rewards/rejected": -19.7017765045166,
|
|
"eval_runtime": 77.6552,
|
|
"eval_samples_per_second": 36.855,
|
|
"eval_steps_per_second": 2.305,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 2.3142857142857144e-07,
|
|
"logits/chosen": -1.194087266921997,
|
|
"logits/rejected": -1.1440551280975342,
|
|
"logps/chosen": -87.83335876464844,
|
|
"logps/rejected": -106.5591812133789,
|
|
"loss": 0.015,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5759541988372803,
|
|
"rewards/margins": 20.1513614654541,
|
|
"rewards/rejected": -18.575407028198242,
|
|
"step": 2010
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 2.3092138630600167e-07,
|
|
"logits/chosen": -1.1671911478042603,
|
|
"logits/rejected": -1.1248931884765625,
|
|
"logps/chosen": -82.89833068847656,
|
|
"logps/rejected": -110.85752868652344,
|
|
"loss": 0.0005,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.971392035484314,
|
|
"rewards/margins": 20.262601852416992,
|
|
"rewards/rejected": -19.291210174560547,
|
|
"step": 2020
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 2.3041420118343192e-07,
|
|
"logits/chosen": -1.1789249181747437,
|
|
"logits/rejected": -1.1308996677398682,
|
|
"logps/chosen": -85.59063720703125,
|
|
"logps/rejected": -98.32166290283203,
|
|
"loss": 0.0064,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.9662765264511108,
|
|
"rewards/margins": 20.44546890258789,
|
|
"rewards/rejected": -18.479190826416016,
|
|
"step": 2030
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 2.299070160608622e-07,
|
|
"logits/chosen": -1.1901233196258545,
|
|
"logits/rejected": -1.1428929567337036,
|
|
"logps/chosen": -83.4649887084961,
|
|
"logps/rejected": -101.49374389648438,
|
|
"loss": 0.0036,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.49424687027931213,
|
|
"rewards/margins": 19.59086036682129,
|
|
"rewards/rejected": -19.09661293029785,
|
|
"step": 2040
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 2.2939983093829248e-07,
|
|
"logits/chosen": -1.211072325706482,
|
|
"logits/rejected": -1.1505995988845825,
|
|
"logps/chosen": -92.88377380371094,
|
|
"logps/rejected": -105.16926574707031,
|
|
"loss": 0.0069,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.585486650466919,
|
|
"rewards/margins": 20.312013626098633,
|
|
"rewards/rejected": -18.72652816772461,
|
|
"step": 2050
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 2.2889264581572274e-07,
|
|
"logits/chosen": -1.1979122161865234,
|
|
"logits/rejected": -1.1408374309539795,
|
|
"logps/chosen": -88.95879364013672,
|
|
"logps/rejected": -100.50065612792969,
|
|
"loss": 0.0088,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.2848825454711914,
|
|
"rewards/margins": 20.595218658447266,
|
|
"rewards/rejected": -18.31033706665039,
|
|
"step": 2060
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 2.2838546069315297e-07,
|
|
"logits/chosen": -1.1775312423706055,
|
|
"logits/rejected": -1.1326141357421875,
|
|
"logps/chosen": -83.72702026367188,
|
|
"logps/rejected": -104.96478271484375,
|
|
"loss": 0.0017,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2204118967056274,
|
|
"rewards/margins": 20.910999298095703,
|
|
"rewards/rejected": -19.69058609008789,
|
|
"step": 2070
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 2.2787827557058325e-07,
|
|
"logits/chosen": -1.1710788011550903,
|
|
"logits/rejected": -1.1334991455078125,
|
|
"logps/chosen": -84.7524185180664,
|
|
"logps/rejected": -103.06710052490234,
|
|
"loss": 0.0025,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.717494249343872,
|
|
"rewards/margins": 20.584232330322266,
|
|
"rewards/rejected": -18.866741180419922,
|
|
"step": 2080
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 2.273710904480135e-07,
|
|
"logits/chosen": -1.1447703838348389,
|
|
"logits/rejected": -1.1037436723709106,
|
|
"logps/chosen": -88.0915298461914,
|
|
"logps/rejected": -101.42835998535156,
|
|
"loss": 0.0026,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3874223232269287,
|
|
"rewards/margins": 19.40856170654297,
|
|
"rewards/rejected": -18.02113914489746,
|
|
"step": 2090
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 2.2686390532544378e-07,
|
|
"logits/chosen": -1.1998827457427979,
|
|
"logits/rejected": -1.144376277923584,
|
|
"logps/chosen": -87.5056381225586,
|
|
"logps/rejected": -100.17271423339844,
|
|
"loss": 0.0066,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.228088855743408,
|
|
"rewards/margins": 20.235179901123047,
|
|
"rewards/rejected": -18.007089614868164,
|
|
"step": 2100
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"eval_logits/chosen": -1.2238779067993164,
|
|
"eval_logits/rejected": -1.1694415807724,
|
|
"eval_logps/chosen": -86.23553466796875,
|
|
"eval_logps/rejected": -102.94633483886719,
|
|
"eval_loss": 0.0070498245768249035,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.5984489917755127,
|
|
"eval_rewards/margins": 20.400758743286133,
|
|
"eval_rewards/rejected": -18.802310943603516,
|
|
"eval_runtime": 62.0998,
|
|
"eval_samples_per_second": 46.087,
|
|
"eval_steps_per_second": 2.882,
|
|
"step": 2100
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 2.2635672020287406e-07,
|
|
"logits/chosen": -1.1710069179534912,
|
|
"logits/rejected": -1.1208176612854004,
|
|
"logps/chosen": -89.8357162475586,
|
|
"logps/rejected": -104.9393310546875,
|
|
"loss": 0.0051,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2994686365127563,
|
|
"rewards/margins": 20.46290397644043,
|
|
"rewards/rejected": -19.163436889648438,
|
|
"step": 2110
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 2.258495350803043e-07,
|
|
"logits/chosen": -1.167830228805542,
|
|
"logits/rejected": -1.1247303485870361,
|
|
"logps/chosen": -88.80085754394531,
|
|
"logps/rejected": -101.0454330444336,
|
|
"loss": 0.0029,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7379745244979858,
|
|
"rewards/margins": 20.091503143310547,
|
|
"rewards/rejected": -18.353527069091797,
|
|
"step": 2120
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 2.2534234995773454e-07,
|
|
"logits/chosen": -1.1950472593307495,
|
|
"logits/rejected": -1.144486427307129,
|
|
"logps/chosen": -91.32661437988281,
|
|
"logps/rejected": -110.21795654296875,
|
|
"loss": 0.0077,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.46089300513267517,
|
|
"rewards/margins": 21.63949966430664,
|
|
"rewards/rejected": -21.178604125976562,
|
|
"step": 2130
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 2.2483516483516483e-07,
|
|
"logits/chosen": -1.1894288063049316,
|
|
"logits/rejected": -1.1427417993545532,
|
|
"logps/chosen": -85.63072967529297,
|
|
"logps/rejected": -102.2844467163086,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.9994566440582275,
|
|
"rewards/margins": 21.046451568603516,
|
|
"rewards/rejected": -19.0469913482666,
|
|
"step": 2140
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 2.243279797125951e-07,
|
|
"logits/chosen": -1.1913617849349976,
|
|
"logits/rejected": -1.1351690292358398,
|
|
"logps/chosen": -86.342529296875,
|
|
"logps/rejected": -105.4500503540039,
|
|
"loss": 0.0041,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3321704864501953,
|
|
"rewards/margins": 20.112749099731445,
|
|
"rewards/rejected": -18.780582427978516,
|
|
"step": 2150
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 2.2382079459002536e-07,
|
|
"logits/chosen": -1.200955867767334,
|
|
"logits/rejected": -1.141643762588501,
|
|
"logps/chosen": -93.14427185058594,
|
|
"logps/rejected": -100.25897979736328,
|
|
"loss": 0.0028,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.7122005224227905,
|
|
"rewards/margins": 20.42462158203125,
|
|
"rewards/rejected": -18.712421417236328,
|
|
"step": 2160
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 2.233136094674556e-07,
|
|
"logits/chosen": -1.2126085758209229,
|
|
"logits/rejected": -1.1542718410491943,
|
|
"logps/chosen": -94.84847259521484,
|
|
"logps/rejected": -105.94117736816406,
|
|
"loss": 0.0035,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.69661545753479,
|
|
"rewards/margins": 20.98930549621582,
|
|
"rewards/rejected": -19.29269027709961,
|
|
"step": 2170
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 2.2280642434488587e-07,
|
|
"logits/chosen": -1.2024385929107666,
|
|
"logits/rejected": -1.15294349193573,
|
|
"logps/chosen": -81.07176208496094,
|
|
"logps/rejected": -103.16294860839844,
|
|
"loss": 0.0027,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.119349479675293,
|
|
"rewards/margins": 21.580371856689453,
|
|
"rewards/rejected": -19.461023330688477,
|
|
"step": 2180
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.2229923922231615e-07,
|
|
"logits/chosen": -1.1685454845428467,
|
|
"logits/rejected": -1.1278364658355713,
|
|
"logps/chosen": -91.21316528320312,
|
|
"logps/rejected": -108.0694580078125,
|
|
"loss": 0.0002,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.229431629180908,
|
|
"rewards/margins": 21.74981689453125,
|
|
"rewards/rejected": -19.5203857421875,
|
|
"step": 2190
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 2.217920540997464e-07,
|
|
"logits/chosen": -1.1836879253387451,
|
|
"logits/rejected": -1.1423550844192505,
|
|
"logps/chosen": -88.64768981933594,
|
|
"logps/rejected": -109.5053482055664,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5388556718826294,
|
|
"rewards/margins": 20.56551742553711,
|
|
"rewards/rejected": -20.026660919189453,
|
|
"step": 2200
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"eval_logits/chosen": -1.2711361646652222,
|
|
"eval_logits/rejected": -1.2118490934371948,
|
|
"eval_logps/chosen": -86.72749328613281,
|
|
"eval_logps/rejected": -104.70652770996094,
|
|
"eval_loss": 0.0074717202223837376,
|
|
"eval_rewards/accuracies": 0.994413435459137,
|
|
"eval_rewards/chosen": 1.352469801902771,
|
|
"eval_rewards/margins": 21.034873962402344,
|
|
"eval_rewards/rejected": -19.682405471801758,
|
|
"eval_runtime": 64.1917,
|
|
"eval_samples_per_second": 44.585,
|
|
"eval_steps_per_second": 2.789,
|
|
"step": 2200
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.2128486897717668e-07,
|
|
"logits/chosen": -1.2156295776367188,
|
|
"logits/rejected": -1.1702911853790283,
|
|
"logps/chosen": -84.73490142822266,
|
|
"logps/rejected": -109.97856140136719,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.8699300289154053,
|
|
"rewards/margins": 22.461772918701172,
|
|
"rewards/rejected": -20.591842651367188,
|
|
"step": 2210
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 2.207776838546069e-07,
|
|
"logits/chosen": -1.2317909002304077,
|
|
"logits/rejected": -1.1759793758392334,
|
|
"logps/chosen": -88.38722229003906,
|
|
"logps/rejected": -110.55941009521484,
|
|
"loss": 0.0031,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.2832095623016357,
|
|
"rewards/margins": 21.7716007232666,
|
|
"rewards/rejected": -19.488391876220703,
|
|
"step": 2220
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.202704987320372e-07,
|
|
"logits/chosen": -1.240782618522644,
|
|
"logits/rejected": -1.1896260976791382,
|
|
"logps/chosen": -90.6281509399414,
|
|
"logps/rejected": -107.0479736328125,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5835065841674805,
|
|
"rewards/margins": 21.230846405029297,
|
|
"rewards/rejected": -19.6473388671875,
|
|
"step": 2230
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 2.1976331360946745e-07,
|
|
"logits/chosen": -1.2149629592895508,
|
|
"logits/rejected": -1.1640093326568604,
|
|
"logps/chosen": -90.72506713867188,
|
|
"logps/rejected": -106.3701171875,
|
|
"loss": 0.0035,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3114402294158936,
|
|
"rewards/margins": 21.860370635986328,
|
|
"rewards/rejected": -20.54892921447754,
|
|
"step": 2240
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 2.1925612848689773e-07,
|
|
"logits/chosen": -1.207849383354187,
|
|
"logits/rejected": -1.1694178581237793,
|
|
"logps/chosen": -88.57085418701172,
|
|
"logps/rejected": -107.42121887207031,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3758482933044434,
|
|
"rewards/margins": 21.30167007446289,
|
|
"rewards/rejected": -19.925823211669922,
|
|
"step": 2250
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 2.1874894336432796e-07,
|
|
"logits/chosen": -1.221719741821289,
|
|
"logits/rejected": -1.1561182737350464,
|
|
"logps/chosen": -92.80986022949219,
|
|
"logps/rejected": -109.31239318847656,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.313194513320923,
|
|
"rewards/margins": 23.222518920898438,
|
|
"rewards/rejected": -20.909326553344727,
|
|
"step": 2260
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 2.182417582417582e-07,
|
|
"logits/chosen": -1.2283899784088135,
|
|
"logits/rejected": -1.1857209205627441,
|
|
"logps/chosen": -86.95245361328125,
|
|
"logps/rejected": -110.2240219116211,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5342823266983032,
|
|
"rewards/margins": 21.846160888671875,
|
|
"rewards/rejected": -20.311880111694336,
|
|
"step": 2270
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 2.177345731191885e-07,
|
|
"logits/chosen": -1.2207567691802979,
|
|
"logits/rejected": -1.1733354330062866,
|
|
"logps/chosen": -89.40998077392578,
|
|
"logps/rejected": -108.59842681884766,
|
|
"loss": 0.005,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.084045648574829,
|
|
"rewards/margins": 21.383882522583008,
|
|
"rewards/rejected": -20.29983901977539,
|
|
"step": 2280
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 2.1722738799661877e-07,
|
|
"logits/chosen": -1.2484939098358154,
|
|
"logits/rejected": -1.186006784439087,
|
|
"logps/chosen": -91.48284149169922,
|
|
"logps/rejected": -102.3342514038086,
|
|
"loss": 0.0071,
|
|
"rewards/accuracies": 0.9624999761581421,
|
|
"rewards/chosen": 1.3055270910263062,
|
|
"rewards/margins": 20.427013397216797,
|
|
"rewards/rejected": -19.12148666381836,
|
|
"step": 2290
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 2.1672020287404903e-07,
|
|
"logits/chosen": -1.2090219259262085,
|
|
"logits/rejected": -1.1676021814346313,
|
|
"logps/chosen": -89.25651550292969,
|
|
"logps/rejected": -110.95772552490234,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.608730673789978,
|
|
"rewards/margins": 20.88467025756836,
|
|
"rewards/rejected": -20.275938034057617,
|
|
"step": 2300
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"eval_logits/chosen": -1.3087431192398071,
|
|
"eval_logits/rejected": -1.2488420009613037,
|
|
"eval_logps/chosen": -87.20025634765625,
|
|
"eval_logps/rejected": -105.1304702758789,
|
|
"eval_loss": 0.007476452272385359,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.1160887479782104,
|
|
"eval_rewards/margins": 21.010465621948242,
|
|
"eval_rewards/rejected": -19.894376754760742,
|
|
"eval_runtime": 68.119,
|
|
"eval_samples_per_second": 42.015,
|
|
"eval_steps_per_second": 2.628,
|
|
"step": 2300
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 2.1621301775147925e-07,
|
|
"logits/chosen": -1.2305896282196045,
|
|
"logits/rejected": -1.1756011247634888,
|
|
"logps/chosen": -85.5489501953125,
|
|
"logps/rejected": -109.3161849975586,
|
|
"loss": 0.0017,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.7630363702774048,
|
|
"rewards/margins": 21.09680938720703,
|
|
"rewards/rejected": -20.333772659301758,
|
|
"step": 2310
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 2.1570583262890953e-07,
|
|
"logits/chosen": -1.2177748680114746,
|
|
"logits/rejected": -1.1697107553482056,
|
|
"logps/chosen": -85.29845428466797,
|
|
"logps/rejected": -109.9201889038086,
|
|
"loss": 0.0066,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4945674538612366,
|
|
"rewards/margins": 21.288578033447266,
|
|
"rewards/rejected": -20.794010162353516,
|
|
"step": 2320
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 2.1519864750633982e-07,
|
|
"logits/chosen": -1.2171680927276611,
|
|
"logits/rejected": -1.1630264520645142,
|
|
"logps/chosen": -88.26265716552734,
|
|
"logps/rejected": -104.31333923339844,
|
|
"loss": 0.0019,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4045048952102661,
|
|
"rewards/margins": 20.78306770324707,
|
|
"rewards/rejected": -19.378559112548828,
|
|
"step": 2330
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 2.1469146238377007e-07,
|
|
"logits/chosen": -1.2630029916763306,
|
|
"logits/rejected": -1.2106314897537231,
|
|
"logps/chosen": -88.36056518554688,
|
|
"logps/rejected": -110.58036041259766,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7803552150726318,
|
|
"rewards/margins": 22.81916046142578,
|
|
"rewards/rejected": -21.038806915283203,
|
|
"step": 2340
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 2.1418427726120035e-07,
|
|
"logits/chosen": -1.2328459024429321,
|
|
"logits/rejected": -1.1753827333450317,
|
|
"logps/chosen": -90.98271942138672,
|
|
"logps/rejected": -107.30308532714844,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.918079137802124,
|
|
"rewards/margins": 22.566545486450195,
|
|
"rewards/rejected": -20.648466110229492,
|
|
"step": 2350
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 2.1367709213863058e-07,
|
|
"logits/chosen": -1.221885085105896,
|
|
"logits/rejected": -1.1747316122055054,
|
|
"logps/chosen": -83.33541870117188,
|
|
"logps/rejected": -107.0094985961914,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.8313859701156616,
|
|
"rewards/margins": 22.628454208374023,
|
|
"rewards/rejected": -20.797067642211914,
|
|
"step": 2360
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 2.1316990701606086e-07,
|
|
"logits/chosen": -1.2495050430297852,
|
|
"logits/rejected": -1.1890289783477783,
|
|
"logps/chosen": -86.03840637207031,
|
|
"logps/rejected": -112.093505859375,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5868552923202515,
|
|
"rewards/margins": 22.05801010131836,
|
|
"rewards/rejected": -21.471149444580078,
|
|
"step": 2370
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 2.126627218934911e-07,
|
|
"logits/chosen": -1.2298763990402222,
|
|
"logits/rejected": -1.1767555475234985,
|
|
"logps/chosen": -84.1092758178711,
|
|
"logps/rejected": -106.68165588378906,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.08795030415058136,
|
|
"rewards/margins": 21.050338745117188,
|
|
"rewards/rejected": -20.962390899658203,
|
|
"step": 2380
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 2.121555367709214e-07,
|
|
"logits/chosen": -1.2427847385406494,
|
|
"logits/rejected": -1.2060487270355225,
|
|
"logps/chosen": -85.17793273925781,
|
|
"logps/rejected": -109.41981506347656,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.06969626992940903,
|
|
"rewards/margins": 20.437192916870117,
|
|
"rewards/rejected": -20.367496490478516,
|
|
"step": 2390
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 2.1164835164835165e-07,
|
|
"logits/chosen": -1.223024845123291,
|
|
"logits/rejected": -1.1713087558746338,
|
|
"logps/chosen": -85.1490707397461,
|
|
"logps/rejected": -114.04695129394531,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.0163342952728271,
|
|
"rewards/margins": 22.59818458557129,
|
|
"rewards/rejected": -21.581851959228516,
|
|
"step": 2400
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"eval_logits/chosen": -1.3130491971969604,
|
|
"eval_logits/rejected": -1.2457395792007446,
|
|
"eval_logps/chosen": -87.72120666503906,
|
|
"eval_logps/rejected": -107.27316284179688,
|
|
"eval_loss": 0.007567834109067917,
|
|
"eval_rewards/accuracies": 0.994413435459137,
|
|
"eval_rewards/chosen": 0.8556116223335266,
|
|
"eval_rewards/margins": 21.821340560913086,
|
|
"eval_rewards/rejected": -20.965728759765625,
|
|
"eval_runtime": 68.7759,
|
|
"eval_samples_per_second": 41.613,
|
|
"eval_steps_per_second": 2.603,
|
|
"step": 2400
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 2.1114116652578188e-07,
|
|
"logits/chosen": -1.224714756011963,
|
|
"logits/rejected": -1.1649423837661743,
|
|
"logps/chosen": -91.27989959716797,
|
|
"logps/rejected": -109.84356689453125,
|
|
"loss": 0.0066,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.8775809407234192,
|
|
"rewards/margins": 22.14706039428711,
|
|
"rewards/rejected": -21.26947784423828,
|
|
"step": 2410
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 2.1063398140321216e-07,
|
|
"logits/chosen": -1.2352392673492432,
|
|
"logits/rejected": -1.1807770729064941,
|
|
"logps/chosen": -86.58528137207031,
|
|
"logps/rejected": -107.39393615722656,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1014773845672607,
|
|
"rewards/margins": 22.366958618164062,
|
|
"rewards/rejected": -21.265480041503906,
|
|
"step": 2420
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 2.1012679628064244e-07,
|
|
"logits/chosen": -1.271728515625,
|
|
"logits/rejected": -1.218794584274292,
|
|
"logps/chosen": -83.79855346679688,
|
|
"logps/rejected": -115.31538391113281,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1647331714630127,
|
|
"rewards/margins": 23.370630264282227,
|
|
"rewards/rejected": -22.205896377563477,
|
|
"step": 2430
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 2.096196111580727e-07,
|
|
"logits/chosen": -1.2588279247283936,
|
|
"logits/rejected": -1.1938583850860596,
|
|
"logps/chosen": -92.08492279052734,
|
|
"logps/rejected": -108.75785827636719,
|
|
"loss": 0.0005,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8567431569099426,
|
|
"rewards/margins": 22.12790298461914,
|
|
"rewards/rejected": -21.271160125732422,
|
|
"step": 2440
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 2.0911242603550297e-07,
|
|
"logits/chosen": -1.254792332649231,
|
|
"logits/rejected": -1.184197187423706,
|
|
"logps/chosen": -92.77003479003906,
|
|
"logps/rejected": -114.69380950927734,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.006107807159424,
|
|
"rewards/margins": 24.762887954711914,
|
|
"rewards/rejected": -22.756778717041016,
|
|
"step": 2450
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 2.086052409129332e-07,
|
|
"logits/chosen": -1.2276710271835327,
|
|
"logits/rejected": -1.172614336013794,
|
|
"logps/chosen": -92.51925659179688,
|
|
"logps/rejected": -109.40245056152344,
|
|
"loss": 0.0037,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4506779909133911,
|
|
"rewards/margins": 22.675724029541016,
|
|
"rewards/rejected": -21.225048065185547,
|
|
"step": 2460
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 2.0809805579036348e-07,
|
|
"logits/chosen": -1.2417436838150024,
|
|
"logits/rejected": -1.1916046142578125,
|
|
"logps/chosen": -87.77659606933594,
|
|
"logps/rejected": -107.12300109863281,
|
|
"loss": 0.0015,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.3042961359024048,
|
|
"rewards/margins": 21.55357551574707,
|
|
"rewards/rejected": -20.249279022216797,
|
|
"step": 2470
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 2.0759087066779374e-07,
|
|
"logits/chosen": -1.2384653091430664,
|
|
"logits/rejected": -1.19193434715271,
|
|
"logps/chosen": -87.645751953125,
|
|
"logps/rejected": -111.66351318359375,
|
|
"loss": 0.0017,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5307129621505737,
|
|
"rewards/margins": 22.680551528930664,
|
|
"rewards/rejected": -22.149839401245117,
|
|
"step": 2480
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 2.0708368554522402e-07,
|
|
"logits/chosen": -1.239863634109497,
|
|
"logits/rejected": -1.1838042736053467,
|
|
"logps/chosen": -82.25237274169922,
|
|
"logps/rejected": -107.54081726074219,
|
|
"loss": 0.0004,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.317577600479126,
|
|
"rewards/margins": 21.645660400390625,
|
|
"rewards/rejected": -20.328083038330078,
|
|
"step": 2490
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 2.0657650042265424e-07,
|
|
"logits/chosen": -1.238755464553833,
|
|
"logits/rejected": -1.1892435550689697,
|
|
"logps/chosen": -91.93738555908203,
|
|
"logps/rejected": -105.78620910644531,
|
|
"loss": 0.0083,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8200355768203735,
|
|
"rewards/margins": 21.127065658569336,
|
|
"rewards/rejected": -20.307029724121094,
|
|
"step": 2500
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"eval_logits/chosen": -1.2944495677947998,
|
|
"eval_logits/rejected": -1.2309415340423584,
|
|
"eval_logps/chosen": -87.02932739257812,
|
|
"eval_logps/rejected": -104.29837799072266,
|
|
"eval_loss": 0.006961911916732788,
|
|
"eval_rewards/accuracies": 0.994413435459137,
|
|
"eval_rewards/chosen": 1.2015457153320312,
|
|
"eval_rewards/margins": 20.67987823486328,
|
|
"eval_rewards/rejected": -19.478334426879883,
|
|
"eval_runtime": 71.1702,
|
|
"eval_samples_per_second": 40.213,
|
|
"eval_steps_per_second": 2.515,
|
|
"step": 2500
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 2.0606931530008452e-07,
|
|
"logits/chosen": -1.2269400358200073,
|
|
"logits/rejected": -1.165895700454712,
|
|
"logps/chosen": -89.00814056396484,
|
|
"logps/rejected": -106.82181549072266,
|
|
"loss": 0.0048,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1590993404388428,
|
|
"rewards/margins": 21.093416213989258,
|
|
"rewards/rejected": -19.934314727783203,
|
|
"step": 2510
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 2.0556213017751478e-07,
|
|
"logits/chosen": -1.244373083114624,
|
|
"logits/rejected": -1.1799051761627197,
|
|
"logps/chosen": -86.1784896850586,
|
|
"logps/rejected": -108.1005859375,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9702383279800415,
|
|
"rewards/margins": 21.26332664489746,
|
|
"rewards/rejected": -20.293087005615234,
|
|
"step": 2520
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 2.0505494505494506e-07,
|
|
"logits/chosen": -1.239007830619812,
|
|
"logits/rejected": -1.188663363456726,
|
|
"logps/chosen": -89.08403015136719,
|
|
"logps/rejected": -109.1912612915039,
|
|
"loss": 0.004,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8398208618164062,
|
|
"rewards/margins": 20.724300384521484,
|
|
"rewards/rejected": -19.884479522705078,
|
|
"step": 2530
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 2.0454775993237531e-07,
|
|
"logits/chosen": -1.259961485862732,
|
|
"logits/rejected": -1.1992539167404175,
|
|
"logps/chosen": -88.00000762939453,
|
|
"logps/rejected": -105.28775787353516,
|
|
"loss": 0.0014,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7344595193862915,
|
|
"rewards/margins": 22.018203735351562,
|
|
"rewards/rejected": -20.283742904663086,
|
|
"step": 2540
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 2.0404057480980554e-07,
|
|
"logits/chosen": -1.258159875869751,
|
|
"logits/rejected": -1.1919949054718018,
|
|
"logps/chosen": -91.0265121459961,
|
|
"logps/rejected": -109.3829345703125,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2989047765731812,
|
|
"rewards/margins": 22.686710357666016,
|
|
"rewards/rejected": -21.387805938720703,
|
|
"step": 2550
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 2.0353338968723582e-07,
|
|
"logits/chosen": -1.216018557548523,
|
|
"logits/rejected": -1.1775795221328735,
|
|
"logps/chosen": -84.73078918457031,
|
|
"logps/rejected": -113.40096282958984,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.12488274276256561,
|
|
"rewards/margins": 21.136188507080078,
|
|
"rewards/rejected": -21.261070251464844,
|
|
"step": 2560
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 2.030262045646661e-07,
|
|
"logits/chosen": -1.2377723455429077,
|
|
"logits/rejected": -1.1810615062713623,
|
|
"logps/chosen": -93.79662322998047,
|
|
"logps/rejected": -107.40061950683594,
|
|
"loss": 0.0061,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.11829134076833725,
|
|
"rewards/margins": 20.636348724365234,
|
|
"rewards/rejected": -20.754640579223633,
|
|
"step": 2570
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 2.0251901944209636e-07,
|
|
"logits/chosen": -1.2281471490859985,
|
|
"logits/rejected": -1.1688311100006104,
|
|
"logps/chosen": -91.79940795898438,
|
|
"logps/rejected": -107.32728576660156,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.22485239803791046,
|
|
"rewards/margins": 20.858522415161133,
|
|
"rewards/rejected": -20.633668899536133,
|
|
"step": 2580
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 2.0201183431952664e-07,
|
|
"logits/chosen": -1.197676658630371,
|
|
"logits/rejected": -1.1506736278533936,
|
|
"logps/chosen": -81.2101821899414,
|
|
"logps/rejected": -109.7088623046875,
|
|
"loss": 0.0055,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.2717623710632324,
|
|
"rewards/margins": 21.169933319091797,
|
|
"rewards/rejected": -21.441696166992188,
|
|
"step": 2590
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 2.0150464919695687e-07,
|
|
"logits/chosen": -1.2182395458221436,
|
|
"logits/rejected": -1.177594542503357,
|
|
"logps/chosen": -80.82548522949219,
|
|
"logps/rejected": -109.75965881347656,
|
|
"loss": 0.0024,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1711565256118774,
|
|
"rewards/margins": 23.289653778076172,
|
|
"rewards/rejected": -22.118499755859375,
|
|
"step": 2600
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"eval_logits/chosen": -1.2811179161071777,
|
|
"eval_logits/rejected": -1.2195229530334473,
|
|
"eval_logps/chosen": -87.84212493896484,
|
|
"eval_logps/rejected": -107.58037567138672,
|
|
"eval_loss": 0.007497187703847885,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.7951509952545166,
|
|
"eval_rewards/margins": 21.91447639465332,
|
|
"eval_rewards/rejected": -21.119325637817383,
|
|
"eval_runtime": 67.5783,
|
|
"eval_samples_per_second": 42.351,
|
|
"eval_steps_per_second": 2.649,
|
|
"step": 2600
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 2.0099746407438715e-07,
|
|
"logits/chosen": -1.2012989521026611,
|
|
"logits/rejected": -1.173628807067871,
|
|
"logps/chosen": -84.23133087158203,
|
|
"logps/rejected": -108.86763000488281,
|
|
"loss": 0.0055,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.148093581199646,
|
|
"rewards/margins": 21.286386489868164,
|
|
"rewards/rejected": -20.138290405273438,
|
|
"step": 2610
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 2.004902789518174e-07,
|
|
"logits/chosen": -1.236003041267395,
|
|
"logits/rejected": -1.1768968105316162,
|
|
"logps/chosen": -91.87477111816406,
|
|
"logps/rejected": -112.02117919921875,
|
|
"loss": 0.0038,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2967796325683594,
|
|
"rewards/margins": 22.90878677368164,
|
|
"rewards/rejected": -21.612009048461914,
|
|
"step": 2620
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 1.9998309382924768e-07,
|
|
"logits/chosen": -1.2255749702453613,
|
|
"logits/rejected": -1.169988989830017,
|
|
"logps/chosen": -89.92630004882812,
|
|
"logps/rejected": -109.48509216308594,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.11678309738636017,
|
|
"rewards/margins": 20.700876235961914,
|
|
"rewards/rejected": -20.584091186523438,
|
|
"step": 2630
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 1.9947590870667794e-07,
|
|
"logits/chosen": -1.2192347049713135,
|
|
"logits/rejected": -1.159459114074707,
|
|
"logps/chosen": -96.32530212402344,
|
|
"logps/rejected": -110.26582336425781,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.2365353107452393,
|
|
"rewards/margins": 22.127044677734375,
|
|
"rewards/rejected": -19.8905086517334,
|
|
"step": 2640
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 1.989687235841082e-07,
|
|
"logits/chosen": -1.2065198421478271,
|
|
"logits/rejected": -1.147120714187622,
|
|
"logps/chosen": -88.47337341308594,
|
|
"logps/rejected": -106.5348129272461,
|
|
"loss": 0.004,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9000838994979858,
|
|
"rewards/margins": 20.618175506591797,
|
|
"rewards/rejected": -19.718093872070312,
|
|
"step": 2650
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 1.9846153846153844e-07,
|
|
"logits/chosen": -1.195985198020935,
|
|
"logits/rejected": -1.1491913795471191,
|
|
"logps/chosen": -84.20733642578125,
|
|
"logps/rejected": -109.67060852050781,
|
|
"loss": 0.0028,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2371699810028076,
|
|
"rewards/margins": 22.225109100341797,
|
|
"rewards/rejected": -20.98794174194336,
|
|
"step": 2660
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 1.9795435333896873e-07,
|
|
"logits/chosen": -1.1977856159210205,
|
|
"logits/rejected": -1.1537766456604004,
|
|
"logps/chosen": -91.16226196289062,
|
|
"logps/rejected": -111.68495178222656,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.734151601791382,
|
|
"rewards/margins": 23.330568313598633,
|
|
"rewards/rejected": -20.596416473388672,
|
|
"step": 2670
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 1.9744716821639898e-07,
|
|
"logits/chosen": -1.224413275718689,
|
|
"logits/rejected": -1.1784846782684326,
|
|
"logps/chosen": -87.34507751464844,
|
|
"logps/rejected": -109.75943756103516,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3192319869995117,
|
|
"rewards/margins": 22.662378311157227,
|
|
"rewards/rejected": -21.343143463134766,
|
|
"step": 2680
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 1.9693998309382926e-07,
|
|
"logits/chosen": -1.208308219909668,
|
|
"logits/rejected": -1.1619625091552734,
|
|
"logps/chosen": -88.23912048339844,
|
|
"logps/rejected": -107.49666595458984,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.7308362126350403,
|
|
"rewards/margins": 20.509281158447266,
|
|
"rewards/rejected": -19.778446197509766,
|
|
"step": 2690
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 1.964327979712595e-07,
|
|
"logits/chosen": -1.2484124898910522,
|
|
"logits/rejected": -1.1881217956542969,
|
|
"logps/chosen": -90.5484390258789,
|
|
"logps/rejected": -105.3963851928711,
|
|
"loss": 0.0048,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.22395610809326172,
|
|
"rewards/margins": 21.083341598510742,
|
|
"rewards/rejected": -21.30729866027832,
|
|
"step": 2700
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"eval_logits/chosen": -1.2836943864822388,
|
|
"eval_logits/rejected": -1.2240569591522217,
|
|
"eval_logps/chosen": -88.02273559570312,
|
|
"eval_logps/rejected": -107.38671875,
|
|
"eval_loss": 0.007047051563858986,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.704842746257782,
|
|
"eval_rewards/margins": 21.727344512939453,
|
|
"eval_rewards/rejected": -21.022504806518555,
|
|
"eval_runtime": 66.849,
|
|
"eval_samples_per_second": 42.813,
|
|
"eval_steps_per_second": 2.678,
|
|
"step": 2700
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 1.9592561284868977e-07,
|
|
"logits/chosen": -1.2299364805221558,
|
|
"logits/rejected": -1.1793487071990967,
|
|
"logps/chosen": -92.32845306396484,
|
|
"logps/rejected": -109.91078186035156,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9196799993515015,
|
|
"rewards/margins": 21.892927169799805,
|
|
"rewards/rejected": -20.973247528076172,
|
|
"step": 2710
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 1.9541842772612002e-07,
|
|
"logits/chosen": -1.2148702144622803,
|
|
"logits/rejected": -1.1717993021011353,
|
|
"logps/chosen": -88.74729919433594,
|
|
"logps/rejected": -109.92634582519531,
|
|
"loss": 0.0017,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4501422047615051,
|
|
"rewards/margins": 20.923179626464844,
|
|
"rewards/rejected": -20.47303581237793,
|
|
"step": 2720
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 1.949112426035503e-07,
|
|
"logits/chosen": -1.2163503170013428,
|
|
"logits/rejected": -1.1693785190582275,
|
|
"logps/chosen": -87.26588439941406,
|
|
"logps/rejected": -110.53138732910156,
|
|
"loss": 0.0058,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.8547931909561157,
|
|
"rewards/margins": 21.771400451660156,
|
|
"rewards/rejected": -22.626192092895508,
|
|
"step": 2730
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 1.9440405748098056e-07,
|
|
"logits/chosen": -1.2386237382888794,
|
|
"logits/rejected": -1.1867671012878418,
|
|
"logps/chosen": -89.03327941894531,
|
|
"logps/rejected": -112.5365219116211,
|
|
"loss": 0.0001,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6293159127235413,
|
|
"rewards/margins": 22.378183364868164,
|
|
"rewards/rejected": -21.74886703491211,
|
|
"step": 2740
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 1.938968723584108e-07,
|
|
"logits/chosen": -1.241208791732788,
|
|
"logits/rejected": -1.1873838901519775,
|
|
"logps/chosen": -88.9361801147461,
|
|
"logps/rejected": -111.81880187988281,
|
|
"loss": 0.0056,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0905447006225586,
|
|
"rewards/margins": 23.877498626708984,
|
|
"rewards/rejected": -22.78695297241211,
|
|
"step": 2750
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 1.9338968723584107e-07,
|
|
"logits/chosen": -1.2293826341629028,
|
|
"logits/rejected": -1.1767637729644775,
|
|
"logps/chosen": -84.17843627929688,
|
|
"logps/rejected": -110.3796157836914,
|
|
"loss": 0.0046,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.36438900232315063,
|
|
"rewards/margins": 22.36743927001953,
|
|
"rewards/rejected": -22.731830596923828,
|
|
"step": 2760
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 1.9288250211327135e-07,
|
|
"logits/chosen": -1.2488285303115845,
|
|
"logits/rejected": -1.192888617515564,
|
|
"logps/chosen": -94.11280822753906,
|
|
"logps/rejected": -113.39571380615234,
|
|
"loss": 0.0036,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.8494230508804321,
|
|
"rewards/margins": 21.94849967956543,
|
|
"rewards/rejected": -22.797924041748047,
|
|
"step": 2770
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 1.923753169907016e-07,
|
|
"logits/chosen": -1.2614643573760986,
|
|
"logits/rejected": -1.2098888158798218,
|
|
"logps/chosen": -89.56803894042969,
|
|
"logps/rejected": -112.6842041015625,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.0361621156334877,
|
|
"rewards/margins": 23.073833465576172,
|
|
"rewards/rejected": -23.10999298095703,
|
|
"step": 2780
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 1.9186813186813186e-07,
|
|
"logits/chosen": -1.2534093856811523,
|
|
"logits/rejected": -1.1944758892059326,
|
|
"logps/chosen": -93.7116470336914,
|
|
"logps/rejected": -106.58601379394531,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6133219003677368,
|
|
"rewards/margins": 21.842636108398438,
|
|
"rewards/rejected": -21.22931671142578,
|
|
"step": 2790
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 1.913609467455621e-07,
|
|
"logits/chosen": -1.245892882347107,
|
|
"logits/rejected": -1.1841986179351807,
|
|
"logps/chosen": -90.11885070800781,
|
|
"logps/rejected": -108.6605224609375,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8591505885124207,
|
|
"rewards/margins": 23.346960067749023,
|
|
"rewards/rejected": -22.487812042236328,
|
|
"step": 2800
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"eval_logits/chosen": -1.3369088172912598,
|
|
"eval_logits/rejected": -1.2667419910430908,
|
|
"eval_logps/chosen": -88.43457794189453,
|
|
"eval_logps/rejected": -109.54808807373047,
|
|
"eval_loss": 0.007251236122101545,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.4989199936389923,
|
|
"eval_rewards/margins": 22.602100372314453,
|
|
"eval_rewards/rejected": -22.103178024291992,
|
|
"eval_runtime": 65.6934,
|
|
"eval_samples_per_second": 43.566,
|
|
"eval_steps_per_second": 2.725,
|
|
"step": 2800
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 1.908537616229924e-07,
|
|
"logits/chosen": -1.258395791053772,
|
|
"logits/rejected": -1.2066611051559448,
|
|
"logps/chosen": -88.32936096191406,
|
|
"logps/rejected": -113.46788024902344,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.060993265360593796,
|
|
"rewards/margins": 23.58555030822754,
|
|
"rewards/rejected": -23.52455711364746,
|
|
"step": 2810
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 1.9034657650042265e-07,
|
|
"logits/chosen": -1.2754733562469482,
|
|
"logits/rejected": -1.1980760097503662,
|
|
"logps/chosen": -93.66505432128906,
|
|
"logps/rejected": -109.4578857421875,
|
|
"loss": 0.0028,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4866414964199066,
|
|
"rewards/margins": 22.506893157958984,
|
|
"rewards/rejected": -22.020254135131836,
|
|
"step": 2820
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 1.8983939137785293e-07,
|
|
"logits/chosen": -1.2679641246795654,
|
|
"logits/rejected": -1.2068798542022705,
|
|
"logps/chosen": -89.0657958984375,
|
|
"logps/rejected": -111.84321594238281,
|
|
"loss": 0.0007,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.362867146730423,
|
|
"rewards/margins": 23.33441162109375,
|
|
"rewards/rejected": -23.697277069091797,
|
|
"step": 2830
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 1.8933220625528315e-07,
|
|
"logits/chosen": -1.2522757053375244,
|
|
"logits/rejected": -1.179939866065979,
|
|
"logps/chosen": -93.52415466308594,
|
|
"logps/rejected": -112.2892074584961,
|
|
"loss": 0.003,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2486017942428589,
|
|
"rewards/margins": 23.161243438720703,
|
|
"rewards/rejected": -21.912641525268555,
|
|
"step": 2840
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 1.8882502113271343e-07,
|
|
"logits/chosen": -1.2578299045562744,
|
|
"logits/rejected": -1.193433165550232,
|
|
"logps/chosen": -89.56175994873047,
|
|
"logps/rejected": -114.70777893066406,
|
|
"loss": 0.0008,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8356913328170776,
|
|
"rewards/margins": 23.007110595703125,
|
|
"rewards/rejected": -22.171415328979492,
|
|
"step": 2850
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 1.883178360101437e-07,
|
|
"logits/chosen": -1.279767632484436,
|
|
"logits/rejected": -1.2183212041854858,
|
|
"logps/chosen": -89.8489990234375,
|
|
"logps/rejected": -108.4361572265625,
|
|
"loss": 0.0078,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6225892305374146,
|
|
"rewards/margins": 21.247386932373047,
|
|
"rewards/rejected": -20.624794006347656,
|
|
"step": 2860
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 1.8781065088757397e-07,
|
|
"logits/chosen": -1.3064444065093994,
|
|
"logits/rejected": -1.251308798789978,
|
|
"logps/chosen": -88.62841033935547,
|
|
"logps/rejected": -114.151123046875,
|
|
"loss": 0.0046,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6001108288764954,
|
|
"rewards/margins": 23.617412567138672,
|
|
"rewards/rejected": -23.017301559448242,
|
|
"step": 2870
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 1.8730346576500422e-07,
|
|
"logits/chosen": -1.2963197231292725,
|
|
"logits/rejected": -1.2211467027664185,
|
|
"logps/chosen": -89.53153991699219,
|
|
"logps/rejected": -116.00457763671875,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3336631059646606,
|
|
"rewards/margins": 24.389131546020508,
|
|
"rewards/rejected": -23.055469512939453,
|
|
"step": 2880
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 1.8679628064243448e-07,
|
|
"logits/chosen": -1.2721803188323975,
|
|
"logits/rejected": -1.209424614906311,
|
|
"logps/chosen": -90.9636459350586,
|
|
"logps/rejected": -112.62530517578125,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8648780584335327,
|
|
"rewards/margins": 23.844661712646484,
|
|
"rewards/rejected": -22.97978401184082,
|
|
"step": 2890
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 1.8628909551986473e-07,
|
|
"logits/chosen": -1.304807186126709,
|
|
"logits/rejected": -1.2427805662155151,
|
|
"logps/chosen": -83.58341979980469,
|
|
"logps/rejected": -114.4560546875,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.09369596093893051,
|
|
"rewards/margins": 23.40268325805664,
|
|
"rewards/rejected": -23.496379852294922,
|
|
"step": 2900
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"eval_logits/chosen": -1.3647960424423218,
|
|
"eval_logits/rejected": -1.2919734716415405,
|
|
"eval_logps/chosen": -88.49767303466797,
|
|
"eval_logps/rejected": -111.2530517578125,
|
|
"eval_loss": 0.007645368576049805,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.4673759341239929,
|
|
"eval_rewards/margins": 23.423046112060547,
|
|
"eval_rewards/rejected": -22.95566749572754,
|
|
"eval_runtime": 68.7318,
|
|
"eval_samples_per_second": 41.64,
|
|
"eval_steps_per_second": 2.604,
|
|
"step": 2900
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 1.8578191039729501e-07,
|
|
"logits/chosen": -1.2682093381881714,
|
|
"logits/rejected": -1.2079681158065796,
|
|
"logps/chosen": -94.27190399169922,
|
|
"logps/rejected": -113.67183685302734,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.7304227352142334,
|
|
"rewards/margins": 24.069181442260742,
|
|
"rewards/rejected": -22.338756561279297,
|
|
"step": 2910
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 1.8527472527472527e-07,
|
|
"logits/chosen": -1.2731274366378784,
|
|
"logits/rejected": -1.2099798917770386,
|
|
"logps/chosen": -90.7890625,
|
|
"logps/rejected": -106.4814224243164,
|
|
"loss": 0.0063,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.650849461555481,
|
|
"rewards/margins": 22.963346481323242,
|
|
"rewards/rejected": -22.312496185302734,
|
|
"step": 2920
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 1.8476754015215555e-07,
|
|
"logits/chosen": -1.292454481124878,
|
|
"logits/rejected": -1.225913166999817,
|
|
"logps/chosen": -85.83187103271484,
|
|
"logps/rejected": -110.36527252197266,
|
|
"loss": 0.0001,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2073954343795776,
|
|
"rewards/margins": 24.3538818359375,
|
|
"rewards/rejected": -23.146484375,
|
|
"step": 2930
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 1.8426035502958578e-07,
|
|
"logits/chosen": -1.2939157485961914,
|
|
"logits/rejected": -1.241677165031433,
|
|
"logps/chosen": -84.26634216308594,
|
|
"logps/rejected": -108.85639953613281,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.16938212513923645,
|
|
"rewards/margins": 22.67632293701172,
|
|
"rewards/rejected": -22.506940841674805,
|
|
"step": 2940
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 1.8375316990701606e-07,
|
|
"logits/chosen": -1.2990548610687256,
|
|
"logits/rejected": -1.2253262996673584,
|
|
"logps/chosen": -85.23284149169922,
|
|
"logps/rejected": -111.3265609741211,
|
|
"loss": 0.0002,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6182515025138855,
|
|
"rewards/margins": 23.335525512695312,
|
|
"rewards/rejected": -22.717273712158203,
|
|
"step": 2950
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 1.832459847844463e-07,
|
|
"logits/chosen": -1.3059203624725342,
|
|
"logits/rejected": -1.2436602115631104,
|
|
"logps/chosen": -90.0726089477539,
|
|
"logps/rejected": -115.29248046875,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.11955726146698,
|
|
"rewards/margins": 24.451736450195312,
|
|
"rewards/rejected": -23.332178115844727,
|
|
"step": 2960
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 1.827387996618766e-07,
|
|
"logits/chosen": -1.2917280197143555,
|
|
"logits/rejected": -1.2361105680465698,
|
|
"logps/chosen": -83.81163024902344,
|
|
"logps/rejected": -110.36248779296875,
|
|
"loss": 0.0035,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.2116299867630005,
|
|
"rewards/margins": 23.077804565429688,
|
|
"rewards/rejected": -22.866174697875977,
|
|
"step": 2970
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 1.8223161453930685e-07,
|
|
"logits/chosen": -1.2799623012542725,
|
|
"logits/rejected": -1.2227824926376343,
|
|
"logps/chosen": -87.18734741210938,
|
|
"logps/rejected": -111.6829605102539,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.030847549438476562,
|
|
"rewards/margins": 22.31467056274414,
|
|
"rewards/rejected": -22.283823013305664,
|
|
"step": 2980
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 1.817244294167371e-07,
|
|
"logits/chosen": -1.2903432846069336,
|
|
"logits/rejected": -1.2311654090881348,
|
|
"logps/chosen": -92.79168701171875,
|
|
"logps/rejected": -119.3042984008789,
|
|
"loss": 0.0148,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6965764164924622,
|
|
"rewards/margins": 24.21768569946289,
|
|
"rewards/rejected": -23.52111053466797,
|
|
"step": 2990
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 1.8121724429416736e-07,
|
|
"logits/chosen": -1.2526795864105225,
|
|
"logits/rejected": -1.1851316690444946,
|
|
"logps/chosen": -89.67386627197266,
|
|
"logps/rejected": -108.59903717041016,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0026484727859497,
|
|
"rewards/margins": 23.317981719970703,
|
|
"rewards/rejected": -22.315330505371094,
|
|
"step": 3000
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"eval_logits/chosen": -1.3374384641647339,
|
|
"eval_logits/rejected": -1.2689754962921143,
|
|
"eval_logps/chosen": -87.6075668334961,
|
|
"eval_logps/rejected": -110.42879486083984,
|
|
"eval_loss": 0.007348579820245504,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.9124311804771423,
|
|
"eval_rewards/margins": 23.45596694946289,
|
|
"eval_rewards/rejected": -22.543535232543945,
|
|
"eval_runtime": 63.1636,
|
|
"eval_samples_per_second": 45.311,
|
|
"eval_steps_per_second": 2.834,
|
|
"step": 3000
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 1.8071005917159764e-07,
|
|
"logits/chosen": -1.2701785564422607,
|
|
"logits/rejected": -1.2133753299713135,
|
|
"logps/chosen": -89.51029968261719,
|
|
"logps/rejected": -107.2208023071289,
|
|
"loss": 0.001,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4126553535461426,
|
|
"rewards/margins": 23.328351974487305,
|
|
"rewards/rejected": -21.915699005126953,
|
|
"step": 3010
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 1.802028740490279e-07,
|
|
"logits/chosen": -1.2436578273773193,
|
|
"logits/rejected": -1.178425908088684,
|
|
"logps/chosen": -87.91740417480469,
|
|
"logps/rejected": -108.53468322753906,
|
|
"loss": 0.0027,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.169164776802063,
|
|
"rewards/margins": 23.27883529663086,
|
|
"rewards/rejected": -22.109668731689453,
|
|
"step": 3020
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 1.7969568892645814e-07,
|
|
"logits/chosen": -1.2551841735839844,
|
|
"logits/rejected": -1.1954948902130127,
|
|
"logps/chosen": -86.35130310058594,
|
|
"logps/rejected": -113.94222259521484,
|
|
"loss": 0.0014,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.3147376775741577,
|
|
"rewards/margins": 22.975399017333984,
|
|
"rewards/rejected": -22.660661697387695,
|
|
"step": 3030
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 1.791885038038884e-07,
|
|
"logits/chosen": -1.2619261741638184,
|
|
"logits/rejected": -1.1995004415512085,
|
|
"logps/chosen": -89.64755249023438,
|
|
"logps/rejected": -115.27685546875,
|
|
"loss": 0.0014,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.632409691810608,
|
|
"rewards/margins": 25.825876235961914,
|
|
"rewards/rejected": -24.193464279174805,
|
|
"step": 3040
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 1.7868131868131868e-07,
|
|
"logits/chosen": -1.2092974185943604,
|
|
"logits/rejected": -1.1581265926361084,
|
|
"logps/chosen": -94.53260803222656,
|
|
"logps/rejected": -113.39241790771484,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.7243481278419495,
|
|
"rewards/margins": 24.363618850708008,
|
|
"rewards/rejected": -23.639272689819336,
|
|
"step": 3050
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 1.7817413355874893e-07,
|
|
"logits/chosen": -1.2218453884124756,
|
|
"logits/rejected": -1.1740856170654297,
|
|
"logps/chosen": -87.55658721923828,
|
|
"logps/rejected": -112.15279388427734,
|
|
"loss": 0.0024,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.7927554845809937,
|
|
"rewards/margins": 24.38910484313965,
|
|
"rewards/rejected": -23.59635353088379,
|
|
"step": 3060
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 1.7766694843617921e-07,
|
|
"logits/chosen": -1.272851586341858,
|
|
"logits/rejected": -1.206168293952942,
|
|
"logps/chosen": -83.55747985839844,
|
|
"logps/rejected": -120.4216079711914,
|
|
"loss": 0.0,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5537545084953308,
|
|
"rewards/margins": 24.476547241210938,
|
|
"rewards/rejected": -23.922794342041016,
|
|
"step": 3070
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 1.7715976331360944e-07,
|
|
"logits/chosen": -1.2406799793243408,
|
|
"logits/rejected": -1.1959577798843384,
|
|
"logps/chosen": -88.22968292236328,
|
|
"logps/rejected": -118.29911041259766,
|
|
"loss": 0.0018,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6699932813644409,
|
|
"rewards/margins": 24.441137313842773,
|
|
"rewards/rejected": -23.77114486694336,
|
|
"step": 3080
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 1.7665257819103972e-07,
|
|
"logits/chosen": -1.2692559957504272,
|
|
"logits/rejected": -1.1990225315093994,
|
|
"logps/chosen": -91.19334411621094,
|
|
"logps/rejected": -115.9480209350586,
|
|
"loss": 0.0058,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4864407777786255,
|
|
"rewards/margins": 24.107561111450195,
|
|
"rewards/rejected": -22.62112045288086,
|
|
"step": 3090
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 1.7614539306846998e-07,
|
|
"logits/chosen": -1.2366522550582886,
|
|
"logits/rejected": -1.1838405132293701,
|
|
"logps/chosen": -89.22175598144531,
|
|
"logps/rejected": -109.91280364990234,
|
|
"loss": 0.0038,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0529274940490723,
|
|
"rewards/margins": 23.17164421081543,
|
|
"rewards/rejected": -22.118717193603516,
|
|
"step": 3100
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"eval_logits/chosen": -1.3056086301803589,
|
|
"eval_logits/rejected": -1.2372196912765503,
|
|
"eval_logps/chosen": -87.62897491455078,
|
|
"eval_logps/rejected": -110.38069915771484,
|
|
"eval_loss": 0.007260460406541824,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.9017306566238403,
|
|
"eval_rewards/margins": 23.421215057373047,
|
|
"eval_rewards/rejected": -22.519485473632812,
|
|
"eval_runtime": 74.4757,
|
|
"eval_samples_per_second": 38.429,
|
|
"eval_steps_per_second": 2.403,
|
|
"step": 3100
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 1.7563820794590026e-07,
|
|
"logits/chosen": -1.2446845769882202,
|
|
"logits/rejected": -1.1912556886672974,
|
|
"logps/chosen": -87.75724792480469,
|
|
"logps/rejected": -117.93830871582031,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.7617045640945435,
|
|
"rewards/margins": 24.722881317138672,
|
|
"rewards/rejected": -23.961177825927734,
|
|
"step": 3110
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 1.751310228233305e-07,
|
|
"logits/chosen": -1.2533457279205322,
|
|
"logits/rejected": -1.1890594959259033,
|
|
"logps/chosen": -94.13309478759766,
|
|
"logps/rejected": -112.0062026977539,
|
|
"loss": 0.0059,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5000749826431274,
|
|
"rewards/margins": 22.833240509033203,
|
|
"rewards/rejected": -22.333166122436523,
|
|
"step": 3120
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 1.7462383770076077e-07,
|
|
"logits/chosen": -1.2367146015167236,
|
|
"logits/rejected": -1.178846001625061,
|
|
"logps/chosen": -88.32302856445312,
|
|
"logps/rejected": -115.77542877197266,
|
|
"loss": 0.0076,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.37687280774116516,
|
|
"rewards/margins": 24.140419006347656,
|
|
"rewards/rejected": -23.763545989990234,
|
|
"step": 3130
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 1.7411665257819102e-07,
|
|
"logits/chosen": -1.2372697591781616,
|
|
"logits/rejected": -1.1662713289260864,
|
|
"logps/chosen": -96.11444091796875,
|
|
"logps/rejected": -114.61024475097656,
|
|
"loss": 0.0026,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8727655410766602,
|
|
"rewards/margins": 24.095346450805664,
|
|
"rewards/rejected": -23.222579956054688,
|
|
"step": 3140
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 1.736094674556213e-07,
|
|
"logits/chosen": -1.2260363101959229,
|
|
"logits/rejected": -1.169950246810913,
|
|
"logps/chosen": -87.82176208496094,
|
|
"logps/rejected": -114.6717529296875,
|
|
"loss": 0.0014,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6339221000671387,
|
|
"rewards/margins": 23.64284324645996,
|
|
"rewards/rejected": -23.00891876220703,
|
|
"step": 3150
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 1.7310228233305156e-07,
|
|
"logits/chosen": -1.2591478824615479,
|
|
"logits/rejected": -1.2040024995803833,
|
|
"logps/chosen": -89.02254486083984,
|
|
"logps/rejected": -113.30888366699219,
|
|
"loss": 0.0065,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9198814630508423,
|
|
"rewards/margins": 23.625865936279297,
|
|
"rewards/rejected": -22.705984115600586,
|
|
"step": 3160
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 1.7259509721048184e-07,
|
|
"logits/chosen": -1.2532278299331665,
|
|
"logits/rejected": -1.2039217948913574,
|
|
"logps/chosen": -82.79835510253906,
|
|
"logps/rejected": -113.0461654663086,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.6827878952026367,
|
|
"rewards/margins": 23.46468734741211,
|
|
"rewards/rejected": -22.781898498535156,
|
|
"step": 3170
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 1.7208791208791206e-07,
|
|
"logits/chosen": -1.272351622581482,
|
|
"logits/rejected": -1.2235658168792725,
|
|
"logps/chosen": -83.52709197998047,
|
|
"logps/rejected": -115.59749603271484,
|
|
"loss": 0.0035,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9059032201766968,
|
|
"rewards/margins": 24.44829559326172,
|
|
"rewards/rejected": -23.54239273071289,
|
|
"step": 3180
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 1.7158072696534235e-07,
|
|
"logits/chosen": -1.2341419458389282,
|
|
"logits/rejected": -1.178829550743103,
|
|
"logps/chosen": -90.00852966308594,
|
|
"logps/rejected": -108.0736312866211,
|
|
"loss": 0.0024,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.2999484539031982,
|
|
"rewards/margins": 22.940885543823242,
|
|
"rewards/rejected": -21.64093589782715,
|
|
"step": 3190
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 1.710735418427726e-07,
|
|
"logits/chosen": -1.2229827642440796,
|
|
"logits/rejected": -1.1729035377502441,
|
|
"logps/chosen": -88.7994384765625,
|
|
"logps/rejected": -115.1880111694336,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.18852441012859344,
|
|
"rewards/margins": 22.59259033203125,
|
|
"rewards/rejected": -22.781112670898438,
|
|
"step": 3200
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"eval_logits/chosen": -1.3309955596923828,
|
|
"eval_logits/rejected": -1.2624685764312744,
|
|
"eval_logps/chosen": -87.6993408203125,
|
|
"eval_logps/rejected": -111.36738586425781,
|
|
"eval_loss": 0.0076772235333919525,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.866541862487793,
|
|
"eval_rewards/margins": 23.879375457763672,
|
|
"eval_rewards/rejected": -23.012832641601562,
|
|
"eval_runtime": 71.9768,
|
|
"eval_samples_per_second": 39.763,
|
|
"eval_steps_per_second": 2.487,
|
|
"step": 3200
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 1.7056635672020288e-07,
|
|
"logits/chosen": -1.2811453342437744,
|
|
"logits/rejected": -1.2151391506195068,
|
|
"logps/chosen": -92.49798583984375,
|
|
"logps/rejected": -116.91839599609375,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1861944198608398,
|
|
"rewards/margins": 25.033100128173828,
|
|
"rewards/rejected": -23.846908569335938,
|
|
"step": 3210
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 1.7005917159763313e-07,
|
|
"logits/chosen": -1.2741590738296509,
|
|
"logits/rejected": -1.2121143341064453,
|
|
"logps/chosen": -88.5379409790039,
|
|
"logps/rejected": -116.83935546875,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.4937240481376648,
|
|
"rewards/margins": 23.87834930419922,
|
|
"rewards/rejected": -23.384624481201172,
|
|
"step": 3220
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 1.695519864750634e-07,
|
|
"logits/chosen": -1.2720822095870972,
|
|
"logits/rejected": -1.2110103368759155,
|
|
"logps/chosen": -90.10042572021484,
|
|
"logps/rejected": -108.63752746582031,
|
|
"loss": 0.0071,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.890106201171875,
|
|
"rewards/margins": 22.506465911865234,
|
|
"rewards/rejected": -21.61635971069336,
|
|
"step": 3230
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 1.6904480135249364e-07,
|
|
"logits/chosen": -1.2414791584014893,
|
|
"logits/rejected": -1.1807258129119873,
|
|
"logps/chosen": -90.51213073730469,
|
|
"logps/rejected": -115.80250549316406,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1200735569000244,
|
|
"rewards/margins": 25.453067779541016,
|
|
"rewards/rejected": -24.332996368408203,
|
|
"step": 3240
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 1.6853761622992392e-07,
|
|
"logits/chosen": -1.2474098205566406,
|
|
"logits/rejected": -1.203169584274292,
|
|
"logps/chosen": -85.99417114257812,
|
|
"logps/rejected": -115.08979797363281,
|
|
"loss": 0.0028,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.23325538635253906,
|
|
"rewards/margins": 23.269723892211914,
|
|
"rewards/rejected": -23.036466598510742,
|
|
"step": 3250
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 1.6803043110735418e-07,
|
|
"logits/chosen": -1.2636672258377075,
|
|
"logits/rejected": -1.197353482246399,
|
|
"logps/chosen": -84.11290740966797,
|
|
"logps/rejected": -111.87858581542969,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2561546564102173,
|
|
"rewards/margins": 24.462989807128906,
|
|
"rewards/rejected": -23.206836700439453,
|
|
"step": 3260
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 1.6752324598478443e-07,
|
|
"logits/chosen": -1.2593055963516235,
|
|
"logits/rejected": -1.191775918006897,
|
|
"logps/chosen": -86.7191390991211,
|
|
"logps/rejected": -121.62376403808594,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.4610673785209656,
|
|
"rewards/margins": 25.324344635009766,
|
|
"rewards/rejected": -25.785411834716797,
|
|
"step": 3270
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 1.6701606086221469e-07,
|
|
"logits/chosen": -1.2484862804412842,
|
|
"logits/rejected": -1.1960498094558716,
|
|
"logps/chosen": -91.21693420410156,
|
|
"logps/rejected": -115.73272705078125,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.2348252534866333,
|
|
"rewards/margins": 23.737483978271484,
|
|
"rewards/rejected": -23.502655029296875,
|
|
"step": 3280
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 1.6650887573964497e-07,
|
|
"logits/chosen": -1.247483253479004,
|
|
"logits/rejected": -1.1900913715362549,
|
|
"logps/chosen": -89.03797149658203,
|
|
"logps/rejected": -114.04032135009766,
|
|
"loss": 0.0014,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4475841522216797,
|
|
"rewards/margins": 24.663578033447266,
|
|
"rewards/rejected": -23.215991973876953,
|
|
"step": 3290
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 1.6600169061707522e-07,
|
|
"logits/chosen": -1.3016645908355713,
|
|
"logits/rejected": -1.2365801334381104,
|
|
"logps/chosen": -94.5399398803711,
|
|
"logps/rejected": -124.09773254394531,
|
|
"loss": 0.0019,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6547442078590393,
|
|
"rewards/margins": 25.51272201538086,
|
|
"rewards/rejected": -24.857975006103516,
|
|
"step": 3300
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"eval_logits/chosen": -1.3517431020736694,
|
|
"eval_logits/rejected": -1.2811920642852783,
|
|
"eval_logps/chosen": -87.5091781616211,
|
|
"eval_logps/rejected": -111.33209991455078,
|
|
"eval_loss": 0.007596523035317659,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.9616219401359558,
|
|
"eval_rewards/margins": 23.956809997558594,
|
|
"eval_rewards/rejected": -22.995187759399414,
|
|
"eval_runtime": 59.4047,
|
|
"eval_samples_per_second": 48.178,
|
|
"eval_steps_per_second": 3.013,
|
|
"step": 3300
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 1.654945054945055e-07,
|
|
"logits/chosen": -1.2787243127822876,
|
|
"logits/rejected": -1.2220463752746582,
|
|
"logps/chosen": -85.47830963134766,
|
|
"logps/rejected": -120.3558349609375,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.1819784939289093,
|
|
"rewards/margins": 24.017372131347656,
|
|
"rewards/rejected": -23.835391998291016,
|
|
"step": 3310
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 1.6498732037193573e-07,
|
|
"logits/chosen": -1.2893264293670654,
|
|
"logits/rejected": -1.2257001399993896,
|
|
"logps/chosen": -85.7628402709961,
|
|
"logps/rejected": -111.8202896118164,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0795629024505615,
|
|
"rewards/margins": 24.557388305664062,
|
|
"rewards/rejected": -23.477825164794922,
|
|
"step": 3320
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 1.64480135249366e-07,
|
|
"logits/chosen": -1.2428733110427856,
|
|
"logits/rejected": -1.1967283487319946,
|
|
"logps/chosen": -90.3432846069336,
|
|
"logps/rejected": -121.13531494140625,
|
|
"loss": 0.0039,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1183116436004639,
|
|
"rewards/margins": 25.94301414489746,
|
|
"rewards/rejected": -24.824703216552734,
|
|
"step": 3330
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 1.6397295012679627e-07,
|
|
"logits/chosen": -1.2334848642349243,
|
|
"logits/rejected": -1.1695020198822021,
|
|
"logps/chosen": -89.45833587646484,
|
|
"logps/rejected": -115.14144134521484,
|
|
"loss": 0.007,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 0.8301811218261719,
|
|
"rewards/margins": 23.684751510620117,
|
|
"rewards/rejected": -22.854568481445312,
|
|
"step": 3340
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 1.6346576500422655e-07,
|
|
"logits/chosen": -1.261792778968811,
|
|
"logits/rejected": -1.2009265422821045,
|
|
"logps/chosen": -91.06483459472656,
|
|
"logps/rejected": -116.6786880493164,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8044708371162415,
|
|
"rewards/margins": 24.910701751708984,
|
|
"rewards/rejected": -24.10622787475586,
|
|
"step": 3350
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 1.629585798816568e-07,
|
|
"logits/chosen": -1.2624332904815674,
|
|
"logits/rejected": -1.2024790048599243,
|
|
"logps/chosen": -92.38378143310547,
|
|
"logps/rejected": -116.35567474365234,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.27976271510124207,
|
|
"rewards/margins": 23.846553802490234,
|
|
"rewards/rejected": -24.12631607055664,
|
|
"step": 3360
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 1.6245139475908705e-07,
|
|
"logits/chosen": -1.2883548736572266,
|
|
"logits/rejected": -1.2222377061843872,
|
|
"logps/chosen": -88.30838775634766,
|
|
"logps/rejected": -112.32978820800781,
|
|
"loss": 0.0028,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.10080569982528687,
|
|
"rewards/margins": 22.96854019165039,
|
|
"rewards/rejected": -22.867733001708984,
|
|
"step": 3370
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 1.619442096365173e-07,
|
|
"logits/chosen": -1.3186638355255127,
|
|
"logits/rejected": -1.245281457901001,
|
|
"logps/chosen": -87.22611999511719,
|
|
"logps/rejected": -113.5263671875,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.1867640018463135,
|
|
"rewards/margins": 25.683629989624023,
|
|
"rewards/rejected": -24.49686622619629,
|
|
"step": 3380
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 1.614370245139476e-07,
|
|
"logits/chosen": -1.2895255088806152,
|
|
"logits/rejected": -1.229871392250061,
|
|
"logps/chosen": -85.41172790527344,
|
|
"logps/rejected": -117.4696273803711,
|
|
"loss": 0.0001,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6948531866073608,
|
|
"rewards/margins": 25.749099731445312,
|
|
"rewards/rejected": -25.054248809814453,
|
|
"step": 3390
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 1.6092983939137784e-07,
|
|
"logits/chosen": -1.2895100116729736,
|
|
"logits/rejected": -1.2215838432312012,
|
|
"logps/chosen": -92.64399719238281,
|
|
"logps/rejected": -113.7071304321289,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.008798837661743,
|
|
"rewards/margins": 25.56492805480957,
|
|
"rewards/rejected": -23.556129455566406,
|
|
"step": 3400
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"eval_logits/chosen": -1.352734923362732,
|
|
"eval_logits/rejected": -1.2827072143554688,
|
|
"eval_logps/chosen": -87.6773681640625,
|
|
"eval_logps/rejected": -113.02869415283203,
|
|
"eval_loss": 0.007990003563463688,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.8775248527526855,
|
|
"eval_rewards/margins": 24.72101593017578,
|
|
"eval_rewards/rejected": -23.843490600585938,
|
|
"eval_runtime": 59.8847,
|
|
"eval_samples_per_second": 47.792,
|
|
"eval_steps_per_second": 2.989,
|
|
"step": 3400
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 1.6042265426880812e-07,
|
|
"logits/chosen": -1.2344064712524414,
|
|
"logits/rejected": -1.1821552515029907,
|
|
"logps/chosen": -87.24339294433594,
|
|
"logps/rejected": -113.21418762207031,
|
|
"loss": 0.0042,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5112111568450928,
|
|
"rewards/margins": 24.323862075805664,
|
|
"rewards/rejected": -22.81264877319336,
|
|
"step": 3410
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 1.5991546914623835e-07,
|
|
"logits/chosen": -1.2430731058120728,
|
|
"logits/rejected": -1.1846189498901367,
|
|
"logps/chosen": -86.92017364501953,
|
|
"logps/rejected": -111.88777160644531,
|
|
"loss": 0.0036,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5619323253631592,
|
|
"rewards/margins": 25.513439178466797,
|
|
"rewards/rejected": -23.951505661010742,
|
|
"step": 3420
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 1.5940828402366863e-07,
|
|
"logits/chosen": -1.2778552770614624,
|
|
"logits/rejected": -1.2257239818572998,
|
|
"logps/chosen": -89.74571228027344,
|
|
"logps/rejected": -122.40673828125,
|
|
"loss": 0.0026,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9483524560928345,
|
|
"rewards/margins": 26.498586654663086,
|
|
"rewards/rejected": -25.550235748291016,
|
|
"step": 3430
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 1.589010989010989e-07,
|
|
"logits/chosen": -1.2671973705291748,
|
|
"logits/rejected": -1.1991872787475586,
|
|
"logps/chosen": -95.77862548828125,
|
|
"logps/rejected": -120.44307708740234,
|
|
"loss": 0.0017,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5015150308609009,
|
|
"rewards/margins": 25.031169891357422,
|
|
"rewards/rejected": -23.529653549194336,
|
|
"step": 3440
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 1.5839391377852917e-07,
|
|
"logits/chosen": -1.283717393875122,
|
|
"logits/rejected": -1.2063400745391846,
|
|
"logps/chosen": -93.32177734375,
|
|
"logps/rejected": -118.2795639038086,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5961825847625732,
|
|
"rewards/margins": 25.887725830078125,
|
|
"rewards/rejected": -24.29154396057129,
|
|
"step": 3450
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 1.5788672865595942e-07,
|
|
"logits/chosen": -1.2777965068817139,
|
|
"logits/rejected": -1.2139501571655273,
|
|
"logps/chosen": -86.36488342285156,
|
|
"logps/rejected": -112.5156021118164,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4618368148803711,
|
|
"rewards/margins": 24.505504608154297,
|
|
"rewards/rejected": -24.043668746948242,
|
|
"step": 3460
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 1.5737954353338968e-07,
|
|
"logits/chosen": -1.2852230072021484,
|
|
"logits/rejected": -1.2266560792922974,
|
|
"logps/chosen": -85.028076171875,
|
|
"logps/rejected": -111.90921783447266,
|
|
"loss": 0.0054,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.6333560943603516,
|
|
"rewards/margins": 25.04153060913086,
|
|
"rewards/rejected": -23.40817642211914,
|
|
"step": 3470
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 1.5687235841081993e-07,
|
|
"logits/chosen": -1.2799792289733887,
|
|
"logits/rejected": -1.216094970703125,
|
|
"logps/chosen": -88.36021423339844,
|
|
"logps/rejected": -112.99951171875,
|
|
"loss": 0.0067,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.2346739768981934,
|
|
"rewards/margins": 25.258432388305664,
|
|
"rewards/rejected": -24.023757934570312,
|
|
"step": 3480
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 1.563651732882502e-07,
|
|
"logits/chosen": -1.2925515174865723,
|
|
"logits/rejected": -1.2144296169281006,
|
|
"logps/chosen": -93.78471374511719,
|
|
"logps/rejected": -115.75556945800781,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4354177713394165,
|
|
"rewards/margins": 26.061655044555664,
|
|
"rewards/rejected": -24.626237869262695,
|
|
"step": 3490
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 1.5585798816568047e-07,
|
|
"logits/chosen": -1.2529594898223877,
|
|
"logits/rejected": -1.2021251916885376,
|
|
"logps/chosen": -85.00303649902344,
|
|
"logps/rejected": -112.36210632324219,
|
|
"loss": 0.0003,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6849428415298462,
|
|
"rewards/margins": 23.76913070678711,
|
|
"rewards/rejected": -22.08418846130371,
|
|
"step": 3500
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"eval_logits/chosen": -1.347659945487976,
|
|
"eval_logits/rejected": -1.2767060995101929,
|
|
"eval_logps/chosen": -86.58101654052734,
|
|
"eval_logps/rejected": -111.87799835205078,
|
|
"eval_loss": 0.007571995258331299,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.4257049560546875,
|
|
"eval_rewards/margins": 24.693849563598633,
|
|
"eval_rewards/rejected": -23.268142700195312,
|
|
"eval_runtime": 61.6155,
|
|
"eval_samples_per_second": 46.449,
|
|
"eval_steps_per_second": 2.905,
|
|
"step": 3500
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 1.5535080304311072e-07,
|
|
"logits/chosen": -1.3136241436004639,
|
|
"logits/rejected": -1.2429287433624268,
|
|
"logps/chosen": -89.40177917480469,
|
|
"logps/rejected": -115.84840393066406,
|
|
"loss": 0.0039,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.674997091293335,
|
|
"rewards/margins": 24.841594696044922,
|
|
"rewards/rejected": -23.166595458984375,
|
|
"step": 3510
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 1.5484361792054097e-07,
|
|
"logits/chosen": -1.2944393157958984,
|
|
"logits/rejected": -1.2253139019012451,
|
|
"logps/chosen": -89.92081451416016,
|
|
"logps/rejected": -117.0937271118164,
|
|
"loss": 0.0036,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 0.8267130851745605,
|
|
"rewards/margins": 26.268218994140625,
|
|
"rewards/rejected": -25.441509246826172,
|
|
"step": 3520
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 1.5433643279797126e-07,
|
|
"logits/chosen": -1.2919762134552002,
|
|
"logits/rejected": -1.2337154150009155,
|
|
"logps/chosen": -83.75240325927734,
|
|
"logps/rejected": -113.67716979980469,
|
|
"loss": 0.0041,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2121895551681519,
|
|
"rewards/margins": 25.531957626342773,
|
|
"rewards/rejected": -24.319766998291016,
|
|
"step": 3530
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 1.538292476754015e-07,
|
|
"logits/chosen": -1.2944412231445312,
|
|
"logits/rejected": -1.2345895767211914,
|
|
"logps/chosen": -87.15584564208984,
|
|
"logps/rejected": -115.55804443359375,
|
|
"loss": 0.0001,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.4887280464172363,
|
|
"rewards/margins": 26.8122501373291,
|
|
"rewards/rejected": -24.323522567749023,
|
|
"step": 3540
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 1.533220625528318e-07,
|
|
"logits/chosen": -1.2804213762283325,
|
|
"logits/rejected": -1.2162938117980957,
|
|
"logps/chosen": -89.55908203125,
|
|
"logps/rejected": -120.44891357421875,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0079282522201538,
|
|
"rewards/margins": 26.581329345703125,
|
|
"rewards/rejected": -25.573402404785156,
|
|
"step": 3550
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 1.5281487743026202e-07,
|
|
"logits/chosen": -1.3119621276855469,
|
|
"logits/rejected": -1.227165937423706,
|
|
"logps/chosen": -96.74497985839844,
|
|
"logps/rejected": -121.7210464477539,
|
|
"loss": 0.0035,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5257515907287598,
|
|
"rewards/margins": 27.811294555664062,
|
|
"rewards/rejected": -26.28554344177246,
|
|
"step": 3560
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 1.523076923076923e-07,
|
|
"logits/chosen": -1.279449224472046,
|
|
"logits/rejected": -1.22541081905365,
|
|
"logps/chosen": -84.21186828613281,
|
|
"logps/rejected": -113.58978271484375,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9092862010002136,
|
|
"rewards/margins": 25.38943099975586,
|
|
"rewards/rejected": -24.480144500732422,
|
|
"step": 3570
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 1.5180050718512255e-07,
|
|
"logits/chosen": -1.2649726867675781,
|
|
"logits/rejected": -1.1972577571868896,
|
|
"logps/chosen": -91.45044708251953,
|
|
"logps/rejected": -119.5804443359375,
|
|
"loss": 0.0015,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.8337907791137695,
|
|
"rewards/margins": 28.143224716186523,
|
|
"rewards/rejected": -25.309436798095703,
|
|
"step": 3580
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 1.5129332206255283e-07,
|
|
"logits/chosen": -1.2978521585464478,
|
|
"logits/rejected": -1.252637505531311,
|
|
"logps/chosen": -90.22964477539062,
|
|
"logps/rejected": -125.02496337890625,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.8120530843734741,
|
|
"rewards/margins": 26.514413833618164,
|
|
"rewards/rejected": -24.702360153198242,
|
|
"step": 3590
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 1.507861369399831e-07,
|
|
"logits/chosen": -1.2789386510849,
|
|
"logits/rejected": -1.2158098220825195,
|
|
"logps/chosen": -90.51055908203125,
|
|
"logps/rejected": -118.33430480957031,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.5107016563415527,
|
|
"rewards/margins": 28.50537109375,
|
|
"rewards/rejected": -25.994670867919922,
|
|
"step": 3600
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"eval_logits/chosen": -1.3777239322662354,
|
|
"eval_logits/rejected": -1.3042701482772827,
|
|
"eval_logps/chosen": -87.28597259521484,
|
|
"eval_logps/rejected": -114.07413482666016,
|
|
"eval_loss": 0.007842887192964554,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.0732280015945435,
|
|
"eval_rewards/margins": 25.439434051513672,
|
|
"eval_rewards/rejected": -24.3662052154541,
|
|
"eval_runtime": 63.3589,
|
|
"eval_samples_per_second": 45.171,
|
|
"eval_steps_per_second": 2.825,
|
|
"step": 3600
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 1.5027895181741334e-07,
|
|
"logits/chosen": -1.370045781135559,
|
|
"logits/rejected": -1.283416748046875,
|
|
"logps/chosen": -90.81478118896484,
|
|
"logps/rejected": -112.16239929199219,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.0164520740509033,
|
|
"rewards/margins": 23.93118667602539,
|
|
"rewards/rejected": -22.91473388671875,
|
|
"step": 3610
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 1.497717666948436e-07,
|
|
"logits/chosen": -1.2713721990585327,
|
|
"logits/rejected": -1.216667890548706,
|
|
"logps/chosen": -84.43434143066406,
|
|
"logps/rejected": -113.650146484375,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.3617621064186096,
|
|
"rewards/margins": 24.290042877197266,
|
|
"rewards/rejected": -23.92828369140625,
|
|
"step": 3620
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 1.4926458157227388e-07,
|
|
"logits/chosen": -1.316611647605896,
|
|
"logits/rejected": -1.2415025234222412,
|
|
"logps/chosen": -88.00056457519531,
|
|
"logps/rejected": -118.13746643066406,
|
|
"loss": 0.0058,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.7822587490081787,
|
|
"rewards/margins": 27.109432220458984,
|
|
"rewards/rejected": -25.327173233032227,
|
|
"step": 3630
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 1.4875739644970413e-07,
|
|
"logits/chosen": -1.2669055461883545,
|
|
"logits/rejected": -1.198326587677002,
|
|
"logps/chosen": -92.14048767089844,
|
|
"logps/rejected": -115.46183776855469,
|
|
"loss": 0.0052,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0600082874298096,
|
|
"rewards/margins": 25.716211318969727,
|
|
"rewards/rejected": -24.656200408935547,
|
|
"step": 3640
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 1.4825021132713439e-07,
|
|
"logits/chosen": -1.2855212688446045,
|
|
"logits/rejected": -1.2362643480300903,
|
|
"logps/chosen": -85.3121337890625,
|
|
"logps/rejected": -116.80509948730469,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.1400699615478516,
|
|
"rewards/margins": 25.843036651611328,
|
|
"rewards/rejected": -23.70296859741211,
|
|
"step": 3650
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 1.4774302620456467e-07,
|
|
"logits/chosen": -1.2816253900527954,
|
|
"logits/rejected": -1.2244489192962646,
|
|
"logps/chosen": -86.08045959472656,
|
|
"logps/rejected": -120.78515625,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.7160287499427795,
|
|
"rewards/margins": 25.913681030273438,
|
|
"rewards/rejected": -25.197650909423828,
|
|
"step": 3660
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 1.4723584108199492e-07,
|
|
"logits/chosen": -1.3148205280303955,
|
|
"logits/rejected": -1.2522845268249512,
|
|
"logps/chosen": -88.29036712646484,
|
|
"logps/rejected": -117.13260650634766,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6190974712371826,
|
|
"rewards/margins": 26.740299224853516,
|
|
"rewards/rejected": -25.121198654174805,
|
|
"step": 3670
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 1.4672865595942518e-07,
|
|
"logits/chosen": -1.3006482124328613,
|
|
"logits/rejected": -1.2300279140472412,
|
|
"logps/chosen": -87.94609069824219,
|
|
"logps/rejected": -120.0467300415039,
|
|
"loss": 0.0049,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.505680799484253,
|
|
"rewards/margins": 26.633773803710938,
|
|
"rewards/rejected": -24.128095626831055,
|
|
"step": 3680
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 1.4622147083685546e-07,
|
|
"logits/chosen": -1.3202269077301025,
|
|
"logits/rejected": -1.252046823501587,
|
|
"logps/chosen": -95.1660385131836,
|
|
"logps/rejected": -121.23722839355469,
|
|
"loss": 0.0057,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.524958848953247,
|
|
"rewards/margins": 27.411617279052734,
|
|
"rewards/rejected": -25.886661529541016,
|
|
"step": 3690
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 1.457142857142857e-07,
|
|
"logits/chosen": -1.2965933084487915,
|
|
"logits/rejected": -1.2345640659332275,
|
|
"logps/chosen": -83.83021545410156,
|
|
"logps/rejected": -117.0789794921875,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.2809300422668457,
|
|
"rewards/margins": 26.77134132385254,
|
|
"rewards/rejected": -24.49040985107422,
|
|
"step": 3700
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"eval_logits/chosen": -1.3869707584381104,
|
|
"eval_logits/rejected": -1.3094760179519653,
|
|
"eval_logps/chosen": -86.7977066040039,
|
|
"eval_logps/rejected": -115.47274017333984,
|
|
"eval_loss": 0.008259255439043045,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.3173613548278809,
|
|
"eval_rewards/margins": 26.382875442504883,
|
|
"eval_rewards/rejected": -25.065513610839844,
|
|
"eval_runtime": 63.3424,
|
|
"eval_samples_per_second": 45.183,
|
|
"eval_steps_per_second": 2.826,
|
|
"step": 3700
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 1.4520710059171596e-07,
|
|
"logits/chosen": -1.2985725402832031,
|
|
"logits/rejected": -1.2297552824020386,
|
|
"logps/chosen": -90.36856079101562,
|
|
"logps/rejected": -116.03338623046875,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.916562557220459,
|
|
"rewards/margins": 27.512588500976562,
|
|
"rewards/rejected": -24.596027374267578,
|
|
"step": 3710
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 1.4469991546914622e-07,
|
|
"logits/chosen": -1.3176157474517822,
|
|
"logits/rejected": -1.247491478919983,
|
|
"logps/chosen": -85.81680297851562,
|
|
"logps/rejected": -112.21038818359375,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.835550308227539,
|
|
"rewards/margins": 26.416223526000977,
|
|
"rewards/rejected": -24.580673217773438,
|
|
"step": 3720
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 1.441927303465765e-07,
|
|
"logits/chosen": -1.3307037353515625,
|
|
"logits/rejected": -1.2499626874923706,
|
|
"logps/chosen": -90.99395751953125,
|
|
"logps/rejected": -117.75846099853516,
|
|
"loss": 0.0055,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8873962163925171,
|
|
"rewards/margins": 25.621952056884766,
|
|
"rewards/rejected": -24.734556198120117,
|
|
"step": 3730
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 1.4368554522400675e-07,
|
|
"logits/chosen": -1.2795710563659668,
|
|
"logits/rejected": -1.2092183828353882,
|
|
"logps/chosen": -90.55952453613281,
|
|
"logps/rejected": -121.86543273925781,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.7004776000976562,
|
|
"rewards/margins": 29.03326988220215,
|
|
"rewards/rejected": -26.332794189453125,
|
|
"step": 3740
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 1.43178360101437e-07,
|
|
"logits/chosen": -1.3224431276321411,
|
|
"logits/rejected": -1.2512967586517334,
|
|
"logps/chosen": -84.958984375,
|
|
"logps/rejected": -113.85465240478516,
|
|
"loss": 0.0016,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.3535866737365723,
|
|
"rewards/margins": 26.663818359375,
|
|
"rewards/rejected": -24.310230255126953,
|
|
"step": 3750
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 1.426711749788673e-07,
|
|
"logits/chosen": -1.3394898176193237,
|
|
"logits/rejected": -1.2797186374664307,
|
|
"logps/chosen": -83.2117919921875,
|
|
"logps/rejected": -118.57608795166016,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2878071069717407,
|
|
"rewards/margins": 26.771793365478516,
|
|
"rewards/rejected": -25.483983993530273,
|
|
"step": 3760
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 1.4216398985629754e-07,
|
|
"logits/chosen": -1.2791587114334106,
|
|
"logits/rejected": -1.2220796346664429,
|
|
"logps/chosen": -87.02328491210938,
|
|
"logps/rejected": -123.32960510253906,
|
|
"loss": 0.0007,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.185002088546753,
|
|
"rewards/margins": 27.750438690185547,
|
|
"rewards/rejected": -25.565439224243164,
|
|
"step": 3770
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 1.416568047337278e-07,
|
|
"logits/chosen": -1.3284387588500977,
|
|
"logits/rejected": -1.2588412761688232,
|
|
"logps/chosen": -88.56816101074219,
|
|
"logps/rejected": -118.0064468383789,
|
|
"loss": 0.0,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4277937412261963,
|
|
"rewards/margins": 26.984973907470703,
|
|
"rewards/rejected": -25.557178497314453,
|
|
"step": 3780
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 1.4114961961115805e-07,
|
|
"logits/chosen": -1.3133352994918823,
|
|
"logits/rejected": -1.248510718345642,
|
|
"logps/chosen": -92.50769805908203,
|
|
"logps/rejected": -114.99954986572266,
|
|
"loss": 0.0046,
|
|
"rewards/accuracies": 0.9624999761581421,
|
|
"rewards/chosen": 1.1458348035812378,
|
|
"rewards/margins": 24.03805160522461,
|
|
"rewards/rejected": -22.89221954345703,
|
|
"step": 3790
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 1.4064243448858833e-07,
|
|
"logits/chosen": -1.302869200706482,
|
|
"logits/rejected": -1.2390058040618896,
|
|
"logps/chosen": -82.003662109375,
|
|
"logps/rejected": -112.65030670166016,
|
|
"loss": 0.0098,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.544621706008911,
|
|
"rewards/margins": 26.4320068359375,
|
|
"rewards/rejected": -23.887386322021484,
|
|
"step": 3800
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"eval_logits/chosen": -1.3750476837158203,
|
|
"eval_logits/rejected": -1.29920494556427,
|
|
"eval_logps/chosen": -85.6637191772461,
|
|
"eval_logps/rejected": -111.03407287597656,
|
|
"eval_loss": 0.007216113153845072,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.884352207183838,
|
|
"eval_rewards/margins": 24.73053741455078,
|
|
"eval_rewards/rejected": -22.84618377685547,
|
|
"eval_runtime": 62.1547,
|
|
"eval_samples_per_second": 46.046,
|
|
"eval_steps_per_second": 2.88,
|
|
"step": 3800
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 1.401352493660186e-07,
|
|
"logits/chosen": -1.2911725044250488,
|
|
"logits/rejected": -1.2250169515609741,
|
|
"logps/chosen": -84.48226928710938,
|
|
"logps/rejected": -108.87812805175781,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.1833953857421875,
|
|
"rewards/margins": 23.128978729248047,
|
|
"rewards/rejected": -21.94558334350586,
|
|
"step": 3810
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 1.3962806424344884e-07,
|
|
"logits/chosen": -1.320176362991333,
|
|
"logits/rejected": -1.2417948246002197,
|
|
"logps/chosen": -87.6998519897461,
|
|
"logps/rejected": -110.061279296875,
|
|
"loss": 0.0007,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.8369624614715576,
|
|
"rewards/margins": 24.469478607177734,
|
|
"rewards/rejected": -22.63251495361328,
|
|
"step": 3820
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 1.3912087912087912e-07,
|
|
"logits/chosen": -1.3057355880737305,
|
|
"logits/rejected": -1.23427414894104,
|
|
"logps/chosen": -84.95487976074219,
|
|
"logps/rejected": -113.34101867675781,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.752861976623535,
|
|
"rewards/margins": 27.136266708374023,
|
|
"rewards/rejected": -24.383405685424805,
|
|
"step": 3830
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 1.3861369399830938e-07,
|
|
"logits/chosen": -1.3030786514282227,
|
|
"logits/rejected": -1.236664056777954,
|
|
"logps/chosen": -83.6253890991211,
|
|
"logps/rejected": -113.64739990234375,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3076999187469482,
|
|
"rewards/margins": 25.728729248046875,
|
|
"rewards/rejected": -24.421030044555664,
|
|
"step": 3840
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 1.3810650887573963e-07,
|
|
"logits/chosen": -1.30088210105896,
|
|
"logits/rejected": -1.2418001890182495,
|
|
"logps/chosen": -85.37004852294922,
|
|
"logps/rejected": -113.1008071899414,
|
|
"loss": 0.006,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.105379104614258,
|
|
"rewards/margins": 25.393503189086914,
|
|
"rewards/rejected": -23.288122177124023,
|
|
"step": 3850
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 1.3759932375316989e-07,
|
|
"logits/chosen": -1.2641175985336304,
|
|
"logits/rejected": -1.20892333984375,
|
|
"logps/chosen": -87.78987121582031,
|
|
"logps/rejected": -111.98150634765625,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.467369556427002,
|
|
"rewards/margins": 25.168516159057617,
|
|
"rewards/rejected": -22.70115089416504,
|
|
"step": 3860
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 1.3709213863060017e-07,
|
|
"logits/chosen": -1.328049898147583,
|
|
"logits/rejected": -1.2603827714920044,
|
|
"logps/chosen": -81.99150085449219,
|
|
"logps/rejected": -113.47479248046875,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.4597467184066772,
|
|
"rewards/margins": 24.159908294677734,
|
|
"rewards/rejected": -22.70016098022461,
|
|
"step": 3870
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 1.3658495350803042e-07,
|
|
"logits/chosen": -1.2881757020950317,
|
|
"logits/rejected": -1.219227910041809,
|
|
"logps/chosen": -89.5770034790039,
|
|
"logps/rejected": -117.85916900634766,
|
|
"loss": 0.0024,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.6704366207122803,
|
|
"rewards/margins": 26.334217071533203,
|
|
"rewards/rejected": -24.663782119750977,
|
|
"step": 3880
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 1.3607776838546067e-07,
|
|
"logits/chosen": -1.2890180349349976,
|
|
"logits/rejected": -1.2281222343444824,
|
|
"logps/chosen": -87.47191619873047,
|
|
"logps/rejected": -117.46723937988281,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.8343310356140137,
|
|
"rewards/margins": 27.57296371459961,
|
|
"rewards/rejected": -24.738632202148438,
|
|
"step": 3890
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 1.3557058326289096e-07,
|
|
"logits/chosen": -1.2887780666351318,
|
|
"logits/rejected": -1.2126020193099976,
|
|
"logps/chosen": -92.53477478027344,
|
|
"logps/rejected": -118.53013610839844,
|
|
"loss": 0.004,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.5731879472732544,
|
|
"rewards/margins": 27.09688949584961,
|
|
"rewards/rejected": -25.52370262145996,
|
|
"step": 3900
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"eval_logits/chosen": -1.3683326244354248,
|
|
"eval_logits/rejected": -1.2940618991851807,
|
|
"eval_logps/chosen": -86.10755920410156,
|
|
"eval_logps/rejected": -113.04136657714844,
|
|
"eval_loss": 0.00731794023886323,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.6624343395233154,
|
|
"eval_rewards/margins": 25.512258529663086,
|
|
"eval_rewards/rejected": -23.849822998046875,
|
|
"eval_runtime": 60.9365,
|
|
"eval_samples_per_second": 46.967,
|
|
"eval_steps_per_second": 2.937,
|
|
"step": 3900
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 1.350633981403212e-07,
|
|
"logits/chosen": -1.2927218675613403,
|
|
"logits/rejected": -1.2228248119354248,
|
|
"logps/chosen": -83.60839080810547,
|
|
"logps/rejected": -112.75862121582031,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.901503324508667,
|
|
"rewards/margins": 24.796140670776367,
|
|
"rewards/rejected": -22.894638061523438,
|
|
"step": 3910
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 1.3455621301775146e-07,
|
|
"logits/chosen": -1.2360032796859741,
|
|
"logits/rejected": -1.1930327415466309,
|
|
"logps/chosen": -85.62734985351562,
|
|
"logps/rejected": -117.69222259521484,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.328904390335083,
|
|
"rewards/margins": 26.03916358947754,
|
|
"rewards/rejected": -24.710262298583984,
|
|
"step": 3920
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 1.3404902789518174e-07,
|
|
"logits/chosen": -1.299557089805603,
|
|
"logits/rejected": -1.2226934432983398,
|
|
"logps/chosen": -95.00071716308594,
|
|
"logps/rejected": -120.6982650756836,
|
|
"loss": 0.0046,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4814507961273193,
|
|
"rewards/margins": 25.5311222076416,
|
|
"rewards/rejected": -24.049671173095703,
|
|
"step": 3930
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 1.33541842772612e-07,
|
|
"logits/chosen": -1.2686429023742676,
|
|
"logits/rejected": -1.2006398439407349,
|
|
"logps/chosen": -87.60845184326172,
|
|
"logps/rejected": -113.25736999511719,
|
|
"loss": 0.0067,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.330335021018982,
|
|
"rewards/margins": 24.519290924072266,
|
|
"rewards/rejected": -23.188955307006836,
|
|
"step": 3940
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 1.3303465765004225e-07,
|
|
"logits/chosen": -1.3029606342315674,
|
|
"logits/rejected": -1.2397031784057617,
|
|
"logps/chosen": -86.67044067382812,
|
|
"logps/rejected": -116.4280776977539,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.413769245147705,
|
|
"rewards/margins": 26.642818450927734,
|
|
"rewards/rejected": -24.22905158996582,
|
|
"step": 3950
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 1.325274725274725e-07,
|
|
"logits/chosen": -1.288098931312561,
|
|
"logits/rejected": -1.22249436378479,
|
|
"logps/chosen": -84.44660186767578,
|
|
"logps/rejected": -116.7946548461914,
|
|
"loss": 0.0016,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.941779375076294,
|
|
"rewards/margins": 26.12166404724121,
|
|
"rewards/rejected": -24.17988395690918,
|
|
"step": 3960
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 1.320202874049028e-07,
|
|
"logits/chosen": -1.2721738815307617,
|
|
"logits/rejected": -1.2212003469467163,
|
|
"logps/chosen": -86.49989318847656,
|
|
"logps/rejected": -116.10292053222656,
|
|
"loss": 0.002,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.964007019996643,
|
|
"rewards/margins": 25.622446060180664,
|
|
"rewards/rejected": -23.658437728881836,
|
|
"step": 3970
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 1.3151310228233304e-07,
|
|
"logits/chosen": -1.2734956741333008,
|
|
"logits/rejected": -1.2074804306030273,
|
|
"logps/chosen": -87.59144592285156,
|
|
"logps/rejected": -116.40534973144531,
|
|
"loss": 0.0056,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6690824031829834,
|
|
"rewards/margins": 26.692928314208984,
|
|
"rewards/rejected": -25.02384376525879,
|
|
"step": 3980
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 1.310059171597633e-07,
|
|
"logits/chosen": -1.3229997158050537,
|
|
"logits/rejected": -1.2673537731170654,
|
|
"logps/chosen": -88.17668151855469,
|
|
"logps/rejected": -119.3404769897461,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.873644232749939,
|
|
"rewards/margins": 26.401302337646484,
|
|
"rewards/rejected": -24.527660369873047,
|
|
"step": 3990
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 1.3049873203719358e-07,
|
|
"logits/chosen": -1.3080793619155884,
|
|
"logits/rejected": -1.2402544021606445,
|
|
"logps/chosen": -90.626953125,
|
|
"logps/rejected": -115.0836410522461,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6485347747802734,
|
|
"rewards/margins": 25.51757049560547,
|
|
"rewards/rejected": -23.86903953552246,
|
|
"step": 4000
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"eval_logits/chosen": -1.388069987297058,
|
|
"eval_logits/rejected": -1.311712622642517,
|
|
"eval_logps/chosen": -86.8644790649414,
|
|
"eval_logps/rejected": -114.29080963134766,
|
|
"eval_loss": 0.007556203752756119,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.283972144126892,
|
|
"eval_rewards/margins": 25.75852394104004,
|
|
"eval_rewards/rejected": -24.474552154541016,
|
|
"eval_runtime": 61.4383,
|
|
"eval_samples_per_second": 46.583,
|
|
"eval_steps_per_second": 2.913,
|
|
"step": 4000
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 1.2999154691462383e-07,
|
|
"logits/chosen": -1.2937663793563843,
|
|
"logits/rejected": -1.2333722114562988,
|
|
"logps/chosen": -93.04609680175781,
|
|
"logps/rejected": -114.9658203125,
|
|
"loss": 0.0066,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.066020131111145,
|
|
"rewards/margins": 25.072368621826172,
|
|
"rewards/rejected": -24.00634765625,
|
|
"step": 4010
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 1.2948436179205409e-07,
|
|
"logits/chosen": -1.3368138074874878,
|
|
"logits/rejected": -1.2809922695159912,
|
|
"logps/chosen": -82.21131134033203,
|
|
"logps/rejected": -116.16499328613281,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2117726802825928,
|
|
"rewards/margins": 25.279598236083984,
|
|
"rewards/rejected": -24.067827224731445,
|
|
"step": 4020
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 1.2897717666948434e-07,
|
|
"logits/chosen": -1.275615930557251,
|
|
"logits/rejected": -1.2142200469970703,
|
|
"logps/chosen": -93.66702270507812,
|
|
"logps/rejected": -120.89031982421875,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.48200541734695435,
|
|
"rewards/margins": 24.838642120361328,
|
|
"rewards/rejected": -24.356637954711914,
|
|
"step": 4030
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 1.2846999154691462e-07,
|
|
"logits/chosen": -1.3146488666534424,
|
|
"logits/rejected": -1.2405660152435303,
|
|
"logps/chosen": -96.49334716796875,
|
|
"logps/rejected": -119.23211669921875,
|
|
"loss": 0.0001,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.9090890884399414,
|
|
"rewards/margins": 27.04913330078125,
|
|
"rewards/rejected": -25.140043258666992,
|
|
"step": 4040
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 1.2796280642434488e-07,
|
|
"logits/chosen": -1.307680606842041,
|
|
"logits/rejected": -1.2320573329925537,
|
|
"logps/chosen": -95.1331787109375,
|
|
"logps/rejected": -114.7768325805664,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.5242226123809814,
|
|
"rewards/margins": 27.13631820678711,
|
|
"rewards/rejected": -24.61209487915039,
|
|
"step": 4050
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 1.2745562130177513e-07,
|
|
"logits/chosen": -1.3010895252227783,
|
|
"logits/rejected": -1.2413908243179321,
|
|
"logps/chosen": -93.52859497070312,
|
|
"logps/rejected": -118.62088775634766,
|
|
"loss": 0.0031,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5129265785217285,
|
|
"rewards/margins": 25.957473754882812,
|
|
"rewards/rejected": -24.444549560546875,
|
|
"step": 4060
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 1.269484361792054e-07,
|
|
"logits/chosen": -1.2945194244384766,
|
|
"logits/rejected": -1.2368277311325073,
|
|
"logps/chosen": -90.33158874511719,
|
|
"logps/rejected": -120.31858825683594,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.025995206087827682,
|
|
"rewards/margins": 24.914783477783203,
|
|
"rewards/rejected": -24.940778732299805,
|
|
"step": 4070
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 1.2644125105663566e-07,
|
|
"logits/chosen": -1.298654317855835,
|
|
"logits/rejected": -1.2463370561599731,
|
|
"logps/chosen": -85.01954650878906,
|
|
"logps/rejected": -115.01509094238281,
|
|
"loss": 0.0028,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.912619411945343,
|
|
"rewards/margins": 25.490772247314453,
|
|
"rewards/rejected": -24.57815170288086,
|
|
"step": 4080
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 1.2593406593406592e-07,
|
|
"logits/chosen": -1.282814383506775,
|
|
"logits/rejected": -1.2446370124816895,
|
|
"logps/chosen": -85.000732421875,
|
|
"logps/rejected": -119.21089935302734,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7353527545928955,
|
|
"rewards/margins": 26.011398315429688,
|
|
"rewards/rejected": -24.276042938232422,
|
|
"step": 4090
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 1.2542688081149617e-07,
|
|
"logits/chosen": -1.2971045970916748,
|
|
"logits/rejected": -1.2302577495574951,
|
|
"logps/chosen": -90.2288589477539,
|
|
"logps/rejected": -119.21665954589844,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2725168466567993,
|
|
"rewards/margins": 25.95538902282715,
|
|
"rewards/rejected": -24.68286895751953,
|
|
"step": 4100
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"eval_logits/chosen": -1.3849560022354126,
|
|
"eval_logits/rejected": -1.30857515335083,
|
|
"eval_logps/chosen": -86.99442291259766,
|
|
"eval_logps/rejected": -114.5916748046875,
|
|
"eval_loss": 0.007762798108160496,
|
|
"eval_rewards/accuracies": 0.994413435459137,
|
|
"eval_rewards/chosen": 1.2190001010894775,
|
|
"eval_rewards/margins": 25.843978881835938,
|
|
"eval_rewards/rejected": -24.624980926513672,
|
|
"eval_runtime": 60.6192,
|
|
"eval_samples_per_second": 47.213,
|
|
"eval_steps_per_second": 2.953,
|
|
"step": 4100
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 1.2491969568892645e-07,
|
|
"logits/chosen": -1.2938053607940674,
|
|
"logits/rejected": -1.2201104164123535,
|
|
"logps/chosen": -91.01365661621094,
|
|
"logps/rejected": -116.42720031738281,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.568861961364746,
|
|
"rewards/margins": 26.685710906982422,
|
|
"rewards/rejected": -24.116849899291992,
|
|
"step": 4110
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 1.244125105663567e-07,
|
|
"logits/chosen": -1.3405344486236572,
|
|
"logits/rejected": -1.2672364711761475,
|
|
"logps/chosen": -91.08586120605469,
|
|
"logps/rejected": -121.0147705078125,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.5391464233398438,
|
|
"rewards/margins": 27.237112045288086,
|
|
"rewards/rejected": -25.69796371459961,
|
|
"step": 4120
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 1.2390532544378696e-07,
|
|
"logits/chosen": -1.2898027896881104,
|
|
"logits/rejected": -1.2514622211456299,
|
|
"logps/chosen": -82.71153259277344,
|
|
"logps/rejected": -122.4542236328125,
|
|
"loss": 0.0055,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.03917119652032852,
|
|
"rewards/margins": 26.051074981689453,
|
|
"rewards/rejected": -26.01190185546875,
|
|
"step": 4130
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 1.2339814032121724e-07,
|
|
"logits/chosen": -1.3003642559051514,
|
|
"logits/rejected": -1.2193737030029297,
|
|
"logps/chosen": -92.31513214111328,
|
|
"logps/rejected": -114.4735336303711,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.3771212100982666,
|
|
"rewards/margins": 25.401681900024414,
|
|
"rewards/rejected": -24.02456283569336,
|
|
"step": 4140
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 1.228909551986475e-07,
|
|
"logits/chosen": -1.3193917274475098,
|
|
"logits/rejected": -1.2444701194763184,
|
|
"logps/chosen": -88.08500671386719,
|
|
"logps/rejected": -111.34246826171875,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7157142162322998,
|
|
"rewards/margins": 25.313156127929688,
|
|
"rewards/rejected": -23.59743881225586,
|
|
"step": 4150
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 1.2238377007607775e-07,
|
|
"logits/chosen": -1.2875083684921265,
|
|
"logits/rejected": -1.2169303894042969,
|
|
"logps/chosen": -90.44905090332031,
|
|
"logps/rejected": -113.9192886352539,
|
|
"loss": 0.0016,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2316787242889404,
|
|
"rewards/margins": 25.797632217407227,
|
|
"rewards/rejected": -24.56595230102539,
|
|
"step": 4160
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 1.2187658495350803e-07,
|
|
"logits/chosen": -1.319364309310913,
|
|
"logits/rejected": -1.2405459880828857,
|
|
"logps/chosen": -94.96125793457031,
|
|
"logps/rejected": -118.59950256347656,
|
|
"loss": 0.0024,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.9828643798828125,
|
|
"rewards/margins": 25.974903106689453,
|
|
"rewards/rejected": -23.99203872680664,
|
|
"step": 4170
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 1.213693998309383e-07,
|
|
"logits/chosen": -1.3195867538452148,
|
|
"logits/rejected": -1.2445495128631592,
|
|
"logps/chosen": -90.23202514648438,
|
|
"logps/rejected": -114.99327087402344,
|
|
"loss": 0.0049,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.1389697790145874,
|
|
"rewards/margins": 25.39142608642578,
|
|
"rewards/rejected": -25.252456665039062,
|
|
"step": 4180
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 1.2086221470836854e-07,
|
|
"logits/chosen": -1.3217099905014038,
|
|
"logits/rejected": -1.248329997062683,
|
|
"logps/chosen": -89.2808609008789,
|
|
"logps/rejected": -123.01396179199219,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.657078742980957,
|
|
"rewards/margins": 27.067142486572266,
|
|
"rewards/rejected": -26.410064697265625,
|
|
"step": 4190
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 1.203550295857988e-07,
|
|
"logits/chosen": -1.324310064315796,
|
|
"logits/rejected": -1.257927417755127,
|
|
"logps/chosen": -86.30921173095703,
|
|
"logps/rejected": -114.84613037109375,
|
|
"loss": 0.0006,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0646955966949463,
|
|
"rewards/margins": 25.24355697631836,
|
|
"rewards/rejected": -24.178863525390625,
|
|
"step": 4200
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"eval_logits/chosen": -1.4128456115722656,
|
|
"eval_logits/rejected": -1.331875205039978,
|
|
"eval_logps/chosen": -87.6678466796875,
|
|
"eval_logps/rejected": -114.6224594116211,
|
|
"eval_loss": 0.007396237924695015,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.8822879791259766,
|
|
"eval_rewards/margins": 25.522655487060547,
|
|
"eval_rewards/rejected": -24.640369415283203,
|
|
"eval_runtime": 63.2696,
|
|
"eval_samples_per_second": 45.235,
|
|
"eval_steps_per_second": 2.829,
|
|
"step": 4200
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 1.1984784446322908e-07,
|
|
"logits/chosen": -1.3565280437469482,
|
|
"logits/rejected": -1.2859523296356201,
|
|
"logps/chosen": -85.88087463378906,
|
|
"logps/rejected": -116.72319030761719,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.10283108055591583,
|
|
"rewards/margins": 24.91754150390625,
|
|
"rewards/rejected": -25.020374298095703,
|
|
"step": 4210
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 1.1934065934065933e-07,
|
|
"logits/chosen": -1.3157243728637695,
|
|
"logits/rejected": -1.253631830215454,
|
|
"logps/chosen": -88.73226165771484,
|
|
"logps/rejected": -118.5372543334961,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.7852961421012878,
|
|
"rewards/margins": 26.198617935180664,
|
|
"rewards/rejected": -25.41332244873047,
|
|
"step": 4220
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 1.188334742180896e-07,
|
|
"logits/chosen": -1.3382046222686768,
|
|
"logits/rejected": -1.2745530605316162,
|
|
"logps/chosen": -87.59422302246094,
|
|
"logps/rejected": -120.77787017822266,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4094334542751312,
|
|
"rewards/margins": 26.14259910583496,
|
|
"rewards/rejected": -25.733165740966797,
|
|
"step": 4230
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 1.1832628909551987e-07,
|
|
"logits/chosen": -1.3534907102584839,
|
|
"logits/rejected": -1.2823253870010376,
|
|
"logps/chosen": -84.35427856445312,
|
|
"logps/rejected": -119.73152923583984,
|
|
"loss": 0.0029,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.9467264413833618,
|
|
"rewards/margins": 28.15091323852539,
|
|
"rewards/rejected": -26.20418357849121,
|
|
"step": 4240
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 1.1781910397295012e-07,
|
|
"logits/chosen": -1.325655221939087,
|
|
"logits/rejected": -1.2503167390823364,
|
|
"logps/chosen": -89.13822937011719,
|
|
"logps/rejected": -122.24861145019531,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.596404790878296,
|
|
"rewards/margins": 26.64560317993164,
|
|
"rewards/rejected": -25.0491943359375,
|
|
"step": 4250
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 1.1731191885038039e-07,
|
|
"logits/chosen": -1.2953070402145386,
|
|
"logits/rejected": -1.2415964603424072,
|
|
"logps/chosen": -89.16068267822266,
|
|
"logps/rejected": -121.19921875,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6085556745529175,
|
|
"rewards/margins": 25.759998321533203,
|
|
"rewards/rejected": -25.15144157409668,
|
|
"step": 4260
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 1.1680473372781064e-07,
|
|
"logits/chosen": -1.314546823501587,
|
|
"logits/rejected": -1.2527072429656982,
|
|
"logps/chosen": -89.053466796875,
|
|
"logps/rejected": -118.50617980957031,
|
|
"loss": 0.0079,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 0.5143841505050659,
|
|
"rewards/margins": 25.380834579467773,
|
|
"rewards/rejected": -24.866455078125,
|
|
"step": 4270
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 1.1629754860524091e-07,
|
|
"logits/chosen": -1.29738450050354,
|
|
"logits/rejected": -1.256644606590271,
|
|
"logps/chosen": -91.60551452636719,
|
|
"logps/rejected": -119.3724136352539,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.4654719829559326,
|
|
"rewards/margins": 25.70328140258789,
|
|
"rewards/rejected": -24.237810134887695,
|
|
"step": 4280
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 1.1579036348267118e-07,
|
|
"logits/chosen": -1.3026115894317627,
|
|
"logits/rejected": -1.2353935241699219,
|
|
"logps/chosen": -88.94530487060547,
|
|
"logps/rejected": -110.451904296875,
|
|
"loss": 0.0014,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.12701252102851868,
|
|
"rewards/margins": 23.56112289428711,
|
|
"rewards/rejected": -23.43410873413086,
|
|
"step": 4290
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 1.1528317836010143e-07,
|
|
"logits/chosen": -1.311436414718628,
|
|
"logits/rejected": -1.2566239833831787,
|
|
"logps/chosen": -85.2987289428711,
|
|
"logps/rejected": -115.095947265625,
|
|
"loss": 0.0046,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.6091933846473694,
|
|
"rewards/margins": 25.31754493713379,
|
|
"rewards/rejected": -24.70834732055664,
|
|
"step": 4300
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"eval_logits/chosen": -1.4097875356674194,
|
|
"eval_logits/rejected": -1.3324825763702393,
|
|
"eval_logps/chosen": -87.3016357421875,
|
|
"eval_logps/rejected": -111.95087432861328,
|
|
"eval_loss": 0.006749654188752174,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.0653921365737915,
|
|
"eval_rewards/margins": 24.369966506958008,
|
|
"eval_rewards/rejected": -23.30457305908203,
|
|
"eval_runtime": 121.9318,
|
|
"eval_samples_per_second": 23.472,
|
|
"eval_steps_per_second": 1.468,
|
|
"step": 4300
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 1.147759932375317e-07,
|
|
"logits/chosen": -1.3246426582336426,
|
|
"logits/rejected": -1.2590562105178833,
|
|
"logps/chosen": -88.42132568359375,
|
|
"logps/rejected": -115.56095886230469,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7354822158813477,
|
|
"rewards/margins": 25.661792755126953,
|
|
"rewards/rejected": -23.926311492919922,
|
|
"step": 4310
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 1.1426880811496195e-07,
|
|
"logits/chosen": -1.3025567531585693,
|
|
"logits/rejected": -1.227550745010376,
|
|
"logps/chosen": -93.46360778808594,
|
|
"logps/rejected": -112.05072021484375,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7918212413787842,
|
|
"rewards/margins": 25.10645866394043,
|
|
"rewards/rejected": -23.314638137817383,
|
|
"step": 4320
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 1.1376162299239222e-07,
|
|
"logits/chosen": -1.3441492319107056,
|
|
"logits/rejected": -1.2833201885223389,
|
|
"logps/chosen": -87.40711975097656,
|
|
"logps/rejected": -111.749267578125,
|
|
"loss": 0.0073,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4917819499969482,
|
|
"rewards/margins": 24.15626335144043,
|
|
"rewards/rejected": -22.664485931396484,
|
|
"step": 4330
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 1.1325443786982247e-07,
|
|
"logits/chosen": -1.2675365209579468,
|
|
"logits/rejected": -1.2221524715423584,
|
|
"logps/chosen": -87.13706970214844,
|
|
"logps/rejected": -113.51161193847656,
|
|
"loss": 0.005,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4677601754665375,
|
|
"rewards/margins": 23.889986038208008,
|
|
"rewards/rejected": -23.42222785949707,
|
|
"step": 4340
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 1.1274725274725274e-07,
|
|
"logits/chosen": -1.2820374965667725,
|
|
"logits/rejected": -1.224862813949585,
|
|
"logps/chosen": -86.58576965332031,
|
|
"logps/rejected": -113.4376220703125,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2789913415908813,
|
|
"rewards/margins": 24.847978591918945,
|
|
"rewards/rejected": -23.568988800048828,
|
|
"step": 4350
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 1.1224006762468301e-07,
|
|
"logits/chosen": -1.3038794994354248,
|
|
"logits/rejected": -1.2235076427459717,
|
|
"logps/chosen": -91.44725036621094,
|
|
"logps/rejected": -112.098876953125,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9584652185440063,
|
|
"rewards/margins": 24.83078384399414,
|
|
"rewards/rejected": -23.872318267822266,
|
|
"step": 4360
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 1.1173288250211326e-07,
|
|
"logits/chosen": -1.3245998620986938,
|
|
"logits/rejected": -1.2639939785003662,
|
|
"logps/chosen": -88.39584350585938,
|
|
"logps/rejected": -113.70811462402344,
|
|
"loss": 0.0067,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3476972579956055,
|
|
"rewards/margins": 23.937400817871094,
|
|
"rewards/rejected": -22.589706420898438,
|
|
"step": 4370
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"learning_rate": 1.1122569737954353e-07,
|
|
"logits/chosen": -1.3038419485092163,
|
|
"logits/rejected": -1.2349718809127808,
|
|
"logps/chosen": -88.70793151855469,
|
|
"logps/rejected": -110.3454360961914,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.9998445510864258,
|
|
"rewards/margins": 24.013072967529297,
|
|
"rewards/rejected": -22.013225555419922,
|
|
"step": 4380
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"learning_rate": 1.1071851225697379e-07,
|
|
"logits/chosen": -1.3599039316177368,
|
|
"logits/rejected": -1.282843828201294,
|
|
"logps/chosen": -95.58447265625,
|
|
"logps/rejected": -119.10038757324219,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9651339650154114,
|
|
"rewards/margins": 25.86013412475586,
|
|
"rewards/rejected": -24.89499855041504,
|
|
"step": 4390
|
|
},
|
|
{
|
|
"epoch": 2.01,
|
|
"learning_rate": 1.1021132713440405e-07,
|
|
"logits/chosen": -1.2892515659332275,
|
|
"logits/rejected": -1.2328523397445679,
|
|
"logps/chosen": -88.33984375,
|
|
"logps/rejected": -111.24787902832031,
|
|
"loss": 0.0024,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.816326916217804,
|
|
"rewards/margins": 23.1959171295166,
|
|
"rewards/rejected": -22.379589080810547,
|
|
"step": 4400
|
|
},
|
|
{
|
|
"epoch": 2.01,
|
|
"eval_logits/chosen": -1.4092082977294922,
|
|
"eval_logits/rejected": -1.3323308229446411,
|
|
"eval_logps/chosen": -86.91348266601562,
|
|
"eval_logps/rejected": -111.49284362792969,
|
|
"eval_loss": 0.006608007475733757,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.2594695091247559,
|
|
"eval_rewards/margins": 24.33502960205078,
|
|
"eval_rewards/rejected": -23.075557708740234,
|
|
"eval_runtime": 62.0198,
|
|
"eval_samples_per_second": 46.147,
|
|
"eval_steps_per_second": 2.886,
|
|
"step": 4400
|
|
},
|
|
{
|
|
"epoch": 2.01,
|
|
"learning_rate": 1.0970414201183432e-07,
|
|
"logits/chosen": -1.2805674076080322,
|
|
"logits/rejected": -1.221649408340454,
|
|
"logps/chosen": -94.44866943359375,
|
|
"logps/rejected": -120.46419525146484,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9686487317085266,
|
|
"rewards/margins": 25.624675750732422,
|
|
"rewards/rejected": -24.65602684020996,
|
|
"step": 4410
|
|
},
|
|
{
|
|
"epoch": 2.02,
|
|
"learning_rate": 1.0919695688926457e-07,
|
|
"logits/chosen": -1.3352880477905273,
|
|
"logits/rejected": -1.2853882312774658,
|
|
"logps/chosen": -86.63716125488281,
|
|
"logps/rejected": -115.88981628417969,
|
|
"loss": 0.0079,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9782888293266296,
|
|
"rewards/margins": 23.240489959716797,
|
|
"rewards/rejected": -22.262203216552734,
|
|
"step": 4420
|
|
},
|
|
{
|
|
"epoch": 2.02,
|
|
"learning_rate": 1.0868977176669484e-07,
|
|
"logits/chosen": -1.3116331100463867,
|
|
"logits/rejected": -1.2609537839889526,
|
|
"logps/chosen": -85.02494812011719,
|
|
"logps/rejected": -115.8293228149414,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2943823337554932,
|
|
"rewards/margins": 24.715749740600586,
|
|
"rewards/rejected": -23.421369552612305,
|
|
"step": 4430
|
|
},
|
|
{
|
|
"epoch": 2.03,
|
|
"learning_rate": 1.081825866441251e-07,
|
|
"logits/chosen": -1.3299988508224487,
|
|
"logits/rejected": -1.266905665397644,
|
|
"logps/chosen": -90.40599060058594,
|
|
"logps/rejected": -116.70439147949219,
|
|
"loss": 0.0026,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.15012526512146,
|
|
"rewards/margins": 24.405237197875977,
|
|
"rewards/rejected": -23.25511360168457,
|
|
"step": 4440
|
|
},
|
|
{
|
|
"epoch": 2.03,
|
|
"learning_rate": 1.0767540152155536e-07,
|
|
"logits/chosen": -1.324263572692871,
|
|
"logits/rejected": -1.2538119554519653,
|
|
"logps/chosen": -83.78913116455078,
|
|
"logps/rejected": -115.94393157958984,
|
|
"loss": 0.0027,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9307165145874023,
|
|
"rewards/margins": 24.19430160522461,
|
|
"rewards/rejected": -23.263586044311523,
|
|
"step": 4450
|
|
},
|
|
{
|
|
"epoch": 2.04,
|
|
"learning_rate": 1.0716821639898562e-07,
|
|
"logits/chosen": -1.340303659439087,
|
|
"logits/rejected": -1.2662428617477417,
|
|
"logps/chosen": -93.44474792480469,
|
|
"logps/rejected": -110.15861511230469,
|
|
"loss": 0.0001,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.513580560684204,
|
|
"rewards/margins": 24.58599853515625,
|
|
"rewards/rejected": -23.072418212890625,
|
|
"step": 4460
|
|
},
|
|
{
|
|
"epoch": 2.04,
|
|
"learning_rate": 1.0666103127641589e-07,
|
|
"logits/chosen": -1.3335578441619873,
|
|
"logits/rejected": -1.2744941711425781,
|
|
"logps/chosen": -81.96378326416016,
|
|
"logps/rejected": -116.51377868652344,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.511701762676239,
|
|
"rewards/margins": 25.025680541992188,
|
|
"rewards/rejected": -24.51398277282715,
|
|
"step": 4470
|
|
},
|
|
{
|
|
"epoch": 2.04,
|
|
"learning_rate": 1.0615384615384615e-07,
|
|
"logits/chosen": -1.3683226108551025,
|
|
"logits/rejected": -1.2874228954315186,
|
|
"logps/chosen": -84.69481658935547,
|
|
"logps/rejected": -113.8635482788086,
|
|
"loss": 0.0002,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6471071243286133,
|
|
"rewards/margins": 25.367168426513672,
|
|
"rewards/rejected": -23.720062255859375,
|
|
"step": 4480
|
|
},
|
|
{
|
|
"epoch": 2.05,
|
|
"learning_rate": 1.0564666103127641e-07,
|
|
"logits/chosen": -1.3335193395614624,
|
|
"logits/rejected": -1.2593390941619873,
|
|
"logps/chosen": -89.44517517089844,
|
|
"logps/rejected": -113.9664535522461,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6522144079208374,
|
|
"rewards/margins": 24.980897903442383,
|
|
"rewards/rejected": -23.328683853149414,
|
|
"step": 4490
|
|
},
|
|
{
|
|
"epoch": 2.05,
|
|
"learning_rate": 1.0513947590870668e-07,
|
|
"logits/chosen": -1.3129395246505737,
|
|
"logits/rejected": -1.2438514232635498,
|
|
"logps/chosen": -89.5646743774414,
|
|
"logps/rejected": -110.78108215332031,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.7508271932601929,
|
|
"rewards/margins": 24.931434631347656,
|
|
"rewards/rejected": -24.180606842041016,
|
|
"step": 4500
|
|
},
|
|
{
|
|
"epoch": 2.05,
|
|
"eval_logits/chosen": -1.4130858182907104,
|
|
"eval_logits/rejected": -1.3323482275009155,
|
|
"eval_logps/chosen": -87.22586059570312,
|
|
"eval_logps/rejected": -112.64375305175781,
|
|
"eval_loss": 0.0067411912605166435,
|
|
"eval_rewards/accuracies": 0.994413435459137,
|
|
"eval_rewards/chosen": 1.1032801866531372,
|
|
"eval_rewards/margins": 24.75430679321289,
|
|
"eval_rewards/rejected": -23.651025772094727,
|
|
"eval_runtime": 68.0364,
|
|
"eval_samples_per_second": 42.066,
|
|
"eval_steps_per_second": 2.631,
|
|
"step": 4500
|
|
},
|
|
{
|
|
"epoch": 2.06,
|
|
"learning_rate": 1.0463229078613693e-07,
|
|
"logits/chosen": -1.338761568069458,
|
|
"logits/rejected": -1.2733943462371826,
|
|
"logps/chosen": -93.2102279663086,
|
|
"logps/rejected": -115.36387634277344,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.200361728668213,
|
|
"rewards/margins": 24.453805923461914,
|
|
"rewards/rejected": -23.25344467163086,
|
|
"step": 4510
|
|
},
|
|
{
|
|
"epoch": 2.06,
|
|
"learning_rate": 1.041251056635672e-07,
|
|
"logits/chosen": -1.3229678869247437,
|
|
"logits/rejected": -1.258108377456665,
|
|
"logps/chosen": -88.09375762939453,
|
|
"logps/rejected": -119.7026596069336,
|
|
"loss": 0.0055,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.234534740447998,
|
|
"rewards/margins": 26.626230239868164,
|
|
"rewards/rejected": -24.39169692993164,
|
|
"step": 4520
|
|
},
|
|
{
|
|
"epoch": 2.07,
|
|
"learning_rate": 1.0361792054099746e-07,
|
|
"logits/chosen": -1.344191312789917,
|
|
"logits/rejected": -1.2628560066223145,
|
|
"logps/chosen": -94.99612426757812,
|
|
"logps/rejected": -112.13182067871094,
|
|
"loss": 0.0055,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.6532214879989624,
|
|
"rewards/margins": 25.270252227783203,
|
|
"rewards/rejected": -23.61703109741211,
|
|
"step": 4530
|
|
},
|
|
{
|
|
"epoch": 2.07,
|
|
"learning_rate": 1.0311073541842772e-07,
|
|
"logits/chosen": -1.3356167078018188,
|
|
"logits/rejected": -1.274715542793274,
|
|
"logps/chosen": -85.35197448730469,
|
|
"logps/rejected": -114.82647705078125,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3368360996246338,
|
|
"rewards/margins": 25.388629913330078,
|
|
"rewards/rejected": -24.051794052124023,
|
|
"step": 4540
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"learning_rate": 1.0260355029585799e-07,
|
|
"logits/chosen": -1.3028538227081299,
|
|
"logits/rejected": -1.243327021598816,
|
|
"logps/chosen": -83.73817443847656,
|
|
"logps/rejected": -119.57144927978516,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.675358235836029,
|
|
"rewards/margins": 25.816675186157227,
|
|
"rewards/rejected": -25.141313552856445,
|
|
"step": 4550
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"learning_rate": 1.0209636517328824e-07,
|
|
"logits/chosen": -1.3460599184036255,
|
|
"logits/rejected": -1.2746741771697998,
|
|
"logps/chosen": -86.41990661621094,
|
|
"logps/rejected": -114.42486572265625,
|
|
"loss": 0.0076,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9956108331680298,
|
|
"rewards/margins": 26.080272674560547,
|
|
"rewards/rejected": -25.084659576416016,
|
|
"step": 4560
|
|
},
|
|
{
|
|
"epoch": 2.09,
|
|
"learning_rate": 1.0158918005071851e-07,
|
|
"logits/chosen": -1.33854079246521,
|
|
"logits/rejected": -1.2720884084701538,
|
|
"logps/chosen": -88.3348617553711,
|
|
"logps/rejected": -110.352783203125,
|
|
"loss": 0.0055,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.0480217933654785,
|
|
"rewards/margins": 24.384031295776367,
|
|
"rewards/rejected": -23.336009979248047,
|
|
"step": 4570
|
|
},
|
|
{
|
|
"epoch": 2.09,
|
|
"learning_rate": 1.0108199492814876e-07,
|
|
"logits/chosen": -1.3189232349395752,
|
|
"logits/rejected": -1.2537434101104736,
|
|
"logps/chosen": -93.42473602294922,
|
|
"logps/rejected": -117.38157653808594,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.220496416091919,
|
|
"rewards/margins": 25.104259490966797,
|
|
"rewards/rejected": -23.88376235961914,
|
|
"step": 4580
|
|
},
|
|
{
|
|
"epoch": 2.09,
|
|
"learning_rate": 1.0057480980557903e-07,
|
|
"logits/chosen": -1.346653938293457,
|
|
"logits/rejected": -1.2888588905334473,
|
|
"logps/chosen": -84.58158111572266,
|
|
"logps/rejected": -116.2260971069336,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9570630788803101,
|
|
"rewards/margins": 24.75169563293457,
|
|
"rewards/rejected": -23.794628143310547,
|
|
"step": 4590
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"learning_rate": 1.000676246830093e-07,
|
|
"logits/chosen": -1.3445775508880615,
|
|
"logits/rejected": -1.2613394260406494,
|
|
"logps/chosen": -88.18348693847656,
|
|
"logps/rejected": -117.32752990722656,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.2772257328033447,
|
|
"rewards/margins": 27.420055389404297,
|
|
"rewards/rejected": -25.1428279876709,
|
|
"step": 4600
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"eval_logits/chosen": -1.4255520105361938,
|
|
"eval_logits/rejected": -1.3487104177474976,
|
|
"eval_logps/chosen": -87.3498306274414,
|
|
"eval_logps/rejected": -113.354248046875,
|
|
"eval_loss": 0.006937822792679071,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 1.0413002967834473,
|
|
"eval_rewards/margins": 25.047561645507812,
|
|
"eval_rewards/rejected": -24.006263732910156,
|
|
"eval_runtime": 70.6268,
|
|
"eval_samples_per_second": 40.523,
|
|
"eval_steps_per_second": 2.534,
|
|
"step": 4600
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"learning_rate": 9.956043956043955e-08,
|
|
"logits/chosen": -1.3094035387039185,
|
|
"logits/rejected": -1.2451339960098267,
|
|
"logps/chosen": -89.12696838378906,
|
|
"logps/rejected": -114.3353271484375,
|
|
"loss": 0.0036,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 0.26013678312301636,
|
|
"rewards/margins": 23.884912490844727,
|
|
"rewards/rejected": -23.624773025512695,
|
|
"step": 4610
|
|
},
|
|
{
|
|
"epoch": 2.11,
|
|
"learning_rate": 9.905325443786982e-08,
|
|
"logits/chosen": -1.3309595584869385,
|
|
"logits/rejected": -1.2776004076004028,
|
|
"logps/chosen": -86.14755249023438,
|
|
"logps/rejected": -116.1485366821289,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.36527928709983826,
|
|
"rewards/margins": 24.457759857177734,
|
|
"rewards/rejected": -24.092479705810547,
|
|
"step": 4620
|
|
},
|
|
{
|
|
"epoch": 2.11,
|
|
"learning_rate": 9.854606931530007e-08,
|
|
"logits/chosen": -1.3390415906906128,
|
|
"logits/rejected": -1.2725872993469238,
|
|
"logps/chosen": -91.37030029296875,
|
|
"logps/rejected": -115.81642150878906,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0659124851226807,
|
|
"rewards/margins": 25.772014617919922,
|
|
"rewards/rejected": -24.706104278564453,
|
|
"step": 4630
|
|
},
|
|
{
|
|
"epoch": 2.12,
|
|
"learning_rate": 9.803888419273034e-08,
|
|
"logits/chosen": -1.3206666707992554,
|
|
"logits/rejected": -1.2614423036575317,
|
|
"logps/chosen": -83.52031707763672,
|
|
"logps/rejected": -116.8504867553711,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.7927557826042175,
|
|
"rewards/margins": 25.795413970947266,
|
|
"rewards/rejected": -25.002655029296875,
|
|
"step": 4640
|
|
},
|
|
{
|
|
"epoch": 2.12,
|
|
"learning_rate": 9.753169907016061e-08,
|
|
"logits/chosen": -1.3370873928070068,
|
|
"logits/rejected": -1.2674705982208252,
|
|
"logps/chosen": -87.10755157470703,
|
|
"logps/rejected": -119.5748519897461,
|
|
"loss": 0.0054,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.217125654220581,
|
|
"rewards/margins": 27.1856632232666,
|
|
"rewards/rejected": -24.968536376953125,
|
|
"step": 4650
|
|
},
|
|
{
|
|
"epoch": 2.13,
|
|
"learning_rate": 9.702451394759086e-08,
|
|
"logits/chosen": -1.3084181547164917,
|
|
"logits/rejected": -1.2384984493255615,
|
|
"logps/chosen": -87.81915283203125,
|
|
"logps/rejected": -113.61048889160156,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1231558322906494,
|
|
"rewards/margins": 25.29172706604004,
|
|
"rewards/rejected": -24.168569564819336,
|
|
"step": 4660
|
|
},
|
|
{
|
|
"epoch": 2.13,
|
|
"learning_rate": 9.651732882502113e-08,
|
|
"logits/chosen": -1.31204092502594,
|
|
"logits/rejected": -1.2542951107025146,
|
|
"logps/chosen": -92.28341674804688,
|
|
"logps/rejected": -115.6882553100586,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2532768249511719,
|
|
"rewards/margins": 24.862823486328125,
|
|
"rewards/rejected": -23.609546661376953,
|
|
"step": 4670
|
|
},
|
|
{
|
|
"epoch": 2.14,
|
|
"learning_rate": 9.601014370245138e-08,
|
|
"logits/chosen": -1.3849248886108398,
|
|
"logits/rejected": -1.3047869205474854,
|
|
"logps/chosen": -88.0652084350586,
|
|
"logps/rejected": -116.75108337402344,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.82236385345459,
|
|
"rewards/margins": 27.45639419555664,
|
|
"rewards/rejected": -24.634029388427734,
|
|
"step": 4680
|
|
},
|
|
{
|
|
"epoch": 2.14,
|
|
"learning_rate": 9.550295857988165e-08,
|
|
"logits/chosen": -1.3264617919921875,
|
|
"logits/rejected": -1.2588824033737183,
|
|
"logps/chosen": -90.63042449951172,
|
|
"logps/rejected": -117.13484954833984,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8007968664169312,
|
|
"rewards/margins": 25.313175201416016,
|
|
"rewards/rejected": -24.512378692626953,
|
|
"step": 4690
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 9.49957734573119e-08,
|
|
"logits/chosen": -1.319097876548767,
|
|
"logits/rejected": -1.2523882389068604,
|
|
"logps/chosen": -91.26343536376953,
|
|
"logps/rejected": -122.9940414428711,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.29426923394203186,
|
|
"rewards/margins": 26.019664764404297,
|
|
"rewards/rejected": -25.72539710998535,
|
|
"step": 4700
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"eval_logits/chosen": -1.431301236152649,
|
|
"eval_logits/rejected": -1.3519829511642456,
|
|
"eval_logps/chosen": -87.77189636230469,
|
|
"eval_logps/rejected": -114.70515441894531,
|
|
"eval_loss": 0.007204956840723753,
|
|
"eval_rewards/accuracies": 0.994413435459137,
|
|
"eval_rewards/chosen": 0.8302635550498962,
|
|
"eval_rewards/margins": 25.51198387145996,
|
|
"eval_rewards/rejected": -24.681718826293945,
|
|
"eval_runtime": 71.5845,
|
|
"eval_samples_per_second": 39.981,
|
|
"eval_steps_per_second": 2.501,
|
|
"step": 4700
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 9.448858833474217e-08,
|
|
"logits/chosen": -1.3294093608856201,
|
|
"logits/rejected": -1.255859613418579,
|
|
"logps/chosen": -87.07064819335938,
|
|
"logps/rejected": -119.63526916503906,
|
|
"loss": 0.0001,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.91957026720047,
|
|
"rewards/margins": 25.611663818359375,
|
|
"rewards/rejected": -24.692096710205078,
|
|
"step": 4710
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 9.398140321217244e-08,
|
|
"logits/chosen": -1.3767739534378052,
|
|
"logits/rejected": -1.3070622682571411,
|
|
"logps/chosen": -91.01148986816406,
|
|
"logps/rejected": -119.0385971069336,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5754287242889404,
|
|
"rewards/margins": 25.93698501586914,
|
|
"rewards/rejected": -24.361553192138672,
|
|
"step": 4720
|
|
},
|
|
{
|
|
"epoch": 2.16,
|
|
"learning_rate": 9.34742180896027e-08,
|
|
"logits/chosen": -1.3255687952041626,
|
|
"logits/rejected": -1.2571724653244019,
|
|
"logps/chosen": -92.34770202636719,
|
|
"logps/rejected": -119.81795501708984,
|
|
"loss": 0.0067,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": -0.29987969994544983,
|
|
"rewards/margins": 24.60814094543457,
|
|
"rewards/rejected": -24.90802001953125,
|
|
"step": 4730
|
|
},
|
|
{
|
|
"epoch": 2.16,
|
|
"learning_rate": 9.296703296703296e-08,
|
|
"logits/chosen": -1.364595651626587,
|
|
"logits/rejected": -1.2936162948608398,
|
|
"logps/chosen": -88.7773666381836,
|
|
"logps/rejected": -116.1494140625,
|
|
"loss": 0.0024,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.7363441586494446,
|
|
"rewards/margins": 25.116535186767578,
|
|
"rewards/rejected": -25.852880477905273,
|
|
"step": 4740
|
|
},
|
|
{
|
|
"epoch": 2.17,
|
|
"learning_rate": 9.245984784446322e-08,
|
|
"logits/chosen": -1.3715788125991821,
|
|
"logits/rejected": -1.3029972314834595,
|
|
"logps/chosen": -93.72126770019531,
|
|
"logps/rejected": -118.23783111572266,
|
|
"loss": 0.0054,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3293616771697998,
|
|
"rewards/margins": 25.474594116210938,
|
|
"rewards/rejected": -24.145231246948242,
|
|
"step": 4750
|
|
},
|
|
{
|
|
"epoch": 2.17,
|
|
"learning_rate": 9.195266272189349e-08,
|
|
"logits/chosen": -1.3523025512695312,
|
|
"logits/rejected": -1.2690999507904053,
|
|
"logps/chosen": -93.32905578613281,
|
|
"logps/rejected": -118.7462158203125,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9021531343460083,
|
|
"rewards/margins": 26.36541748046875,
|
|
"rewards/rejected": -25.463268280029297,
|
|
"step": 4760
|
|
},
|
|
{
|
|
"epoch": 2.18,
|
|
"learning_rate": 9.144547759932375e-08,
|
|
"logits/chosen": -1.294425368309021,
|
|
"logits/rejected": -1.2407166957855225,
|
|
"logps/chosen": -90.24799346923828,
|
|
"logps/rejected": -123.93524169921875,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4086612462997437,
|
|
"rewards/margins": 25.56521224975586,
|
|
"rewards/rejected": -24.15654945373535,
|
|
"step": 4770
|
|
},
|
|
{
|
|
"epoch": 2.18,
|
|
"learning_rate": 9.093829247675401e-08,
|
|
"logits/chosen": -1.3438308238983154,
|
|
"logits/rejected": -1.275816798210144,
|
|
"logps/chosen": -91.6666488647461,
|
|
"logps/rejected": -119.28436279296875,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.383312702178955,
|
|
"rewards/margins": 27.003463745117188,
|
|
"rewards/rejected": -24.620147705078125,
|
|
"step": 4780
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"learning_rate": 9.043110735418427e-08,
|
|
"logits/chosen": -1.3451048135757446,
|
|
"logits/rejected": -1.2772449254989624,
|
|
"logps/chosen": -94.4778823852539,
|
|
"logps/rejected": -116.2429428100586,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.2315725088119507,
|
|
"rewards/margins": 25.15644073486328,
|
|
"rewards/rejected": -23.924867630004883,
|
|
"step": 4790
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"learning_rate": 8.992392223161453e-08,
|
|
"logits/chosen": -1.2999621629714966,
|
|
"logits/rejected": -1.2457572221755981,
|
|
"logps/chosen": -88.31067657470703,
|
|
"logps/rejected": -120.50489807128906,
|
|
"loss": 0.0067,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.5283034443855286,
|
|
"rewards/margins": 25.667068481445312,
|
|
"rewards/rejected": -25.13875961303711,
|
|
"step": 4800
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"eval_logits/chosen": -1.4411685466766357,
|
|
"eval_logits/rejected": -1.3603928089141846,
|
|
"eval_logps/chosen": -87.89783477783203,
|
|
"eval_logps/rejected": -114.8197250366211,
|
|
"eval_loss": 0.007284725550562143,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.7672966718673706,
|
|
"eval_rewards/margins": 25.506298065185547,
|
|
"eval_rewards/rejected": -24.739002227783203,
|
|
"eval_runtime": 61.4965,
|
|
"eval_samples_per_second": 46.539,
|
|
"eval_steps_per_second": 2.911,
|
|
"step": 4800
|
|
},
|
|
{
|
|
"epoch": 2.2,
|
|
"learning_rate": 8.94167371090448e-08,
|
|
"logits/chosen": -1.3604772090911865,
|
|
"logits/rejected": -1.291609287261963,
|
|
"logps/chosen": -87.88396453857422,
|
|
"logps/rejected": -116.81595611572266,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6294358968734741,
|
|
"rewards/margins": 25.354211807250977,
|
|
"rewards/rejected": -24.724775314331055,
|
|
"step": 4810
|
|
},
|
|
{
|
|
"epoch": 2.2,
|
|
"learning_rate": 8.890955198647506e-08,
|
|
"logits/chosen": -1.327947735786438,
|
|
"logits/rejected": -1.2544294595718384,
|
|
"logps/chosen": -87.47128295898438,
|
|
"logps/rejected": -121.36673736572266,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.21699205040931702,
|
|
"rewards/margins": 25.810955047607422,
|
|
"rewards/rejected": -25.593963623046875,
|
|
"step": 4820
|
|
},
|
|
{
|
|
"epoch": 2.2,
|
|
"learning_rate": 8.840236686390532e-08,
|
|
"logits/chosen": -1.3332288265228271,
|
|
"logits/rejected": -1.2797290086746216,
|
|
"logps/chosen": -84.39314270019531,
|
|
"logps/rejected": -118.59745788574219,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0452686548233032,
|
|
"rewards/margins": 26.805339813232422,
|
|
"rewards/rejected": -25.760074615478516,
|
|
"step": 4830
|
|
},
|
|
{
|
|
"epoch": 2.21,
|
|
"learning_rate": 8.789518174133559e-08,
|
|
"logits/chosen": -1.3392903804779053,
|
|
"logits/rejected": -1.273781418800354,
|
|
"logps/chosen": -87.46485137939453,
|
|
"logps/rejected": -118.17325592041016,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.027283096686005592,
|
|
"rewards/margins": 25.02812385559082,
|
|
"rewards/rejected": -25.05540657043457,
|
|
"step": 4840
|
|
},
|
|
{
|
|
"epoch": 2.21,
|
|
"learning_rate": 8.738799661876584e-08,
|
|
"logits/chosen": -1.36063551902771,
|
|
"logits/rejected": -1.275144338607788,
|
|
"logps/chosen": -93.5079116821289,
|
|
"logps/rejected": -117.1824951171875,
|
|
"loss": 0.0014,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.3975207209587097,
|
|
"rewards/margins": 25.79745101928711,
|
|
"rewards/rejected": -25.399932861328125,
|
|
"step": 4850
|
|
},
|
|
{
|
|
"epoch": 2.22,
|
|
"learning_rate": 8.688081149619611e-08,
|
|
"logits/chosen": -1.3441591262817383,
|
|
"logits/rejected": -1.2624187469482422,
|
|
"logps/chosen": -91.46955108642578,
|
|
"logps/rejected": -112.102783203125,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 2.1940736770629883,
|
|
"rewards/margins": 25.44129753112793,
|
|
"rewards/rejected": -23.247224807739258,
|
|
"step": 4860
|
|
},
|
|
{
|
|
"epoch": 2.22,
|
|
"learning_rate": 8.637362637362636e-08,
|
|
"logits/chosen": -1.371538758277893,
|
|
"logits/rejected": -1.3088654279708862,
|
|
"logps/chosen": -86.52335357666016,
|
|
"logps/rejected": -117.4716567993164,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9726071357727051,
|
|
"rewards/margins": 26.16326332092285,
|
|
"rewards/rejected": -25.190654754638672,
|
|
"step": 4870
|
|
},
|
|
{
|
|
"epoch": 2.23,
|
|
"learning_rate": 8.586644125105663e-08,
|
|
"logits/chosen": -1.3846619129180908,
|
|
"logits/rejected": -1.307712197303772,
|
|
"logps/chosen": -87.76756286621094,
|
|
"logps/rejected": -115.74507141113281,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.3747789263725281,
|
|
"rewards/margins": 25.810970306396484,
|
|
"rewards/rejected": -25.436189651489258,
|
|
"step": 4880
|
|
},
|
|
{
|
|
"epoch": 2.23,
|
|
"learning_rate": 8.53592561284869e-08,
|
|
"logits/chosen": -1.3542847633361816,
|
|
"logits/rejected": -1.288946270942688,
|
|
"logps/chosen": -91.2932357788086,
|
|
"logps/rejected": -116.8852767944336,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8335541486740112,
|
|
"rewards/margins": 25.681522369384766,
|
|
"rewards/rejected": -24.847965240478516,
|
|
"step": 4890
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"learning_rate": 8.485207100591715e-08,
|
|
"logits/chosen": -1.3608229160308838,
|
|
"logits/rejected": -1.2844916582107544,
|
|
"logps/chosen": -88.2170181274414,
|
|
"logps/rejected": -120.30814361572266,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6866910457611084,
|
|
"rewards/margins": 27.874353408813477,
|
|
"rewards/rejected": -26.187664031982422,
|
|
"step": 4900
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"eval_logits/chosen": -1.461911678314209,
|
|
"eval_logits/rejected": -1.3814847469329834,
|
|
"eval_logps/chosen": -87.90396881103516,
|
|
"eval_logps/rejected": -115.87995147705078,
|
|
"eval_loss": 0.007340571843087673,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.7642290592193604,
|
|
"eval_rewards/margins": 26.033344268798828,
|
|
"eval_rewards/rejected": -25.26911735534668,
|
|
"eval_runtime": 68.6122,
|
|
"eval_samples_per_second": 41.713,
|
|
"eval_steps_per_second": 2.609,
|
|
"step": 4900
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"learning_rate": 8.434488588334742e-08,
|
|
"logits/chosen": -1.4112756252288818,
|
|
"logits/rejected": -1.3500181436538696,
|
|
"logps/chosen": -89.21757507324219,
|
|
"logps/rejected": -117.94172668457031,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4118437170982361,
|
|
"rewards/margins": 24.942344665527344,
|
|
"rewards/rejected": -24.530498504638672,
|
|
"step": 4910
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 8.383770076077767e-08,
|
|
"logits/chosen": -1.3511896133422852,
|
|
"logits/rejected": -1.2767199277877808,
|
|
"logps/chosen": -95.7844009399414,
|
|
"logps/rejected": -117.9591293334961,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.7757484912872314,
|
|
"rewards/margins": 26.916187286376953,
|
|
"rewards/rejected": -25.14043617248535,
|
|
"step": 4920
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 8.333051563820794e-08,
|
|
"logits/chosen": -1.3630616664886475,
|
|
"logits/rejected": -1.2954555749893188,
|
|
"logps/chosen": -91.95133972167969,
|
|
"logps/rejected": -123.22148132324219,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.0093523263931274,
|
|
"rewards/margins": 27.7642822265625,
|
|
"rewards/rejected": -26.754928588867188,
|
|
"step": 4930
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 8.282333051563821e-08,
|
|
"logits/chosen": -1.3652262687683105,
|
|
"logits/rejected": -1.3002612590789795,
|
|
"logps/chosen": -89.00601959228516,
|
|
"logps/rejected": -117.67951965332031,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.025102138519287,
|
|
"rewards/margins": 26.249094009399414,
|
|
"rewards/rejected": -25.223987579345703,
|
|
"step": 4940
|
|
},
|
|
{
|
|
"epoch": 2.26,
|
|
"learning_rate": 8.231614539306846e-08,
|
|
"logits/chosen": -1.3879879713058472,
|
|
"logits/rejected": -1.3240267038345337,
|
|
"logps/chosen": -87.9173812866211,
|
|
"logps/rejected": -117.83023834228516,
|
|
"loss": 0.0055,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.08251948654651642,
|
|
"rewards/margins": 24.835966110229492,
|
|
"rewards/rejected": -24.75344467163086,
|
|
"step": 4950
|
|
},
|
|
{
|
|
"epoch": 2.26,
|
|
"learning_rate": 8.180896027049873e-08,
|
|
"logits/chosen": -1.372811198234558,
|
|
"logits/rejected": -1.2928178310394287,
|
|
"logps/chosen": -90.76931762695312,
|
|
"logps/rejected": -118.8313217163086,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.388689398765564,
|
|
"rewards/margins": 26.92373275756836,
|
|
"rewards/rejected": -25.535043716430664,
|
|
"step": 4960
|
|
},
|
|
{
|
|
"epoch": 2.27,
|
|
"learning_rate": 8.130177514792898e-08,
|
|
"logits/chosen": -1.3881031274795532,
|
|
"logits/rejected": -1.314194917678833,
|
|
"logps/chosen": -87.59803771972656,
|
|
"logps/rejected": -118.09504699707031,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4410366117954254,
|
|
"rewards/margins": 26.20395278930664,
|
|
"rewards/rejected": -25.762914657592773,
|
|
"step": 4970
|
|
},
|
|
{
|
|
"epoch": 2.27,
|
|
"learning_rate": 8.079459002535925e-08,
|
|
"logits/chosen": -1.3640211820602417,
|
|
"logits/rejected": -1.2896654605865479,
|
|
"logps/chosen": -87.9101333618164,
|
|
"logps/rejected": -117.8978500366211,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.964639663696289,
|
|
"rewards/margins": 28.1193790435791,
|
|
"rewards/rejected": -26.154743194580078,
|
|
"step": 4980
|
|
},
|
|
{
|
|
"epoch": 2.28,
|
|
"learning_rate": 8.02874049027895e-08,
|
|
"logits/chosen": -1.377068281173706,
|
|
"logits/rejected": -1.3123360872268677,
|
|
"logps/chosen": -87.87947845458984,
|
|
"logps/rejected": -125.39814758300781,
|
|
"loss": 0.0002,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9372149705886841,
|
|
"rewards/margins": 27.093730926513672,
|
|
"rewards/rejected": -26.156518936157227,
|
|
"step": 4990
|
|
},
|
|
{
|
|
"epoch": 2.28,
|
|
"learning_rate": 7.978021978021977e-08,
|
|
"logits/chosen": -1.3721301555633545,
|
|
"logits/rejected": -1.302929162979126,
|
|
"logps/chosen": -88.95343780517578,
|
|
"logps/rejected": -122.36506652832031,
|
|
"loss": 0.0016,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1758254766464233,
|
|
"rewards/margins": 27.310205459594727,
|
|
"rewards/rejected": -26.134380340576172,
|
|
"step": 5000
|
|
},
|
|
{
|
|
"epoch": 2.28,
|
|
"eval_logits/chosen": -1.4729403257369995,
|
|
"eval_logits/rejected": -1.3894855976104736,
|
|
"eval_logps/chosen": -88.23526763916016,
|
|
"eval_logps/rejected": -117.04609680175781,
|
|
"eval_loss": 0.0077310591004788876,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.5985845923423767,
|
|
"eval_rewards/margins": 26.450777053833008,
|
|
"eval_rewards/rejected": -25.852190017700195,
|
|
"eval_runtime": 65.1848,
|
|
"eval_samples_per_second": 43.906,
|
|
"eval_steps_per_second": 2.746,
|
|
"step": 5000
|
|
},
|
|
{
|
|
"epoch": 2.29,
|
|
"learning_rate": 7.927303465765004e-08,
|
|
"logits/chosen": -1.4071820974349976,
|
|
"logits/rejected": -1.325002908706665,
|
|
"logps/chosen": -92.00203704833984,
|
|
"logps/rejected": -121.43309020996094,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.40916553139686584,
|
|
"rewards/margins": 25.54302978515625,
|
|
"rewards/rejected": -25.133861541748047,
|
|
"step": 5010
|
|
},
|
|
{
|
|
"epoch": 2.29,
|
|
"learning_rate": 7.87658495350803e-08,
|
|
"logits/chosen": -1.382490873336792,
|
|
"logits/rejected": -1.290702223777771,
|
|
"logps/chosen": -91.81244659423828,
|
|
"logps/rejected": -116.6902847290039,
|
|
"loss": 0.0065,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.020965814590454,
|
|
"rewards/margins": 27.167251586914062,
|
|
"rewards/rejected": -26.146282196044922,
|
|
"step": 5020
|
|
},
|
|
{
|
|
"epoch": 2.3,
|
|
"learning_rate": 7.825866441251056e-08,
|
|
"logits/chosen": -1.359053373336792,
|
|
"logits/rejected": -1.290919542312622,
|
|
"logps/chosen": -86.95582580566406,
|
|
"logps/rejected": -120.83614349365234,
|
|
"loss": 0.0,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.065791368484497,
|
|
"rewards/margins": 29.087757110595703,
|
|
"rewards/rejected": -27.021968841552734,
|
|
"step": 5030
|
|
},
|
|
{
|
|
"epoch": 2.3,
|
|
"learning_rate": 7.775147928994082e-08,
|
|
"logits/chosen": -1.369732141494751,
|
|
"logits/rejected": -1.3190780878067017,
|
|
"logps/chosen": -88.25080108642578,
|
|
"logps/rejected": -122.08647155761719,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4733394682407379,
|
|
"rewards/margins": 26.999135971069336,
|
|
"rewards/rejected": -26.52579689025879,
|
|
"step": 5040
|
|
},
|
|
{
|
|
"epoch": 2.3,
|
|
"learning_rate": 7.724429416737108e-08,
|
|
"logits/chosen": -1.3806296586990356,
|
|
"logits/rejected": -1.314660668373108,
|
|
"logps/chosen": -82.57357025146484,
|
|
"logps/rejected": -113.80106353759766,
|
|
"loss": 0.0024,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6215075254440308,
|
|
"rewards/margins": 25.933462142944336,
|
|
"rewards/rejected": -25.311954498291016,
|
|
"step": 5050
|
|
},
|
|
{
|
|
"epoch": 2.31,
|
|
"learning_rate": 7.673710904480135e-08,
|
|
"logits/chosen": -1.3427412509918213,
|
|
"logits/rejected": -1.277625560760498,
|
|
"logps/chosen": -88.83329772949219,
|
|
"logps/rejected": -113.25550842285156,
|
|
"loss": 0.0018,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.39279159903526306,
|
|
"rewards/margins": 24.52106475830078,
|
|
"rewards/rejected": -24.128273010253906,
|
|
"step": 5060
|
|
},
|
|
{
|
|
"epoch": 2.31,
|
|
"learning_rate": 7.62299239222316e-08,
|
|
"logits/chosen": -1.3516267538070679,
|
|
"logits/rejected": -1.290647268295288,
|
|
"logps/chosen": -93.58750915527344,
|
|
"logps/rejected": -119.11576843261719,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.06238238885998726,
|
|
"rewards/margins": 25.322040557861328,
|
|
"rewards/rejected": -25.384424209594727,
|
|
"step": 5070
|
|
},
|
|
{
|
|
"epoch": 2.32,
|
|
"learning_rate": 7.572273879966187e-08,
|
|
"logits/chosen": -1.3757238388061523,
|
|
"logits/rejected": -1.306443452835083,
|
|
"logps/chosen": -93.31622314453125,
|
|
"logps/rejected": -117.56782531738281,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3956282138824463,
|
|
"rewards/margins": 26.834218978881836,
|
|
"rewards/rejected": -25.43859100341797,
|
|
"step": 5080
|
|
},
|
|
{
|
|
"epoch": 2.32,
|
|
"learning_rate": 7.521555367709213e-08,
|
|
"logits/chosen": -1.3766396045684814,
|
|
"logits/rejected": -1.3114879131317139,
|
|
"logps/chosen": -86.86039733886719,
|
|
"logps/rejected": -118.87870025634766,
|
|
"loss": 0.0055,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0653996467590332,
|
|
"rewards/margins": 26.928573608398438,
|
|
"rewards/rejected": -25.863174438476562,
|
|
"step": 5090
|
|
},
|
|
{
|
|
"epoch": 2.33,
|
|
"learning_rate": 7.47083685545224e-08,
|
|
"logits/chosen": -1.38792884349823,
|
|
"logits/rejected": -1.31496262550354,
|
|
"logps/chosen": -86.50321197509766,
|
|
"logps/rejected": -119.1225357055664,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.047428846359253,
|
|
"rewards/margins": 26.675806045532227,
|
|
"rewards/rejected": -25.628376007080078,
|
|
"step": 5100
|
|
},
|
|
{
|
|
"epoch": 2.33,
|
|
"eval_logits/chosen": -1.4878861904144287,
|
|
"eval_logits/rejected": -1.4035993814468384,
|
|
"eval_logps/chosen": -88.03448486328125,
|
|
"eval_logps/rejected": -116.85780334472656,
|
|
"eval_loss": 0.007282613776624203,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.6989699006080627,
|
|
"eval_rewards/margins": 26.457014083862305,
|
|
"eval_rewards/rejected": -25.758041381835938,
|
|
"eval_runtime": 63.4536,
|
|
"eval_samples_per_second": 45.104,
|
|
"eval_steps_per_second": 2.821,
|
|
"step": 5100
|
|
},
|
|
{
|
|
"epoch": 2.33,
|
|
"learning_rate": 7.420118343195266e-08,
|
|
"logits/chosen": -1.3945187330245972,
|
|
"logits/rejected": -1.311867594718933,
|
|
"logps/chosen": -94.6859359741211,
|
|
"logps/rejected": -113.41032409667969,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1795772314071655,
|
|
"rewards/margins": 26.022451400756836,
|
|
"rewards/rejected": -24.842870712280273,
|
|
"step": 5110
|
|
},
|
|
{
|
|
"epoch": 2.34,
|
|
"learning_rate": 7.369399830938292e-08,
|
|
"logits/chosen": -1.3820443153381348,
|
|
"logits/rejected": -1.3166067600250244,
|
|
"logps/chosen": -83.39116668701172,
|
|
"logps/rejected": -115.35809326171875,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4604426920413971,
|
|
"rewards/margins": 26.040401458740234,
|
|
"rewards/rejected": -25.579957962036133,
|
|
"step": 5120
|
|
},
|
|
{
|
|
"epoch": 2.34,
|
|
"learning_rate": 7.318681318681318e-08,
|
|
"logits/chosen": -1.3669214248657227,
|
|
"logits/rejected": -1.3000479936599731,
|
|
"logps/chosen": -90.93084716796875,
|
|
"logps/rejected": -114.6349868774414,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9245045781135559,
|
|
"rewards/margins": 26.640636444091797,
|
|
"rewards/rejected": -25.716135025024414,
|
|
"step": 5130
|
|
},
|
|
{
|
|
"epoch": 2.35,
|
|
"learning_rate": 7.267962806424344e-08,
|
|
"logits/chosen": -1.3965649604797363,
|
|
"logits/rejected": -1.3084334135055542,
|
|
"logps/chosen": -97.05128479003906,
|
|
"logps/rejected": -123.03981018066406,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9901596903800964,
|
|
"rewards/margins": 26.782718658447266,
|
|
"rewards/rejected": -25.792556762695312,
|
|
"step": 5140
|
|
},
|
|
{
|
|
"epoch": 2.35,
|
|
"learning_rate": 7.21724429416737e-08,
|
|
"logits/chosen": -1.3820029497146606,
|
|
"logits/rejected": -1.2966136932373047,
|
|
"logps/chosen": -90.78387451171875,
|
|
"logps/rejected": -119.2174301147461,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5053150653839111,
|
|
"rewards/margins": 27.21875,
|
|
"rewards/rejected": -25.713436126708984,
|
|
"step": 5150
|
|
},
|
|
{
|
|
"epoch": 2.36,
|
|
"learning_rate": 7.166525781910397e-08,
|
|
"logits/chosen": -1.3862712383270264,
|
|
"logits/rejected": -1.315342903137207,
|
|
"logps/chosen": -88.68326568603516,
|
|
"logps/rejected": -119.2471923828125,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": -0.5128425359725952,
|
|
"rewards/margins": 24.922550201416016,
|
|
"rewards/rejected": -25.435388565063477,
|
|
"step": 5160
|
|
},
|
|
{
|
|
"epoch": 2.36,
|
|
"learning_rate": 7.115807269653423e-08,
|
|
"logits/chosen": -1.358323335647583,
|
|
"logits/rejected": -1.2905691862106323,
|
|
"logps/chosen": -88.0818862915039,
|
|
"logps/rejected": -118.68809509277344,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.4642322063446045,
|
|
"rewards/margins": 26.244770050048828,
|
|
"rewards/rejected": -24.78053855895996,
|
|
"step": 5170
|
|
},
|
|
{
|
|
"epoch": 2.36,
|
|
"learning_rate": 7.06508875739645e-08,
|
|
"logits/chosen": -1.412334680557251,
|
|
"logits/rejected": -1.33579683303833,
|
|
"logps/chosen": -93.82710266113281,
|
|
"logps/rejected": -117.13541412353516,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.39411622285842896,
|
|
"rewards/margins": 25.49358367919922,
|
|
"rewards/rejected": -25.099470138549805,
|
|
"step": 5180
|
|
},
|
|
{
|
|
"epoch": 2.37,
|
|
"learning_rate": 7.014370245139475e-08,
|
|
"logits/chosen": -1.3709720373153687,
|
|
"logits/rejected": -1.2929903268814087,
|
|
"logps/chosen": -91.2157974243164,
|
|
"logps/rejected": -124.2120132446289,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4855771064758301,
|
|
"rewards/margins": 27.795948028564453,
|
|
"rewards/rejected": -27.31036949157715,
|
|
"step": 5190
|
|
},
|
|
{
|
|
"epoch": 2.37,
|
|
"learning_rate": 6.963651732882502e-08,
|
|
"logits/chosen": -1.3689887523651123,
|
|
"logits/rejected": -1.2964820861816406,
|
|
"logps/chosen": -86.01536560058594,
|
|
"logps/rejected": -118.3643798828125,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1772948503494263,
|
|
"rewards/margins": 27.771255493164062,
|
|
"rewards/rejected": -26.59395980834961,
|
|
"step": 5200
|
|
},
|
|
{
|
|
"epoch": 2.37,
|
|
"eval_logits/chosen": -1.4872238636016846,
|
|
"eval_logits/rejected": -1.399623990058899,
|
|
"eval_logps/chosen": -88.03882598876953,
|
|
"eval_logps/rejected": -117.7686996459961,
|
|
"eval_loss": 0.007167758885771036,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.6967981457710266,
|
|
"eval_rewards/margins": 26.910293579101562,
|
|
"eval_rewards/rejected": -26.213491439819336,
|
|
"eval_runtime": 70.0341,
|
|
"eval_samples_per_second": 40.866,
|
|
"eval_steps_per_second": 2.556,
|
|
"step": 5200
|
|
},
|
|
{
|
|
"epoch": 2.38,
|
|
"learning_rate": 6.912933220625529e-08,
|
|
"logits/chosen": -1.4059669971466064,
|
|
"logits/rejected": -1.3302024602890015,
|
|
"logps/chosen": -88.3302230834961,
|
|
"logps/rejected": -118.16276550292969,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0470592975616455,
|
|
"rewards/margins": 26.200448989868164,
|
|
"rewards/rejected": -25.153392791748047,
|
|
"step": 5210
|
|
},
|
|
{
|
|
"epoch": 2.38,
|
|
"learning_rate": 6.862214708368554e-08,
|
|
"logits/chosen": -1.397664189338684,
|
|
"logits/rejected": -1.3320751190185547,
|
|
"logps/chosen": -85.62290954589844,
|
|
"logps/rejected": -119.31951904296875,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.21012182533740997,
|
|
"rewards/margins": 25.69754409790039,
|
|
"rewards/rejected": -25.907669067382812,
|
|
"step": 5220
|
|
},
|
|
{
|
|
"epoch": 2.39,
|
|
"learning_rate": 6.811496196111581e-08,
|
|
"logits/chosen": -1.4220311641693115,
|
|
"logits/rejected": -1.3425581455230713,
|
|
"logps/chosen": -88.67344665527344,
|
|
"logps/rejected": -117.9163589477539,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3023020029067993,
|
|
"rewards/margins": 27.779937744140625,
|
|
"rewards/rejected": -26.477636337280273,
|
|
"step": 5230
|
|
},
|
|
{
|
|
"epoch": 2.39,
|
|
"learning_rate": 6.760777683854606e-08,
|
|
"logits/chosen": -1.3980010747909546,
|
|
"logits/rejected": -1.3161672353744507,
|
|
"logps/chosen": -94.78350830078125,
|
|
"logps/rejected": -118.20045471191406,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4626989364624023,
|
|
"rewards/margins": 27.554616928100586,
|
|
"rewards/rejected": -26.091922760009766,
|
|
"step": 5240
|
|
},
|
|
{
|
|
"epoch": 2.4,
|
|
"learning_rate": 6.710059171597633e-08,
|
|
"logits/chosen": -1.3799418210983276,
|
|
"logits/rejected": -1.3152214288711548,
|
|
"logps/chosen": -90.38761901855469,
|
|
"logps/rejected": -122.6275863647461,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 0.18404527008533478,
|
|
"rewards/margins": 27.751272201538086,
|
|
"rewards/rejected": -27.56722640991211,
|
|
"step": 5250
|
|
},
|
|
{
|
|
"epoch": 2.4,
|
|
"learning_rate": 6.659340659340658e-08,
|
|
"logits/chosen": -1.3793939352035522,
|
|
"logits/rejected": -1.3092762231826782,
|
|
"logps/chosen": -90.41889953613281,
|
|
"logps/rejected": -118.99210357666016,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.34509414434432983,
|
|
"rewards/margins": 27.324649810791016,
|
|
"rewards/rejected": -26.97955322265625,
|
|
"step": 5260
|
|
},
|
|
{
|
|
"epoch": 2.41,
|
|
"learning_rate": 6.608622147083685e-08,
|
|
"logits/chosen": -1.3813196420669556,
|
|
"logits/rejected": -1.3037269115447998,
|
|
"logps/chosen": -93.68791198730469,
|
|
"logps/rejected": -128.4936981201172,
|
|
"loss": 0.0015,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1084160804748535,
|
|
"rewards/margins": 28.909637451171875,
|
|
"rewards/rejected": -27.801223754882812,
|
|
"step": 5270
|
|
},
|
|
{
|
|
"epoch": 2.41,
|
|
"learning_rate": 6.557903634826712e-08,
|
|
"logits/chosen": -1.3911049365997314,
|
|
"logits/rejected": -1.3268201351165771,
|
|
"logps/chosen": -87.09625244140625,
|
|
"logps/rejected": -122.59407806396484,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.46324244141578674,
|
|
"rewards/margins": 27.979909896850586,
|
|
"rewards/rejected": -27.51666831970215,
|
|
"step": 5280
|
|
},
|
|
{
|
|
"epoch": 2.41,
|
|
"learning_rate": 6.507185122569737e-08,
|
|
"logits/chosen": -1.3915858268737793,
|
|
"logits/rejected": -1.3138229846954346,
|
|
"logps/chosen": -93.52770233154297,
|
|
"logps/rejected": -121.19853210449219,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6541392803192139,
|
|
"rewards/margins": 29.604787826538086,
|
|
"rewards/rejected": -27.95064926147461,
|
|
"step": 5290
|
|
},
|
|
{
|
|
"epoch": 2.42,
|
|
"learning_rate": 6.456466610312764e-08,
|
|
"logits/chosen": -1.4070789813995361,
|
|
"logits/rejected": -1.3349624872207642,
|
|
"logps/chosen": -91.31390380859375,
|
|
"logps/rejected": -119.59687805175781,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.320601224899292,
|
|
"rewards/margins": 28.13943862915039,
|
|
"rewards/rejected": -26.818838119506836,
|
|
"step": 5300
|
|
},
|
|
{
|
|
"epoch": 2.42,
|
|
"eval_logits/chosen": -1.501613736152649,
|
|
"eval_logits/rejected": -1.4148671627044678,
|
|
"eval_logps/chosen": -88.82596588134766,
|
|
"eval_logps/rejected": -119.5730209350586,
|
|
"eval_loss": 0.007715919055044651,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.30323532223701477,
|
|
"eval_rewards/margins": 27.418886184692383,
|
|
"eval_rewards/rejected": -27.115652084350586,
|
|
"eval_runtime": 60.0405,
|
|
"eval_samples_per_second": 47.668,
|
|
"eval_steps_per_second": 2.981,
|
|
"step": 5300
|
|
},
|
|
{
|
|
"epoch": 2.42,
|
|
"learning_rate": 6.40574809805579e-08,
|
|
"logits/chosen": -1.388594150543213,
|
|
"logits/rejected": -1.314540147781372,
|
|
"logps/chosen": -90.13387298583984,
|
|
"logps/rejected": -122.98262023925781,
|
|
"loss": 0.0047,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6017175316810608,
|
|
"rewards/margins": 28.00093650817871,
|
|
"rewards/rejected": -27.399221420288086,
|
|
"step": 5310
|
|
},
|
|
{
|
|
"epoch": 2.43,
|
|
"learning_rate": 6.355029585798816e-08,
|
|
"logits/chosen": -1.4170560836791992,
|
|
"logits/rejected": -1.3456897735595703,
|
|
"logps/chosen": -88.29602813720703,
|
|
"logps/rejected": -122.81431579589844,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3829694986343384,
|
|
"rewards/margins": 29.370956420898438,
|
|
"rewards/rejected": -27.987987518310547,
|
|
"step": 5320
|
|
},
|
|
{
|
|
"epoch": 2.43,
|
|
"learning_rate": 6.304311073541843e-08,
|
|
"logits/chosen": -1.3887192010879517,
|
|
"logits/rejected": -1.3243298530578613,
|
|
"logps/chosen": -88.75332641601562,
|
|
"logps/rejected": -120.09466552734375,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.2215611934661865,
|
|
"rewards/margins": 26.91391944885254,
|
|
"rewards/rejected": -25.692358016967773,
|
|
"step": 5330
|
|
},
|
|
{
|
|
"epoch": 2.44,
|
|
"learning_rate": 6.253592561284868e-08,
|
|
"logits/chosen": -1.4018067121505737,
|
|
"logits/rejected": -1.3270535469055176,
|
|
"logps/chosen": -93.98942565917969,
|
|
"logps/rejected": -125.354248046875,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.2943921089172363,
|
|
"rewards/margins": 28.14287757873535,
|
|
"rewards/rejected": -26.84848403930664,
|
|
"step": 5340
|
|
},
|
|
{
|
|
"epoch": 2.44,
|
|
"learning_rate": 6.202874049027895e-08,
|
|
"logits/chosen": -1.3864609003067017,
|
|
"logits/rejected": -1.3177391290664673,
|
|
"logps/chosen": -91.09075927734375,
|
|
"logps/rejected": -124.11067962646484,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8371120691299438,
|
|
"rewards/margins": 28.26397705078125,
|
|
"rewards/rejected": -27.426868438720703,
|
|
"step": 5350
|
|
},
|
|
{
|
|
"epoch": 2.45,
|
|
"learning_rate": 6.15215553677092e-08,
|
|
"logits/chosen": -1.4059492349624634,
|
|
"logits/rejected": -1.3294079303741455,
|
|
"logps/chosen": -93.74061584472656,
|
|
"logps/rejected": -119.74947357177734,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4449167251586914,
|
|
"rewards/margins": 27.587207794189453,
|
|
"rewards/rejected": -26.142292022705078,
|
|
"step": 5360
|
|
},
|
|
{
|
|
"epoch": 2.45,
|
|
"learning_rate": 6.101437024513947e-08,
|
|
"logits/chosen": -1.3770349025726318,
|
|
"logits/rejected": -1.3040486574172974,
|
|
"logps/chosen": -92.32209014892578,
|
|
"logps/rejected": -123.76420593261719,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0093997716903687,
|
|
"rewards/margins": 27.536449432373047,
|
|
"rewards/rejected": -26.527050018310547,
|
|
"step": 5370
|
|
},
|
|
{
|
|
"epoch": 2.46,
|
|
"learning_rate": 6.050718512256973e-08,
|
|
"logits/chosen": -1.3874174356460571,
|
|
"logits/rejected": -1.3079521656036377,
|
|
"logps/chosen": -89.07852172851562,
|
|
"logps/rejected": -120.56114196777344,
|
|
"loss": 0.0001,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3143458366394043,
|
|
"rewards/margins": 27.301443099975586,
|
|
"rewards/rejected": -25.987096786499023,
|
|
"step": 5380
|
|
},
|
|
{
|
|
"epoch": 2.46,
|
|
"learning_rate": 6e-08,
|
|
"logits/chosen": -1.4424034357070923,
|
|
"logits/rejected": -1.3676373958587646,
|
|
"logps/chosen": -93.13945770263672,
|
|
"logps/rejected": -124.12135314941406,
|
|
"loss": 0.006,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.4384705424308777,
|
|
"rewards/margins": 27.193965911865234,
|
|
"rewards/rejected": -26.755496978759766,
|
|
"step": 5390
|
|
},
|
|
{
|
|
"epoch": 2.46,
|
|
"learning_rate": 5.949281487743026e-08,
|
|
"logits/chosen": -1.4014787673950195,
|
|
"logits/rejected": -1.3362300395965576,
|
|
"logps/chosen": -88.39823150634766,
|
|
"logps/rejected": -119.18223571777344,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5621711611747742,
|
|
"rewards/margins": 27.4252872467041,
|
|
"rewards/rejected": -26.863113403320312,
|
|
"step": 5400
|
|
},
|
|
{
|
|
"epoch": 2.46,
|
|
"eval_logits/chosen": -1.5089606046676636,
|
|
"eval_logits/rejected": -1.4168381690979004,
|
|
"eval_logps/chosen": -88.5813217163086,
|
|
"eval_logps/rejected": -119.76658630371094,
|
|
"eval_loss": 0.00763333635404706,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.42555317282676697,
|
|
"eval_rewards/margins": 27.637985229492188,
|
|
"eval_rewards/rejected": -27.212432861328125,
|
|
"eval_runtime": 62.6634,
|
|
"eval_samples_per_second": 45.673,
|
|
"eval_steps_per_second": 2.857,
|
|
"step": 5400
|
|
},
|
|
{
|
|
"epoch": 2.47,
|
|
"learning_rate": 5.898562975486052e-08,
|
|
"logits/chosen": -1.4362767934799194,
|
|
"logits/rejected": -1.3550790548324585,
|
|
"logps/chosen": -89.90331268310547,
|
|
"logps/rejected": -124.781982421875,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.522152304649353,
|
|
"rewards/margins": 29.26023292541504,
|
|
"rewards/rejected": -27.738079071044922,
|
|
"step": 5410
|
|
},
|
|
{
|
|
"epoch": 2.47,
|
|
"learning_rate": 5.8478444632290784e-08,
|
|
"logits/chosen": -1.4165668487548828,
|
|
"logits/rejected": -1.343462586402893,
|
|
"logps/chosen": -90.41630554199219,
|
|
"logps/rejected": -127.66859436035156,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.15914836525917053,
|
|
"rewards/margins": 29.304555892944336,
|
|
"rewards/rejected": -29.145410537719727,
|
|
"step": 5420
|
|
},
|
|
{
|
|
"epoch": 2.48,
|
|
"learning_rate": 5.7971259509721045e-08,
|
|
"logits/chosen": -1.3894436359405518,
|
|
"logits/rejected": -1.317000150680542,
|
|
"logps/chosen": -88.46882629394531,
|
|
"logps/rejected": -120.24174499511719,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4189438819885254,
|
|
"rewards/margins": 27.224233627319336,
|
|
"rewards/rejected": -26.805288314819336,
|
|
"step": 5430
|
|
},
|
|
{
|
|
"epoch": 2.48,
|
|
"learning_rate": 5.7464074387151306e-08,
|
|
"logits/chosen": -1.4001535177230835,
|
|
"logits/rejected": -1.3215277194976807,
|
|
"logps/chosen": -87.59601593017578,
|
|
"logps/rejected": -119.82057189941406,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.6788640022277832,
|
|
"rewards/margins": 27.242944717407227,
|
|
"rewards/rejected": -26.5640811920166,
|
|
"step": 5440
|
|
},
|
|
{
|
|
"epoch": 2.49,
|
|
"learning_rate": 5.695688926458157e-08,
|
|
"logits/chosen": -1.4465348720550537,
|
|
"logits/rejected": -1.3477718830108643,
|
|
"logps/chosen": -89.69527435302734,
|
|
"logps/rejected": -118.54779052734375,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.376774549484253,
|
|
"rewards/margins": 28.213756561279297,
|
|
"rewards/rejected": -25.836984634399414,
|
|
"step": 5450
|
|
},
|
|
{
|
|
"epoch": 2.49,
|
|
"learning_rate": 5.6449704142011834e-08,
|
|
"logits/chosen": -1.440271019935608,
|
|
"logits/rejected": -1.3544074296951294,
|
|
"logps/chosen": -85.5431900024414,
|
|
"logps/rejected": -122.612060546875,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.067859172821045,
|
|
"rewards/margins": 29.92409896850586,
|
|
"rewards/rejected": -28.85623550415039,
|
|
"step": 5460
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 5.5942519019442095e-08,
|
|
"logits/chosen": -1.39127779006958,
|
|
"logits/rejected": -1.325758695602417,
|
|
"logps/chosen": -83.56703186035156,
|
|
"logps/rejected": -124.7584457397461,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 0.6307471990585327,
|
|
"rewards/margins": 28.204599380493164,
|
|
"rewards/rejected": -27.5738468170166,
|
|
"step": 5470
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 5.5435333896872356e-08,
|
|
"logits/chosen": -1.4168808460235596,
|
|
"logits/rejected": -1.3478556871414185,
|
|
"logps/chosen": -89.90568542480469,
|
|
"logps/rejected": -121.96623229980469,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0229051113128662,
|
|
"rewards/margins": 28.127361297607422,
|
|
"rewards/rejected": -27.10445785522461,
|
|
"step": 5480
|
|
},
|
|
{
|
|
"epoch": 2.51,
|
|
"learning_rate": 5.492814877430262e-08,
|
|
"logits/chosen": -1.432664394378662,
|
|
"logits/rejected": -1.359386920928955,
|
|
"logps/chosen": -86.13729858398438,
|
|
"logps/rejected": -120.99588775634766,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0249274969100952,
|
|
"rewards/margins": 28.6977481842041,
|
|
"rewards/rejected": -27.67281723022461,
|
|
"step": 5490
|
|
},
|
|
{
|
|
"epoch": 2.51,
|
|
"learning_rate": 5.442096365173288e-08,
|
|
"logits/chosen": -1.4274625778198242,
|
|
"logits/rejected": -1.3294579982757568,
|
|
"logps/chosen": -94.29920959472656,
|
|
"logps/rejected": -124.21602630615234,
|
|
"loss": 0.0,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.37506765127182007,
|
|
"rewards/margins": 28.83444595336914,
|
|
"rewards/rejected": -28.45937728881836,
|
|
"step": 5500
|
|
},
|
|
{
|
|
"epoch": 2.51,
|
|
"eval_logits/chosen": -1.530104637145996,
|
|
"eval_logits/rejected": -1.4378955364227295,
|
|
"eval_logps/chosen": -88.8484115600586,
|
|
"eval_logps/rejected": -120.21839141845703,
|
|
"eval_loss": 0.007588541135191917,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.2920094132423401,
|
|
"eval_rewards/margins": 27.730342864990234,
|
|
"eval_rewards/rejected": -27.438335418701172,
|
|
"eval_runtime": 63.525,
|
|
"eval_samples_per_second": 45.053,
|
|
"eval_steps_per_second": 2.818,
|
|
"step": 5500
|
|
},
|
|
{
|
|
"epoch": 2.51,
|
|
"learning_rate": 5.3913778529163145e-08,
|
|
"logits/chosen": -1.4144002199172974,
|
|
"logits/rejected": -1.337476372718811,
|
|
"logps/chosen": -86.93864440917969,
|
|
"logps/rejected": -125.42852783203125,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5241835117340088,
|
|
"rewards/margins": 29.53707504272461,
|
|
"rewards/rejected": -29.012889862060547,
|
|
"step": 5510
|
|
},
|
|
{
|
|
"epoch": 2.52,
|
|
"learning_rate": 5.3406593406593406e-08,
|
|
"logits/chosen": -1.3884155750274658,
|
|
"logits/rejected": -1.324973464012146,
|
|
"logps/chosen": -83.95980834960938,
|
|
"logps/rejected": -120.20356750488281,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.15396860241889954,
|
|
"rewards/margins": 26.457748413085938,
|
|
"rewards/rejected": -26.61171531677246,
|
|
"step": 5520
|
|
},
|
|
{
|
|
"epoch": 2.52,
|
|
"learning_rate": 5.289940828402367e-08,
|
|
"logits/chosen": -1.4237573146820068,
|
|
"logits/rejected": -1.3433820009231567,
|
|
"logps/chosen": -94.42403411865234,
|
|
"logps/rejected": -125.9416732788086,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.3251895606517792,
|
|
"rewards/margins": 28.843847274780273,
|
|
"rewards/rejected": -28.51865577697754,
|
|
"step": 5530
|
|
},
|
|
{
|
|
"epoch": 2.53,
|
|
"learning_rate": 5.239222316145393e-08,
|
|
"logits/chosen": -1.4355299472808838,
|
|
"logits/rejected": -1.363010048866272,
|
|
"logps/chosen": -88.18751525878906,
|
|
"logps/rejected": -125.2325668334961,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.37160930037498474,
|
|
"rewards/margins": 28.638696670532227,
|
|
"rewards/rejected": -28.26708984375,
|
|
"step": 5540
|
|
},
|
|
{
|
|
"epoch": 2.53,
|
|
"learning_rate": 5.188503803888419e-08,
|
|
"logits/chosen": -1.4135535955429077,
|
|
"logits/rejected": -1.331364631652832,
|
|
"logps/chosen": -94.18409729003906,
|
|
"logps/rejected": -123.58796691894531,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.10003487765789032,
|
|
"rewards/margins": 28.183786392211914,
|
|
"rewards/rejected": -28.083749771118164,
|
|
"step": 5550
|
|
},
|
|
{
|
|
"epoch": 2.54,
|
|
"learning_rate": 5.137785291631445e-08,
|
|
"logits/chosen": -1.4515702724456787,
|
|
"logits/rejected": -1.3703655004501343,
|
|
"logps/chosen": -86.55001068115234,
|
|
"logps/rejected": -129.2012481689453,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.22907297313213348,
|
|
"rewards/margins": 29.383153915405273,
|
|
"rewards/rejected": -29.15407943725586,
|
|
"step": 5560
|
|
},
|
|
{
|
|
"epoch": 2.54,
|
|
"learning_rate": 5.087066779374472e-08,
|
|
"logits/chosen": -1.3880281448364258,
|
|
"logits/rejected": -1.3039556741714478,
|
|
"logps/chosen": -93.82992553710938,
|
|
"logps/rejected": -118.81803131103516,
|
|
"loss": 0.0076,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1777141094207764,
|
|
"rewards/margins": 27.936527252197266,
|
|
"rewards/rejected": -26.758813858032227,
|
|
"step": 5570
|
|
},
|
|
{
|
|
"epoch": 2.55,
|
|
"learning_rate": 5.036348267117498e-08,
|
|
"logits/chosen": -1.4143790006637573,
|
|
"logits/rejected": -1.3327438831329346,
|
|
"logps/chosen": -90.89906311035156,
|
|
"logps/rejected": -125.50030517578125,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.46656179428100586,
|
|
"rewards/margins": 29.756689071655273,
|
|
"rewards/rejected": -29.290124893188477,
|
|
"step": 5580
|
|
},
|
|
{
|
|
"epoch": 2.55,
|
|
"learning_rate": 4.985629754860524e-08,
|
|
"logits/chosen": -1.3946322202682495,
|
|
"logits/rejected": -1.3158382177352905,
|
|
"logps/chosen": -93.13166809082031,
|
|
"logps/rejected": -120.45831298828125,
|
|
"loss": 0.0014,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.6907525062561035,
|
|
"rewards/margins": 27.038888931274414,
|
|
"rewards/rejected": -27.72964096069336,
|
|
"step": 5590
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"learning_rate": 4.93491124260355e-08,
|
|
"logits/chosen": -1.444240927696228,
|
|
"logits/rejected": -1.3477418422698975,
|
|
"logps/chosen": -92.03813171386719,
|
|
"logps/rejected": -123.6197280883789,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.03295283392071724,
|
|
"rewards/margins": 28.512805938720703,
|
|
"rewards/rejected": -28.479854583740234,
|
|
"step": 5600
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"eval_logits/chosen": -1.5288245677947998,
|
|
"eval_logits/rejected": -1.4369823932647705,
|
|
"eval_logps/chosen": -89.01703643798828,
|
|
"eval_logps/rejected": -120.62100982666016,
|
|
"eval_loss": 0.007766298484057188,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.2076922357082367,
|
|
"eval_rewards/margins": 27.84734535217285,
|
|
"eval_rewards/rejected": -27.639650344848633,
|
|
"eval_runtime": 65.4543,
|
|
"eval_samples_per_second": 43.725,
|
|
"eval_steps_per_second": 2.735,
|
|
"step": 5600
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"learning_rate": 4.884192730346576e-08,
|
|
"logits/chosen": -1.4230222702026367,
|
|
"logits/rejected": -1.3513391017913818,
|
|
"logps/chosen": -90.27880096435547,
|
|
"logps/rejected": -119.94172668457031,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.44236135482788086,
|
|
"rewards/margins": 28.575183868408203,
|
|
"rewards/rejected": -28.132823944091797,
|
|
"step": 5610
|
|
},
|
|
{
|
|
"epoch": 2.57,
|
|
"learning_rate": 4.833474218089602e-08,
|
|
"logits/chosen": -1.4552013874053955,
|
|
"logits/rejected": -1.3836065530776978,
|
|
"logps/chosen": -87.5635986328125,
|
|
"logps/rejected": -122.162353515625,
|
|
"loss": 0.0065,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -1.6596953868865967,
|
|
"rewards/margins": 26.969858169555664,
|
|
"rewards/rejected": -28.629552841186523,
|
|
"step": 5620
|
|
},
|
|
{
|
|
"epoch": 2.57,
|
|
"learning_rate": 4.782755705832629e-08,
|
|
"logits/chosen": -1.4242037534713745,
|
|
"logits/rejected": -1.3452680110931396,
|
|
"logps/chosen": -88.09815979003906,
|
|
"logps/rejected": -124.96726989746094,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3257520198822021,
|
|
"rewards/margins": 29.73336410522461,
|
|
"rewards/rejected": -28.407611846923828,
|
|
"step": 5630
|
|
},
|
|
{
|
|
"epoch": 2.57,
|
|
"learning_rate": 4.732037193575655e-08,
|
|
"logits/chosen": -1.4236841201782227,
|
|
"logits/rejected": -1.3491275310516357,
|
|
"logps/chosen": -91.39635467529297,
|
|
"logps/rejected": -121.4232406616211,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.2293356955051422,
|
|
"rewards/margins": 27.425186157226562,
|
|
"rewards/rejected": -27.195850372314453,
|
|
"step": 5640
|
|
},
|
|
{
|
|
"epoch": 2.58,
|
|
"learning_rate": 4.681318681318681e-08,
|
|
"logits/chosen": -1.4207613468170166,
|
|
"logits/rejected": -1.3308521509170532,
|
|
"logps/chosen": -91.08123779296875,
|
|
"logps/rejected": -118.93770599365234,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.274003028869629,
|
|
"rewards/margins": 28.229358673095703,
|
|
"rewards/rejected": -26.955352783203125,
|
|
"step": 5650
|
|
},
|
|
{
|
|
"epoch": 2.58,
|
|
"learning_rate": 4.630600169061707e-08,
|
|
"logits/chosen": -1.4237556457519531,
|
|
"logits/rejected": -1.3543760776519775,
|
|
"logps/chosen": -85.66486358642578,
|
|
"logps/rejected": -124.79557037353516,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.14971943199634552,
|
|
"rewards/margins": 28.228811264038086,
|
|
"rewards/rejected": -28.079092025756836,
|
|
"step": 5660
|
|
},
|
|
{
|
|
"epoch": 2.59,
|
|
"learning_rate": 4.579881656804733e-08,
|
|
"logits/chosen": -1.452664852142334,
|
|
"logits/rejected": -1.3631222248077393,
|
|
"logps/chosen": -93.4820556640625,
|
|
"logps/rejected": -123.2723617553711,
|
|
"loss": 0.0035,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.7186456918716431,
|
|
"rewards/margins": 28.007343292236328,
|
|
"rewards/rejected": -28.72598648071289,
|
|
"step": 5670
|
|
},
|
|
{
|
|
"epoch": 2.59,
|
|
"learning_rate": 4.5291631445477594e-08,
|
|
"logits/chosen": -1.4738563299179077,
|
|
"logits/rejected": -1.396863341331482,
|
|
"logps/chosen": -85.08771514892578,
|
|
"logps/rejected": -125.86419677734375,
|
|
"loss": 0.0014,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.7788127660751343,
|
|
"rewards/margins": 29.58379554748535,
|
|
"rewards/rejected": -28.804983139038086,
|
|
"step": 5680
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"learning_rate": 4.478444632290786e-08,
|
|
"logits/chosen": -1.4112271070480347,
|
|
"logits/rejected": -1.342129111289978,
|
|
"logps/chosen": -93.12382507324219,
|
|
"logps/rejected": -132.10922241210938,
|
|
"loss": 0.0,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.42668676376342773,
|
|
"rewards/margins": 29.515426635742188,
|
|
"rewards/rejected": -29.0887393951416,
|
|
"step": 5690
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"learning_rate": 4.427726120033812e-08,
|
|
"logits/chosen": -1.4182324409484863,
|
|
"logits/rejected": -1.3355042934417725,
|
|
"logps/chosen": -95.10716247558594,
|
|
"logps/rejected": -121.40716552734375,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.24203085899353027,
|
|
"rewards/margins": 27.118215560913086,
|
|
"rewards/rejected": -27.360248565673828,
|
|
"step": 5700
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"eval_logits/chosen": -1.521308422088623,
|
|
"eval_logits/rejected": -1.4287875890731812,
|
|
"eval_logps/chosen": -88.81271362304688,
|
|
"eval_logps/rejected": -120.16915893554688,
|
|
"eval_loss": 0.007557415869086981,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.3098601996898651,
|
|
"eval_rewards/margins": 27.723575592041016,
|
|
"eval_rewards/rejected": -27.413715362548828,
|
|
"eval_runtime": 64.7947,
|
|
"eval_samples_per_second": 44.17,
|
|
"eval_steps_per_second": 2.763,
|
|
"step": 5700
|
|
},
|
|
{
|
|
"epoch": 2.61,
|
|
"learning_rate": 4.377007607776838e-08,
|
|
"logits/chosen": -1.4278671741485596,
|
|
"logits/rejected": -1.3431296348571777,
|
|
"logps/chosen": -88.24993133544922,
|
|
"logps/rejected": -121.37858581542969,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.6779876947402954,
|
|
"rewards/margins": 29.29281997680664,
|
|
"rewards/rejected": -27.614831924438477,
|
|
"step": 5710
|
|
},
|
|
{
|
|
"epoch": 2.61,
|
|
"learning_rate": 4.3262890955198644e-08,
|
|
"logits/chosen": -1.3866733312606812,
|
|
"logits/rejected": -1.3263394832611084,
|
|
"logps/chosen": -89.70895385742188,
|
|
"logps/rejected": -118.44575500488281,
|
|
"loss": 0.0,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.36240631341934204,
|
|
"rewards/margins": 26.25311851501465,
|
|
"rewards/rejected": -25.89071273803711,
|
|
"step": 5720
|
|
},
|
|
{
|
|
"epoch": 2.62,
|
|
"learning_rate": 4.2755705832628905e-08,
|
|
"logits/chosen": -1.4577974081039429,
|
|
"logits/rejected": -1.3561092615127563,
|
|
"logps/chosen": -90.57540893554688,
|
|
"logps/rejected": -122.50830078125,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9288503527641296,
|
|
"rewards/margins": 29.591562271118164,
|
|
"rewards/rejected": -28.662710189819336,
|
|
"step": 5730
|
|
},
|
|
{
|
|
"epoch": 2.62,
|
|
"learning_rate": 4.2248520710059166e-08,
|
|
"logits/chosen": -1.430964708328247,
|
|
"logits/rejected": -1.362330675125122,
|
|
"logps/chosen": -93.49796295166016,
|
|
"logps/rejected": -127.06739807128906,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.9624999761581421,
|
|
"rewards/chosen": 0.5318681001663208,
|
|
"rewards/margins": 28.357601165771484,
|
|
"rewards/rejected": -27.82573890686035,
|
|
"step": 5740
|
|
},
|
|
{
|
|
"epoch": 2.62,
|
|
"learning_rate": 4.1741335587489433e-08,
|
|
"logits/chosen": -1.4357731342315674,
|
|
"logits/rejected": -1.3574306964874268,
|
|
"logps/chosen": -88.33089447021484,
|
|
"logps/rejected": -124.7070083618164,
|
|
"loss": 0.0024,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.2313622683286667,
|
|
"rewards/margins": 29.021398544311523,
|
|
"rewards/rejected": -28.790035247802734,
|
|
"step": 5750
|
|
},
|
|
{
|
|
"epoch": 2.63,
|
|
"learning_rate": 4.1234150464919694e-08,
|
|
"logits/chosen": -1.4213526248931885,
|
|
"logits/rejected": -1.364757776260376,
|
|
"logps/chosen": -81.57777404785156,
|
|
"logps/rejected": -119.826171875,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9536989331245422,
|
|
"rewards/margins": 27.143407821655273,
|
|
"rewards/rejected": -26.189708709716797,
|
|
"step": 5760
|
|
},
|
|
{
|
|
"epoch": 2.63,
|
|
"learning_rate": 4.0726965342349955e-08,
|
|
"logits/chosen": -1.4012161493301392,
|
|
"logits/rejected": -1.3225289583206177,
|
|
"logps/chosen": -91.90386199951172,
|
|
"logps/rejected": -119.8282241821289,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.7034690380096436,
|
|
"rewards/margins": 28.835796356201172,
|
|
"rewards/rejected": -27.132328033447266,
|
|
"step": 5770
|
|
},
|
|
{
|
|
"epoch": 2.64,
|
|
"learning_rate": 4.0219780219780216e-08,
|
|
"logits/chosen": -1.4080169200897217,
|
|
"logits/rejected": -1.338749647140503,
|
|
"logps/chosen": -94.41007995605469,
|
|
"logps/rejected": -125.92796325683594,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.140974998474121,
|
|
"rewards/margins": 26.483749389648438,
|
|
"rewards/rejected": -25.342771530151367,
|
|
"step": 5780
|
|
},
|
|
{
|
|
"epoch": 2.64,
|
|
"learning_rate": 3.971259509721048e-08,
|
|
"logits/chosen": -1.4085487127304077,
|
|
"logits/rejected": -1.327182650566101,
|
|
"logps/chosen": -88.75342559814453,
|
|
"logps/rejected": -119.1405029296875,
|
|
"loss": 0.0054,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0149381160736084,
|
|
"rewards/margins": 27.902973175048828,
|
|
"rewards/rejected": -26.888031005859375,
|
|
"step": 5790
|
|
},
|
|
{
|
|
"epoch": 2.65,
|
|
"learning_rate": 3.920540997464074e-08,
|
|
"logits/chosen": -1.433281421661377,
|
|
"logits/rejected": -1.3512113094329834,
|
|
"logps/chosen": -90.19625854492188,
|
|
"logps/rejected": -126.81318664550781,
|
|
"loss": 0.0054,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6478980779647827,
|
|
"rewards/margins": 29.77631187438965,
|
|
"rewards/rejected": -29.12841796875,
|
|
"step": 5800
|
|
},
|
|
{
|
|
"epoch": 2.65,
|
|
"eval_logits/chosen": -1.5241363048553467,
|
|
"eval_logits/rejected": -1.4326359033584595,
|
|
"eval_logps/chosen": -88.73265838623047,
|
|
"eval_logps/rejected": -120.24764251708984,
|
|
"eval_loss": 0.007415792904794216,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.3498854637145996,
|
|
"eval_rewards/margins": 27.802854537963867,
|
|
"eval_rewards/rejected": -27.452966690063477,
|
|
"eval_runtime": 66.3571,
|
|
"eval_samples_per_second": 43.13,
|
|
"eval_steps_per_second": 2.698,
|
|
"step": 5800
|
|
},
|
|
{
|
|
"epoch": 2.65,
|
|
"learning_rate": 3.8698224852071005e-08,
|
|
"logits/chosen": -1.3824363946914673,
|
|
"logits/rejected": -1.3102766275405884,
|
|
"logps/chosen": -90.2943115234375,
|
|
"logps/rejected": -120.10107421875,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4802871346473694,
|
|
"rewards/margins": 27.660263061523438,
|
|
"rewards/rejected": -27.179973602294922,
|
|
"step": 5810
|
|
},
|
|
{
|
|
"epoch": 2.66,
|
|
"learning_rate": 3.8191039729501266e-08,
|
|
"logits/chosen": -1.405815601348877,
|
|
"logits/rejected": -1.3217995166778564,
|
|
"logps/chosen": -83.80717468261719,
|
|
"logps/rejected": -120.4073715209961,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.3315509557724,
|
|
"rewards/margins": 28.07989501953125,
|
|
"rewards/rejected": -26.748342514038086,
|
|
"step": 5820
|
|
},
|
|
{
|
|
"epoch": 2.66,
|
|
"learning_rate": 3.768385460693153e-08,
|
|
"logits/chosen": -1.400803565979004,
|
|
"logits/rejected": -1.3341869115829468,
|
|
"logps/chosen": -85.88481140136719,
|
|
"logps/rejected": -122.70805358886719,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.08723556995391846,
|
|
"rewards/margins": 27.86008071899414,
|
|
"rewards/rejected": -27.947315216064453,
|
|
"step": 5830
|
|
},
|
|
{
|
|
"epoch": 2.67,
|
|
"learning_rate": 3.717666948436179e-08,
|
|
"logits/chosen": -1.4672653675079346,
|
|
"logits/rejected": -1.3654979467391968,
|
|
"logps/chosen": -88.60813903808594,
|
|
"logps/rejected": -124.53550720214844,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.0332581996917725,
|
|
"rewards/margins": 29.575796127319336,
|
|
"rewards/rejected": -28.54253578186035,
|
|
"step": 5840
|
|
},
|
|
{
|
|
"epoch": 2.67,
|
|
"learning_rate": 3.6669484361792056e-08,
|
|
"logits/chosen": -1.4773852825164795,
|
|
"logits/rejected": -1.3942975997924805,
|
|
"logps/chosen": -84.61897277832031,
|
|
"logps/rejected": -118.28257751464844,
|
|
"loss": 0.0059,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 1.075499415397644,
|
|
"rewards/margins": 27.52908706665039,
|
|
"rewards/rejected": -26.45358657836914,
|
|
"step": 5850
|
|
},
|
|
{
|
|
"epoch": 2.67,
|
|
"learning_rate": 3.6162299239222316e-08,
|
|
"logits/chosen": -1.4367072582244873,
|
|
"logits/rejected": -1.3531558513641357,
|
|
"logps/chosen": -92.48080444335938,
|
|
"logps/rejected": -124.60569763183594,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.9651349782943726,
|
|
"rewards/margins": 30.93218421936035,
|
|
"rewards/rejected": -28.9670467376709,
|
|
"step": 5860
|
|
},
|
|
{
|
|
"epoch": 2.68,
|
|
"learning_rate": 3.565511411665258e-08,
|
|
"logits/chosen": -1.4036922454833984,
|
|
"logits/rejected": -1.3271372318267822,
|
|
"logps/chosen": -91.8263168334961,
|
|
"logps/rejected": -123.1822509765625,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.49861449003219604,
|
|
"rewards/margins": 27.064804077148438,
|
|
"rewards/rejected": -26.56618881225586,
|
|
"step": 5870
|
|
},
|
|
{
|
|
"epoch": 2.68,
|
|
"learning_rate": 3.514792899408284e-08,
|
|
"logits/chosen": -1.3846232891082764,
|
|
"logits/rejected": -1.3122532367706299,
|
|
"logps/chosen": -84.51948547363281,
|
|
"logps/rejected": -121.88045501708984,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.880649209022522,
|
|
"rewards/margins": 28.404842376708984,
|
|
"rewards/rejected": -27.524194717407227,
|
|
"step": 5880
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"learning_rate": 3.46407438715131e-08,
|
|
"logits/chosen": -1.4396214485168457,
|
|
"logits/rejected": -1.3636229038238525,
|
|
"logps/chosen": -93.21275329589844,
|
|
"logps/rejected": -124.11541748046875,
|
|
"loss": 0.0087,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": 0.49375027418136597,
|
|
"rewards/margins": 28.062625885009766,
|
|
"rewards/rejected": -27.568878173828125,
|
|
"step": 5890
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"learning_rate": 3.413355874894336e-08,
|
|
"logits/chosen": -1.4009182453155518,
|
|
"logits/rejected": -1.3155372142791748,
|
|
"logps/chosen": -90.7492446899414,
|
|
"logps/rejected": -121.24101257324219,
|
|
"loss": 0.0035,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.570563018321991,
|
|
"rewards/margins": 28.15814781188965,
|
|
"rewards/rejected": -27.587589263916016,
|
|
"step": 5900
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"eval_logits/chosen": -1.5298057794570923,
|
|
"eval_logits/rejected": -1.4371873140335083,
|
|
"eval_logps/chosen": -88.87010955810547,
|
|
"eval_logps/rejected": -120.50233459472656,
|
|
"eval_loss": 0.007439317647367716,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.28115907311439514,
|
|
"eval_rewards/margins": 27.861465454101562,
|
|
"eval_rewards/rejected": -27.580307006835938,
|
|
"eval_runtime": 61.2948,
|
|
"eval_samples_per_second": 46.692,
|
|
"eval_steps_per_second": 2.92,
|
|
"step": 5900
|
|
},
|
|
{
|
|
"epoch": 2.7,
|
|
"learning_rate": 3.362637362637363e-08,
|
|
"logits/chosen": -1.386823058128357,
|
|
"logits/rejected": -1.317825198173523,
|
|
"logps/chosen": -88.96937561035156,
|
|
"logps/rejected": -118.41639709472656,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.38685184717178345,
|
|
"rewards/margins": 26.644840240478516,
|
|
"rewards/rejected": -26.257986068725586,
|
|
"step": 5910
|
|
},
|
|
{
|
|
"epoch": 2.7,
|
|
"learning_rate": 3.311918850380389e-08,
|
|
"logits/chosen": -1.4166361093521118,
|
|
"logits/rejected": -1.3327529430389404,
|
|
"logps/chosen": -91.54283905029297,
|
|
"logps/rejected": -125.4769515991211,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.6538678407669067,
|
|
"rewards/margins": 29.736963272094727,
|
|
"rewards/rejected": -29.083093643188477,
|
|
"step": 5920
|
|
},
|
|
{
|
|
"epoch": 2.71,
|
|
"learning_rate": 3.261200338123415e-08,
|
|
"logits/chosen": -1.4290189743041992,
|
|
"logits/rejected": -1.3503470420837402,
|
|
"logps/chosen": -93.33061218261719,
|
|
"logps/rejected": -120.43190002441406,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5092540979385376,
|
|
"rewards/margins": 27.29727554321289,
|
|
"rewards/rejected": -25.788021087646484,
|
|
"step": 5930
|
|
},
|
|
{
|
|
"epoch": 2.71,
|
|
"learning_rate": 3.210481825866441e-08,
|
|
"logits/chosen": -1.4096345901489258,
|
|
"logits/rejected": -1.3365631103515625,
|
|
"logps/chosen": -92.1152114868164,
|
|
"logps/rejected": -129.61373901367188,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.6155846118927002,
|
|
"rewards/margins": 27.218673706054688,
|
|
"rewards/rejected": -27.83425521850586,
|
|
"step": 5940
|
|
},
|
|
{
|
|
"epoch": 2.72,
|
|
"learning_rate": 3.159763313609467e-08,
|
|
"logits/chosen": -1.4408024549484253,
|
|
"logits/rejected": -1.358703851699829,
|
|
"logps/chosen": -92.8883056640625,
|
|
"logps/rejected": -127.37141418457031,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.4508030414581299,
|
|
"rewards/margins": 29.662633895874023,
|
|
"rewards/rejected": -28.211828231811523,
|
|
"step": 5950
|
|
},
|
|
{
|
|
"epoch": 2.72,
|
|
"learning_rate": 3.109044801352493e-08,
|
|
"logits/chosen": -1.4268407821655273,
|
|
"logits/rejected": -1.3649609088897705,
|
|
"logps/chosen": -91.47252655029297,
|
|
"logps/rejected": -127.54573059082031,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9097731709480286,
|
|
"rewards/margins": 28.423721313476562,
|
|
"rewards/rejected": -27.51395034790039,
|
|
"step": 5960
|
|
},
|
|
{
|
|
"epoch": 2.72,
|
|
"learning_rate": 3.05832628909552e-08,
|
|
"logits/chosen": -1.427631139755249,
|
|
"logits/rejected": -1.349169373512268,
|
|
"logps/chosen": -88.80760192871094,
|
|
"logps/rejected": -119.60186767578125,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.3762190341949463,
|
|
"rewards/margins": 28.201892852783203,
|
|
"rewards/rejected": -26.825674057006836,
|
|
"step": 5970
|
|
},
|
|
{
|
|
"epoch": 2.73,
|
|
"learning_rate": 3.007607776838546e-08,
|
|
"logits/chosen": -1.433571219444275,
|
|
"logits/rejected": -1.3482530117034912,
|
|
"logps/chosen": -92.32679748535156,
|
|
"logps/rejected": -124.57658386230469,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.08425948768854141,
|
|
"rewards/margins": 28.501611709594727,
|
|
"rewards/rejected": -28.4173526763916,
|
|
"step": 5980
|
|
},
|
|
{
|
|
"epoch": 2.73,
|
|
"learning_rate": 2.956889264581572e-08,
|
|
"logits/chosen": -1.4247334003448486,
|
|
"logits/rejected": -1.3412024974822998,
|
|
"logps/chosen": -97.79167938232422,
|
|
"logps/rejected": -129.82369995117188,
|
|
"loss": 0.0045,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.341404676437378,
|
|
"rewards/margins": 30.7274169921875,
|
|
"rewards/rejected": -29.38601303100586,
|
|
"step": 5990
|
|
},
|
|
{
|
|
"epoch": 2.74,
|
|
"learning_rate": 2.9061707523245986e-08,
|
|
"logits/chosen": -1.4527391195297241,
|
|
"logits/rejected": -1.3852466344833374,
|
|
"logps/chosen": -84.54109191894531,
|
|
"logps/rejected": -123.23868560791016,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4699872136116028,
|
|
"rewards/margins": 28.9207706451416,
|
|
"rewards/rejected": -28.450780868530273,
|
|
"step": 6000
|
|
},
|
|
{
|
|
"epoch": 2.74,
|
|
"eval_logits/chosen": -1.5361018180847168,
|
|
"eval_logits/rejected": -1.4470702409744263,
|
|
"eval_logps/chosen": -88.81427764892578,
|
|
"eval_logps/rejected": -120.10942077636719,
|
|
"eval_loss": 0.0073595428839325905,
|
|
"eval_rewards/accuracies": 0.994413435459137,
|
|
"eval_rewards/chosen": 0.30907416343688965,
|
|
"eval_rewards/margins": 27.692928314208984,
|
|
"eval_rewards/rejected": -27.383853912353516,
|
|
"eval_runtime": 66.1237,
|
|
"eval_samples_per_second": 43.283,
|
|
"eval_steps_per_second": 2.707,
|
|
"step": 6000
|
|
},
|
|
{
|
|
"epoch": 2.74,
|
|
"learning_rate": 2.8554522400676247e-08,
|
|
"logits/chosen": -1.437772512435913,
|
|
"logits/rejected": -1.3702958822250366,
|
|
"logps/chosen": -88.56996154785156,
|
|
"logps/rejected": -120.36534118652344,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.8493738174438477,
|
|
"rewards/margins": 25.939929962158203,
|
|
"rewards/rejected": -26.789306640625,
|
|
"step": 6010
|
|
},
|
|
{
|
|
"epoch": 2.75,
|
|
"learning_rate": 2.8047337278106507e-08,
|
|
"logits/chosen": -1.3986080884933472,
|
|
"logits/rejected": -1.3344132900238037,
|
|
"logps/chosen": -87.32596588134766,
|
|
"logps/rejected": -124.3438949584961,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.7483171820640564,
|
|
"rewards/margins": 28.974384307861328,
|
|
"rewards/rejected": -28.2260684967041,
|
|
"step": 6020
|
|
},
|
|
{
|
|
"epoch": 2.75,
|
|
"learning_rate": 2.754015215553677e-08,
|
|
"logits/chosen": -1.4268220663070679,
|
|
"logits/rejected": -1.3452110290527344,
|
|
"logps/chosen": -91.21931457519531,
|
|
"logps/rejected": -122.6042709350586,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.31938114762306213,
|
|
"rewards/margins": 27.742395401000977,
|
|
"rewards/rejected": -27.423009872436523,
|
|
"step": 6030
|
|
},
|
|
{
|
|
"epoch": 2.76,
|
|
"learning_rate": 2.7032967032967033e-08,
|
|
"logits/chosen": -1.4078203439712524,
|
|
"logits/rejected": -1.341101884841919,
|
|
"logps/chosen": -88.9620132446289,
|
|
"logps/rejected": -126.86651611328125,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.03730924054980278,
|
|
"rewards/margins": 28.27874755859375,
|
|
"rewards/rejected": -28.24143409729004,
|
|
"step": 6040
|
|
},
|
|
{
|
|
"epoch": 2.76,
|
|
"learning_rate": 2.6525781910397293e-08,
|
|
"logits/chosen": -1.4218459129333496,
|
|
"logits/rejected": -1.3417071104049683,
|
|
"logps/chosen": -94.39894104003906,
|
|
"logps/rejected": -126.17645263671875,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.41776663064956665,
|
|
"rewards/margins": 27.610363006591797,
|
|
"rewards/rejected": -27.192596435546875,
|
|
"step": 6050
|
|
},
|
|
{
|
|
"epoch": 2.77,
|
|
"learning_rate": 2.6018596787827558e-08,
|
|
"logits/chosen": -1.3829963207244873,
|
|
"logits/rejected": -1.3349157571792603,
|
|
"logps/chosen": -81.25392150878906,
|
|
"logps/rejected": -116.95680236816406,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4236263334751129,
|
|
"rewards/margins": 25.6158447265625,
|
|
"rewards/rejected": -25.192218780517578,
|
|
"step": 6060
|
|
},
|
|
{
|
|
"epoch": 2.77,
|
|
"learning_rate": 2.551141166525782e-08,
|
|
"logits/chosen": -1.4441430568695068,
|
|
"logits/rejected": -1.3542683124542236,
|
|
"logps/chosen": -94.20352172851562,
|
|
"logps/rejected": -124.4933090209961,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.7039328813552856,
|
|
"rewards/margins": 27.073434829711914,
|
|
"rewards/rejected": -26.369503021240234,
|
|
"step": 6070
|
|
},
|
|
{
|
|
"epoch": 2.77,
|
|
"learning_rate": 2.500422654268808e-08,
|
|
"logits/chosen": -1.4258487224578857,
|
|
"logits/rejected": -1.357102632522583,
|
|
"logps/chosen": -86.0672607421875,
|
|
"logps/rejected": -125.6698226928711,
|
|
"loss": 0.0034,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5518980026245117,
|
|
"rewards/margins": 29.156414031982422,
|
|
"rewards/rejected": -28.604516983032227,
|
|
"step": 6080
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"learning_rate": 2.4497041420118344e-08,
|
|
"logits/chosen": -1.44197678565979,
|
|
"logits/rejected": -1.3703984022140503,
|
|
"logps/chosen": -84.29279327392578,
|
|
"logps/rejected": -122.15272521972656,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0316746234893799,
|
|
"rewards/margins": 28.804452896118164,
|
|
"rewards/rejected": -27.772777557373047,
|
|
"step": 6090
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"learning_rate": 2.3989856297548605e-08,
|
|
"logits/chosen": -1.4140945672988892,
|
|
"logits/rejected": -1.3373467922210693,
|
|
"logps/chosen": -92.00602722167969,
|
|
"logps/rejected": -121.0610122680664,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.4240033030509949,
|
|
"rewards/margins": 27.450969696044922,
|
|
"rewards/rejected": -27.026966094970703,
|
|
"step": 6100
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"eval_logits/chosen": -1.528950810432434,
|
|
"eval_logits/rejected": -1.4393686056137085,
|
|
"eval_logps/chosen": -88.54615783691406,
|
|
"eval_logps/rejected": -119.571533203125,
|
|
"eval_loss": 0.007185075432062149,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.4431401193141937,
|
|
"eval_rewards/margins": 27.55805015563965,
|
|
"eval_rewards/rejected": -27.114912033081055,
|
|
"eval_runtime": 67.6666,
|
|
"eval_samples_per_second": 42.296,
|
|
"eval_steps_per_second": 2.645,
|
|
"step": 6100
|
|
},
|
|
{
|
|
"epoch": 2.79,
|
|
"learning_rate": 2.3482671174978865e-08,
|
|
"logits/chosen": -1.3959143161773682,
|
|
"logits/rejected": -1.329087495803833,
|
|
"logps/chosen": -90.89431762695312,
|
|
"logps/rejected": -123.38334655761719,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.9750000238418579,
|
|
"rewards/chosen": -1.0483791828155518,
|
|
"rewards/margins": 25.339763641357422,
|
|
"rewards/rejected": -26.388141632080078,
|
|
"step": 6110
|
|
},
|
|
{
|
|
"epoch": 2.79,
|
|
"learning_rate": 2.297548605240913e-08,
|
|
"logits/chosen": -1.4264185428619385,
|
|
"logits/rejected": -1.3347035646438599,
|
|
"logps/chosen": -95.05891418457031,
|
|
"logps/rejected": -122.20552825927734,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.40551090240478516,
|
|
"rewards/margins": 27.485652923583984,
|
|
"rewards/rejected": -27.080142974853516,
|
|
"step": 6120
|
|
},
|
|
{
|
|
"epoch": 2.8,
|
|
"learning_rate": 2.246830092983939e-08,
|
|
"logits/chosen": -1.4183063507080078,
|
|
"logits/rejected": -1.3475987911224365,
|
|
"logps/chosen": -90.90924835205078,
|
|
"logps/rejected": -123.7392349243164,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.11213073879480362,
|
|
"rewards/margins": 26.713592529296875,
|
|
"rewards/rejected": -26.82572364807129,
|
|
"step": 6130
|
|
},
|
|
{
|
|
"epoch": 2.8,
|
|
"learning_rate": 2.196111580726965e-08,
|
|
"logits/chosen": -1.3835358619689941,
|
|
"logits/rejected": -1.304900884628296,
|
|
"logps/chosen": -94.07560729980469,
|
|
"logps/rejected": -119.58497619628906,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.7558839321136475,
|
|
"rewards/margins": 26.48470115661621,
|
|
"rewards/rejected": -25.72881507873535,
|
|
"step": 6140
|
|
},
|
|
{
|
|
"epoch": 2.81,
|
|
"learning_rate": 2.1453930684699916e-08,
|
|
"logits/chosen": -1.4276055097579956,
|
|
"logits/rejected": -1.356466293334961,
|
|
"logps/chosen": -92.86775207519531,
|
|
"logps/rejected": -124.24592590332031,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.07010960578918457,
|
|
"rewards/margins": 28.186676025390625,
|
|
"rewards/rejected": -28.25678062438965,
|
|
"step": 6150
|
|
},
|
|
{
|
|
"epoch": 2.81,
|
|
"learning_rate": 2.0946745562130177e-08,
|
|
"logits/chosen": -1.3897449970245361,
|
|
"logits/rejected": -1.3085477352142334,
|
|
"logps/chosen": -89.38758850097656,
|
|
"logps/rejected": -123.3631820678711,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.502859890460968,
|
|
"rewards/margins": 29.507598876953125,
|
|
"rewards/rejected": -29.00473976135254,
|
|
"step": 6160
|
|
},
|
|
{
|
|
"epoch": 2.82,
|
|
"learning_rate": 2.0439560439560437e-08,
|
|
"logits/chosen": -1.407470464706421,
|
|
"logits/rejected": -1.3406095504760742,
|
|
"logps/chosen": -93.42985534667969,
|
|
"logps/rejected": -124.22257995605469,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.7225341200828552,
|
|
"rewards/margins": 28.852397918701172,
|
|
"rewards/rejected": -28.12986183166504,
|
|
"step": 6170
|
|
},
|
|
{
|
|
"epoch": 2.82,
|
|
"learning_rate": 1.99323753169907e-08,
|
|
"logits/chosen": -1.3827449083328247,
|
|
"logits/rejected": -1.3094285726547241,
|
|
"logps/chosen": -93.71525573730469,
|
|
"logps/rejected": -119.5619125366211,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.6188391447067261,
|
|
"rewards/margins": 26.20868492126465,
|
|
"rewards/rejected": -25.58984375,
|
|
"step": 6180
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"learning_rate": 1.9425190194420963e-08,
|
|
"logits/chosen": -1.3776720762252808,
|
|
"logits/rejected": -1.3083680868148804,
|
|
"logps/chosen": -91.71514129638672,
|
|
"logps/rejected": -120.59306335449219,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.6132274270057678,
|
|
"rewards/margins": 26.796030044555664,
|
|
"rewards/rejected": -27.409259796142578,
|
|
"step": 6190
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"learning_rate": 1.8918005071851223e-08,
|
|
"logits/chosen": -1.4177929162979126,
|
|
"logits/rejected": -1.3377645015716553,
|
|
"logps/chosen": -87.91746520996094,
|
|
"logps/rejected": -118.75276184082031,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8456829786300659,
|
|
"rewards/margins": 27.677001953125,
|
|
"rewards/rejected": -26.831317901611328,
|
|
"step": 6200
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"eval_logits/chosen": -1.5227910280227661,
|
|
"eval_logits/rejected": -1.4297844171524048,
|
|
"eval_logps/chosen": -88.48534393310547,
|
|
"eval_logps/rejected": -119.60254669189453,
|
|
"eval_loss": 0.00707243150100112,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.47354137897491455,
|
|
"eval_rewards/margins": 27.603960037231445,
|
|
"eval_rewards/rejected": -27.130416870117188,
|
|
"eval_runtime": 65.0475,
|
|
"eval_samples_per_second": 43.999,
|
|
"eval_steps_per_second": 2.752,
|
|
"step": 6200
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"learning_rate": 1.8410819949281488e-08,
|
|
"logits/chosen": -1.4211013317108154,
|
|
"logits/rejected": -1.336397409439087,
|
|
"logps/chosen": -89.54491424560547,
|
|
"logps/rejected": -122.1443862915039,
|
|
"loss": 0.0054,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.5423094034194946,
|
|
"rewards/margins": 28.126476287841797,
|
|
"rewards/rejected": -26.58416748046875,
|
|
"step": 6210
|
|
},
|
|
{
|
|
"epoch": 2.84,
|
|
"learning_rate": 1.790363482671175e-08,
|
|
"logits/chosen": -1.4224960803985596,
|
|
"logits/rejected": -1.3450286388397217,
|
|
"logps/chosen": -91.13011169433594,
|
|
"logps/rejected": -124.55561828613281,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.6848041415214539,
|
|
"rewards/margins": 27.732412338256836,
|
|
"rewards/rejected": -27.047607421875,
|
|
"step": 6220
|
|
},
|
|
{
|
|
"epoch": 2.84,
|
|
"learning_rate": 1.7396449704142013e-08,
|
|
"logits/chosen": -1.4057667255401611,
|
|
"logits/rejected": -1.3270689249038696,
|
|
"logps/chosen": -94.62673950195312,
|
|
"logps/rejected": -121.52132415771484,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.3413497507572174,
|
|
"rewards/margins": 27.458669662475586,
|
|
"rewards/rejected": -27.117321014404297,
|
|
"step": 6230
|
|
},
|
|
{
|
|
"epoch": 2.85,
|
|
"learning_rate": 1.6889264581572274e-08,
|
|
"logits/chosen": -1.433934211730957,
|
|
"logits/rejected": -1.343518614768982,
|
|
"logps/chosen": -90.16001892089844,
|
|
"logps/rejected": -122.25347900390625,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.359809547662735,
|
|
"rewards/margins": 28.131118774414062,
|
|
"rewards/rejected": -27.771310806274414,
|
|
"step": 6240
|
|
},
|
|
{
|
|
"epoch": 2.85,
|
|
"learning_rate": 1.6382079459002535e-08,
|
|
"logits/chosen": -1.4110838174819946,
|
|
"logits/rejected": -1.3541514873504639,
|
|
"logps/chosen": -91.34024810791016,
|
|
"logps/rejected": -128.32180786132812,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": -0.36720380187034607,
|
|
"rewards/margins": 27.321441650390625,
|
|
"rewards/rejected": -27.688648223876953,
|
|
"step": 6250
|
|
},
|
|
{
|
|
"epoch": 2.86,
|
|
"learning_rate": 1.58748943364328e-08,
|
|
"logits/chosen": -1.389146089553833,
|
|
"logits/rejected": -1.3081094026565552,
|
|
"logps/chosen": -88.67632293701172,
|
|
"logps/rejected": -115.4472885131836,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.8905091285705566,
|
|
"rewards/margins": 27.710601806640625,
|
|
"rewards/rejected": -26.82008934020996,
|
|
"step": 6260
|
|
},
|
|
{
|
|
"epoch": 2.86,
|
|
"learning_rate": 1.536770921386306e-08,
|
|
"logits/chosen": -1.3963115215301514,
|
|
"logits/rejected": -1.303027868270874,
|
|
"logps/chosen": -94.36498260498047,
|
|
"logps/rejected": -124.65643310546875,
|
|
"loss": 0.0012,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.32212838530540466,
|
|
"rewards/margins": 29.300546646118164,
|
|
"rewards/rejected": -28.97841453552246,
|
|
"step": 6270
|
|
},
|
|
{
|
|
"epoch": 2.87,
|
|
"learning_rate": 1.486052409129332e-08,
|
|
"logits/chosen": -1.4071121215820312,
|
|
"logits/rejected": -1.324487566947937,
|
|
"logps/chosen": -91.63459777832031,
|
|
"logps/rejected": -124.7200698852539,
|
|
"loss": 0.0013,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.18130116164684296,
|
|
"rewards/margins": 28.662633895874023,
|
|
"rewards/rejected": -28.481334686279297,
|
|
"step": 6280
|
|
},
|
|
{
|
|
"epoch": 2.87,
|
|
"learning_rate": 1.4353338968723583e-08,
|
|
"logits/chosen": -1.3938955068588257,
|
|
"logits/rejected": -1.3304332494735718,
|
|
"logps/chosen": -86.27269744873047,
|
|
"logps/rejected": -119.1937255859375,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.13392703235149384,
|
|
"rewards/margins": 26.717966079711914,
|
|
"rewards/rejected": -26.584041595458984,
|
|
"step": 6290
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"learning_rate": 1.3846153846153846e-08,
|
|
"logits/chosen": -1.4227640628814697,
|
|
"logits/rejected": -1.3536622524261475,
|
|
"logps/chosen": -89.34703063964844,
|
|
"logps/rejected": -123.28599548339844,
|
|
"loss": 0.0026,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.276293992996216,
|
|
"rewards/margins": 29.977630615234375,
|
|
"rewards/rejected": -27.701339721679688,
|
|
"step": 6300
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"eval_logits/chosen": -1.5174129009246826,
|
|
"eval_logits/rejected": -1.4249118566513062,
|
|
"eval_logps/chosen": -88.27791595458984,
|
|
"eval_logps/rejected": -119.14039611816406,
|
|
"eval_loss": 0.007073402404785156,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.5772566795349121,
|
|
"eval_rewards/margins": 27.476600646972656,
|
|
"eval_rewards/rejected": -26.89933967590332,
|
|
"eval_runtime": 65.0621,
|
|
"eval_samples_per_second": 43.989,
|
|
"eval_steps_per_second": 2.751,
|
|
"step": 6300
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"learning_rate": 1.3338968723584107e-08,
|
|
"logits/chosen": -1.4048776626586914,
|
|
"logits/rejected": -1.3209692239761353,
|
|
"logps/chosen": -94.9172592163086,
|
|
"logps/rejected": -121.06196594238281,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9303966760635376,
|
|
"rewards/margins": 27.744037628173828,
|
|
"rewards/rejected": -26.813640594482422,
|
|
"step": 6310
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"learning_rate": 1.2831783601014369e-08,
|
|
"logits/chosen": -1.4251675605773926,
|
|
"logits/rejected": -1.3501653671264648,
|
|
"logps/chosen": -87.30450439453125,
|
|
"logps/rejected": -124.9731674194336,
|
|
"loss": 0.0023,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.17765673995018005,
|
|
"rewards/margins": 27.48089599609375,
|
|
"rewards/rejected": -27.303241729736328,
|
|
"step": 6320
|
|
},
|
|
{
|
|
"epoch": 2.89,
|
|
"learning_rate": 1.2324598478444632e-08,
|
|
"logits/chosen": -1.3993405103683472,
|
|
"logits/rejected": -1.3213218450546265,
|
|
"logps/chosen": -89.23213958740234,
|
|
"logps/rejected": -119.50053405761719,
|
|
"loss": 0.0077,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.9308843612670898,
|
|
"rewards/margins": 29.413715362548828,
|
|
"rewards/rejected": -28.482830047607422,
|
|
"step": 6330
|
|
},
|
|
{
|
|
"epoch": 2.89,
|
|
"learning_rate": 1.1817413355874893e-08,
|
|
"logits/chosen": -1.3939188718795776,
|
|
"logits/rejected": -1.326827883720398,
|
|
"logps/chosen": -90.73262786865234,
|
|
"logps/rejected": -124.04630279541016,
|
|
"loss": 0.0043,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.2764853835105896,
|
|
"rewards/margins": 27.666738510131836,
|
|
"rewards/rejected": -27.390254974365234,
|
|
"step": 6340
|
|
},
|
|
{
|
|
"epoch": 2.9,
|
|
"learning_rate": 1.1310228233305155e-08,
|
|
"logits/chosen": -1.3985464572906494,
|
|
"logits/rejected": -1.3156406879425049,
|
|
"logps/chosen": -86.709716796875,
|
|
"logps/rejected": -120.27693176269531,
|
|
"loss": 0.0025,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.8232513666152954,
|
|
"rewards/margins": 27.919193267822266,
|
|
"rewards/rejected": -27.0959415435791,
|
|
"step": 6350
|
|
},
|
|
{
|
|
"epoch": 2.9,
|
|
"learning_rate": 1.080304311073542e-08,
|
|
"logits/chosen": -1.3861408233642578,
|
|
"logits/rejected": -1.3142074346542358,
|
|
"logps/chosen": -89.95799255371094,
|
|
"logps/rejected": -118.38623046875,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": -0.4884513318538666,
|
|
"rewards/margins": 26.449548721313477,
|
|
"rewards/rejected": -26.937997817993164,
|
|
"step": 6360
|
|
},
|
|
{
|
|
"epoch": 2.91,
|
|
"learning_rate": 1.0295857988165679e-08,
|
|
"logits/chosen": -1.4367835521697998,
|
|
"logits/rejected": -1.3561252355575562,
|
|
"logps/chosen": -91.15764617919922,
|
|
"logps/rejected": -122.2125473022461,
|
|
"loss": 0.0087,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 1.2589377164840698,
|
|
"rewards/margins": 28.840377807617188,
|
|
"rewards/rejected": -27.581439971923828,
|
|
"step": 6370
|
|
},
|
|
{
|
|
"epoch": 2.91,
|
|
"learning_rate": 9.788672865595943e-09,
|
|
"logits/chosen": -1.4210089445114136,
|
|
"logits/rejected": -1.3463882207870483,
|
|
"logps/chosen": -85.81571197509766,
|
|
"logps/rejected": -118.3035659790039,
|
|
"loss": 0.0055,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9070496559143066,
|
|
"rewards/margins": 27.036890029907227,
|
|
"rewards/rejected": -26.129840850830078,
|
|
"step": 6380
|
|
},
|
|
{
|
|
"epoch": 2.92,
|
|
"learning_rate": 9.281487743026204e-09,
|
|
"logits/chosen": -1.4335598945617676,
|
|
"logits/rejected": -1.355756163597107,
|
|
"logps/chosen": -90.24996185302734,
|
|
"logps/rejected": -118.03495025634766,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.1046497821807861,
|
|
"rewards/margins": 27.585540771484375,
|
|
"rewards/rejected": -26.480892181396484,
|
|
"step": 6390
|
|
},
|
|
{
|
|
"epoch": 2.92,
|
|
"learning_rate": 8.774302620456466e-09,
|
|
"logits/chosen": -1.3927547931671143,
|
|
"logits/rejected": -1.3146473169326782,
|
|
"logps/chosen": -85.73943328857422,
|
|
"logps/rejected": -122.47743225097656,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 2.633237600326538,
|
|
"rewards/margins": 30.070592880249023,
|
|
"rewards/rejected": -27.43735694885254,
|
|
"step": 6400
|
|
},
|
|
{
|
|
"epoch": 2.92,
|
|
"eval_logits/chosen": -1.5202581882476807,
|
|
"eval_logits/rejected": -1.428078293800354,
|
|
"eval_logps/chosen": -88.32048034667969,
|
|
"eval_logps/rejected": -119.28140258789062,
|
|
"eval_loss": 0.007134352345019579,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.5559725165367126,
|
|
"eval_rewards/margins": 27.525819778442383,
|
|
"eval_rewards/rejected": -26.969846725463867,
|
|
"eval_runtime": 69.0433,
|
|
"eval_samples_per_second": 41.452,
|
|
"eval_steps_per_second": 2.593,
|
|
"step": 6400
|
|
},
|
|
{
|
|
"epoch": 2.93,
|
|
"learning_rate": 8.267117497886729e-09,
|
|
"logits/chosen": -1.4425251483917236,
|
|
"logits/rejected": -1.3791589736938477,
|
|
"logps/chosen": -87.10059356689453,
|
|
"logps/rejected": -123.00959777832031,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5705853700637817,
|
|
"rewards/margins": 27.75478744506836,
|
|
"rewards/rejected": -27.1842041015625,
|
|
"step": 6410
|
|
},
|
|
{
|
|
"epoch": 2.93,
|
|
"learning_rate": 7.75993237531699e-09,
|
|
"logits/chosen": -1.3837153911590576,
|
|
"logits/rejected": -1.322196364402771,
|
|
"logps/chosen": -84.60411071777344,
|
|
"logps/rejected": -121.41426849365234,
|
|
"loss": 0.0044,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.2655871510505676,
|
|
"rewards/margins": 26.094432830810547,
|
|
"rewards/rejected": -25.828845977783203,
|
|
"step": 6420
|
|
},
|
|
{
|
|
"epoch": 2.93,
|
|
"learning_rate": 7.252747252747252e-09,
|
|
"logits/chosen": -1.4268747568130493,
|
|
"logits/rejected": -1.349467158317566,
|
|
"logps/chosen": -89.53246307373047,
|
|
"logps/rejected": -123.66633605957031,
|
|
"loss": 0.0036,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.6228126883506775,
|
|
"rewards/margins": 29.124929428100586,
|
|
"rewards/rejected": -28.502117156982422,
|
|
"step": 6430
|
|
},
|
|
{
|
|
"epoch": 2.94,
|
|
"learning_rate": 6.745562130177514e-09,
|
|
"logits/chosen": -1.3894002437591553,
|
|
"logits/rejected": -1.328368067741394,
|
|
"logps/chosen": -85.84336853027344,
|
|
"logps/rejected": -120.75773620605469,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.8587135076522827,
|
|
"rewards/margins": 26.4877986907959,
|
|
"rewards/rejected": -25.62908363342285,
|
|
"step": 6440
|
|
},
|
|
{
|
|
"epoch": 2.94,
|
|
"learning_rate": 6.238377007607776e-09,
|
|
"logits/chosen": -1.4132146835327148,
|
|
"logits/rejected": -1.3355062007904053,
|
|
"logps/chosen": -87.57127380371094,
|
|
"logps/rejected": -124.08438873291016,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.5653873682022095,
|
|
"rewards/margins": 28.114559173583984,
|
|
"rewards/rejected": -27.549169540405273,
|
|
"step": 6450
|
|
},
|
|
{
|
|
"epoch": 2.95,
|
|
"learning_rate": 5.731191885038039e-09,
|
|
"logits/chosen": -1.4229376316070557,
|
|
"logits/rejected": -1.3501025438308716,
|
|
"logps/chosen": -88.9939956665039,
|
|
"logps/rejected": -123.60148620605469,
|
|
"loss": 0.0022,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.9622406959533691,
|
|
"rewards/margins": 28.20614242553711,
|
|
"rewards/rejected": -27.243900299072266,
|
|
"step": 6460
|
|
},
|
|
{
|
|
"epoch": 2.95,
|
|
"learning_rate": 5.224006762468301e-09,
|
|
"logits/chosen": -1.3894935846328735,
|
|
"logits/rejected": -1.3000738620758057,
|
|
"logps/chosen": -93.95217895507812,
|
|
"logps/rejected": -121.9637451171875,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 1.0586116313934326,
|
|
"rewards/margins": 28.589069366455078,
|
|
"rewards/rejected": -27.530452728271484,
|
|
"step": 6470
|
|
},
|
|
{
|
|
"epoch": 2.96,
|
|
"learning_rate": 4.7168216398985626e-09,
|
|
"logits/chosen": -1.396301507949829,
|
|
"logits/rejected": -1.3185522556304932,
|
|
"logps/chosen": -89.41260528564453,
|
|
"logps/rejected": -119.04069519042969,
|
|
"loss": 0.0033,
|
|
"rewards/accuracies": 0.987500011920929,
|
|
"rewards/chosen": 0.45848971605300903,
|
|
"rewards/margins": 27.00503158569336,
|
|
"rewards/rejected": -26.54654312133789,
|
|
"step": 6480
|
|
},
|
|
{
|
|
"epoch": 2.96,
|
|
"learning_rate": 4.209636517328825e-09,
|
|
"logits/chosen": -1.3974330425262451,
|
|
"logits/rejected": -1.3252642154693604,
|
|
"logps/chosen": -85.48316955566406,
|
|
"logps/rejected": -120.0402603149414,
|
|
"loss": 0.0014,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.14276909828186035,
|
|
"rewards/margins": 27.301326751708984,
|
|
"rewards/rejected": -27.158557891845703,
|
|
"step": 6490
|
|
},
|
|
{
|
|
"epoch": 2.97,
|
|
"learning_rate": 3.702451394759087e-09,
|
|
"logits/chosen": -1.4150840044021606,
|
|
"logits/rejected": -1.333387017250061,
|
|
"logps/chosen": -89.85456085205078,
|
|
"logps/rejected": -121.59066009521484,
|
|
"loss": 0.0011,
|
|
"rewards/accuracies": 1.0,
|
|
"rewards/chosen": 0.14147010445594788,
|
|
"rewards/margins": 26.220012664794922,
|
|
"rewards/rejected": -26.078542709350586,
|
|
"step": 6500
|
|
},
|
|
{
|
|
"epoch": 2.97,
|
|
"eval_logits/chosen": -1.524153232574463,
|
|
"eval_logits/rejected": -1.4333622455596924,
|
|
"eval_logps/chosen": -88.29774475097656,
|
|
"eval_logps/rejected": -119.24208068847656,
|
|
"eval_loss": 0.006867639254778624,
|
|
"eval_rewards/accuracies": 0.9916201233863831,
|
|
"eval_rewards/chosen": 0.5673406720161438,
|
|
"eval_rewards/margins": 27.51752281188965,
|
|
"eval_rewards/rejected": -26.95018196105957,
|
|
"eval_runtime": 78.1379,
|
|
"eval_samples_per_second": 36.628,
|
|
"eval_steps_per_second": 2.291,
|
|
"step": 6500
|
|
}
|
|
],
|
|
"logging_steps": 10,
|
|
"max_steps": 6573,
|
|
"num_train_epochs": 3,
|
|
"save_steps": 500,
|
|
"total_flos": 0.0,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|