phi-2-gpo-v25-i1 / trainer_state.json
lole25's picture
Model save
123afc1 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.9968,
"eval_steps": 500,
"global_step": 936,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.0638297872340426e-07,
"logits/chosen": 0.1422731578350067,
"logits/rejected": 0.3313992917537689,
"logps/chosen": -382.77789306640625,
"logps/rejected": -331.4576416015625,
"loss": 0.4045,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 1.0638297872340427e-06,
"logits/chosen": 0.11944054067134857,
"logits/rejected": 0.18962062895298004,
"logps/chosen": -326.4944763183594,
"logps/rejected": -291.0369873046875,
"loss": 0.3779,
"rewards/accuracies": 0.4236111044883728,
"rewards/chosen": 5.325177335180342e-05,
"rewards/margins": 5.4291256674332544e-05,
"rewards/rejected": -1.039480139297666e-06,
"step": 10
},
{
"epoch": 0.04,
"learning_rate": 2.1276595744680853e-06,
"logits/chosen": 0.12437532097101212,
"logits/rejected": 0.17575038969516754,
"logps/chosen": -355.1729736328125,
"logps/rejected": -326.6689453125,
"loss": 0.379,
"rewards/accuracies": 0.4749999940395355,
"rewards/chosen": 0.0001221647544298321,
"rewards/margins": 0.00025352081866003573,
"rewards/rejected": -0.0001313560496782884,
"step": 20
},
{
"epoch": 0.06,
"learning_rate": 3.191489361702128e-06,
"logits/chosen": 0.14455276727676392,
"logits/rejected": 0.17237640917301178,
"logps/chosen": -338.4366455078125,
"logps/rejected": -344.5356140136719,
"loss": 0.3714,
"rewards/accuracies": 0.512499988079071,
"rewards/chosen": -8.463774429401383e-05,
"rewards/margins": 0.00026104756398126483,
"rewards/rejected": -0.00034568531555123627,
"step": 30
},
{
"epoch": 0.09,
"learning_rate": 4.255319148936171e-06,
"logits/chosen": 0.09771748632192612,
"logits/rejected": 0.20755767822265625,
"logps/chosen": -372.2322692871094,
"logps/rejected": -351.06793212890625,
"loss": 0.3867,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -4.432376954355277e-05,
"rewards/margins": 0.0008093343349173665,
"rewards/rejected": -0.00085365807171911,
"step": 40
},
{
"epoch": 0.11,
"learning_rate": 4.999373573764188e-06,
"logits/chosen": 0.07760350406169891,
"logits/rejected": 0.154631569981575,
"logps/chosen": -344.04986572265625,
"logps/rejected": -310.27630615234375,
"loss": 0.3668,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": 0.0006088110385462642,
"rewards/margins": 0.0018962211906909943,
"rewards/rejected": -0.0012874103849753737,
"step": 50
},
{
"epoch": 0.13,
"learning_rate": 4.988245838331339e-06,
"logits/chosen": 0.14376303553581238,
"logits/rejected": 0.1840187907218933,
"logps/chosen": -348.7585754394531,
"logps/rejected": -347.42633056640625,
"loss": 0.3701,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": 0.0007757244748063385,
"rewards/margins": 0.0029361725319176912,
"rewards/rejected": -0.00216044788248837,
"step": 60
},
{
"epoch": 0.15,
"learning_rate": 4.963268819535228e-06,
"logits/chosen": 0.10374955832958221,
"logits/rejected": 0.20326051115989685,
"logps/chosen": -330.6069641113281,
"logps/rejected": -329.6276550292969,
"loss": 0.3494,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": 0.0009276155615225434,
"rewards/margins": 0.005808635149151087,
"rewards/rejected": -0.004881019704043865,
"step": 70
},
{
"epoch": 0.17,
"learning_rate": 4.9245815365216115e-06,
"logits/chosen": 0.11984933912754059,
"logits/rejected": 0.19475053250789642,
"logps/chosen": -341.2612609863281,
"logps/rejected": -341.21063232421875,
"loss": 0.3772,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.0007902610814198852,
"rewards/margins": 0.014523616060614586,
"rewards/rejected": -0.013733354397118092,
"step": 80
},
{
"epoch": 0.19,
"learning_rate": 4.872399318152594e-06,
"logits/chosen": 0.09717285633087158,
"logits/rejected": 0.19757065176963806,
"logps/chosen": -383.49658203125,
"logps/rejected": -375.48980712890625,
"loss": 0.3443,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.013060875236988068,
"rewards/margins": 0.018941307440400124,
"rewards/rejected": -0.03200218454003334,
"step": 90
},
{
"epoch": 0.21,
"learning_rate": 4.807012604511542e-06,
"logits/chosen": 0.09459100663661957,
"logits/rejected": 0.172474667429924,
"logps/chosen": -414.6573181152344,
"logps/rejected": -423.6068420410156,
"loss": 0.3207,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.0480206236243248,
"rewards/margins": 0.05255536362528801,
"rewards/rejected": -0.10057598352432251,
"step": 100
},
{
"epoch": 0.23,
"learning_rate": 4.728785330347771e-06,
"logits/chosen": 0.17286503314971924,
"logits/rejected": 0.2384227067232132,
"logps/chosen": -535.0108032226562,
"logps/rejected": -592.4139404296875,
"loss": 0.3274,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.14901921153068542,
"rewards/margins": 0.07979042828083038,
"rewards/rejected": -0.22880962491035461,
"step": 110
},
{
"epoch": 0.26,
"learning_rate": 4.63815289945858e-06,
"logits/chosen": 0.17661482095718384,
"logits/rejected": 0.20918302237987518,
"logps/chosen": -542.838134765625,
"logps/rejected": -599.6415405273438,
"loss": 0.3159,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.1861373484134674,
"rewards/margins": 0.09151609241962433,
"rewards/rejected": -0.27765345573425293,
"step": 120
},
{
"epoch": 0.28,
"learning_rate": 4.535619761282989e-06,
"logits/chosen": 0.13282707333564758,
"logits/rejected": 0.18375340104103088,
"logps/chosen": -484.95953369140625,
"logps/rejected": -634.0150146484375,
"loss": 0.2988,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.15236204862594604,
"rewards/margins": 0.14143863320350647,
"rewards/rejected": -0.2938006520271301,
"step": 130
},
{
"epoch": 0.3,
"learning_rate": 4.42175660319555e-06,
"logits/chosen": 0.18953588604927063,
"logits/rejected": 0.19973225891590118,
"logps/chosen": -518.3944091796875,
"logps/rejected": -572.0684814453125,
"loss": 0.3261,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.17030271887779236,
"rewards/margins": 0.08826728165149689,
"rewards/rejected": -0.25857001543045044,
"step": 140
},
{
"epoch": 0.32,
"learning_rate": 4.297197174127619e-06,
"logits/chosen": 0.14177818596363068,
"logits/rejected": 0.19645562767982483,
"logps/chosen": -483.99383544921875,
"logps/rejected": -587.1974487304688,
"loss": 0.3197,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.1454397737979889,
"rewards/margins": 0.09491118788719177,
"rewards/rejected": -0.24035099148750305,
"step": 150
},
{
"epoch": 0.34,
"learning_rate": 4.162634757195418e-06,
"logits/chosen": 0.1273745745420456,
"logits/rejected": 0.21790139377117157,
"logps/chosen": -491.3045959472656,
"logps/rejected": -556.7537841796875,
"loss": 0.3097,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.15721774101257324,
"rewards/margins": 0.08056008815765381,
"rewards/rejected": -0.23777782917022705,
"step": 160
},
{
"epoch": 0.36,
"learning_rate": 4.018818310967843e-06,
"logits/chosen": 0.16942057013511658,
"logits/rejected": 0.25549784302711487,
"logps/chosen": -509.8619079589844,
"logps/rejected": -641.41748046875,
"loss": 0.3034,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.1676618903875351,
"rewards/margins": 0.12202789634466171,
"rewards/rejected": -0.2896897792816162,
"step": 170
},
{
"epoch": 0.38,
"learning_rate": 3.866548300851254e-06,
"logits/chosen": 0.15910522639751434,
"logits/rejected": 0.1666603833436966,
"logps/chosen": -499.43865966796875,
"logps/rejected": -589.9558715820312,
"loss": 0.3048,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.17921218276023865,
"rewards/margins": 0.09624715149402618,
"rewards/rejected": -0.275459349155426,
"step": 180
},
{
"epoch": 0.41,
"learning_rate": 3.706672243793271e-06,
"logits/chosen": 0.11776936054229736,
"logits/rejected": 0.15903571248054504,
"logps/chosen": -492.0912170410156,
"logps/rejected": -582.832275390625,
"loss": 0.3032,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.18043668568134308,
"rewards/margins": 0.09369800984859467,
"rewards/rejected": -0.27413469552993774,
"step": 190
},
{
"epoch": 0.43,
"learning_rate": 3.5400799911032357e-06,
"logits/chosen": 0.1436721682548523,
"logits/rejected": 0.1706967055797577,
"logps/chosen": -506.0414123535156,
"logps/rejected": -571.4830322265625,
"loss": 0.3173,
"rewards/accuracies": 0.581250011920929,
"rewards/chosen": -0.15588083863258362,
"rewards/margins": 0.07620591670274734,
"rewards/rejected": -0.23208674788475037,
"step": 200
},
{
"epoch": 0.45,
"learning_rate": 3.3676987756445894e-06,
"logits/chosen": 0.1567540317773819,
"logits/rejected": 0.21821483969688416,
"logps/chosen": -516.2056884765625,
"logps/rejected": -605.7388916015625,
"loss": 0.2942,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.16281944513320923,
"rewards/margins": 0.11313279718160629,
"rewards/rejected": -0.2759522497653961,
"step": 210
},
{
"epoch": 0.47,
"learning_rate": 3.1904880509659397e-06,
"logits/chosen": 0.09412068128585815,
"logits/rejected": 0.26981237530708313,
"logps/chosen": -573.1748046875,
"logps/rejected": -600.4771728515625,
"loss": 0.3023,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.1811535358428955,
"rewards/margins": 0.09422539919614792,
"rewards/rejected": -0.275378942489624,
"step": 220
},
{
"epoch": 0.49,
"learning_rate": 3.0094341510955697e-06,
"logits/chosen": 0.11837532371282578,
"logits/rejected": 0.19376713037490845,
"logps/chosen": -569.38037109375,
"logps/rejected": -616.151123046875,
"loss": 0.3027,
"rewards/accuracies": 0.5687500238418579,
"rewards/chosen": -0.21316485106945038,
"rewards/margins": 0.07764045149087906,
"rewards/rejected": -0.29080528020858765,
"step": 230
},
{
"epoch": 0.51,
"learning_rate": 2.825544800722376e-06,
"logits/chosen": 0.1659005582332611,
"logits/rejected": 0.2277708798646927,
"logps/chosen": -571.8699951171875,
"logps/rejected": -628.5322875976562,
"loss": 0.2852,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.20772342383861542,
"rewards/margins": 0.10460670292377472,
"rewards/rejected": -0.31233012676239014,
"step": 240
},
{
"epoch": 0.53,
"learning_rate": 2.639843506318899e-06,
"logits/chosen": 0.1882917284965515,
"logits/rejected": 0.2411785125732422,
"logps/chosen": -557.4056396484375,
"logps/rejected": -673.3493041992188,
"loss": 0.2959,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.20086392760276794,
"rewards/margins": 0.11475809663534164,
"rewards/rejected": -0.3156220316886902,
"step": 250
},
{
"epoch": 0.55,
"learning_rate": 2.4533638594248094e-06,
"logits/chosen": 0.13379403948783875,
"logits/rejected": 0.19422146677970886,
"logps/chosen": -565.0435180664062,
"logps/rejected": -694.2730712890625,
"loss": 0.2895,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.23057802021503448,
"rewards/margins": 0.11758030951023102,
"rewards/rejected": -0.3481583297252655,
"step": 260
},
{
"epoch": 0.58,
"learning_rate": 2.2671437837980943e-06,
"logits/chosen": 0.1472124606370926,
"logits/rejected": 0.21264033019542694,
"logps/chosen": -553.3011474609375,
"logps/rejected": -670.6328125,
"loss": 0.285,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.21868601441383362,
"rewards/margins": 0.12228033691644669,
"rewards/rejected": -0.3409663736820221,
"step": 270
},
{
"epoch": 0.6,
"learning_rate": 2.082219758453629e-06,
"logits/chosen": 0.15410706400871277,
"logits/rejected": 0.17184455692768097,
"logps/chosen": -515.2255859375,
"logps/rejected": -623.633544921875,
"loss": 0.296,
"rewards/accuracies": 0.606249988079071,
"rewards/chosen": -0.21407786011695862,
"rewards/margins": 0.105091392993927,
"rewards/rejected": -0.3191692531108856,
"step": 280
},
{
"epoch": 0.62,
"learning_rate": 1.899621048743019e-06,
"logits/chosen": 0.08964169770479202,
"logits/rejected": 0.16084159910678864,
"logps/chosen": -526.2240600585938,
"logps/rejected": -663.767822265625,
"loss": 0.3056,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.21246090531349182,
"rewards/margins": 0.12431824207305908,
"rewards/rejected": -0.3367791175842285,
"step": 290
},
{
"epoch": 0.64,
"learning_rate": 1.7203639775848423e-06,
"logits/chosen": 0.06412859261035919,
"logits/rejected": 0.1526188850402832,
"logps/chosen": -508.3192443847656,
"logps/rejected": -625.1074829101562,
"loss": 0.2862,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.18599674105644226,
"rewards/margins": 0.11884822696447372,
"rewards/rejected": -0.3048449456691742,
"step": 300
},
{
"epoch": 0.66,
"learning_rate": 1.5454462687309445e-06,
"logits/chosen": 0.12014339864253998,
"logits/rejected": 0.17034398019313812,
"logps/chosen": -528.6392822265625,
"logps/rejected": -658.5401611328125,
"loss": 0.3008,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.2073144167661667,
"rewards/margins": 0.1150335818529129,
"rewards/rejected": -0.3223479390144348,
"step": 310
},
{
"epoch": 0.68,
"learning_rate": 1.3758414935535147e-06,
"logits/chosen": 0.1499626636505127,
"logits/rejected": 0.1570507287979126,
"logps/chosen": -577.6300659179688,
"logps/rejected": -674.1231079101562,
"loss": 0.2964,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.2138519585132599,
"rewards/margins": 0.10956285148859024,
"rewards/rejected": -0.32341480255126953,
"step": 320
},
{
"epoch": 0.7,
"learning_rate": 1.2124936522614622e-06,
"logits/chosen": 0.08589409291744232,
"logits/rejected": 0.18073198199272156,
"logps/chosen": -520.1068115234375,
"logps/rejected": -660.3935546875,
"loss": 0.284,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.17694085836410522,
"rewards/margins": 0.15071728825569153,
"rewards/rejected": -0.32765814661979675,
"step": 330
},
{
"epoch": 0.73,
"learning_rate": 1.0563119197063934e-06,
"logits/chosen": 0.09931635111570358,
"logits/rejected": 0.19644983112812042,
"logps/chosen": -560.4113159179688,
"logps/rejected": -655.904296875,
"loss": 0.2886,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.21122217178344727,
"rewards/margins": 0.09574859589338303,
"rewards/rejected": -0.3069707751274109,
"step": 340
},
{
"epoch": 0.75,
"learning_rate": 9.081655850224449e-07,
"logits/chosen": 0.10707017034292221,
"logits/rejected": 0.1817469596862793,
"logps/chosen": -537.7971801757812,
"logps/rejected": -607.695068359375,
"loss": 0.3062,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.21373042464256287,
"rewards/margins": 0.07750532776117325,
"rewards/rejected": -0.2912357449531555,
"step": 350
},
{
"epoch": 0.77,
"learning_rate": 7.688792132653111e-07,
"logits/chosen": 0.09279724210500717,
"logits/rejected": 0.14600899815559387,
"logps/chosen": -518.6719970703125,
"logps/rejected": -613.5050048828125,
"loss": 0.3102,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.20901672542095184,
"rewards/margins": 0.10807285457849503,
"rewards/rejected": -0.31708958745002747,
"step": 360
},
{
"epoch": 0.79,
"learning_rate": 6.392280559802341e-07,
"logits/chosen": 0.09189110994338989,
"logits/rejected": 0.14240947365760803,
"logps/chosen": -499.902587890625,
"logps/rejected": -645.5587158203125,
"loss": 0.2901,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.19295868277549744,
"rewards/margins": 0.1267748475074768,
"rewards/rejected": -0.31973353028297424,
"step": 370
},
{
"epoch": 0.81,
"learning_rate": 5.199337362431792e-07,
"logits/chosen": 0.10996093600988388,
"logits/rejected": 0.17150555551052094,
"logps/chosen": -541.5203857421875,
"logps/rejected": -660.8944091796875,
"loss": 0.2869,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.19876644015312195,
"rewards/margins": 0.13654473423957825,
"rewards/rejected": -0.3353111743927002,
"step": 380
},
{
"epoch": 0.83,
"learning_rate": 4.1166023219176176e-07,
"logits/chosen": 0.08110027015209198,
"logits/rejected": 0.1370091438293457,
"logps/chosen": -551.21923828125,
"logps/rejected": -639.4774780273438,
"loss": 0.2989,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.2069006711244583,
"rewards/margins": 0.1049899011850357,
"rewards/rejected": -0.31189054250717163,
"step": 390
},
{
"epoch": 0.85,
"learning_rate": 3.150101814011136e-07,
"logits/chosen": 0.09940965473651886,
"logits/rejected": 0.14982450008392334,
"logps/chosen": -575.6122436523438,
"logps/rejected": -670.1174926757812,
"loss": 0.3034,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.2270297110080719,
"rewards/margins": 0.09964431822299957,
"rewards/rejected": -0.3266740143299103,
"step": 400
},
{
"epoch": 0.87,
"learning_rate": 3.4545537312690565e-06,
"logits/chosen": 0.1286882907152176,
"logits/rejected": 0.13096146285533905,
"logps/chosen": -515.2964477539062,
"logps/rejected": -708.9202270507812,
"loss": 0.2656,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.19799669086933136,
"rewards/margins": 0.1586463451385498,
"rewards/rejected": -0.35664302110671997,
"step": 410
},
{
"epoch": 0.9,
"learning_rate": 3.3676987756445894e-06,
"logits/chosen": 0.13132785260677338,
"logits/rejected": 0.13156192004680634,
"logps/chosen": -523.5271606445312,
"logps/rejected": -717.6295166015625,
"loss": 0.3083,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.2122844010591507,
"rewards/margins": 0.15919235348701477,
"rewards/rejected": -0.3714767396450043,
"step": 420
},
{
"epoch": 0.92,
"learning_rate": 3.2796360224151587e-06,
"logits/chosen": 0.1404297947883606,
"logits/rejected": 0.13226640224456787,
"logps/chosen": -557.2911987304688,
"logps/rejected": -633.6729736328125,
"loss": 0.2935,
"rewards/accuracies": 0.612500011920929,
"rewards/chosen": -0.22657549381256104,
"rewards/margins": 0.11841969192028046,
"rewards/rejected": -0.3449952006340027,
"step": 430
},
{
"epoch": 0.94,
"learning_rate": 3.1904880509659397e-06,
"logits/chosen": 0.12834230065345764,
"logits/rejected": 0.22366562485694885,
"logps/chosen": -551.0172119140625,
"logps/rejected": -710.9666748046875,
"loss": 0.2803,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.22843453288078308,
"rewards/margins": 0.14828652143478394,
"rewards/rejected": -0.376721054315567,
"step": 440
},
{
"epoch": 0.96,
"learning_rate": 3.100378951256981e-06,
"logits/chosen": 0.11333411931991577,
"logits/rejected": 0.18246471881866455,
"logps/chosen": -568.0606689453125,
"logps/rejected": -714.0242309570312,
"loss": 0.2998,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.24061532318592072,
"rewards/margins": 0.12436308711767197,
"rewards/rejected": -0.3649784326553345,
"step": 450
},
{
"epoch": 0.98,
"learning_rate": 3.0094341510955697e-06,
"logits/chosen": 0.14087212085723877,
"logits/rejected": 0.2087050974369049,
"logps/chosen": -552.45947265625,
"logps/rejected": -642.1560668945312,
"loss": 0.2889,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.20147624611854553,
"rewards/margins": 0.1283041536808014,
"rewards/rejected": -0.32978036999702454,
"step": 460
},
{
"epoch": 1.0,
"learning_rate": 2.9177802415463714e-06,
"logits/chosen": 0.16158871352672577,
"logits/rejected": 0.23096832633018494,
"logps/chosen": -573.5584716796875,
"logps/rejected": -712.2202758789062,
"loss": 0.2774,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.22124040126800537,
"rewards/margins": 0.14029137790203094,
"rewards/rejected": -0.3615317940711975,
"step": 470
},
{
"epoch": 1.02,
"learning_rate": 2.825544800722376e-06,
"logits/chosen": 0.1098761335015297,
"logits/rejected": 0.2546694278717041,
"logps/chosen": -615.3218994140625,
"logps/rejected": -713.9712524414062,
"loss": 0.2699,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.23255982995033264,
"rewards/margins": 0.13802029192447662,
"rewards/rejected": -0.3705800771713257,
"step": 480
},
{
"epoch": 1.05,
"learning_rate": 2.732856216201906e-06,
"logits/chosen": 0.08082452416419983,
"logits/rejected": 0.1814456284046173,
"logps/chosen": -614.1076049804688,
"logps/rejected": -726.0872192382812,
"loss": 0.2972,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.24807974696159363,
"rewards/margins": 0.13322189450263977,
"rewards/rejected": -0.3813017010688782,
"step": 490
},
{
"epoch": 1.07,
"learning_rate": 2.639843506318899e-06,
"logits/chosen": 0.1431538164615631,
"logits/rejected": 0.16376420855522156,
"logps/chosen": -603.7203369140625,
"logps/rejected": -724.11669921875,
"loss": 0.2644,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.24805834889411926,
"rewards/margins": 0.13241049647331238,
"rewards/rejected": -0.38046884536743164,
"step": 500
},
{
"epoch": 1.09,
"learning_rate": 2.5466361405751914e-06,
"logits/chosen": 0.09880776703357697,
"logits/rejected": 0.15729175508022308,
"logps/chosen": -563.4605712890625,
"logps/rejected": -697.0057373046875,
"loss": 0.3066,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.22475488483905792,
"rewards/margins": 0.13303039968013763,
"rewards/rejected": -0.35778528451919556,
"step": 510
},
{
"epoch": 1.11,
"learning_rate": 2.4533638594248094e-06,
"logits/chosen": 0.08443008363246918,
"logits/rejected": 0.15599215030670166,
"logps/chosen": -553.6532592773438,
"logps/rejected": -658.5653076171875,
"loss": 0.2935,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.22460079193115234,
"rewards/margins": 0.11079039424657822,
"rewards/rejected": -0.3353911340236664,
"step": 520
},
{
"epoch": 1.13,
"learning_rate": 2.360156493681102e-06,
"logits/chosen": 0.10354779660701752,
"logits/rejected": 0.14787225425243378,
"logps/chosen": -592.5408325195312,
"logps/rejected": -705.7700805664062,
"loss": 0.2888,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.25297555327415466,
"rewards/margins": 0.1141052097082138,
"rewards/rejected": -0.36708077788352966,
"step": 530
},
{
"epoch": 1.15,
"learning_rate": 2.2671437837980943e-06,
"logits/chosen": 0.08793269097805023,
"logits/rejected": 0.17164576053619385,
"logps/chosen": -603.9788208007812,
"logps/rejected": -723.2301025390625,
"loss": 0.284,
"rewards/accuracies": 0.643750011920929,
"rewards/chosen": -0.2590700089931488,
"rewards/margins": 0.13130523264408112,
"rewards/rejected": -0.3903752863407135,
"step": 540
},
{
"epoch": 1.17,
"learning_rate": 2.1744551992776247e-06,
"logits/chosen": 0.09502136707305908,
"logits/rejected": 0.12443023920059204,
"logps/chosen": -556.7479858398438,
"logps/rejected": -695.36669921875,
"loss": 0.2874,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.24774689972400665,
"rewards/margins": 0.13886813819408417,
"rewards/rejected": -0.3866150975227356,
"step": 550
},
{
"epoch": 1.19,
"learning_rate": 2.082219758453629e-06,
"logits/chosen": 0.03692219406366348,
"logits/rejected": 0.1840008795261383,
"logps/chosen": -606.75,
"logps/rejected": -705.9107666015625,
"loss": 0.2798,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.27475666999816895,
"rewards/margins": 0.11195492744445801,
"rewards/rejected": -0.38671156764030457,
"step": 560
},
{
"epoch": 1.22,
"learning_rate": 1.990565848904431e-06,
"logits/chosen": 0.07638466358184814,
"logits/rejected": 0.12104539573192596,
"logps/chosen": -611.66650390625,
"logps/rejected": -736.2400512695312,
"loss": 0.3029,
"rewards/accuracies": 0.6499999761581421,
"rewards/chosen": -0.2728550434112549,
"rewards/margins": 0.14484602212905884,
"rewards/rejected": -0.4177010655403137,
"step": 570
},
{
"epoch": 1.24,
"learning_rate": 1.899621048743019e-06,
"logits/chosen": 0.12088838964700699,
"logits/rejected": 0.07981579750776291,
"logps/chosen": -601.095947265625,
"logps/rejected": -716.2054443359375,
"loss": 0.2954,
"rewards/accuracies": 0.6312500238418579,
"rewards/chosen": -0.2841927409172058,
"rewards/margins": 0.11602810770273209,
"rewards/rejected": -0.4002207815647125,
"step": 580
},
{
"epoch": 1.26,
"learning_rate": 1.8095119490340618e-06,
"logits/chosen": 0.1057354211807251,
"logits/rejected": 0.20554998517036438,
"logps/chosen": -656.5835571289062,
"logps/rejected": -745.124267578125,
"loss": 0.2951,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.28658175468444824,
"rewards/margins": 0.13431188464164734,
"rewards/rejected": -0.42089366912841797,
"step": 590
},
{
"epoch": 1.28,
"learning_rate": 1.7203639775848423e-06,
"logits/chosen": 0.08960368484258652,
"logits/rejected": 0.1500837355852127,
"logps/chosen": -605.3450317382812,
"logps/rejected": -729.7435913085938,
"loss": 0.2824,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.26503902673721313,
"rewards/margins": 0.13953007757663727,
"rewards/rejected": -0.4045690894126892,
"step": 600
},
{
"epoch": 1.3,
"learning_rate": 1.632301224355411e-06,
"logits/chosen": 0.0958481878042221,
"logits/rejected": 0.14852683246135712,
"logps/chosen": -632.3499145507812,
"logps/rejected": -763.3027954101562,
"loss": 0.2912,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.26938048005104065,
"rewards/margins": 0.15196532011032104,
"rewards/rejected": -0.4213457703590393,
"step": 610
},
{
"epoch": 1.32,
"learning_rate": 1.5454462687309445e-06,
"logits/chosen": 0.07703549414873123,
"logits/rejected": 0.14966604113578796,
"logps/chosen": -574.7545166015625,
"logps/rejected": -733.3427734375,
"loss": 0.2753,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.2730166018009186,
"rewards/margins": 0.15502889454364777,
"rewards/rejected": -0.42804545164108276,
"step": 620
},
{
"epoch": 1.34,
"learning_rate": 1.4599200088967652e-06,
"logits/chosen": 0.06834821403026581,
"logits/rejected": 0.09840573370456696,
"logps/chosen": -588.8328857421875,
"logps/rejected": -734.0791625976562,
"loss": 0.2827,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.29259082674980164,
"rewards/margins": 0.11803393065929413,
"rewards/rejected": -0.41062480211257935,
"step": 630
},
{
"epoch": 1.37,
"learning_rate": 1.3758414935535147e-06,
"logits/chosen": 0.05101291462779045,
"logits/rejected": 0.19463148713111877,
"logps/chosen": -635.7034912109375,
"logps/rejected": -769.3609619140625,
"loss": 0.287,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": -0.27811720967292786,
"rewards/margins": 0.15194621682167053,
"rewards/rejected": -0.4300634264945984,
"step": 640
},
{
"epoch": 1.39,
"learning_rate": 1.293327756206729e-06,
"logits/chosen": 0.06910277158021927,
"logits/rejected": 0.08267343044281006,
"logps/chosen": -630.0482177734375,
"logps/rejected": -760.6334228515625,
"loss": 0.2618,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.28737881779670715,
"rewards/margins": 0.14742901921272278,
"rewards/rejected": -0.4348078668117523,
"step": 650
},
{
"epoch": 1.41,
"learning_rate": 1.2124936522614622e-06,
"logits/chosen": 0.08306999504566193,
"logits/rejected": 0.18402306735515594,
"logps/chosen": -614.7750244140625,
"logps/rejected": -746.0162353515625,
"loss": 0.243,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.2686184048652649,
"rewards/margins": 0.14612407982349396,
"rewards/rejected": -0.41474246978759766,
"step": 660
},
{
"epoch": 1.43,
"learning_rate": 1.1334516991487473e-06,
"logits/chosen": 0.10118447244167328,
"logits/rejected": 0.17648069560527802,
"logps/chosen": -568.6295166015625,
"logps/rejected": -708.2867431640625,
"loss": 0.29,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.2447134256362915,
"rewards/margins": 0.14246779680252075,
"rewards/rejected": -0.38718122243881226,
"step": 670
},
{
"epoch": 1.45,
"learning_rate": 1.0563119197063934e-06,
"logits/chosen": 0.09860675781965256,
"logits/rejected": 0.16828308999538422,
"logps/chosen": -644.8552856445312,
"logps/rejected": -772.2024536132812,
"loss": 0.2985,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.2842712104320526,
"rewards/margins": 0.14040768146514893,
"rewards/rejected": -0.42467889189720154,
"step": 680
},
{
"epoch": 1.47,
"learning_rate": 9.81181689032158e-07,
"logits/chosen": 0.05354565382003784,
"logits/rejected": 0.14584000408649445,
"logps/chosen": -637.1915893554688,
"logps/rejected": -777.9390869140625,
"loss": 0.2788,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.28641483187675476,
"rewards/margins": 0.14343242347240448,
"rewards/rejected": -0.4298473000526428,
"step": 690
},
{
"epoch": 1.49,
"learning_rate": 9.081655850224449e-07,
"logits/chosen": 0.05313466861844063,
"logits/rejected": 0.1454547941684723,
"logps/chosen": -599.1509399414062,
"logps/rejected": -701.4476318359375,
"loss": 0.2915,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.2644391655921936,
"rewards/margins": 0.11923892796039581,
"rewards/rejected": -0.3836781084537506,
"step": 700
},
{
"epoch": 1.51,
"learning_rate": 8.373652428045831e-07,
"logits/chosen": 0.08170448243618011,
"logits/rejected": 0.17704859375953674,
"logps/chosen": -674.1668701171875,
"logps/rejected": -791.4132080078125,
"loss": 0.28,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.2925952076911926,
"rewards/margins": 0.15293267369270325,
"rewards/rejected": -0.4455278813838959,
"step": 710
},
{
"epoch": 1.54,
"learning_rate": 7.688792132653111e-07,
"logits/chosen": 0.13877324759960175,
"logits/rejected": 0.11766058206558228,
"logps/chosen": -636.9558715820312,
"logps/rejected": -835.0140380859375,
"loss": 0.244,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.26449164748191833,
"rewards/margins": 0.2034236639738083,
"rewards/rejected": -0.46791529655456543,
"step": 720
},
{
"epoch": 1.56,
"learning_rate": 7.028028258723818e-07,
"logits/chosen": 0.07956353574991226,
"logits/rejected": 0.15025413036346436,
"logps/chosen": -635.7805786132812,
"logps/rejected": -744.8255615234375,
"loss": 0.2902,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.29231467843055725,
"rewards/margins": 0.12245229631662369,
"rewards/rejected": -0.41476696729660034,
"step": 730
},
{
"epoch": 1.58,
"learning_rate": 6.392280559802341e-07,
"logits/chosen": 0.0947863757610321,
"logits/rejected": 0.14226123690605164,
"logps/chosen": -580.0279541015625,
"logps/rejected": -725.8198852539062,
"loss": 0.2555,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -0.23229190707206726,
"rewards/margins": 0.16553141176700592,
"rewards/rejected": -0.3978233337402344,
"step": 740
},
{
"epoch": 1.6,
"learning_rate": 5.782433968044495e-07,
"logits/chosen": 0.09284155815839767,
"logits/rejected": 0.19596152007579803,
"logps/chosen": -648.7670288085938,
"logps/rejected": -761.2144775390625,
"loss": 0.2766,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": -0.27112412452697754,
"rewards/margins": 0.1452399045228958,
"rewards/rejected": -0.4163641035556793,
"step": 750
},
{
"epoch": 1.62,
"learning_rate": 5.199337362431792e-07,
"logits/chosen": 0.08202387392520905,
"logits/rejected": 0.1395001858472824,
"logps/chosen": -638.7788696289062,
"logps/rejected": -739.9161376953125,
"loss": 0.2912,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.2824631333351135,
"rewards/margins": 0.1440524160861969,
"rewards/rejected": -0.42651551961898804,
"step": 760
},
{
"epoch": 1.64,
"learning_rate": 4.643802387170118e-07,
"logits/chosen": 0.031006410717964172,
"logits/rejected": 0.20315364003181458,
"logps/chosen": -635.1582641601562,
"logps/rejected": -805.4976196289062,
"loss": 0.2579,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.28585267066955566,
"rewards/margins": 0.1737949550151825,
"rewards/rejected": -0.45964759588241577,
"step": 770
},
{
"epoch": 1.66,
"learning_rate": 4.1166023219176176e-07,
"logits/chosen": 0.11464326083660126,
"logits/rejected": 0.14991115033626556,
"logps/chosen": -589.2257690429688,
"logps/rejected": -750.9774169921875,
"loss": 0.2713,
"rewards/accuracies": 0.637499988079071,
"rewards/chosen": -0.2886643409729004,
"rewards/margins": 0.15789894759655,
"rewards/rejected": -0.4465633034706116,
"step": 780
},
{
"epoch": 1.69,
"learning_rate": 3.618471005414215e-07,
"logits/chosen": 0.07082124054431915,
"logits/rejected": 0.11936893314123154,
"logps/chosen": -596.2206420898438,
"logps/rejected": -772.8016357421875,
"loss": 0.2372,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": -0.2700226902961731,
"rewards/margins": 0.17619924247264862,
"rewards/rejected": -0.4462219178676605,
"step": 790
},
{
"epoch": 1.71,
"learning_rate": 3.150101814011136e-07,
"logits/chosen": 0.060673005878925323,
"logits/rejected": 0.15200214087963104,
"logps/chosen": -641.08056640625,
"logps/rejected": -811.6611328125,
"loss": 0.2697,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.2966248095035553,
"rewards/margins": 0.16086846590042114,
"rewards/rejected": -0.45749321579933167,
"step": 800
},
{
"epoch": 1.73,
"learning_rate": 2.712146696522305e-07,
"logits/chosen": 0.0920618399977684,
"logits/rejected": 0.11318673193454742,
"logps/chosen": -665.89599609375,
"logps/rejected": -780.0318603515625,
"loss": 0.2911,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.3211720585823059,
"rewards/margins": 0.14485250413417816,
"rewards/rejected": -0.46602457761764526,
"step": 810
},
{
"epoch": 1.75,
"learning_rate": 2.3052152667409289e-07,
"logits/chosen": 0.10792404413223267,
"logits/rejected": 0.1413693130016327,
"logps/chosen": -645.6889038085938,
"logps/rejected": -751.7999877929688,
"loss": 0.3049,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.30210328102111816,
"rewards/margins": 0.12790395319461823,
"rewards/rejected": -0.43000727891921997,
"step": 820
},
{
"epoch": 1.77,
"learning_rate": 1.9298739548845813e-07,
"logits/chosen": 0.09733234345912933,
"logits/rejected": 0.16931495070457458,
"logps/chosen": -679.776123046875,
"logps/rejected": -784.3840942382812,
"loss": 0.2662,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.30325835943222046,
"rewards/margins": 0.15936632454395294,
"rewards/rejected": -0.4626246392726898,
"step": 830
},
{
"epoch": 1.79,
"learning_rate": 1.5866452191498488e-07,
"logits/chosen": 0.07468713819980621,
"logits/rejected": 0.15481439232826233,
"logps/chosen": -627.0779418945312,
"logps/rejected": -745.9903564453125,
"loss": 0.2667,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.27285081148147583,
"rewards/margins": 0.14448818564414978,
"rewards/rejected": -0.41733893752098083,
"step": 840
},
{
"epoch": 1.81,
"learning_rate": 1.2760068184740597e-07,
"logits/chosen": 0.07968030869960785,
"logits/rejected": 0.14501354098320007,
"logps/chosen": -628.3474731445312,
"logps/rejected": -817.0506591796875,
"loss": 0.2647,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.28495317697525024,
"rewards/margins": 0.17006665468215942,
"rewards/rejected": -0.4550197720527649,
"step": 850
},
{
"epoch": 1.83,
"learning_rate": 9.983911475163727e-08,
"logits/chosen": 0.05757877230644226,
"logits/rejected": 0.20654790103435516,
"logps/chosen": -694.041748046875,
"logps/rejected": -830.3031005859375,
"loss": 0.279,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.3297106921672821,
"rewards/margins": 0.1504661738872528,
"rewards/rejected": -0.4801768362522125,
"step": 860
},
{
"epoch": 1.86,
"learning_rate": 7.541846347838915e-08,
"logits/chosen": 0.07009818404912949,
"logits/rejected": 0.08682046085596085,
"logps/chosen": -622.718505859375,
"logps/rejected": -736.4935302734375,
"loss": 0.2882,
"rewards/accuracies": 0.6937500238418579,
"rewards/chosen": -0.2856903374195099,
"rewards/margins": 0.15370842814445496,
"rewards/rejected": -0.4393988251686096,
"step": 870
},
{
"epoch": 1.88,
"learning_rate": 5.437272047405712e-08,
"logits/chosen": 0.05985601618885994,
"logits/rejected": 0.19025097787380219,
"logps/chosen": -640.6109619140625,
"logps/rejected": -787.9774169921875,
"loss": 0.2732,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": -0.29887470602989197,
"rewards/margins": 0.16820885241031647,
"rewards/rejected": -0.467083603143692,
"step": 880
},
{
"epoch": 1.9,
"learning_rate": 3.673118046477159e-08,
"logits/chosen": 0.07184259593486786,
"logits/rejected": 0.19540506601333618,
"logps/chosen": -646.6096801757812,
"logps/rejected": -782.7240600585938,
"loss": 0.278,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.3017248213291168,
"rewards/margins": 0.1628948152065277,
"rewards/rejected": -0.46461963653564453,
"step": 890
},
{
"epoch": 1.92,
"learning_rate": 2.251839967945535e-08,
"logits/chosen": 0.05988436937332153,
"logits/rejected": 0.11050447076559067,
"logps/chosen": -601.7271728515625,
"logps/rejected": -758.0606689453125,
"loss": 0.2704,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.28261813521385193,
"rewards/margins": 0.15564511716365814,
"rewards/rejected": -0.4382632374763489,
"step": 900
},
{
"epoch": 1.94,
"learning_rate": 1.1754161668660612e-08,
"logits/chosen": 0.07712358981370926,
"logits/rejected": 0.17667053639888763,
"logps/chosen": -672.001708984375,
"logps/rejected": -807.9947509765625,
"loss": 0.2561,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": -0.2962462306022644,
"rewards/margins": 0.16602984070777893,
"rewards/rejected": -0.46227607131004333,
"step": 910
},
{
"epoch": 1.96,
"learning_rate": 4.453449766758933e-09,
"logits/chosen": 0.1141105443239212,
"logits/rejected": 0.20343729853630066,
"logps/chosen": -630.8961181640625,
"logps/rejected": -757.9405517578125,
"loss": 0.2796,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": -0.2968885600566864,
"rewards/margins": 0.12835559248924255,
"rewards/rejected": -0.42524418234825134,
"step": 920
},
{
"epoch": 1.98,
"learning_rate": 6.264262358129936e-10,
"logits/chosen": 0.07077296823263168,
"logits/rejected": 0.14096608757972717,
"logps/chosen": -652.6800537109375,
"logps/rejected": -770.9761352539062,
"loss": 0.291,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.30475863814353943,
"rewards/margins": 0.14434154331684113,
"rewards/rejected": -0.44910019636154175,
"step": 930
},
{
"epoch": 2.0,
"step": 936,
"total_flos": 0.0,
"train_loss": 0.16023516272887206,
"train_runtime": 6530.1152,
"train_samples_per_second": 4.594,
"train_steps_per_second": 0.143
}
],
"logging_steps": 10,
"max_steps": 936,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 100,
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}