LLama-8B-Instruct-v0.1-MI-6e-7 / trainer_state.json
tengxiao1
TX
9969e95
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9982631930527722,
"eval_steps": 400,
"global_step": 467,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01068804275217101,
"grad_norm": 7.822805657905783,
"learning_rate": 6.382978723404255e-08,
"logits/chosen": 0.06214674562215805,
"logits/rejected": 0.03797388821840286,
"logps/chosen": -0.2699491083621979,
"logps/rejected": -0.26826155185699463,
"loss": 1.2748,
"rewards/accuracies": 0.4937500059604645,
"rewards/chosen": -0.2699491083621979,
"rewards/margins": -0.0016875670989975333,
"rewards/rejected": -0.26826155185699463,
"step": 5
},
{
"epoch": 0.02137608550434202,
"grad_norm": 5.0967725327137074,
"learning_rate": 1.276595744680851e-07,
"logits/chosen": -0.010526341386139393,
"logits/rejected": -0.012353870086371899,
"logps/chosen": -0.2696549892425537,
"logps/rejected": -0.2676162123680115,
"loss": 1.2725,
"rewards/accuracies": 0.48750001192092896,
"rewards/chosen": -0.2696549892425537,
"rewards/margins": -0.0020388036500662565,
"rewards/rejected": -0.2676162123680115,
"step": 10
},
{
"epoch": 0.03206412825651302,
"grad_norm": 6.6390016305878055,
"learning_rate": 1.9148936170212767e-07,
"logits/chosen": 0.0009885445469990373,
"logits/rejected": 0.00387256289832294,
"logps/chosen": -0.2789618670940399,
"logps/rejected": -0.2836909592151642,
"loss": 1.2796,
"rewards/accuracies": 0.44999998807907104,
"rewards/chosen": -0.2789618670940399,
"rewards/margins": 0.00472906231880188,
"rewards/rejected": -0.2836909592151642,
"step": 15
},
{
"epoch": 0.04275217100868404,
"grad_norm": 9.068174455913743,
"learning_rate": 2.553191489361702e-07,
"logits/chosen": -0.06325958669185638,
"logits/rejected": -0.06925094127655029,
"logps/chosen": -0.2819739878177643,
"logps/rejected": -0.2899174988269806,
"loss": 1.2759,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.2819739878177643,
"rewards/margins": 0.007943493314087391,
"rewards/rejected": -0.2899174988269806,
"step": 20
},
{
"epoch": 0.053440213760855046,
"grad_norm": 4.867685031446897,
"learning_rate": 3.1914893617021275e-07,
"logits/chosen": -0.06868849694728851,
"logits/rejected": -0.04817543178796768,
"logps/chosen": -0.25565916299819946,
"logps/rejected": -0.2749556303024292,
"loss": 1.2567,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.25565916299819946,
"rewards/margins": 0.01929648406803608,
"rewards/rejected": -0.2749556303024292,
"step": 25
},
{
"epoch": 0.06412825651302605,
"grad_norm": 4.806810486248379,
"learning_rate": 3.8297872340425535e-07,
"logits/chosen": -0.014168953523039818,
"logits/rejected": -0.00634436309337616,
"logps/chosen": -0.2789873480796814,
"logps/rejected": -0.2939203977584839,
"loss": 1.2769,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.2789873480796814,
"rewards/margins": 0.01493304967880249,
"rewards/rejected": -0.2939203977584839,
"step": 30
},
{
"epoch": 0.07481629926519706,
"grad_norm": 7.01406965287447,
"learning_rate": 4.4680851063829783e-07,
"logits/chosen": -0.029415354132652283,
"logits/rejected": -0.009010488167405128,
"logps/chosen": -0.2785240411758423,
"logps/rejected": -0.29580387473106384,
"loss": 1.2752,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.2785240411758423,
"rewards/margins": 0.01727980747818947,
"rewards/rejected": -0.29580387473106384,
"step": 35
},
{
"epoch": 0.08550434201736808,
"grad_norm": 5.373513139182376,
"learning_rate": 5.106382978723404e-07,
"logits/chosen": -0.06608792394399643,
"logits/rejected": -0.07190172374248505,
"logps/chosen": -0.26092082262039185,
"logps/rejected": -0.2700851261615753,
"loss": 1.2674,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.26092082262039185,
"rewards/margins": 0.009164294227957726,
"rewards/rejected": -0.2700851261615753,
"step": 40
},
{
"epoch": 0.09619238476953908,
"grad_norm": 5.609168218812681,
"learning_rate": 5.74468085106383e-07,
"logits/chosen": -0.05338377505540848,
"logits/rejected": -0.01094720046967268,
"logps/chosen": -0.2855256199836731,
"logps/rejected": -0.28623315691947937,
"loss": 1.2738,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.2855256199836731,
"rewards/margins": 0.0007075363537296653,
"rewards/rejected": -0.28623315691947937,
"step": 45
},
{
"epoch": 0.10688042752171009,
"grad_norm": 4.336678776172025,
"learning_rate": 5.999244704827519e-07,
"logits/chosen": -0.017125016078352928,
"logits/rejected": -0.004308671224862337,
"logps/chosen": -0.28561219573020935,
"logps/rejected": -0.299736350774765,
"loss": 1.2688,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.28561219573020935,
"rewards/margins": 0.014124127104878426,
"rewards/rejected": -0.299736350774765,
"step": 50
},
{
"epoch": 0.11756847027388109,
"grad_norm": 5.882470821439722,
"learning_rate": 5.994630389303205e-07,
"logits/chosen": 0.0162811242043972,
"logits/rejected": -0.004544490482658148,
"logps/chosen": -0.2731076776981354,
"logps/rejected": -0.2809983193874359,
"loss": 1.2631,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.2731076776981354,
"rewards/margins": 0.007890653796494007,
"rewards/rejected": -0.2809983193874359,
"step": 55
},
{
"epoch": 0.1282565130260521,
"grad_norm": 6.092605492824151,
"learning_rate": 5.985827812395378e-07,
"logits/chosen": -0.03923701494932175,
"logits/rejected": -0.07081723213195801,
"logps/chosen": -0.2858438491821289,
"logps/rejected": -0.31485337018966675,
"loss": 1.2628,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.2858438491821289,
"rewards/margins": 0.029009530320763588,
"rewards/rejected": -0.31485337018966675,
"step": 60
},
{
"epoch": 0.13894455577822312,
"grad_norm": 5.246427419034069,
"learning_rate": 5.972849285303804e-07,
"logits/chosen": -0.024546677246689796,
"logits/rejected": 0.03360120207071304,
"logps/chosen": -0.29182225465774536,
"logps/rejected": -0.31506821513175964,
"loss": 1.2705,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.29182225465774536,
"rewards/margins": 0.023245956748723984,
"rewards/rejected": -0.31506821513175964,
"step": 65
},
{
"epoch": 0.14963259853039412,
"grad_norm": 5.0996411858772115,
"learning_rate": 5.955712959672177e-07,
"logits/chosen": -0.016444489359855652,
"logits/rejected": -0.020679041743278503,
"logps/chosen": -0.29391151666641235,
"logps/rejected": -0.3471246361732483,
"loss": 1.2543,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.29391151666641235,
"rewards/margins": 0.05321308970451355,
"rewards/rejected": -0.3471246361732483,
"step": 70
},
{
"epoch": 0.16032064128256512,
"grad_norm": 4.491905521876928,
"learning_rate": 5.934442802201417e-07,
"logits/chosen": 0.06254759430885315,
"logits/rejected": 0.10311929881572723,
"logps/chosen": -0.3031434714794159,
"logps/rejected": -0.3355598449707031,
"loss": 1.2679,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.3031434714794159,
"rewards/margins": 0.03241636976599693,
"rewards/rejected": -0.3355598449707031,
"step": 75
},
{
"epoch": 0.17100868403473615,
"grad_norm": 6.478010906716982,
"learning_rate": 5.909068561130061e-07,
"logits/chosen": -0.01297207735478878,
"logits/rejected": -0.004632393829524517,
"logps/chosen": -0.29228898882865906,
"logps/rejected": -0.322248637676239,
"loss": 1.2618,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.29228898882865906,
"rewards/margins": 0.02995964325964451,
"rewards/rejected": -0.322248637676239,
"step": 80
},
{
"epoch": 0.18169672678690715,
"grad_norm": 4.650638031490373,
"learning_rate": 5.879625724628667e-07,
"logits/chosen": 0.005947749130427837,
"logits/rejected": 0.021510040387511253,
"logps/chosen": -0.2952747642993927,
"logps/rejected": -0.3337419927120209,
"loss": 1.257,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.2952747642993927,
"rewards/margins": 0.03846726939082146,
"rewards/rejected": -0.3337419927120209,
"step": 85
},
{
"epoch": 0.19238476953907815,
"grad_norm": 7.530524241094077,
"learning_rate": 5.846155471166399e-07,
"logits/chosen": 0.015343578532338142,
"logits/rejected": 0.03540420904755592,
"logps/chosen": -0.3116888105869293,
"logps/rejected": -0.3682340085506439,
"loss": 1.253,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.3116888105869293,
"rewards/margins": 0.05654525011777878,
"rewards/rejected": -0.3682340085506439,
"step": 90
},
{
"epoch": 0.20307281229124916,
"grad_norm": 6.23873980938112,
"learning_rate": 5.808704611919212e-07,
"logits/chosen": 0.0079043535515666,
"logits/rejected": -0.009995353408157825,
"logps/chosen": -0.30725741386413574,
"logps/rejected": -0.3175857663154602,
"loss": 1.2597,
"rewards/accuracies": 0.48124998807907104,
"rewards/chosen": -0.30725741386413574,
"rewards/margins": 0.010328322649002075,
"rewards/rejected": -0.3175857663154602,
"step": 95
},
{
"epoch": 0.21376085504342018,
"grad_norm": 5.495931507709334,
"learning_rate": 5.767325525300187e-07,
"logits/chosen": 0.012924237176775932,
"logits/rejected": 0.015158179216086864,
"logps/chosen": -0.30597418546676636,
"logps/rejected": -0.3576403558254242,
"loss": 1.2572,
"rewards/accuracies": 0.4937500059604645,
"rewards/chosen": -0.30597418546676636,
"rewards/margins": 0.051666177809238434,
"rewards/rejected": -0.3576403558254242,
"step": 100
},
{
"epoch": 0.22444889779559118,
"grad_norm": 8.704066182889123,
"learning_rate": 5.722076083703594e-07,
"logits/chosen": -0.011864040978252888,
"logits/rejected": -0.015826348215341568,
"logps/chosen": -0.2861265540122986,
"logps/rejected": -0.3439098000526428,
"loss": 1.2455,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": -0.2861265540122986,
"rewards/margins": 0.057783275842666626,
"rewards/rejected": -0.3439098000526428,
"step": 105
},
{
"epoch": 0.23513694054776219,
"grad_norm": 6.081543266921472,
"learning_rate": 5.673019572565103e-07,
"logits/chosen": -0.024934740737080574,
"logits/rejected": -0.036910589784383774,
"logps/chosen": -0.29488444328308105,
"logps/rejected": -0.3499029576778412,
"loss": 1.2384,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.29488444328308105,
"rewards/margins": 0.05501857399940491,
"rewards/rejected": -0.3499029576778412,
"step": 110
},
{
"epoch": 0.2458249832999332,
"grad_norm": 5.083503334201838,
"learning_rate": 5.620224601851389e-07,
"logits/chosen": 0.0035224161110818386,
"logits/rejected": 0.001966515090316534,
"logps/chosen": -0.30457058548927307,
"logps/rejected": -0.35604608058929443,
"loss": 1.2561,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.30457058548927307,
"rewards/margins": 0.051475513726472855,
"rewards/rejected": -0.35604608058929443,
"step": 115
},
{
"epoch": 0.2565130260521042,
"grad_norm": 7.099362190379442,
"learning_rate": 5.563765010102885e-07,
"logits/chosen": -0.06543167680501938,
"logits/rejected": -0.0410967655479908,
"logps/chosen": -0.3293083906173706,
"logps/rejected": -0.3675723075866699,
"loss": 1.2552,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.3293083906173706,
"rewards/margins": 0.03826391324400902,
"rewards/rejected": -0.3675723075866699,
"step": 120
},
{
"epoch": 0.26720106880427524,
"grad_norm": 5.147990697882454,
"learning_rate": 5.503719761163907e-07,
"logits/chosen": -0.10343233495950699,
"logits/rejected": -0.08113230764865875,
"logps/chosen": -0.2962001860141754,
"logps/rejected": -0.35733163356781006,
"loss": 1.234,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.2962001860141754,
"rewards/margins": 0.06113145500421524,
"rewards/rejected": -0.35733163356781006,
"step": 125
},
{
"epoch": 0.27788911155644624,
"grad_norm": 5.070905426510523,
"learning_rate": 5.440172833744582e-07,
"logits/chosen": -0.059284817427396774,
"logits/rejected": -0.020249750465154648,
"logps/chosen": -0.3295074701309204,
"logps/rejected": -0.37299367785453796,
"loss": 1.2644,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.3295074701309204,
"rewards/margins": 0.04348624125123024,
"rewards/rejected": -0.37299367785453796,
"step": 130
},
{
"epoch": 0.28857715430861725,
"grad_norm": 8.434154523088012,
"learning_rate": 5.373213103969024e-07,
"logits/chosen": -0.09271787106990814,
"logits/rejected": -0.0978003442287445,
"logps/chosen": -0.3175578713417053,
"logps/rejected": -0.3870469629764557,
"loss": 1.2466,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.3175578713417053,
"rewards/margins": 0.06948906183242798,
"rewards/rejected": -0.3870469629764557,
"step": 135
},
{
"epoch": 0.29926519706078825,
"grad_norm": 8.050573635697841,
"learning_rate": 5.302934221074033e-07,
"logits/chosen": -0.18472157418727875,
"logits/rejected": -0.18296249210834503,
"logps/chosen": -0.35015708208084106,
"logps/rejected": -0.418283075094223,
"loss": 1.2553,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.35015708208084106,
"rewards/margins": 0.06812603026628494,
"rewards/rejected": -0.418283075094223,
"step": 140
},
{
"epoch": 0.30995323981295925,
"grad_norm": 5.537790942631876,
"learning_rate": 5.229434476432182e-07,
"logits/chosen": -0.04427188262343407,
"logits/rejected": -0.07002754509449005,
"logps/chosen": -0.3104066252708435,
"logps/rejected": -0.36531931161880493,
"loss": 1.2393,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.3104066252708435,
"rewards/margins": 0.05491270869970322,
"rewards/rejected": -0.36531931161880493,
"step": 145
},
{
"epoch": 0.32064128256513025,
"grad_norm": 13.46796536991619,
"learning_rate": 5.152816666082435e-07,
"logits/chosen": -0.09154470264911652,
"logits/rejected": -0.10489149391651154,
"logps/chosen": -0.3233293890953064,
"logps/rejected": -0.42510905861854553,
"loss": 1.2451,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.3233293890953064,
"rewards/margins": 0.10177962481975555,
"rewards/rejected": -0.42510905861854553,
"step": 150
},
{
"epoch": 0.33132932531730125,
"grad_norm": 7.7500923534994985,
"learning_rate": 5.073187946960594e-07,
"logits/chosen": -0.08783230930566788,
"logits/rejected": -0.08713527768850327,
"logps/chosen": -0.3179479241371155,
"logps/rejected": -0.3679467737674713,
"loss": 1.2488,
"rewards/accuracies": 0.4625000059604645,
"rewards/chosen": -0.3179479241371155,
"rewards/margins": 0.04999883845448494,
"rewards/rejected": -0.3679467737674713,
"step": 155
},
{
"epoch": 0.3420173680694723,
"grad_norm": 7.025909512719848,
"learning_rate": 4.990659687030634e-07,
"logits/chosen": -0.1076837033033371,
"logits/rejected": -0.08046683669090271,
"logps/chosen": -0.31321102380752563,
"logps/rejected": -0.3762710690498352,
"loss": 1.2529,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.31321102380752563,
"rewards/margins": 0.06306007504463196,
"rewards/rejected": -0.3762710690498352,
"step": 160
},
{
"epoch": 0.3527054108216433,
"grad_norm": 8.01239629928041,
"learning_rate": 4.905347309526536e-07,
"logits/chosen": -0.06136934086680412,
"logits/rejected": -0.07382142543792725,
"logps/chosen": -0.29660579562187195,
"logps/rejected": -0.37172654271125793,
"loss": 1.2427,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.29660579562187195,
"rewards/margins": 0.07512073218822479,
"rewards/rejected": -0.37172654271125793,
"step": 165
},
{
"epoch": 0.3633934535738143,
"grad_norm": 5.642901073511358,
"learning_rate": 4.817370131522459e-07,
"logits/chosen": -0.0563310906291008,
"logits/rejected": -0.030183713883161545,
"logps/chosen": -0.31852108240127563,
"logps/rejected": -0.39116546511650085,
"loss": 1.2442,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.31852108240127563,
"rewards/margins": 0.07264441251754761,
"rewards/rejected": -0.39116546511650085,
"step": 170
},
{
"epoch": 0.3740814963259853,
"grad_norm": 5.485147120245967,
"learning_rate": 4.7268511970570207e-07,
"logits/chosen": -0.08929944038391113,
"logits/rejected": -0.08389794826507568,
"logps/chosen": -0.30848273634910583,
"logps/rejected": -0.3659656345844269,
"loss": 1.2505,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.30848273634910583,
"rewards/margins": 0.05748288705945015,
"rewards/rejected": -0.3659656345844269,
"step": 175
},
{
"epoch": 0.3847695390781563,
"grad_norm": 7.991274503438784,
"learning_rate": 4.6339171050450815e-07,
"logits/chosen": -0.10145304352045059,
"logits/rejected": -0.09222683310508728,
"logps/chosen": -0.31844446063041687,
"logps/rejected": -0.3609256148338318,
"loss": 1.2546,
"rewards/accuracies": 0.48124998807907104,
"rewards/chosen": -0.31844446063041687,
"rewards/margins": 0.042481135576963425,
"rewards/rejected": -0.3609256148338318,
"step": 180
},
{
"epoch": 0.3954575818303273,
"grad_norm": 5.832967942513168,
"learning_rate": 4.5386978322177184e-07,
"logits/chosen": -0.051486529409885406,
"logits/rejected": -0.07657450437545776,
"logps/chosen": -0.3131783604621887,
"logps/rejected": -0.35496917366981506,
"loss": 1.2627,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.3131783604621887,
"rewards/margins": 0.04179079458117485,
"rewards/rejected": -0.35496917366981506,
"step": 185
},
{
"epoch": 0.4061456245824983,
"grad_norm": 8.218974322062975,
"learning_rate": 4.4413265513380134e-07,
"logits/chosen": -0.08528328686952591,
"logits/rejected": -0.0627092644572258,
"logps/chosen": -0.3054826855659485,
"logps/rejected": -0.37131738662719727,
"loss": 1.2456,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": -0.3054826855659485,
"rewards/margins": 0.06583467870950699,
"rewards/rejected": -0.37131738662719727,
"step": 190
},
{
"epoch": 0.4168336673346693,
"grad_norm": 8.281751397477004,
"learning_rate": 4.3419394449468975e-07,
"logits/chosen": -0.0632157102227211,
"logits/rejected": -0.039062272757291794,
"logps/chosen": -0.3401602506637573,
"logps/rejected": -0.42965516448020935,
"loss": 1.2393,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.3401602506637573,
"rewards/margins": 0.08949492126703262,
"rewards/rejected": -0.42965516448020935,
"step": 195
},
{
"epoch": 0.42752171008684037,
"grad_norm": 5.682547699952222,
"learning_rate": 4.2406755148995617e-07,
"logits/chosen": -0.036120522767305374,
"logits/rejected": -0.00437445193529129,
"logps/chosen": -0.30982089042663574,
"logps/rejected": -0.3825121223926544,
"loss": 1.2418,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.30982089042663574,
"rewards/margins": 0.07269121706485748,
"rewards/rejected": -0.3825121223926544,
"step": 200
},
{
"epoch": 0.43820975283901137,
"grad_norm": 5.711897059557612,
"learning_rate": 4.1376763879587855e-07,
"logits/chosen": -0.08326585590839386,
"logits/rejected": -0.12235681712627411,
"logps/chosen": -0.34261685609817505,
"logps/rejected": -0.4019942283630371,
"loss": 1.2571,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.34261685609817505,
"rewards/margins": 0.05937739089131355,
"rewards/rejected": -0.4019942283630371,
"step": 205
},
{
"epoch": 0.44889779559118237,
"grad_norm": 6.6702507208013895,
"learning_rate": 4.0330861177171046e-07,
"logits/chosen": -0.09191317856311798,
"logits/rejected": -0.07536768168210983,
"logps/chosen": -0.3210485577583313,
"logps/rejected": -0.39176544547080994,
"loss": 1.247,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.3210485577583313,
"rewards/margins": 0.07071693241596222,
"rewards/rejected": -0.39176544547080994,
"step": 210
},
{
"epoch": 0.45958583834335337,
"grad_norm": 6.330049251313848,
"learning_rate": 3.927050983124842e-07,
"logits/chosen": -0.025531485676765442,
"logits/rejected": -0.07240410149097443,
"logps/chosen": -0.29885441064834595,
"logps/rejected": -0.391807496547699,
"loss": 1.238,
"rewards/accuracies": 0.574999988079071,
"rewards/chosen": -0.29885441064834595,
"rewards/margins": 0.09295307099819183,
"rewards/rejected": -0.391807496547699,
"step": 215
},
{
"epoch": 0.47027388109552437,
"grad_norm": 6.737595050647263,
"learning_rate": 3.8197192839057603e-07,
"logits/chosen": -0.1071164608001709,
"logits/rejected": -0.12290854752063751,
"logps/chosen": -0.30930382013320923,
"logps/rejected": -0.4362809658050537,
"loss": 1.2382,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.30930382013320923,
"rewards/margins": 0.12697716057300568,
"rewards/rejected": -0.4362809658050537,
"step": 220
},
{
"epoch": 0.48096192384769537,
"grad_norm": 8.559735052947849,
"learning_rate": 3.7112411331464923e-07,
"logits/chosen": -0.02524995245039463,
"logits/rejected": -0.030149292200803757,
"logps/chosen": -0.3149697184562683,
"logps/rejected": -0.3956434428691864,
"loss": 1.2272,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.3149697184562683,
"rewards/margins": 0.0806737095117569,
"rewards/rejected": -0.3956434428691864,
"step": 225
},
{
"epoch": 0.4916499665998664,
"grad_norm": 6.480287157306168,
"learning_rate": 3.601768247349818e-07,
"logits/chosen": -0.03261668235063553,
"logits/rejected": -0.08516497910022736,
"logps/chosen": -0.3169209659099579,
"logps/rejected": -0.3873901069164276,
"loss": 1.2404,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.3169209659099579,
"rewards/margins": 0.07046912610530853,
"rewards/rejected": -0.3873901069164276,
"step": 230
},
{
"epoch": 0.5023380093520374,
"grad_norm": 8.225368883810985,
"learning_rate": 3.491453734245413e-07,
"logits/chosen": -0.06573788821697235,
"logits/rejected": -0.0159236378967762,
"logps/chosen": -0.3394278287887573,
"logps/rejected": -0.4536859393119812,
"loss": 1.2409,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.3394278287887573,
"rewards/margins": 0.11425812542438507,
"rewards/rejected": -0.4536859393119812,
"step": 235
},
{
"epoch": 0.5130260521042084,
"grad_norm": 6.0501816929552135,
"learning_rate": 3.3804518786548455e-07,
"logits/chosen": -0.09407626837491989,
"logits/rejected": -0.07616542279720306,
"logps/chosen": -0.3101692199707031,
"logps/rejected": -0.428670734167099,
"loss": 1.2453,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.3101692199707031,
"rewards/margins": 0.11850155889987946,
"rewards/rejected": -0.428670734167099,
"step": 240
},
{
"epoch": 0.5237140948563794,
"grad_norm": 5.475578816065332,
"learning_rate": 3.2689179267103006e-07,
"logits/chosen": -0.1301025003194809,
"logits/rejected": -0.12063749134540558,
"logps/chosen": -0.3209839463233948,
"logps/rejected": -0.3626781404018402,
"loss": 1.2299,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.3209839463233948,
"rewards/margins": 0.04169422388076782,
"rewards/rejected": -0.3626781404018402,
"step": 245
},
{
"epoch": 0.5344021376085505,
"grad_norm": 7.848337245875008,
"learning_rate": 3.1570078687288317e-07,
"logits/chosen": -0.07203061133623123,
"logits/rejected": -0.07748202979564667,
"logps/chosen": -0.34228605031967163,
"logps/rejected": -0.46370163559913635,
"loss": 1.2274,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.34228605031967163,
"rewards/margins": 0.12141555547714233,
"rewards/rejected": -0.46370163559913635,
"step": 250
},
{
"epoch": 0.5450901803607214,
"grad_norm": 8.463852280152306,
"learning_rate": 3.0448782210457906e-07,
"logits/chosen": -0.07813692837953568,
"logits/rejected": -0.07056453824043274,
"logps/chosen": -0.3696078360080719,
"logps/rejected": -0.4517177939414978,
"loss": 1.2467,
"rewards/accuracies": 0.550000011920929,
"rewards/chosen": -0.3696078360080719,
"rewards/margins": 0.08210990577936172,
"rewards/rejected": -0.4517177939414978,
"step": 255
},
{
"epoch": 0.5557782231128925,
"grad_norm": 8.673679370712454,
"learning_rate": 2.932685807112585e-07,
"logits/chosen": -0.13425521552562714,
"logits/rejected": -0.13180285692214966,
"logps/chosen": -0.3235534727573395,
"logps/rejected": -0.4138403534889221,
"loss": 1.2392,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.3235534727573395,
"rewards/margins": 0.09028687328100204,
"rewards/rejected": -0.4138403534889221,
"step": 260
},
{
"epoch": 0.5664662658650634,
"grad_norm": 14.773817206579576,
"learning_rate": 2.8205875381648974e-07,
"logits/chosen": -0.10961911827325821,
"logits/rejected": -0.10981354862451553,
"logps/chosen": -0.31433889269828796,
"logps/rejected": -0.40348243713378906,
"loss": 1.2448,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.31433889269828796,
"rewards/margins": 0.08914351463317871,
"rewards/rejected": -0.40348243713378906,
"step": 265
},
{
"epoch": 0.5771543086172345,
"grad_norm": 10.511806133556902,
"learning_rate": 2.708740193768135e-07,
"logits/chosen": -0.08152172714471817,
"logits/rejected": -0.07667910307645798,
"logps/chosen": -0.33281245827674866,
"logps/rejected": -0.4943714141845703,
"loss": 1.2277,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.33281245827674866,
"rewards/margins": 0.16155894100666046,
"rewards/rejected": -0.4943714141845703,
"step": 270
},
{
"epoch": 0.5878423513694054,
"grad_norm": 5.764021925950384,
"learning_rate": 2.597300202547034e-07,
"logits/chosen": -0.0671951025724411,
"logits/rejected": -0.09161119163036346,
"logps/chosen": -0.3234054148197174,
"logps/rejected": -0.36444562673568726,
"loss": 1.2404,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.3234054148197174,
"rewards/margins": 0.04104021191596985,
"rewards/rejected": -0.36444562673568726,
"step": 275
},
{
"epoch": 0.5985303941215765,
"grad_norm": 6.254810248978168,
"learning_rate": 2.4864234234060747e-07,
"logits/chosen": -0.1233711987733841,
"logits/rejected": -0.10507211834192276,
"logps/chosen": -0.326472669839859,
"logps/rejected": -0.4079364836215973,
"loss": 1.2346,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.326472669839859,
"rewards/margins": 0.08146381378173828,
"rewards/rejected": -0.4079364836215973,
"step": 280
},
{
"epoch": 0.6092184368737475,
"grad_norm": 8.145566662679744,
"learning_rate": 2.3762649275467223e-07,
"logits/chosen": -0.11552796512842178,
"logits/rejected": -0.12833945453166962,
"logps/chosen": -0.3270297944545746,
"logps/rejected": -0.40440672636032104,
"loss": 1.2498,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.3270297944545746,
"rewards/margins": 0.07737687975168228,
"rewards/rejected": -0.40440672636032104,
"step": 285
},
{
"epoch": 0.6199064796259185,
"grad_norm": 9.531079036404222,
"learning_rate": 2.2669787815863174e-07,
"logits/chosen": -0.03034001588821411,
"logits/rejected": -0.0528348907828331,
"logps/chosen": -0.3205064833164215,
"logps/rejected": -0.4226464629173279,
"loss": 1.2523,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.3205064833164215,
"rewards/margins": 0.10214000940322876,
"rewards/rejected": -0.4226464629173279,
"step": 290
},
{
"epoch": 0.6305945223780896,
"grad_norm": 5.413246554100629,
"learning_rate": 2.1587178320819919e-07,
"logits/chosen": -0.060756783932447433,
"logits/rejected": -0.0011257051955908537,
"logps/chosen": -0.27187207341194153,
"logps/rejected": -0.3691639006137848,
"loss": 1.2341,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.27187207341194153,
"rewards/margins": 0.09729186445474625,
"rewards/rejected": -0.3691639006137848,
"step": 295
},
{
"epoch": 0.6412825651302605,
"grad_norm": 12.664849670753055,
"learning_rate": 2.0516334917609277e-07,
"logits/chosen": -0.10225675255060196,
"logits/rejected": -0.04843712970614433,
"logps/chosen": -0.3377472758293152,
"logps/rejected": -0.5128234028816223,
"loss": 1.2347,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.3377472758293152,
"rewards/margins": 0.17507611215114594,
"rewards/rejected": -0.5128234028816223,
"step": 300
},
{
"epoch": 0.6519706078824316,
"grad_norm": 7.8017671377878015,
"learning_rate": 1.9458755277559716e-07,
"logits/chosen": -0.11494015157222748,
"logits/rejected": -0.10972355306148529,
"logps/chosen": -0.3163761496543884,
"logps/rejected": -0.4071407914161682,
"loss": 1.2405,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.3163761496543884,
"rewards/margins": 0.09076462686061859,
"rewards/rejected": -0.4071407914161682,
"step": 305
},
{
"epoch": 0.6626586506346025,
"grad_norm": 6.034077165873482,
"learning_rate": 1.8415918521427613e-07,
"logits/chosen": -0.1821509748697281,
"logits/rejected": -0.19082587957382202,
"logps/chosen": -0.31090688705444336,
"logps/rejected": -0.3758618235588074,
"loss": 1.2455,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.31090688705444336,
"rewards/margins": 0.06495492160320282,
"rewards/rejected": -0.3758618235588074,
"step": 310
},
{
"epoch": 0.6733466933867736,
"grad_norm": 10.235181182403219,
"learning_rate": 1.7389283150713038e-07,
"logits/chosen": -0.1251331865787506,
"logits/rejected": -0.11890840530395508,
"logps/chosen": -0.3588525056838989,
"logps/rejected": -0.425645649433136,
"loss": 1.2543,
"rewards/accuracies": 0.48124998807907104,
"rewards/chosen": -0.3588525056838989,
"rewards/margins": 0.06679315119981766,
"rewards/rejected": -0.425645649433136,
"step": 315
},
{
"epoch": 0.6840347361389446,
"grad_norm": 9.723759702295695,
"learning_rate": 1.6380285007813597e-07,
"logits/chosen": -0.1123957484960556,
"logits/rejected": -0.1257510930299759,
"logps/chosen": -0.3145357668399811,
"logps/rejected": -0.3408251404762268,
"loss": 1.2609,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": -0.3145357668399811,
"rewards/margins": 0.026289362460374832,
"rewards/rejected": -0.3408251404762268,
"step": 320
},
{
"epoch": 0.6947227788911156,
"grad_norm": 7.286383815847668,
"learning_rate": 1.539033526786898e-07,
"logits/chosen": -0.1374741941690445,
"logits/rejected": -0.11429701000452042,
"logps/chosen": -0.32331573963165283,
"logps/rejected": -0.4659709930419922,
"loss": 1.2423,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.32331573963165283,
"rewards/margins": 0.14265525341033936,
"rewards/rejected": -0.4659709930419922,
"step": 325
},
{
"epoch": 0.7054108216432866,
"grad_norm": 5.495961241803029,
"learning_rate": 1.4420818465104924e-07,
"logits/chosen": -0.1799645572900772,
"logits/rejected": -0.17759008705615997,
"logps/chosen": -0.30779215693473816,
"logps/rejected": -0.3636534810066223,
"loss": 1.2328,
"rewards/accuracies": 0.543749988079071,
"rewards/chosen": -0.30779215693473816,
"rewards/margins": 0.05586131289601326,
"rewards/rejected": -0.3636534810066223,
"step": 330
},
{
"epoch": 0.7160988643954576,
"grad_norm": 6.732402042819091,
"learning_rate": 1.3473090556436928e-07,
"logits/chosen": -0.09900529682636261,
"logits/rejected": -0.11673985421657562,
"logps/chosen": -0.32374444603919983,
"logps/rejected": -0.42279139161109924,
"loss": 1.2482,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.32374444603919983,
"rewards/margins": 0.0990469679236412,
"rewards/rejected": -0.42279139161109924,
"step": 335
},
{
"epoch": 0.7267869071476286,
"grad_norm": 6.164055292584424,
"learning_rate": 1.2548477025041833e-07,
"logits/chosen": -0.17352089285850525,
"logits/rejected": -0.15723419189453125,
"logps/chosen": -0.31769293546676636,
"logps/rejected": -0.4277707040309906,
"loss": 1.2469,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.31769293546676636,
"rewards/margins": 0.11007778346538544,
"rewards/rejected": -0.4277707040309906,
"step": 340
},
{
"epoch": 0.7374749498997996,
"grad_norm": 7.386683624949419,
"learning_rate": 1.1648271026549805e-07,
"logits/chosen": -0.16501447558403015,
"logits/rejected": -0.17034907639026642,
"logps/chosen": -0.3037567734718323,
"logps/rejected": -0.4147283136844635,
"loss": 1.235,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.3037567734718323,
"rewards/margins": 0.11097153276205063,
"rewards/rejected": -0.4147283136844635,
"step": 345
},
{
"epoch": 0.7481629926519706,
"grad_norm": 8.63440141406496,
"learning_rate": 1.0773731580449275e-07,
"logits/chosen": -0.0861009806394577,
"logits/rejected": -0.10058856010437012,
"logps/chosen": -0.3289971947669983,
"logps/rejected": -0.43288707733154297,
"loss": 1.2271,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.3289971947669983,
"rewards/margins": 0.10388988256454468,
"rewards/rejected": -0.43288707733154297,
"step": 350
},
{
"epoch": 0.7588510354041417,
"grad_norm": 5.491722459194082,
"learning_rate": 9.926081809234262e-08,
"logits/chosen": -0.1492873877286911,
"logits/rejected": -0.14633427560329437,
"logps/chosen": -0.3535214960575104,
"logps/rejected": -0.5062969923019409,
"loss": 1.2331,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.3535214960575104,
"rewards/margins": 0.15277548134326935,
"rewards/rejected": -0.5062969923019409,
"step": 355
},
{
"epoch": 0.7695390781563126,
"grad_norm": 16.598441497777067,
"learning_rate": 9.106507227756998e-08,
"logits/chosen": -0.10592007637023926,
"logits/rejected": -0.1149587631225586,
"logps/chosen": -0.355294406414032,
"logps/rejected": -0.41237178444862366,
"loss": 1.2541,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.355294406414032,
"rewards/margins": 0.057077307254076004,
"rewards/rejected": -0.41237178444862366,
"step": 360
},
{
"epoch": 0.7802271209084837,
"grad_norm": 9.266627411766367,
"learning_rate": 8.316154085178256e-08,
"logits/chosen": -0.1599133014678955,
"logits/rejected": -0.16612327098846436,
"logps/chosen": -0.3537140488624573,
"logps/rejected": -0.45664018392562866,
"loss": 1.2415,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": -0.3537140488624573,
"rewards/margins": 0.10292615741491318,
"rewards/rejected": -0.45664018392562866,
"step": 365
},
{
"epoch": 0.7909151636606546,
"grad_norm": 7.428406479509353,
"learning_rate": 7.55612776183419e-08,
"logits/chosen": -0.10595826804637909,
"logits/rejected": -0.09110520780086517,
"logps/chosen": -0.35074084997177124,
"logps/rejected": -0.40927591919898987,
"loss": 1.2357,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.35074084997177124,
"rewards/margins": 0.05853506922721863,
"rewards/rejected": -0.40927591919898987,
"step": 370
},
{
"epoch": 0.8016032064128257,
"grad_norm": 8.282266775931964,
"learning_rate": 6.827491223262017e-08,
"logits/chosen": -0.14613883197307587,
"logits/rejected": -0.1305559277534485,
"logps/chosen": -0.33350640535354614,
"logps/rejected": -0.3941604495048523,
"loss": 1.2317,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.33350640535354614,
"rewards/margins": 0.06065405532717705,
"rewards/rejected": -0.3941604495048523,
"step": 375
},
{
"epoch": 0.8122912491649966,
"grad_norm": 7.257639288893235,
"learning_rate": 6.131263533546572e-08,
"logits/chosen": -0.13168776035308838,
"logits/rejected": -0.13572274148464203,
"logps/chosen": -0.32469362020492554,
"logps/rejected": -0.4485169053077698,
"loss": 1.2319,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.32469362020492554,
"rewards/margins": 0.12382327020168304,
"rewards/rejected": -0.4485169053077698,
"step": 380
},
{
"epoch": 0.8229792919171677,
"grad_norm": 8.794582511790273,
"learning_rate": 5.468418430067059e-08,
"logits/chosen": -0.13690440356731415,
"logits/rejected": -0.11999843269586563,
"logps/chosen": -0.3403404653072357,
"logps/rejected": -0.41022801399230957,
"loss": 1.2458,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.3403404653072357,
"rewards/margins": 0.06988750398159027,
"rewards/rejected": -0.41022801399230957,
"step": 385
},
{
"epoch": 0.8336673346693386,
"grad_norm": 6.914453118608116,
"learning_rate": 4.839882961637282e-08,
"logits/chosen": -0.14087721705436707,
"logits/rejected": -0.11817269027233124,
"logps/chosen": -0.32691115140914917,
"logps/rejected": -0.39570215344429016,
"loss": 1.2453,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.32691115140914917,
"rewards/margins": 0.0687909945845604,
"rewards/rejected": -0.39570215344429016,
"step": 390
},
{
"epoch": 0.8443553774215097,
"grad_norm": 7.401358248565427,
"learning_rate": 4.2465361919440165e-08,
"logits/chosen": -0.17381078004837036,
"logits/rejected": -0.1655048429965973,
"logps/chosen": -0.32772788405418396,
"logps/rejected": -0.36842280626296997,
"loss": 1.2419,
"rewards/accuracies": 0.518750011920929,
"rewards/chosen": -0.32772788405418396,
"rewards/margins": 0.040694937109947205,
"rewards/rejected": -0.36842280626296997,
"step": 395
},
{
"epoch": 0.8550434201736807,
"grad_norm": 6.499476994247458,
"learning_rate": 3.6892079700970036e-08,
"logits/chosen": -0.19311991333961487,
"logits/rejected": -0.1830570548772812,
"logps/chosen": -0.34295058250427246,
"logps/rejected": -0.37818074226379395,
"loss": 1.2439,
"rewards/accuracies": 0.5562499761581421,
"rewards/chosen": -0.34295058250427246,
"rewards/margins": 0.035230137407779694,
"rewards/rejected": -0.37818074226379395,
"step": 400
},
{
"epoch": 0.8550434201736807,
"eval_logits/chosen": 0.01532436441630125,
"eval_logits/rejected": 0.005963262636214495,
"eval_logps/chosen": -0.3365793526172638,
"eval_logps/rejected": -0.40151944756507874,
"eval_loss": 1.2415482997894287,
"eval_rewards/accuracies": 0.5873983502388,
"eval_rewards/chosen": -0.3365793526172638,
"eval_rewards/margins": 0.06494008004665375,
"eval_rewards/rejected": -0.40151944756507874,
"eval_runtime": 427.7978,
"eval_samples_per_second": 4.584,
"eval_steps_per_second": 0.288,
"step": 400
},
{
"epoch": 0.8657314629258517,
"grad_norm": 8.655152423101507,
"learning_rate": 3.1686777700099e-08,
"logits/chosen": -0.11514046043157578,
"logits/rejected": -0.16686634719371796,
"logps/chosen": -0.3493112325668335,
"logps/rejected": -0.37683025002479553,
"loss": 1.245,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.3493112325668335,
"rewards/margins": 0.027519047260284424,
"rewards/rejected": -0.37683025002479553,
"step": 405
},
{
"epoch": 0.8764195056780227,
"grad_norm": 8.013257979238155,
"learning_rate": 2.685673600235524e-08,
"logits/chosen": -0.10536377131938934,
"logits/rejected": -0.1320020854473114,
"logps/chosen": -0.3498873710632324,
"logps/rejected": -0.39043301343917847,
"loss": 1.2607,
"rewards/accuracies": 0.5375000238418579,
"rewards/chosen": -0.3498873710632324,
"rewards/margins": 0.04054565355181694,
"rewards/rejected": -0.39043301343917847,
"step": 410
},
{
"epoch": 0.8871075484301937,
"grad_norm": 12.61073944196688,
"learning_rate": 2.2408709857800988e-08,
"logits/chosen": -0.12514375150203705,
"logits/rejected": -0.10413704812526703,
"logps/chosen": -0.2854083478450775,
"logps/rejected": -0.3893025517463684,
"loss": 1.233,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.2854083478450775,
"rewards/margins": 0.1038941740989685,
"rewards/rejected": -0.3893025517463684,
"step": 415
},
{
"epoch": 0.8977955911823647,
"grad_norm": 5.802098857336539,
"learning_rate": 1.8348920233204167e-08,
"logits/chosen": -0.08317883312702179,
"logits/rejected": -0.06567595899105072,
"logps/chosen": -0.3321346044540405,
"logps/rejected": -0.4815450608730316,
"loss": 1.2393,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.3321346044540405,
"rewards/margins": 0.14941047132015228,
"rewards/rejected": -0.4815450608730316,
"step": 420
},
{
"epoch": 0.9084836339345357,
"grad_norm": 5.211099670695838,
"learning_rate": 1.468304511145394e-08,
"logits/chosen": -0.02274451218545437,
"logits/rejected": -0.0685218870639801,
"logps/chosen": -0.3114772439002991,
"logps/rejected": -0.4261551797389984,
"loss": 1.2356,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.3114772439002991,
"rewards/margins": 0.11467792093753815,
"rewards/rejected": -0.4261551797389984,
"step": 425
},
{
"epoch": 0.9191716766867067,
"grad_norm": 5.865346057497481,
"learning_rate": 1.1416211550388222e-08,
"logits/chosen": -0.10939434915781021,
"logits/rejected": -0.09104075282812119,
"logps/chosen": -0.29765018820762634,
"logps/rejected": -0.3629956841468811,
"loss": 1.2343,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": -0.29765018820762634,
"rewards/margins": 0.06534545123577118,
"rewards/rejected": -0.3629956841468811,
"step": 430
},
{
"epoch": 0.9298597194388778,
"grad_norm": 7.151775664454521,
"learning_rate": 8.552988512139748e-09,
"logits/chosen": -0.09702922403812408,
"logits/rejected": -0.11269289255142212,
"logps/chosen": -0.3232804238796234,
"logps/rejected": -0.4417162835597992,
"loss": 1.2366,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": -0.3232804238796234,
"rewards/margins": 0.11843589693307877,
"rewards/rejected": -0.4417162835597992,
"step": 435
},
{
"epoch": 0.9405477621910487,
"grad_norm": 19.71341285143898,
"learning_rate": 6.097380473029356e-09,
"logits/chosen": -0.13407650589942932,
"logits/rejected": -0.1464676707983017,
"logps/chosen": -0.33517464995384216,
"logps/rejected": -0.3925517201423645,
"loss": 1.2475,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.33517464995384216,
"rewards/margins": 0.05737708881497383,
"rewards/rejected": -0.3925517201423645,
"step": 440
},
{
"epoch": 0.9512358049432198,
"grad_norm": 7.374118674662329,
"learning_rate": 4.052821822943597e-09,
"logits/chosen": -0.04786144942045212,
"logits/rejected": -0.04718126356601715,
"logps/chosen": -0.3143005967140198,
"logps/rejected": -0.38420677185058594,
"loss": 1.258,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.3143005967140198,
"rewards/margins": 0.06990616768598557,
"rewards/rejected": -0.38420677185058594,
"step": 445
},
{
"epoch": 0.9619238476953907,
"grad_norm": 7.332271422792315,
"learning_rate": 2.4221720620301368e-09,
"logits/chosen": -0.10571523010730743,
"logits/rejected": -0.0989978164434433,
"logps/chosen": -0.321834921836853,
"logps/rejected": -0.40633755922317505,
"loss": 1.2275,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.321834921836853,
"rewards/margins": 0.08450265228748322,
"rewards/rejected": -0.40633755922317505,
"step": 450
},
{
"epoch": 0.9726118904475618,
"grad_norm": 7.032752985388992,
"learning_rate": 1.2077118014282794e-09,
"logits/chosen": -0.06323617696762085,
"logits/rejected": -0.029714446514844894,
"logps/chosen": -0.3421580493450165,
"logps/rejected": -0.4160069525241852,
"loss": 1.2498,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.3421580493450165,
"rewards/margins": 0.0738489031791687,
"rewards/rejected": -0.4160069525241852,
"step": 455
},
{
"epoch": 0.9832999331997327,
"grad_norm": 6.513587335089515,
"learning_rate": 4.1113957362785e-10,
"logits/chosen": -0.05799049139022827,
"logits/rejected": -0.08265287429094315,
"logps/chosen": -0.3295963406562805,
"logps/rejected": -0.37454092502593994,
"loss": 1.2494,
"rewards/accuracies": 0.5249999761581421,
"rewards/chosen": -0.3295963406562805,
"rewards/margins": 0.04494457319378853,
"rewards/rejected": -0.37454092502593994,
"step": 460
},
{
"epoch": 0.9939879759519038,
"grad_norm": 9.565280879421035,
"learning_rate": 3.3569456917970085e-11,
"logits/chosen": -0.030880967155098915,
"logits/rejected": -0.05265098810195923,
"logps/chosen": -0.31436887383461,
"logps/rejected": -0.407276451587677,
"loss": 1.2427,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.31436887383461,
"rewards/margins": 0.0929076224565506,
"rewards/rejected": -0.407276451587677,
"step": 465
},
{
"epoch": 0.9982631930527722,
"step": 467,
"total_flos": 0.0,
"train_loss": 1.2480301234145237,
"train_runtime": 21322.7535,
"train_samples_per_second": 2.808,
"train_steps_per_second": 0.022
}
],
"logging_steps": 5,
"max_steps": 467,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 1000000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}