elichen3051's picture
Model save
e3406fb verified
raw
history blame
227 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.9940933254577673,
"eval_steps": 43,
"global_step": 422,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.004725339633786178,
"grad_norm": 1.1176782705315427,
"learning_rate": 3.846153846153846e-08,
"logits/chosen": -1.5937305688858032,
"logits/rejected": -1.7021960020065308,
"logps/chosen": -247.54559326171875,
"logps/rejected": -179.0218048095703,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.009450679267572357,
"grad_norm": 1.2535773078918948,
"learning_rate": 7.692307692307692e-08,
"logits/chosen": -1.7181015014648438,
"logits/rejected": -1.644026756286621,
"logps/chosen": -259.1505432128906,
"logps/rejected": -241.68020629882812,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 2
},
{
"epoch": 0.014176018901358535,
"grad_norm": 0.995484839907434,
"learning_rate": 1.1538461538461539e-07,
"logits/chosen": -0.8613071441650391,
"logits/rejected": -0.8891040682792664,
"logps/chosen": -230.91070556640625,
"logps/rejected": -219.62979125976562,
"loss": 0.6934,
"rewards/accuracies": 0.390625,
"rewards/chosen": 6.782727723475546e-05,
"rewards/margins": -0.0004302160523366183,
"rewards/rejected": 0.0004980433732271194,
"step": 3
},
{
"epoch": 0.018901358535144713,
"grad_norm": 1.1568688062326662,
"learning_rate": 1.5384615384615385e-07,
"logits/chosen": -1.1649622917175293,
"logits/rejected": -1.131172776222229,
"logps/chosen": -184.93499755859375,
"logps/rejected": -184.127197265625,
"loss": 0.693,
"rewards/accuracies": 0.421875,
"rewards/chosen": -0.0002185619086958468,
"rewards/margins": -0.0006310059688985348,
"rewards/rejected": 0.0004124442348256707,
"step": 4
},
{
"epoch": 0.02362669816893089,
"grad_norm": 1.0155652752384032,
"learning_rate": 1.9230769230769231e-07,
"logits/chosen": -1.8650751113891602,
"logits/rejected": -1.9386688470840454,
"logps/chosen": -193.21636962890625,
"logps/rejected": -175.6696014404297,
"loss": 0.6932,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.00019388733198866248,
"rewards/margins": -0.00023763455101288855,
"rewards/rejected": 0.0004315219703130424,
"step": 5
},
{
"epoch": 0.02835203780271707,
"grad_norm": 1.2022303509332404,
"learning_rate": 2.3076923076923078e-07,
"logits/chosen": -1.8849067687988281,
"logits/rejected": -1.8837637901306152,
"logps/chosen": -234.3896484375,
"logps/rejected": -218.09625244140625,
"loss": 0.6934,
"rewards/accuracies": 0.4375,
"rewards/chosen": 0.0004923291853629053,
"rewards/margins": 0.0006284262635745108,
"rewards/rejected": -0.00013609707821160555,
"step": 6
},
{
"epoch": 0.03307737743650325,
"grad_norm": 1.136952564893261,
"learning_rate": 2.692307692307692e-07,
"logits/chosen": -1.816144585609436,
"logits/rejected": -1.934072494506836,
"logps/chosen": -246.14027404785156,
"logps/rejected": -177.02993774414062,
"loss": 0.6931,
"rewards/accuracies": 0.53125,
"rewards/chosen": -2.0834413589909673e-05,
"rewards/margins": 0.001102915033698082,
"rewards/rejected": -0.0011237493017688394,
"step": 7
},
{
"epoch": 0.03780271707028943,
"grad_norm": 1.1596582232433024,
"learning_rate": 3.076923076923077e-07,
"logits/chosen": -1.8363107442855835,
"logits/rejected": -1.8167006969451904,
"logps/chosen": -239.65370178222656,
"logps/rejected": -221.20333862304688,
"loss": 0.6932,
"rewards/accuracies": 0.4375,
"rewards/chosen": -0.00017343120998702943,
"rewards/margins": -0.0007941695512272418,
"rewards/rejected": 0.0006207384867593646,
"step": 8
},
{
"epoch": 0.042528056704075605,
"grad_norm": 1.1810520949859216,
"learning_rate": 3.461538461538461e-07,
"logits/chosen": -1.6903538703918457,
"logits/rejected": -1.7734307050704956,
"logps/chosen": -239.26535034179688,
"logps/rejected": -193.875244140625,
"loss": 0.6929,
"rewards/accuracies": 0.46875,
"rewards/chosen": 0.0003690614248625934,
"rewards/margins": 0.00022412401449400932,
"rewards/rejected": 0.00014493743947241455,
"step": 9
},
{
"epoch": 0.04725339633786178,
"grad_norm": 1.253373850013781,
"learning_rate": 3.8461538461538463e-07,
"logits/chosen": -1.5372889041900635,
"logits/rejected": -1.5536653995513916,
"logps/chosen": -222.93399047851562,
"logps/rejected": -223.1899871826172,
"loss": 0.6929,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.0005048069870099425,
"rewards/margins": 0.000533790560439229,
"rewards/rejected": -0.0010385976638644934,
"step": 10
},
{
"epoch": 0.05197873597164796,
"grad_norm": 1.1609876885112298,
"learning_rate": 4.2307692307692304e-07,
"logits/chosen": -1.20901620388031,
"logits/rejected": -1.2452011108398438,
"logps/chosen": -274.1497497558594,
"logps/rejected": -227.36790466308594,
"loss": 0.6929,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.0004496507463045418,
"rewards/margins": -0.0006296815699897707,
"rewards/rejected": 0.0001800309109967202,
"step": 11
},
{
"epoch": 0.05670407560543414,
"grad_norm": 1.1166995258772006,
"learning_rate": 4.6153846153846156e-07,
"logits/chosen": -1.8701603412628174,
"logits/rejected": -1.8107975721359253,
"logps/chosen": -219.3397674560547,
"logps/rejected": -232.28269958496094,
"loss": 0.6927,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.0005328843253664672,
"rewards/margins": 0.000161867166752927,
"rewards/rejected": -0.0006947515066713095,
"step": 12
},
{
"epoch": 0.06142941523922032,
"grad_norm": 1.1010862122779934,
"learning_rate": 5e-07,
"logits/chosen": -1.3035281896591187,
"logits/rejected": -1.3319075107574463,
"logps/chosen": -200.9400634765625,
"logps/rejected": -177.5894317626953,
"loss": 0.6924,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.00024598470190539956,
"rewards/margins": 0.001570576336234808,
"rewards/rejected": -0.0013245916925370693,
"step": 13
},
{
"epoch": 0.0661547548730065,
"grad_norm": 1.109450926366569,
"learning_rate": 4.999926250172797e-07,
"logits/chosen": -1.2467422485351562,
"logits/rejected": -1.2619496583938599,
"logps/chosen": -247.1371612548828,
"logps/rejected": -232.03895568847656,
"loss": 0.6925,
"rewards/accuracies": 0.640625,
"rewards/chosen": 0.00045037176460027695,
"rewards/margins": 0.0029086670838296413,
"rewards/rejected": -0.0024582953192293644,
"step": 14
},
{
"epoch": 0.07088009450679268,
"grad_norm": 1.0686336079566285,
"learning_rate": 4.999705005042417e-07,
"logits/chosen": -0.9053488969802856,
"logits/rejected": -0.9105295538902283,
"logps/chosen": -200.528076171875,
"logps/rejected": -191.2373046875,
"loss": 0.6924,
"rewards/accuracies": 0.484375,
"rewards/chosen": -0.0003694885817822069,
"rewards/margins": 0.001977597363293171,
"rewards/rejected": -0.00234708609059453,
"step": 15
},
{
"epoch": 0.07560543414057885,
"grad_norm": 1.1198403704053106,
"learning_rate": 4.999336277662292e-07,
"logits/chosen": -1.382132887840271,
"logits/rejected": -1.3727940320968628,
"logps/chosen": -237.94508361816406,
"logps/rejected": -251.71519470214844,
"loss": 0.6924,
"rewards/accuracies": 0.640625,
"rewards/chosen": -3.6539247957989573e-05,
"rewards/margins": 0.002266494557261467,
"rewards/rejected": -0.002303033834323287,
"step": 16
},
{
"epoch": 0.08033077377436504,
"grad_norm": 1.022581880493408,
"learning_rate": 4.998820089787287e-07,
"logits/chosen": -1.0172172784805298,
"logits/rejected": -1.0724008083343506,
"logps/chosen": -232.03070068359375,
"logps/rejected": -214.65155029296875,
"loss": 0.6919,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.0009355681831948459,
"rewards/margins": 0.002296539256349206,
"rewards/rejected": -0.003232107497751713,
"step": 17
},
{
"epoch": 0.08505611340815121,
"grad_norm": 1.127368600628108,
"learning_rate": 4.998156471872415e-07,
"logits/chosen": -1.6294444799423218,
"logits/rejected": -1.6482771635055542,
"logps/chosen": -226.63442993164062,
"logps/rejected": -209.50799560546875,
"loss": 0.6913,
"rewards/accuracies": 0.671875,
"rewards/chosen": 0.0007066840189509094,
"rewards/margins": 0.005803423933684826,
"rewards/rejected": -0.005096739623695612,
"step": 18
},
{
"epoch": 0.0897814530419374,
"grad_norm": 1.1009230447057081,
"learning_rate": 4.997345463071041e-07,
"logits/chosen": -1.9955631494522095,
"logits/rejected": -1.9177535772323608,
"logps/chosen": -219.0171661376953,
"logps/rejected": -204.72091674804688,
"loss": 0.6906,
"rewards/accuracies": 0.515625,
"rewards/chosen": -0.001279333489947021,
"rewards/margins": 0.003984270617365837,
"rewards/rejected": -0.005263603758066893,
"step": 19
},
{
"epoch": 0.09450679267572357,
"grad_norm": 1.1219432468145276,
"learning_rate": 4.996387111232572e-07,
"logits/chosen": -0.9923038482666016,
"logits/rejected": -1.0317736864089966,
"logps/chosen": -207.70364379882812,
"logps/rejected": -210.37612915039062,
"loss": 0.6903,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.0004186414007563144,
"rewards/margins": 0.007054868154227734,
"rewards/rejected": -0.007473509293049574,
"step": 20
},
{
"epoch": 0.09923213230950975,
"grad_norm": 1.1456255677406275,
"learning_rate": 4.995281472899636e-07,
"logits/chosen": -1.2455811500549316,
"logits/rejected": -1.2986594438552856,
"logps/chosen": -260.4434814453125,
"logps/rejected": -239.2939453125,
"loss": 0.6899,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.001413986086845398,
"rewards/margins": 0.006085187196731567,
"rewards/rejected": -0.007499172817915678,
"step": 21
},
{
"epoch": 0.10395747194329592,
"grad_norm": 1.0796373972108642,
"learning_rate": 4.99402861330474e-07,
"logits/chosen": -1.6776717901229858,
"logits/rejected": -1.6839845180511475,
"logps/chosen": -243.64987182617188,
"logps/rejected": -224.65371704101562,
"loss": 0.6898,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.0021368232555687428,
"rewards/margins": 0.007371032610535622,
"rewards/rejected": -0.009507855400443077,
"step": 22
},
{
"epoch": 0.10868281157708211,
"grad_norm": 1.067721897185461,
"learning_rate": 4.992628606366425e-07,
"logits/chosen": -1.6994775533676147,
"logits/rejected": -1.6823248863220215,
"logps/chosen": -184.07522583007812,
"logps/rejected": -201.24017333984375,
"loss": 0.6892,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.0037174485623836517,
"rewards/margins": 0.005765101406723261,
"rewards/rejected": -0.009482549503445625,
"step": 23
},
{
"epoch": 0.11340815121086828,
"grad_norm": 1.1222480052149575,
"learning_rate": 4.991081534684911e-07,
"logits/chosen": -1.3170721530914307,
"logits/rejected": -1.3209936618804932,
"logps/chosen": -173.28440856933594,
"logps/rejected": -180.39111328125,
"loss": 0.6882,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.0013278307160362601,
"rewards/margins": 0.01227161381393671,
"rewards/rejected": -0.013599444180727005,
"step": 24
},
{
"epoch": 0.11813349084465447,
"grad_norm": 1.1336846530783633,
"learning_rate": 4.98938748953721e-07,
"logits/chosen": -1.2200762033462524,
"logits/rejected": -1.2801724672317505,
"logps/chosen": -228.53701782226562,
"logps/rejected": -209.99066162109375,
"loss": 0.6876,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.0031590620055794716,
"rewards/margins": 0.014617552980780602,
"rewards/rejected": -0.017776615917682648,
"step": 25
},
{
"epoch": 0.12285883047844064,
"grad_norm": 1.1070134845717687,
"learning_rate": 4.987546570871754e-07,
"logits/chosen": -1.7048778533935547,
"logits/rejected": -1.6716248989105225,
"logps/chosen": -237.16851806640625,
"logps/rejected": -234.78970336914062,
"loss": 0.6868,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.006663296837359667,
"rewards/margins": 0.013624398037791252,
"rewards/rejected": -0.02028769627213478,
"step": 26
},
{
"epoch": 0.1275841701122268,
"grad_norm": 1.1180888707264458,
"learning_rate": 4.985558887302488e-07,
"logits/chosen": -1.692581057548523,
"logits/rejected": -1.7662020921707153,
"logps/chosen": -197.77049255371094,
"logps/rejected": -182.52011108398438,
"loss": 0.686,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.003295771311968565,
"rewards/margins": 0.009790323674678802,
"rewards/rejected": -0.01308609452098608,
"step": 27
},
{
"epoch": 0.132309509746013,
"grad_norm": 1.1912060262420856,
"learning_rate": 4.983424556102468e-07,
"logits/chosen": -1.8331196308135986,
"logits/rejected": -1.8659313917160034,
"logps/chosen": -200.51959228515625,
"logps/rejected": -181.59939575195312,
"loss": 0.6841,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.004111767280846834,
"rewards/margins": 0.017347747460007668,
"rewards/rejected": -0.021459516137838364,
"step": 28
},
{
"epoch": 0.13703484937979918,
"grad_norm": 1.1795634739349266,
"learning_rate": 4.981143703196941e-07,
"logits/chosen": -2.1984832286834717,
"logits/rejected": -2.172135591506958,
"logps/chosen": -193.10231018066406,
"logps/rejected": -180.98953247070312,
"loss": 0.6844,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.010889173485338688,
"rewards/margins": 0.014179128222167492,
"rewards/rejected": -0.02506830170750618,
"step": 29
},
{
"epoch": 0.14176018901358536,
"grad_norm": 1.066586203576245,
"learning_rate": 4.978716463155912e-07,
"logits/chosen": -2.06459379196167,
"logits/rejected": -2.0417637825012207,
"logps/chosen": -159.03701782226562,
"logps/rejected": -195.22523498535156,
"loss": 0.6843,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.0062042661011219025,
"rewards/margins": 0.016392884775996208,
"rewards/rejected": -0.02259715273976326,
"step": 30
},
{
"epoch": 0.14648552864737152,
"grad_norm": 1.2110413924457768,
"learning_rate": 4.976142979186209e-07,
"logits/chosen": -1.9228863716125488,
"logits/rejected": -1.9126472473144531,
"logps/chosen": -202.9162139892578,
"logps/rejected": -175.83433532714844,
"loss": 0.6821,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.007678491994738579,
"rewards/margins": 0.02073112316429615,
"rewards/rejected": -0.02840961515903473,
"step": 31
},
{
"epoch": 0.1512108682811577,
"grad_norm": 1.2866090020461936,
"learning_rate": 4.973423403123028e-07,
"logits/chosen": -1.701865792274475,
"logits/rejected": -1.7730942964553833,
"logps/chosen": -235.43563842773438,
"logps/rejected": -229.68411254882812,
"loss": 0.6798,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.0038016163744032383,
"rewards/margins": 0.025650067254900932,
"rewards/rejected": -0.029451683163642883,
"step": 32
},
{
"epoch": 0.1559362079149439,
"grad_norm": 1.1184970866462265,
"learning_rate": 4.970557895420983e-07,
"logits/chosen": -1.7996641397476196,
"logits/rejected": -1.7775640487670898,
"logps/chosen": -172.83163452148438,
"logps/rejected": -201.3394317626953,
"loss": 0.6818,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.014111585915088654,
"rewards/margins": 0.02376263216137886,
"rewards/rejected": -0.03787422180175781,
"step": 33
},
{
"epoch": 0.16066154754873008,
"grad_norm": 1.1028286808678158,
"learning_rate": 4.967546625144633e-07,
"logits/chosen": -1.1831226348876953,
"logits/rejected": -1.1660494804382324,
"logps/chosen": -177.04531860351562,
"logps/rejected": -188.2403564453125,
"loss": 0.6822,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.011519413441419601,
"rewards/margins": 0.02097604051232338,
"rewards/rejected": -0.03249545022845268,
"step": 34
},
{
"epoch": 0.16538688718251623,
"grad_norm": 1.1486945876483394,
"learning_rate": 4.964389769958506e-07,
"logits/chosen": -1.382279634475708,
"logits/rejected": -1.419837474822998,
"logps/chosen": -156.91551208496094,
"logps/rejected": -157.46437072753906,
"loss": 0.6784,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.012027422897517681,
"rewards/margins": 0.030697565525770187,
"rewards/rejected": -0.04272499307990074,
"step": 35
},
{
"epoch": 0.17011222681630242,
"grad_norm": 1.1027256033620638,
"learning_rate": 4.961087516116621e-07,
"logits/chosen": -1.1302804946899414,
"logits/rejected": -1.1647142171859741,
"logps/chosen": -263.8468322753906,
"logps/rejected": -243.76544189453125,
"loss": 0.6793,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.020269813016057014,
"rewards/margins": 0.02116047963500023,
"rewards/rejected": -0.041430290788412094,
"step": 36
},
{
"epoch": 0.1748375664500886,
"grad_norm": 1.0823022111801957,
"learning_rate": 4.957640058451501e-07,
"logits/chosen": -1.5351812839508057,
"logits/rejected": -1.5872442722320557,
"logps/chosen": -204.82644653320312,
"logps/rejected": -176.9189453125,
"loss": 0.6799,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.02254084125161171,
"rewards/margins": 0.017112018540501595,
"rewards/rejected": -0.039652857929468155,
"step": 37
},
{
"epoch": 0.1795629060838748,
"grad_norm": 1.0721194904364362,
"learning_rate": 4.954047600362669e-07,
"logits/chosen": -1.8736214637756348,
"logits/rejected": -1.819676160812378,
"logps/chosen": -178.29270935058594,
"logps/rejected": -191.61605834960938,
"loss": 0.6803,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.02358250319957733,
"rewards/margins": 0.031186437234282494,
"rewards/rejected": -0.054768942296504974,
"step": 38
},
{
"epoch": 0.18428824571766095,
"grad_norm": 1.0629161817956538,
"learning_rate": 4.950310353804659e-07,
"logits/chosen": -1.7329224348068237,
"logits/rejected": -1.780181646347046,
"logps/chosen": -188.88072204589844,
"logps/rejected": -177.74427795410156,
"loss": 0.6784,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.024656984955072403,
"rewards/margins": 0.03351202234625816,
"rewards/rejected": -0.058169007301330566,
"step": 39
},
{
"epoch": 0.18901358535144713,
"grad_norm": 1.0963449505388767,
"learning_rate": 4.946428539274497e-07,
"logits/chosen": -1.920142650604248,
"logits/rejected": -1.9512125253677368,
"logps/chosen": -237.4847412109375,
"logps/rejected": -210.24838256835938,
"loss": 0.6759,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.030252018943428993,
"rewards/margins": 0.040137603878974915,
"rewards/rejected": -0.07038962841033936,
"step": 40
},
{
"epoch": 0.19373892498523332,
"grad_norm": 1.039447855381113,
"learning_rate": 4.942402385798706e-07,
"logits/chosen": -1.246740698814392,
"logits/rejected": -1.281036138534546,
"logps/chosen": -240.5216064453125,
"logps/rejected": -188.30795288085938,
"loss": 0.6762,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.032288651913404465,
"rewards/margins": 0.03159747272729874,
"rewards/rejected": -0.0638861209154129,
"step": 41
},
{
"epoch": 0.1984642646190195,
"grad_norm": 1.0133081629791365,
"learning_rate": 4.938232130919785e-07,
"logits/chosen": -2.049900531768799,
"logits/rejected": -2.0438833236694336,
"logps/chosen": -241.96177673339844,
"logps/rejected": -223.10511779785156,
"loss": 0.6775,
"rewards/accuracies": 0.671875,
"rewards/chosen": -0.05470336228609085,
"rewards/margins": 0.040740497410297394,
"rewards/rejected": -0.09544385224580765,
"step": 42
},
{
"epoch": 0.20318960425280566,
"grad_norm": 1.046962172171356,
"learning_rate": 4.933918020682195e-07,
"logits/chosen": -2.0764129161834717,
"logits/rejected": -1.9940263032913208,
"logps/chosen": -204.7505340576172,
"logps/rejected": -206.99880981445312,
"loss": 0.6727,
"rewards/accuracies": 0.671875,
"rewards/chosen": -0.03313834220170975,
"rewards/margins": 0.06050185114145279,
"rewards/rejected": -0.09364018589258194,
"step": 43
},
{
"epoch": 0.20318960425280566,
"eval_logits/chosen": -2.2005555629730225,
"eval_logits/rejected": -2.2166874408721924,
"eval_logps/chosen": -216.52699279785156,
"eval_logps/rejected": -209.94314575195312,
"eval_loss": 0.6714360117912292,
"eval_rewards/accuracies": 0.5871211886405945,
"eval_rewards/chosen": -0.05296258255839348,
"eval_rewards/margins": 0.046969976276159286,
"eval_rewards/rejected": -0.09993256628513336,
"eval_runtime": 225.63,
"eval_samples_per_second": 16.204,
"eval_steps_per_second": 0.293,
"step": 43
},
{
"epoch": 0.20791494388659185,
"grad_norm": 1.1285157956991194,
"learning_rate": 4.929460309617843e-07,
"logits/chosen": -2.0923304557800293,
"logits/rejected": -2.151911973953247,
"logps/chosen": -252.43092346191406,
"logps/rejected": -221.0844268798828,
"loss": 0.6707,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.04450148344039917,
"rewards/margins": 0.058394819498062134,
"rewards/rejected": -0.10289628803730011,
"step": 44
},
{
"epoch": 0.21264028352037803,
"grad_norm": 1.0651629243936434,
"learning_rate": 4.924859260731066e-07,
"logits/chosen": -2.0476608276367188,
"logits/rejected": -2.174062490463257,
"logps/chosen": -219.19517517089844,
"logps/rejected": -189.25193786621094,
"loss": 0.6718,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.06256880611181259,
"rewards/margins": 0.04902214929461479,
"rewards/rejected": -0.11159095913171768,
"step": 45
},
{
"epoch": 0.21736562315416422,
"grad_norm": 1.0516408835233841,
"learning_rate": 4.920115145483112e-07,
"logits/chosen": -1.602857232093811,
"logits/rejected": -1.6103214025497437,
"logps/chosen": -252.79664611816406,
"logps/rejected": -229.30770874023438,
"loss": 0.6711,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.08149686455726624,
"rewards/margins": 0.037921737879514694,
"rewards/rejected": -0.11941860616207123,
"step": 46
},
{
"epoch": 0.22209096278795037,
"grad_norm": 1.1789732456475222,
"learning_rate": 4.915228243776124e-07,
"logits/chosen": -1.1500588655471802,
"logits/rejected": -1.1534898281097412,
"logps/chosen": -253.38931274414062,
"logps/rejected": -220.110595703125,
"loss": 0.6634,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.06619597226381302,
"rewards/margins": 0.07172124832868576,
"rewards/rejected": -0.13791722059249878,
"step": 47
},
{
"epoch": 0.22681630242173656,
"grad_norm": 1.056306792669834,
"learning_rate": 4.91019884393663e-07,
"logits/chosen": -0.8076485991477966,
"logits/rejected": -0.8253241181373596,
"logps/chosen": -211.2886962890625,
"logps/rejected": -174.4979705810547,
"loss": 0.6668,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.05709861218929291,
"rewards/margins": 0.05306413769721985,
"rewards/rejected": -0.11016274988651276,
"step": 48
},
{
"epoch": 0.23154164205552275,
"grad_norm": 1.1483517794441522,
"learning_rate": 4.905027242698521e-07,
"logits/chosen": -1.5992224216461182,
"logits/rejected": -1.7107007503509521,
"logps/chosen": -258.3623962402344,
"logps/rejected": -208.75918579101562,
"loss": 0.6608,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.07419726252555847,
"rewards/margins": 0.07502313703298569,
"rewards/rejected": -0.14922040700912476,
"step": 49
},
{
"epoch": 0.23626698168930893,
"grad_norm": 1.0512679200538415,
"learning_rate": 4.89971374518556e-07,
"logits/chosen": -2.0465784072875977,
"logits/rejected": -2.033297300338745,
"logps/chosen": -185.704345703125,
"logps/rejected": -203.28013610839844,
"loss": 0.6639,
"rewards/accuracies": 0.578125,
"rewards/chosen": -0.08691324293613434,
"rewards/margins": 0.06635289639234543,
"rewards/rejected": -0.15326614677906036,
"step": 50
},
{
"epoch": 0.2409923213230951,
"grad_norm": 1.112120640811307,
"learning_rate": 4.894258664893363e-07,
"logits/chosen": -1.7012823820114136,
"logits/rejected": -1.7330955266952515,
"logps/chosen": -208.68597412109375,
"logps/rejected": -218.7354278564453,
"loss": 0.6594,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.07796823233366013,
"rewards/margins": 0.038594380021095276,
"rewards/rejected": -0.1165626123547554,
"step": 51
},
{
"epoch": 0.24571766095688127,
"grad_norm": 1.1579361743422294,
"learning_rate": 4.888662323670913e-07,
"logits/chosen": -1.6541762351989746,
"logits/rejected": -1.7235084772109985,
"logps/chosen": -269.59210205078125,
"logps/rejected": -243.711669921875,
"loss": 0.6629,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.06561748683452606,
"rewards/margins": 0.06448770314455032,
"rewards/rejected": -0.1301051825284958,
"step": 52
},
{
"epoch": 0.25044300059066743,
"grad_norm": 1.1521413307343888,
"learning_rate": 4.882925051701568e-07,
"logits/chosen": -1.853175163269043,
"logits/rejected": -1.8978935480117798,
"logps/chosen": -225.21856689453125,
"logps/rejected": -228.90325927734375,
"loss": 0.6578,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.09534727036952972,
"rewards/margins": 0.077830970287323,
"rewards/rejected": -0.17317824065685272,
"step": 53
},
{
"epoch": 0.2551683402244536,
"grad_norm": 1.0619319741777762,
"learning_rate": 4.877047187483582e-07,
"logits/chosen": -1.6998298168182373,
"logits/rejected": -1.7790082693099976,
"logps/chosen": -212.0916748046875,
"logps/rejected": -191.58206176757812,
"loss": 0.6631,
"rewards/accuracies": 0.546875,
"rewards/chosen": -0.10848715156316757,
"rewards/margins": 0.048885174095630646,
"rewards/rejected": -0.1573723405599594,
"step": 54
},
{
"epoch": 0.2598936798582398,
"grad_norm": 1.1160678416368461,
"learning_rate": 4.871029077810132e-07,
"logits/chosen": -1.675370216369629,
"logits/rejected": -1.7553473711013794,
"logps/chosen": -225.4173126220703,
"logps/rejected": -198.70562744140625,
"loss": 0.6558,
"rewards/accuracies": 0.53125,
"rewards/chosen": -0.10019448399543762,
"rewards/margins": 0.08370313048362732,
"rewards/rejected": -0.18389761447906494,
"step": 55
},
{
"epoch": 0.264619019492026,
"grad_norm": 1.180700159151371,
"learning_rate": 4.864871077748857e-07,
"logits/chosen": -2.015566110610962,
"logits/rejected": -2.080444812774658,
"logps/chosen": -229.28094482421875,
"logps/rejected": -215.07102966308594,
"loss": 0.6587,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.1043662577867508,
"rewards/margins": 0.07935845851898193,
"rewards/rejected": -0.18372471630573273,
"step": 56
},
{
"epoch": 0.26934435912581217,
"grad_norm": 1.2229942654627657,
"learning_rate": 4.858573550620908e-07,
"logits/chosen": -2.024144411087036,
"logits/rejected": -2.0165395736694336,
"logps/chosen": -266.3519287109375,
"logps/rejected": -233.6390380859375,
"loss": 0.6454,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.12464563548564911,
"rewards/margins": 0.12967334687709808,
"rewards/rejected": -0.2543190121650696,
"step": 57
},
{
"epoch": 0.27406969875959836,
"grad_norm": 1.0980503837148203,
"learning_rate": 4.852136867979515e-07,
"logits/chosen": -2.3049449920654297,
"logits/rejected": -2.286456346511841,
"logps/chosen": -187.29776000976562,
"logps/rejected": -195.98187255859375,
"loss": 0.6559,
"rewards/accuracies": 0.515625,
"rewards/chosen": -0.13345083594322205,
"rewards/margins": 0.05029616504907608,
"rewards/rejected": -0.18374700844287872,
"step": 58
},
{
"epoch": 0.27879503839338454,
"grad_norm": 1.222277907011155,
"learning_rate": 4.845561409588065e-07,
"logits/chosen": -2.3418726921081543,
"logits/rejected": -2.2416603565216064,
"logps/chosen": -184.75540161132812,
"logps/rejected": -186.7329864501953,
"loss": 0.6424,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.10166566073894501,
"rewards/margins": 0.12095416337251663,
"rewards/rejected": -0.22261981666088104,
"step": 59
},
{
"epoch": 0.2835203780271707,
"grad_norm": 1.0214927771783529,
"learning_rate": 4.838847563397693e-07,
"logits/chosen": -1.5123400688171387,
"logits/rejected": -1.6280531883239746,
"logps/chosen": -237.2841796875,
"logps/rejected": -212.3114013671875,
"loss": 0.6583,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.1481347680091858,
"rewards/margins": 0.09539347887039185,
"rewards/rejected": -0.24352826178073883,
"step": 60
},
{
"epoch": 0.28824571766095686,
"grad_norm": 1.1567243525055237,
"learning_rate": 4.831995725524398e-07,
"logits/chosen": -2.612375497817993,
"logits/rejected": -2.534623861312866,
"logps/chosen": -163.80526733398438,
"logps/rejected": -201.24952697753906,
"loss": 0.6481,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.14992724359035492,
"rewards/margins": 0.10543593764305115,
"rewards/rejected": -0.2553631663322449,
"step": 61
},
{
"epoch": 0.29297105729474304,
"grad_norm": 1.2299920626415959,
"learning_rate": 4.825006300225665e-07,
"logits/chosen": -2.0585803985595703,
"logits/rejected": -2.1235456466674805,
"logps/chosen": -218.53924560546875,
"logps/rejected": -219.55938720703125,
"loss": 0.6439,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.15479950606822968,
"rewards/margins": 0.11182530224323273,
"rewards/rejected": -0.2666248381137848,
"step": 62
},
{
"epoch": 0.2976963969285292,
"grad_norm": 1.203647663122135,
"learning_rate": 4.817879699876622e-07,
"logits/chosen": -1.9584053754806519,
"logits/rejected": -2.047368288040161,
"logps/chosen": -180.12673950195312,
"logps/rejected": -169.58712768554688,
"loss": 0.6365,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.15420033037662506,
"rewards/margins": 0.07548267394304276,
"rewards/rejected": -0.2296830117702484,
"step": 63
},
{
"epoch": 0.3024217365623154,
"grad_norm": 1.1158018473166338,
"learning_rate": 4.810616344945705e-07,
"logits/chosen": -1.889503002166748,
"logits/rejected": -1.8674815893173218,
"logps/chosen": -205.70799255371094,
"logps/rejected": -200.00503540039062,
"loss": 0.6564,
"rewards/accuracies": 0.46875,
"rewards/chosen": -0.20557425916194916,
"rewards/margins": 0.07413503527641296,
"rewards/rejected": -0.2797092795372009,
"step": 64
},
{
"epoch": 0.3071470761961016,
"grad_norm": 1.1145309518931334,
"learning_rate": 4.803216663969849e-07,
"logits/chosen": -2.5508382320404053,
"logits/rejected": -2.58968186378479,
"logps/chosen": -222.9154052734375,
"logps/rejected": -205.91091918945312,
"loss": 0.646,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.16130368411540985,
"rewards/margins": 0.11249940097332001,
"rewards/rejected": -0.27380311489105225,
"step": 65
},
{
"epoch": 0.3118724158298878,
"grad_norm": 1.1907287315768882,
"learning_rate": 4.795681093529209e-07,
"logits/chosen": -1.9771151542663574,
"logits/rejected": -1.9131364822387695,
"logps/chosen": -181.78024291992188,
"logps/rejected": -203.05455017089844,
"loss": 0.6367,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.17563070356845856,
"rewards/margins": 0.1282222419977188,
"rewards/rejected": -0.30385297536849976,
"step": 66
},
{
"epoch": 0.31659775546367397,
"grad_norm": 1.0642807389889697,
"learning_rate": 4.7880100782214e-07,
"logits/chosen": -2.139569044113159,
"logits/rejected": -2.139991283416748,
"logps/chosen": -200.4115447998047,
"logps/rejected": -209.0381317138672,
"loss": 0.6515,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.25353577733039856,
"rewards/margins": 0.025102369487285614,
"rewards/rejected": -0.2786381244659424,
"step": 67
},
{
"epoch": 0.32132309509746015,
"grad_norm": 1.2361659779944367,
"learning_rate": 4.780204070635266e-07,
"logits/chosen": -2.3622794151306152,
"logits/rejected": -2.3103740215301514,
"logps/chosen": -241.8636474609375,
"logps/rejected": -253.21530151367188,
"loss": 0.6309,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.2274862378835678,
"rewards/margins": 0.18495142459869385,
"rewards/rejected": -0.41243770718574524,
"step": 68
},
{
"epoch": 0.3260484347312463,
"grad_norm": 1.2216304493799137,
"learning_rate": 4.772263531324172e-07,
"logits/chosen": -2.275869369506836,
"logits/rejected": -2.296611785888672,
"logps/chosen": -255.72305297851562,
"logps/rejected": -237.79794311523438,
"loss": 0.631,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.24048469960689545,
"rewards/margins": 0.11058272421360016,
"rewards/rejected": -0.351067453622818,
"step": 69
},
{
"epoch": 0.33077377436503247,
"grad_norm": 1.2564703689206955,
"learning_rate": 4.764188928778843e-07,
"logits/chosen": -1.9974974393844604,
"logits/rejected": -2.011615514755249,
"logps/chosen": -195.90382385253906,
"logps/rejected": -214.2757110595703,
"loss": 0.6336,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.2238045632839203,
"rewards/margins": 0.1463470458984375,
"rewards/rejected": -0.3701516091823578,
"step": 70
},
{
"epoch": 0.33549911399881865,
"grad_norm": 1.2749948068499388,
"learning_rate": 4.755980739399711e-07,
"logits/chosen": -2.3057668209075928,
"logits/rejected": -2.2394864559173584,
"logps/chosen": -193.60552978515625,
"logps/rejected": -239.27389526367188,
"loss": 0.6525,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.2364463210105896,
"rewards/margins": 0.1507556438446045,
"rewards/rejected": -0.3872019648551941,
"step": 71
},
{
"epoch": 0.34022445363260484,
"grad_norm": 1.3097724846280558,
"learning_rate": 4.747639447468816e-07,
"logits/chosen": -2.2665905952453613,
"logits/rejected": -2.279989719390869,
"logps/chosen": -285.88140869140625,
"logps/rejected": -304.027587890625,
"loss": 0.6232,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.23830385506153107,
"rewards/margins": 0.22441713511943817,
"rewards/rejected": -0.4627210199832916,
"step": 72
},
{
"epoch": 0.344949793266391,
"grad_norm": 1.267134897055487,
"learning_rate": 4.739165545121228e-07,
"logits/chosen": -1.9879568815231323,
"logits/rejected": -1.9682663679122925,
"logps/chosen": -246.3922882080078,
"logps/rejected": -247.4912109375,
"loss": 0.6229,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.3345012366771698,
"rewards/margins": 0.17793008685112,
"rewards/rejected": -0.5124313235282898,
"step": 73
},
{
"epoch": 0.3496751329001772,
"grad_norm": 1.2766840060333855,
"learning_rate": 4.730559532316014e-07,
"logits/chosen": -2.3778185844421387,
"logits/rejected": -2.4101202487945557,
"logps/chosen": -221.90524291992188,
"logps/rejected": -241.57962036132812,
"loss": 0.6395,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.32600241899490356,
"rewards/margins": 0.14937394857406616,
"rewards/rejected": -0.4753763973712921,
"step": 74
},
{
"epoch": 0.3544004725339634,
"grad_norm": 1.3566530419485627,
"learning_rate": 4.721821916806741e-07,
"logits/chosen": -2.4602699279785156,
"logits/rejected": -2.4688634872436523,
"logps/chosen": -268.510498046875,
"logps/rejected": -286.26165771484375,
"loss": 0.6237,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.31091028451919556,
"rewards/margins": 0.29035893082618713,
"rewards/rejected": -0.6012692451477051,
"step": 75
},
{
"epoch": 0.3591258121677496,
"grad_norm": 1.4205593895904036,
"learning_rate": 4.7129532141115145e-07,
"logits/chosen": -2.5466036796569824,
"logits/rejected": -2.5945773124694824,
"logps/chosen": -318.98370361328125,
"logps/rejected": -309.7763366699219,
"loss": 0.6112,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.3561764359474182,
"rewards/margins": 0.18557003140449524,
"rewards/rejected": -0.5417464375495911,
"step": 76
},
{
"epoch": 0.3638511518015357,
"grad_norm": 1.528828829984918,
"learning_rate": 4.7039539474825683e-07,
"logits/chosen": -2.2787909507751465,
"logits/rejected": -2.3680973052978516,
"logps/chosen": -305.41925048828125,
"logps/rejected": -288.8333435058594,
"loss": 0.6006,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.39139410853385925,
"rewards/margins": 0.2935311794281006,
"rewards/rejected": -0.6849253177642822,
"step": 77
},
{
"epoch": 0.3685764914353219,
"grad_norm": 1.6985476984319592,
"learning_rate": 4.6948246478753903e-07,
"logits/chosen": -2.261338472366333,
"logits/rejected": -2.323387622833252,
"logps/chosen": -247.09033203125,
"logps/rejected": -257.6181640625,
"loss": 0.6314,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.5037412047386169,
"rewards/margins": 0.1979576051235199,
"rewards/rejected": -0.7016987800598145,
"step": 78
},
{
"epoch": 0.3733018310691081,
"grad_norm": 1.729313425365047,
"learning_rate": 4.6855658539173946e-07,
"logits/chosen": -2.698389768600464,
"logits/rejected": -2.646217107772827,
"logps/chosen": -314.0598449707031,
"logps/rejected": -283.12530517578125,
"loss": 0.6206,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.5247927904129028,
"rewards/margins": 0.22078382968902588,
"rewards/rejected": -0.7455766201019287,
"step": 79
},
{
"epoch": 0.37802717070289427,
"grad_norm": 1.66621636166947,
"learning_rate": 4.6761781118761446e-07,
"logits/chosen": -2.343153238296509,
"logits/rejected": -2.3706839084625244,
"logps/chosen": -296.689453125,
"logps/rejected": -309.9665222167969,
"loss": 0.6257,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.44052672386169434,
"rewards/margins": 0.23874229192733765,
"rewards/rejected": -0.6792689561843872,
"step": 80
},
{
"epoch": 0.38275251033668045,
"grad_norm": 1.5082652270101988,
"learning_rate": 4.666661975627123e-07,
"logits/chosen": -2.4629459381103516,
"logits/rejected": -2.486161470413208,
"logps/chosen": -233.51983642578125,
"logps/rejected": -262.89447021484375,
"loss": 0.6154,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.4172506630420685,
"rewards/margins": 0.3244992196559906,
"rewards/rejected": -0.7417498826980591,
"step": 81
},
{
"epoch": 0.38747784997046664,
"grad_norm": 1.5314338973522612,
"learning_rate": 4.657018006621053e-07,
"logits/chosen": -2.1910758018493652,
"logits/rejected": -2.273641586303711,
"logps/chosen": -225.42889404296875,
"logps/rejected": -239.92913818359375,
"loss": 0.6058,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.4517180323600769,
"rewards/margins": 0.22106432914733887,
"rewards/rejected": -0.6727824211120605,
"step": 82
},
{
"epoch": 0.3922031896042528,
"grad_norm": 1.6363344671283797,
"learning_rate": 4.6472467738507724e-07,
"logits/chosen": -2.4740734100341797,
"logits/rejected": -2.593059539794922,
"logps/chosen": -325.74298095703125,
"logps/rejected": -297.570068359375,
"loss": 0.6088,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.6095322370529175,
"rewards/margins": 0.1409136950969696,
"rewards/rejected": -0.7504459619522095,
"step": 83
},
{
"epoch": 0.396928529238039,
"grad_norm": 1.7032682560793109,
"learning_rate": 4.6373488538176656e-07,
"logits/chosen": -2.627995014190674,
"logits/rejected": -2.585923433303833,
"logps/chosen": -278.2349853515625,
"logps/rejected": -319.6769104003906,
"loss": 0.6177,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.6383107900619507,
"rewards/margins": 0.17859135568141937,
"rewards/rejected": -0.8169021010398865,
"step": 84
},
{
"epoch": 0.40165386887182514,
"grad_norm": 1.6927306950035537,
"learning_rate": 4.627324830497645e-07,
"logits/chosen": -2.3960776329040527,
"logits/rejected": -2.4317002296447754,
"logps/chosen": -217.64886474609375,
"logps/rejected": -255.28555297851562,
"loss": 0.6034,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.581438422203064,
"rewards/margins": 0.2544565200805664,
"rewards/rejected": -0.8358950018882751,
"step": 85
},
{
"epoch": 0.4063792085056113,
"grad_norm": 1.517574514685826,
"learning_rate": 4.617175295306701e-07,
"logits/chosen": -2.342132329940796,
"logits/rejected": -2.317411422729492,
"logps/chosen": -249.52786254882812,
"logps/rejected": -272.2576904296875,
"loss": 0.6056,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.4862441420555115,
"rewards/margins": 0.2543086111545563,
"rewards/rejected": -0.7405527830123901,
"step": 86
},
{
"epoch": 0.4063792085056113,
"eval_logits/chosen": -3.0177347660064697,
"eval_logits/rejected": -3.027693748474121,
"eval_logps/chosen": -269.99395751953125,
"eval_logps/rejected": -288.7347106933594,
"eval_loss": 0.6040579080581665,
"eval_rewards/accuracies": 0.6022727489471436,
"eval_rewards/chosen": -0.5876324772834778,
"eval_rewards/margins": 0.3002159297466278,
"eval_rewards/rejected": -0.8878483772277832,
"eval_runtime": 225.4053,
"eval_samples_per_second": 16.22,
"eval_steps_per_second": 0.293,
"step": 86
},
{
"epoch": 0.4111045481393975,
"grad_norm": 1.828247945135766,
"learning_rate": 4.6069008470660057e-07,
"logits/chosen": -2.776036500930786,
"logits/rejected": -2.8447940349578857,
"logps/chosen": -286.0122985839844,
"logps/rejected": -318.0850830078125,
"loss": 0.6202,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.6256659626960754,
"rewards/margins": 0.2720867395401001,
"rewards/rejected": -0.8977527022361755,
"step": 87
},
{
"epoch": 0.4158298877731837,
"grad_norm": 1.7339677719729851,
"learning_rate": 4.596502091966587e-07,
"logits/chosen": -2.6904196739196777,
"logits/rejected": -2.8044652938842773,
"logps/chosen": -318.77557373046875,
"logps/rejected": -326.3057556152344,
"loss": 0.6133,
"rewards/accuracies": 0.546875,
"rewards/chosen": -0.6397165060043335,
"rewards/margins": 0.23243850469589233,
"rewards/rejected": -0.8721550107002258,
"step": 88
},
{
"epoch": 0.4205552274069699,
"grad_norm": 1.5787512385428173,
"learning_rate": 4.5859796435335575e-07,
"logits/chosen": -2.5359115600585938,
"logits/rejected": -2.5303304195404053,
"logps/chosen": -256.5873718261719,
"logps/rejected": -295.2117004394531,
"loss": 0.6062,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.5649542212486267,
"rewards/margins": 0.2900598347187042,
"rewards/rejected": -0.8550140857696533,
"step": 89
},
{
"epoch": 0.42528056704075606,
"grad_norm": 1.6858517491129223,
"learning_rate": 4.5753341225899195e-07,
"logits/chosen": -2.4564006328582764,
"logits/rejected": -2.470282793045044,
"logps/chosen": -327.185791015625,
"logps/rejected": -319.4944152832031,
"loss": 0.6017,
"rewards/accuracies": 0.546875,
"rewards/chosen": -0.6457672715187073,
"rewards/margins": 0.2363281399011612,
"rewards/rejected": -0.882095456123352,
"step": 90
},
{
"epoch": 0.43000590667454225,
"grad_norm": 1.6776313748940133,
"learning_rate": 4.564566157219938e-07,
"logits/chosen": -2.467501163482666,
"logits/rejected": -2.570307970046997,
"logps/chosen": -318.20867919921875,
"logps/rejected": -302.66949462890625,
"loss": 0.6015,
"rewards/accuracies": 0.578125,
"rewards/chosen": -0.5671995878219604,
"rewards/margins": 0.2143457680940628,
"rewards/rejected": -0.7815454006195068,
"step": 91
},
{
"epoch": 0.43473124630832843,
"grad_norm": 2.0614895261976747,
"learning_rate": 4.5536763827320803e-07,
"logits/chosen": -2.3631057739257812,
"logits/rejected": -2.54693341255188,
"logps/chosen": -273.6968078613281,
"logps/rejected": -246.98641967773438,
"loss": 0.5856,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.4555537700653076,
"rewards/margins": 0.22295869886875153,
"rewards/rejected": -0.6785125136375427,
"step": 92
},
{
"epoch": 0.43945658594211456,
"grad_norm": 1.8016509044142324,
"learning_rate": 4.5426654416215367e-07,
"logits/chosen": -2.71864652633667,
"logits/rejected": -2.7026920318603516,
"logps/chosen": -288.259033203125,
"logps/rejected": -330.29132080078125,
"loss": 0.6038,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.5275314450263977,
"rewards/margins": 0.5003238916397095,
"rewards/rejected": -1.027855396270752,
"step": 93
},
{
"epoch": 0.44418192557590075,
"grad_norm": 2.243457029027687,
"learning_rate": 4.5315339835323095e-07,
"logits/chosen": -2.884897232055664,
"logits/rejected": -2.8569068908691406,
"logps/chosen": -269.7425537109375,
"logps/rejected": -297.4085693359375,
"loss": 0.5748,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.5586040019989014,
"rewards/margins": 0.3271249234676361,
"rewards/rejected": -0.8857288956642151,
"step": 94
},
{
"epoch": 0.44890726520968693,
"grad_norm": 2.296742463720749,
"learning_rate": 4.520282665218889e-07,
"logits/chosen": -2.790522336959839,
"logits/rejected": -2.84653377532959,
"logps/chosen": -265.41339111328125,
"logps/rejected": -315.9186096191406,
"loss": 0.5625,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.5728757977485657,
"rewards/margins": 0.515011191368103,
"rewards/rejected": -1.0878870487213135,
"step": 95
},
{
"epoch": 0.4536326048434731,
"grad_norm": 1.922804376035695,
"learning_rate": 4.5089121505074987e-07,
"logits/chosen": -2.6556386947631836,
"logits/rejected": -2.806910276412964,
"logps/chosen": -244.31153869628906,
"logps/rejected": -247.36322021484375,
"loss": 0.5787,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.5964667797088623,
"rewards/margins": 0.29180973768234253,
"rewards/rejected": -0.8882765173912048,
"step": 96
},
{
"epoch": 0.4583579444772593,
"grad_norm": 1.6125061422080407,
"learning_rate": 4.4974231102569355e-07,
"logits/chosen": -2.7232208251953125,
"logits/rejected": -2.8808493614196777,
"logps/chosen": -272.30206298828125,
"logps/rejected": -273.55487060546875,
"loss": 0.5764,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.5672851800918579,
"rewards/margins": 0.39647072553634644,
"rewards/rejected": -0.9637559056282043,
"step": 97
},
{
"epoch": 0.4630832841110455,
"grad_norm": 2.0273197043813598,
"learning_rate": 4.4858162223189853e-07,
"logits/chosen": -2.691676616668701,
"logits/rejected": -2.7318079471588135,
"logps/chosen": -331.65582275390625,
"logps/rejected": -321.1370544433594,
"loss": 0.6022,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.7305989861488342,
"rewards/margins": 0.20282992720603943,
"rewards/rejected": -0.933428943157196,
"step": 98
},
{
"epoch": 0.4678086237448317,
"grad_norm": 1.772177669462408,
"learning_rate": 4.474092171498434e-07,
"logits/chosen": -2.5423169136047363,
"logits/rejected": -2.5960371494293213,
"logps/chosen": -260.9735107421875,
"logps/rejected": -277.1812744140625,
"loss": 0.5888,
"rewards/accuracies": 0.546875,
"rewards/chosen": -0.6579495668411255,
"rewards/margins": 0.32558560371398926,
"rewards/rejected": -0.9835351705551147,
"step": 99
},
{
"epoch": 0.47253396337861786,
"grad_norm": 2.031475270950697,
"learning_rate": 4.462251649512656e-07,
"logits/chosen": -2.805039167404175,
"logits/rejected": -2.7704780101776123,
"logps/chosen": -235.95053100585938,
"logps/rejected": -292.6195373535156,
"loss": 0.5634,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.612291157245636,
"rewards/margins": 0.49476855993270874,
"rewards/rejected": -1.1070597171783447,
"step": 100
},
{
"epoch": 0.477259303012404,
"grad_norm": 2.8786657503544033,
"learning_rate": 4.4502953549508135e-07,
"logits/chosen": -2.829331398010254,
"logits/rejected": -2.8344359397888184,
"logps/chosen": -309.11285400390625,
"logps/rejected": -331.1981201171875,
"loss": 0.596,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.7622178792953491,
"rewards/margins": 0.48595699667930603,
"rewards/rejected": -1.2481749057769775,
"step": 101
},
{
"epoch": 0.4819846426461902,
"grad_norm": 2.6747243514730163,
"learning_rate": 4.438223993232634e-07,
"logits/chosen": -2.71714186668396,
"logits/rejected": -2.7164933681488037,
"logps/chosen": -319.4408264160156,
"logps/rejected": -335.07025146484375,
"loss": 0.5873,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.7349227666854858,
"rewards/margins": 0.5957677960395813,
"rewards/rejected": -1.3306907415390015,
"step": 102
},
{
"epoch": 0.48670998227997636,
"grad_norm": 2.184532115681571,
"learning_rate": 4.426038276566787e-07,
"logits/chosen": -2.595947027206421,
"logits/rejected": -2.6409239768981934,
"logps/chosen": -302.1788330078125,
"logps/rejected": -300.60736083984375,
"loss": 0.5929,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.6437785625457764,
"rewards/margins": 0.4648054838180542,
"rewards/rejected": -1.1085840463638306,
"step": 103
},
{
"epoch": 0.49143532191376255,
"grad_norm": 2.351448463153027,
"learning_rate": 4.413738923908874e-07,
"logits/chosen": -2.820120334625244,
"logits/rejected": -2.881047248840332,
"logps/chosen": -321.73577880859375,
"logps/rejected": -336.362548828125,
"loss": 0.6113,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.6863315105438232,
"rewards/margins": 0.3078695237636566,
"rewards/rejected": -0.9942010641098022,
"step": 104
},
{
"epoch": 0.49616066154754873,
"grad_norm": 2.0712718540984976,
"learning_rate": 4.4013266609190016e-07,
"logits/chosen": -2.726858139038086,
"logits/rejected": -2.8224053382873535,
"logps/chosen": -318.89129638671875,
"logps/rejected": -321.7113037109375,
"loss": 0.5567,
"rewards/accuracies": 0.578125,
"rewards/chosen": -0.7669743299484253,
"rewards/margins": 0.5382488369941711,
"rewards/rejected": -1.3052233457565308,
"step": 105
},
{
"epoch": 0.5008860011813349,
"grad_norm": 2.7161021836994412,
"learning_rate": 4.3888022199189684e-07,
"logits/chosen": -2.5977838039398193,
"logits/rejected": -2.568969488143921,
"logps/chosen": -270.40380859375,
"logps/rejected": -321.2980041503906,
"loss": 0.6029,
"rewards/accuracies": 0.671875,
"rewards/chosen": -0.6543689966201782,
"rewards/margins": 0.5025177001953125,
"rewards/rejected": -1.1568866968154907,
"step": 106
},
{
"epoch": 0.505611340815121,
"grad_norm": 2.133467086083558,
"learning_rate": 4.3761663398490634e-07,
"logits/chosen": -2.5719194412231445,
"logits/rejected": -2.569828510284424,
"logps/chosen": -278.3515930175781,
"logps/rejected": -290.2523193359375,
"loss": 0.5601,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.6445180773735046,
"rewards/margins": 0.44993308186531067,
"rewards/rejected": -1.0944510698318481,
"step": 107
},
{
"epoch": 0.5103366804489072,
"grad_norm": 2.581357357210363,
"learning_rate": 4.363419766224464e-07,
"logits/chosen": -2.5332443714141846,
"logits/rejected": -2.5566651821136475,
"logps/chosen": -255.7954559326172,
"logps/rejected": -285.38604736328125,
"loss": 0.5701,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.7009232044219971,
"rewards/margins": 0.5014970302581787,
"rewards/rejected": -1.2024202346801758,
"step": 108
},
{
"epoch": 0.5150620200826934,
"grad_norm": 2.0545799713212913,
"learning_rate": 4.3505632510912515e-07,
"logits/chosen": -2.5492563247680664,
"logits/rejected": -2.6753411293029785,
"logps/chosen": -252.5239715576172,
"logps/rejected": -250.57691955566406,
"loss": 0.5745,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.580327033996582,
"rewards/margins": 0.28667935729026794,
"rewards/rejected": -0.8670063614845276,
"step": 109
},
{
"epoch": 0.5197873597164796,
"grad_norm": 2.2034797994763458,
"learning_rate": 4.3375975529820414e-07,
"logits/chosen": -2.492084503173828,
"logits/rejected": -2.432577610015869,
"logps/chosen": -310.48046875,
"logps/rejected": -362.7835388183594,
"loss": 0.5392,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.6999268531799316,
"rewards/margins": 0.7504494190216064,
"rewards/rejected": -1.450376272201538,
"step": 110
},
{
"epoch": 0.5245126993502658,
"grad_norm": 1.9829084137694295,
"learning_rate": 4.3245234368712304e-07,
"logits/chosen": -2.7556090354919434,
"logits/rejected": -2.754206418991089,
"logps/chosen": -270.7232666015625,
"logps/rejected": -324.1688232421875,
"loss": 0.5701,
"rewards/accuracies": 0.671875,
"rewards/chosen": -0.6123411655426025,
"rewards/margins": 0.4949289858341217,
"rewards/rejected": -1.1072702407836914,
"step": 111
},
{
"epoch": 0.529238038984052,
"grad_norm": 2.0360711261149596,
"learning_rate": 4.3113416741298616e-07,
"logits/chosen": -2.659914016723633,
"logits/rejected": -2.673081874847412,
"logps/chosen": -292.6357727050781,
"logps/rejected": -286.8974914550781,
"loss": 0.5785,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.645796537399292,
"rewards/margins": 0.4479163587093353,
"rewards/rejected": -1.0937130451202393,
"step": 112
},
{
"epoch": 0.5339633786178382,
"grad_norm": 2.3215552370547945,
"learning_rate": 4.298053042480114e-07,
"logits/chosen": -2.6102089881896973,
"logits/rejected": -2.666215419769287,
"logps/chosen": -284.019775390625,
"logps/rejected": -313.7834777832031,
"loss": 0.5434,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.6691496968269348,
"rewards/margins": 0.5124155879020691,
"rewards/rejected": -1.181565284729004,
"step": 113
},
{
"epoch": 0.5386887182516243,
"grad_norm": 2.1290543043289434,
"learning_rate": 4.2846583259494185e-07,
"logits/chosen": -2.795818328857422,
"logits/rejected": -2.9482498168945312,
"logps/chosen": -291.4162292480469,
"logps/rejected": -279.486083984375,
"loss": 0.5576,
"rewards/accuracies": 0.5625,
"rewards/chosen": -0.791326105594635,
"rewards/margins": 0.29795482754707336,
"rewards/rejected": -1.0892809629440308,
"step": 114
},
{
"epoch": 0.5434140578854105,
"grad_norm": 2.123022825050298,
"learning_rate": 4.271158314824199e-07,
"logits/chosen": -2.5966644287109375,
"logits/rejected": -2.67663836479187,
"logps/chosen": -286.2615661621094,
"logps/rejected": -300.17669677734375,
"loss": 0.5549,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.6490954160690308,
"rewards/margins": 0.37902897596359253,
"rewards/rejected": -1.028124451637268,
"step": 115
},
{
"epoch": 0.5481393975191967,
"grad_norm": 2.3410590679680854,
"learning_rate": 4.2575538056032446e-07,
"logits/chosen": -2.3392884731292725,
"logits/rejected": -2.3976926803588867,
"logps/chosen": -309.89501953125,
"logps/rejected": -347.65264892578125,
"loss": 0.5519,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.8080885410308838,
"rewards/margins": 0.5476577877998352,
"rewards/rejected": -1.3557462692260742,
"step": 116
},
{
"epoch": 0.5528647371529829,
"grad_norm": 2.1818324792915305,
"learning_rate": 4.2438456009507195e-07,
"logits/chosen": -2.751250743865967,
"logits/rejected": -2.683605909347534,
"logps/chosen": -304.8617248535156,
"logps/rejected": -357.66583251953125,
"loss": 0.5637,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.7783306837081909,
"rewards/margins": 0.6774348020553589,
"rewards/rejected": -1.4557652473449707,
"step": 117
},
{
"epoch": 0.5575900767867691,
"grad_norm": 2.2239562359532994,
"learning_rate": 4.230034509648803e-07,
"logits/chosen": -2.653618335723877,
"logits/rejected": -2.6241607666015625,
"logps/chosen": -311.1373596191406,
"logps/rejected": -379.66473388671875,
"loss": 0.5332,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.6985858678817749,
"rewards/margins": 0.754062294960022,
"rewards/rejected": -1.4526481628417969,
"step": 118
},
{
"epoch": 0.5623154164205553,
"grad_norm": 2.2500972070282024,
"learning_rate": 4.216121346549973e-07,
"logits/chosen": -3.0888874530792236,
"logits/rejected": -2.974677562713623,
"logps/chosen": -266.66705322265625,
"logps/rejected": -335.3798828125,
"loss": 0.5606,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.6862295866012573,
"rewards/margins": 0.5550753474235535,
"rewards/rejected": -1.2413049936294556,
"step": 119
},
{
"epoch": 0.5670407560543415,
"grad_norm": 2.18698230287775,
"learning_rate": 4.202106932528928e-07,
"logits/chosen": -2.6840288639068604,
"logits/rejected": -2.7219927310943604,
"logps/chosen": -349.6735534667969,
"logps/rejected": -351.0025634765625,
"loss": 0.5593,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.9063121676445007,
"rewards/margins": 0.5141624808311462,
"rewards/rejected": -1.420474648475647,
"step": 120
},
{
"epoch": 0.5717660956881275,
"grad_norm": 2.3004644364082734,
"learning_rate": 4.1879920944341593e-07,
"logits/chosen": -2.914316177368164,
"logits/rejected": -2.9891955852508545,
"logps/chosen": -279.8473205566406,
"logps/rejected": -313.1243591308594,
"loss": 0.544,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.754368245601654,
"rewards/margins": 0.5700761079788208,
"rewards/rejected": -1.3244441747665405,
"step": 121
},
{
"epoch": 0.5764914353219137,
"grad_norm": 2.3085447219221504,
"learning_rate": 4.1737776650391625e-07,
"logits/chosen": -2.5704903602600098,
"logits/rejected": -2.6333065032958984,
"logps/chosen": -274.203369140625,
"logps/rejected": -333.9256896972656,
"loss": 0.5438,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.7395837903022766,
"rewards/margins": 0.5535318851470947,
"rewards/rejected": -1.2931156158447266,
"step": 122
},
{
"epoch": 0.5812167749556999,
"grad_norm": 2.1555107869656234,
"learning_rate": 4.1594644829933074e-07,
"logits/chosen": -3.0889954566955566,
"logits/rejected": -3.00903582572937,
"logps/chosen": -287.2713623046875,
"logps/rejected": -352.50244140625,
"loss": 0.5432,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.7744563221931458,
"rewards/margins": 0.6194370985031128,
"rewards/rejected": -1.3938933610916138,
"step": 123
},
{
"epoch": 0.5859421145894861,
"grad_norm": 2.3124214479042147,
"learning_rate": 4.1450533927723563e-07,
"logits/chosen": -2.7649660110473633,
"logits/rejected": -2.7645654678344727,
"logps/chosen": -323.98663330078125,
"logps/rejected": -377.0577697753906,
"loss": 0.5562,
"rewards/accuracies": 0.703125,
"rewards/chosen": -0.9616256952285767,
"rewards/margins": 0.6629120707511902,
"rewards/rejected": -1.624537706375122,
"step": 124
},
{
"epoch": 0.5906674542232723,
"grad_norm": 2.201123781604531,
"learning_rate": 4.130545244628638e-07,
"logits/chosen": -2.8170711994171143,
"logits/rejected": -2.801412582397461,
"logps/chosen": -284.6482849121094,
"logps/rejected": -339.86328125,
"loss": 0.5695,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.9369969964027405,
"rewards/margins": 0.6950640678405762,
"rewards/rejected": -1.6320611238479614,
"step": 125
},
{
"epoch": 0.5953927938570585,
"grad_norm": 2.4679791573291907,
"learning_rate": 4.11594089454089e-07,
"logits/chosen": -2.7288601398468018,
"logits/rejected": -2.732513427734375,
"logps/chosen": -331.1778259277344,
"logps/rejected": -422.12139892578125,
"loss": 0.566,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.8497134447097778,
"rewards/margins": 0.6027945280075073,
"rewards/rejected": -1.4525080919265747,
"step": 126
},
{
"epoch": 0.6001181334908446,
"grad_norm": 2.5170488926988073,
"learning_rate": 4.101241204163748e-07,
"logits/chosen": -2.66646671295166,
"logits/rejected": -2.621904134750366,
"logps/chosen": -301.6587829589844,
"logps/rejected": -365.4261474609375,
"loss": 0.5407,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.9232099056243896,
"rewards/margins": 0.4544805884361267,
"rewards/rejected": -1.377690315246582,
"step": 127
},
{
"epoch": 0.6048434731246308,
"grad_norm": 2.3673200269773718,
"learning_rate": 4.086447040776911e-07,
"logits/chosen": -3.108903169631958,
"logits/rejected": -3.056070327758789,
"logps/chosen": -276.7916259765625,
"logps/rejected": -342.23876953125,
"loss": 0.5449,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.8404449820518494,
"rewards/margins": 0.6569064855575562,
"rewards/rejected": -1.4973516464233398,
"step": 128
},
{
"epoch": 0.609568812758417,
"grad_norm": 3.2943893516301768,
"learning_rate": 4.071559277233975e-07,
"logits/chosen": -3.1619277000427246,
"logits/rejected": -3.1500680446624756,
"logps/chosen": -283.8644104003906,
"logps/rejected": -360.0433044433594,
"loss": 0.573,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.969115674495697,
"rewards/margins": 0.5388011932373047,
"rewards/rejected": -1.507916808128357,
"step": 129
},
{
"epoch": 0.609568812758417,
"eval_logits/chosen": -2.9238109588623047,
"eval_logits/rejected": -2.93009352684021,
"eval_logps/chosen": -304.09130859375,
"eval_logps/rejected": -360.09600830078125,
"eval_loss": 0.5451335906982422,
"eval_rewards/accuracies": 0.6174242496490479,
"eval_rewards/chosen": -0.9286060333251953,
"eval_rewards/margins": 0.6728550791740417,
"eval_rewards/rejected": -1.6014612913131714,
"eval_runtime": 226.5871,
"eval_samples_per_second": 16.135,
"eval_steps_per_second": 0.291,
"step": 129
},
{
"epoch": 0.6142941523922032,
"grad_norm": 2.3717766371975086,
"learning_rate": 4.05657879191093e-07,
"logits/chosen": -2.86085844039917,
"logits/rejected": -3.0779123306274414,
"logps/chosen": -353.9771423339844,
"logps/rejected": -372.76788330078125,
"loss": 0.5611,
"rewards/accuracies": 0.59375,
"rewards/chosen": -0.9203246235847473,
"rewards/margins": 0.6556491851806641,
"rewards/rejected": -1.575973629951477,
"step": 130
},
{
"epoch": 0.6190194920259894,
"grad_norm": 2.9558527636533025,
"learning_rate": 4.04150646865434e-07,
"logits/chosen": -2.6664750576019287,
"logits/rejected": -2.6207780838012695,
"logps/chosen": -283.3083801269531,
"logps/rejected": -348.6871032714844,
"loss": 0.5525,
"rewards/accuracies": 0.6875,
"rewards/chosen": -0.8656440377235413,
"rewards/margins": 0.7297619581222534,
"rewards/rejected": -1.5954060554504395,
"step": 131
},
{
"epoch": 0.6237448316597756,
"grad_norm": 2.7269741124418165,
"learning_rate": 4.0263431967291934e-07,
"logits/chosen": -2.8708412647247314,
"logits/rejected": -2.836123466491699,
"logps/chosen": -251.37966918945312,
"logps/rejected": -302.320068359375,
"loss": 0.5554,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.8969355821609497,
"rewards/margins": 0.4451577365398407,
"rewards/rejected": -1.3420933485031128,
"step": 132
},
{
"epoch": 0.6284701712935618,
"grad_norm": 2.3559260676043547,
"learning_rate": 4.011089870766437e-07,
"logits/chosen": -2.9185516834259033,
"logits/rejected": -3.0862460136413574,
"logps/chosen": -340.1504211425781,
"logps/rejected": -376.01629638671875,
"loss": 0.5223,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.8409585952758789,
"rewards/margins": 0.9160802364349365,
"rewards/rejected": -1.7570387125015259,
"step": 133
},
{
"epoch": 0.6331955109273479,
"grad_norm": 2.551925501157886,
"learning_rate": 3.995747390710196e-07,
"logits/chosen": -2.895480155944824,
"logits/rejected": -2.939868211746216,
"logps/chosen": -323.12164306640625,
"logps/rejected": -397.951171875,
"loss": 0.5385,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.1145464181900024,
"rewards/margins": 1.0304715633392334,
"rewards/rejected": -2.1450178623199463,
"step": 134
},
{
"epoch": 0.6379208505611341,
"grad_norm": 3.0639374327823625,
"learning_rate": 3.98031666176467e-07,
"logits/chosen": -3.221116781234741,
"logits/rejected": -3.125380516052246,
"logps/chosen": -277.20684814453125,
"logps/rejected": -368.4245300292969,
"loss": 0.5453,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.9265705943107605,
"rewards/margins": 0.8099436163902283,
"rewards/rejected": -1.7365142107009888,
"step": 135
},
{
"epoch": 0.6426461901949203,
"grad_norm": 3.0229894732050537,
"learning_rate": 3.9647985943407345e-07,
"logits/chosen": -2.7229156494140625,
"logits/rejected": -2.624408006668091,
"logps/chosen": -296.76507568359375,
"logps/rejected": -366.4635925292969,
"loss": 0.5446,
"rewards/accuracies": 0.671875,
"rewards/chosen": -0.911621630191803,
"rewards/margins": 0.3694719672203064,
"rewards/rejected": -1.2810935974121094,
"step": 136
},
{
"epoch": 0.6473715298287065,
"grad_norm": 2.777240365107515,
"learning_rate": 3.949194104002224e-07,
"logits/chosen": -3.008553981781006,
"logits/rejected": -3.0245308876037598,
"logps/chosen": -278.2191162109375,
"logps/rejected": -372.57672119140625,
"loss": 0.5333,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.9170963168144226,
"rewards/margins": 0.91133052110672,
"rewards/rejected": -1.8284270763397217,
"step": 137
},
{
"epoch": 0.6520968694624926,
"grad_norm": 2.5983010643513222,
"learning_rate": 3.93350411141191e-07,
"logits/chosen": -2.984111785888672,
"logits/rejected": -2.9940693378448486,
"logps/chosen": -256.7328796386719,
"logps/rejected": -304.14361572265625,
"loss": 0.5481,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.8156505227088928,
"rewards/margins": 0.6031564474105835,
"rewards/rejected": -1.418807029724121,
"step": 138
},
{
"epoch": 0.6568222090962788,
"grad_norm": 2.758553605218817,
"learning_rate": 3.917729542277187e-07,
"logits/chosen": -2.739635944366455,
"logits/rejected": -2.8373708724975586,
"logps/chosen": -355.6564025878906,
"logps/rejected": -444.02471923828125,
"loss": 0.5188,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.9943738579750061,
"rewards/margins": 0.8635731339454651,
"rewards/rejected": -1.8579471111297607,
"step": 139
},
{
"epoch": 0.6615475487300649,
"grad_norm": 2.5137621293297605,
"learning_rate": 3.901871327295453e-07,
"logits/chosen": -2.6592538356781006,
"logits/rejected": -2.8847031593322754,
"logps/chosen": -330.49609375,
"logps/rejected": -349.5611572265625,
"loss": 0.5406,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.8463073968887329,
"rewards/margins": 0.6520651578903198,
"rewards/rejected": -1.4983725547790527,
"step": 140
},
{
"epoch": 0.6662728883638511,
"grad_norm": 2.285013562717476,
"learning_rate": 3.885930402099199e-07,
"logits/chosen": -2.5882949829101562,
"logits/rejected": -2.6650753021240234,
"logps/chosen": -335.5899658203125,
"logps/rejected": -360.06451416015625,
"loss": 0.5074,
"rewards/accuracies": 0.65625,
"rewards/chosen": -0.9886775016784668,
"rewards/margins": 0.6825906038284302,
"rewards/rejected": -1.6712682247161865,
"step": 141
},
{
"epoch": 0.6709982279976373,
"grad_norm": 2.554720405760107,
"learning_rate": 3.8699077072008085e-07,
"logits/chosen": -2.8670525550842285,
"logits/rejected": -2.866511344909668,
"logps/chosen": -271.18243408203125,
"logps/rejected": -343.53582763671875,
"loss": 0.5163,
"rewards/accuracies": 0.578125,
"rewards/chosen": -1.0047590732574463,
"rewards/margins": 0.5558298826217651,
"rewards/rejected": -1.5605889558792114,
"step": 142
},
{
"epoch": 0.6757235676314235,
"grad_norm": 2.830180877879511,
"learning_rate": 3.8538041879370657e-07,
"logits/chosen": -3.037707567214966,
"logits/rejected": -3.063495397567749,
"logps/chosen": -347.0140380859375,
"logps/rejected": -392.6544494628906,
"loss": 0.5505,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.9989163875579834,
"rewards/margins": 0.8129914402961731,
"rewards/rejected": -1.8119077682495117,
"step": 143
},
{
"epoch": 0.6804489072652097,
"grad_norm": 2.6000164530484224,
"learning_rate": 3.8376207944133817e-07,
"logits/chosen": -3.087387800216675,
"logits/rejected": -3.079148530960083,
"logps/chosen": -316.996337890625,
"logps/rejected": -361.9954528808594,
"loss": 0.5365,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.111214518547058,
"rewards/margins": 0.5254876017570496,
"rewards/rejected": -1.636702060699463,
"step": 144
},
{
"epoch": 0.6851742468989959,
"grad_norm": 2.6115954536025408,
"learning_rate": 3.8213584814477363e-07,
"logits/chosen": -3.111316442489624,
"logits/rejected": -3.184953451156616,
"logps/chosen": -311.6522521972656,
"logps/rejected": -325.0541076660156,
"loss": 0.4982,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.9516675472259521,
"rewards/margins": 0.6691429018974304,
"rewards/rejected": -1.6208105087280273,
"step": 145
},
{
"epoch": 0.689899586532782,
"grad_norm": 2.976191491134426,
"learning_rate": 3.8050182085143464e-07,
"logits/chosen": -2.9731078147888184,
"logits/rejected": -3.072920083999634,
"logps/chosen": -309.2756042480469,
"logps/rejected": -352.2410888671875,
"loss": 0.5572,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.080316185951233,
"rewards/margins": 0.8104506134986877,
"rewards/rejected": -1.8907668590545654,
"step": 146
},
{
"epoch": 0.6946249261665682,
"grad_norm": 2.837971176112728,
"learning_rate": 3.7886009396870564e-07,
"logits/chosen": -2.8793129920959473,
"logits/rejected": -2.8820691108703613,
"logps/chosen": -288.26116943359375,
"logps/rejected": -363.0924072265625,
"loss": 0.5536,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.1091738939285278,
"rewards/margins": 0.493495374917984,
"rewards/rejected": -1.602669358253479,
"step": 147
},
{
"epoch": 0.6993502658003544,
"grad_norm": 2.928245357032695,
"learning_rate": 3.7721076435824585e-07,
"logits/chosen": -2.7040960788726807,
"logits/rejected": -2.8614137172698975,
"logps/chosen": -415.0562438964844,
"logps/rejected": -420.36297607421875,
"loss": 0.5321,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.0266618728637695,
"rewards/margins": 0.8255325555801392,
"rewards/rejected": -1.8521945476531982,
"step": 148
},
{
"epoch": 0.7040756054341406,
"grad_norm": 2.983452098073201,
"learning_rate": 3.755539293302742e-07,
"logits/chosen": -2.614259958267212,
"logits/rejected": -2.6951889991760254,
"logps/chosen": -352.3388366699219,
"logps/rejected": -375.2248840332031,
"loss": 0.5457,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.1011898517608643,
"rewards/margins": 0.613980233669281,
"rewards/rejected": -1.715169906616211,
"step": 149
},
{
"epoch": 0.7088009450679268,
"grad_norm": 2.6652814589714997,
"learning_rate": 3.738896866378282e-07,
"logits/chosen": -2.7506563663482666,
"logits/rejected": -2.690138339996338,
"logps/chosen": -322.7042236328125,
"logps/rejected": -373.8511047363281,
"loss": 0.516,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.9171172380447388,
"rewards/margins": 0.8513570427894592,
"rewards/rejected": -1.7684742212295532,
"step": 150
},
{
"epoch": 0.713526284701713,
"grad_norm": 2.5934419246508096,
"learning_rate": 3.722181344709969e-07,
"logits/chosen": -2.7295525074005127,
"logits/rejected": -2.855721950531006,
"logps/chosen": -332.5438232421875,
"logps/rejected": -359.502197265625,
"loss": 0.5121,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.9578840136528015,
"rewards/margins": 0.7636557221412659,
"rewards/rejected": -1.7215397357940674,
"step": 151
},
{
"epoch": 0.7182516243354992,
"grad_norm": 3.1206430500674975,
"learning_rate": 3.705393714511268e-07,
"logits/chosen": -2.845468282699585,
"logits/rejected": -2.7837162017822266,
"logps/chosen": -314.1913757324219,
"logps/rejected": -419.7186584472656,
"loss": 0.5321,
"rewards/accuracies": 0.640625,
"rewards/chosen": -0.9525049924850464,
"rewards/margins": 0.6129012703895569,
"rewards/rejected": -1.5654062032699585,
"step": 152
},
{
"epoch": 0.7229769639692853,
"grad_norm": 3.3791075672510336,
"learning_rate": 3.688534966250042e-07,
"logits/chosen": -3.007288932800293,
"logits/rejected": -2.9382283687591553,
"logps/chosen": -297.81622314453125,
"logps/rejected": -355.0559387207031,
"loss": 0.54,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.8055727481842041,
"rewards/margins": 0.7948654890060425,
"rewards/rejected": -1.6004382371902466,
"step": 153
},
{
"epoch": 0.7277023036030714,
"grad_norm": 2.7374735457469384,
"learning_rate": 3.671606094590108e-07,
"logits/chosen": -2.7088348865509033,
"logits/rejected": -2.7453291416168213,
"logps/chosen": -337.4541931152344,
"logps/rejected": -411.91485595703125,
"loss": 0.5429,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.0184537172317505,
"rewards/margins": 0.8295416235923767,
"rewards/rejected": -1.8479952812194824,
"step": 154
},
{
"epoch": 0.7324276432368576,
"grad_norm": 2.977388710936637,
"learning_rate": 3.6546080983325523e-07,
"logits/chosen": -2.824364185333252,
"logits/rejected": -2.911698579788208,
"logps/chosen": -308.72161865234375,
"logps/rejected": -312.67431640625,
"loss": 0.5497,
"rewards/accuracies": 0.5625,
"rewards/chosen": -1.054483413696289,
"rewards/margins": 0.345467209815979,
"rewards/rejected": -1.399950623512268,
"step": 155
},
{
"epoch": 0.7371529828706438,
"grad_norm": 3.297987493267062,
"learning_rate": 3.6375419803568046e-07,
"logits/chosen": -2.938750743865967,
"logits/rejected": -3.12616229057312,
"logps/chosen": -380.7288513183594,
"logps/rejected": -393.94683837890625,
"loss": 0.5366,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.0641953945159912,
"rewards/margins": 0.899326741695404,
"rewards/rejected": -1.96352219581604,
"step": 156
},
{
"epoch": 0.74187832250443,
"grad_norm": 2.5345279316698983,
"learning_rate": 3.6204087475614676e-07,
"logits/chosen": -2.923267364501953,
"logits/rejected": -2.8579440116882324,
"logps/chosen": -302.9458923339844,
"logps/rejected": -387.598388671875,
"loss": 0.5246,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.0088130235671997,
"rewards/margins": 0.8763782382011414,
"rewards/rejected": -1.8851913213729858,
"step": 157
},
{
"epoch": 0.7466036621382162,
"grad_norm": 3.13993186485945,
"learning_rate": 3.603209410804906e-07,
"logits/chosen": -2.7970800399780273,
"logits/rejected": -2.77022123336792,
"logps/chosen": -260.28485107421875,
"logps/rejected": -377.06585693359375,
"loss": 0.5166,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.9031432867050171,
"rewards/margins": 0.9541431665420532,
"rewards/rejected": -1.8572864532470703,
"step": 158
},
{
"epoch": 0.7513290017720023,
"grad_norm": 2.777966751571199,
"learning_rate": 3.5859449848456123e-07,
"logits/chosen": -2.83420991897583,
"logits/rejected": -2.9197449684143066,
"logps/chosen": -270.993896484375,
"logps/rejected": -329.7164001464844,
"loss": 0.5438,
"rewards/accuracies": 0.609375,
"rewards/chosen": -0.9526958465576172,
"rewards/margins": 0.5992559194564819,
"rewards/rejected": -1.5519516468048096,
"step": 159
},
{
"epoch": 0.7560543414057885,
"grad_norm": 2.816810414145774,
"learning_rate": 3.5686164882823313e-07,
"logits/chosen": -2.4739251136779785,
"logits/rejected": -2.5660862922668457,
"logps/chosen": -325.5205383300781,
"logps/rejected": -357.9018859863281,
"loss": 0.5207,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.059773564338684,
"rewards/margins": 0.855665922164917,
"rewards/rejected": -1.9154393672943115,
"step": 160
},
{
"epoch": 0.7607796810395747,
"grad_norm": 2.7264976697102217,
"learning_rate": 3.5512249434939634e-07,
"logits/chosen": -3.020364284515381,
"logits/rejected": -3.1138038635253906,
"logps/chosen": -292.44451904296875,
"logps/rejected": -370.27581787109375,
"loss": 0.5324,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.0052980184555054,
"rewards/margins": 0.916731059551239,
"rewards/rejected": -1.9220290184020996,
"step": 161
},
{
"epoch": 0.7655050206733609,
"grad_norm": 4.143376503088087,
"learning_rate": 3.533771376579249e-07,
"logits/chosen": -2.8414347171783447,
"logits/rejected": -2.8189237117767334,
"logps/chosen": -298.5760192871094,
"logps/rejected": -407.37457275390625,
"loss": 0.5111,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.9188227653503418,
"rewards/margins": 1.0365185737609863,
"rewards/rejected": -1.9553413391113281,
"step": 162
},
{
"epoch": 0.7702303603071471,
"grad_norm": 2.851627509613439,
"learning_rate": 3.5162568172962215e-07,
"logits/chosen": -2.737412214279175,
"logits/rejected": -2.8929431438446045,
"logps/chosen": -325.90631103515625,
"logps/rejected": -372.39410400390625,
"loss": 0.5193,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.134871244430542,
"rewards/margins": 0.6576811075210571,
"rewards/rejected": -1.7925523519515991,
"step": 163
},
{
"epoch": 0.7749556999409333,
"grad_norm": 3.4081328480353985,
"learning_rate": 3.498682299001459e-07,
"logits/chosen": -2.622042179107666,
"logits/rejected": -2.759326696395874,
"logps/chosen": -354.53448486328125,
"logps/rejected": -397.6759338378906,
"loss": 0.5084,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.1062355041503906,
"rewards/margins": 0.7745749354362488,
"rewards/rejected": -1.8808104991912842,
"step": 164
},
{
"epoch": 0.7796810395747195,
"grad_norm": 5.113522770893731,
"learning_rate": 3.4810488585891103e-07,
"logits/chosen": -2.7235350608825684,
"logits/rejected": -2.8394298553466797,
"logps/chosen": -339.3172912597656,
"logps/rejected": -403.5739440917969,
"loss": 0.5256,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.0079370737075806,
"rewards/margins": 0.9361110925674438,
"rewards/rejected": -1.9440481662750244,
"step": 165
},
{
"epoch": 0.7844063792085056,
"grad_norm": 2.845383053447366,
"learning_rate": 3.4633575364297224e-07,
"logits/chosen": -3.0904507637023926,
"logits/rejected": -3.1447291374206543,
"logps/chosen": -308.4082336425781,
"logps/rejected": -403.6357421875,
"loss": 0.5258,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.1904851198196411,
"rewards/margins": 1.1804500818252563,
"rewards/rejected": -2.3709352016448975,
"step": 166
},
{
"epoch": 0.7891317188422918,
"grad_norm": 3.6331311614486483,
"learning_rate": 3.445609376308857e-07,
"logits/chosen": -2.7289986610412598,
"logits/rejected": -2.64235520362854,
"logps/chosen": -351.40728759765625,
"logps/rejected": -435.42010498046875,
"loss": 0.4979,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.273624300956726,
"rewards/margins": 1.0576740503311157,
"rewards/rejected": -2.331298351287842,
"step": 167
},
{
"epoch": 0.793857058476078,
"grad_norm": 3.6152830110317655,
"learning_rate": 3.4278054253655086e-07,
"logits/chosen": -2.8024775981903076,
"logits/rejected": -2.853891611099243,
"logps/chosen": -322.92181396484375,
"logps/rejected": -396.2481689453125,
"loss": 0.5094,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.0655136108398438,
"rewards/margins": 0.8890769481658936,
"rewards/rejected": -1.9545905590057373,
"step": 168
},
{
"epoch": 0.7985823981098642,
"grad_norm": 3.119357172634371,
"learning_rate": 3.4099467340303214e-07,
"logits/chosen": -3.0272624492645264,
"logits/rejected": -3.140334129333496,
"logps/chosen": -307.60479736328125,
"logps/rejected": -376.73590087890625,
"loss": 0.5064,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.1390501260757446,
"rewards/margins": 0.9899504780769348,
"rewards/rejected": -2.129000425338745,
"step": 169
},
{
"epoch": 0.8033077377436503,
"grad_norm": 3.457360676304788,
"learning_rate": 3.392034355963614e-07,
"logits/chosen": -2.8242180347442627,
"logits/rejected": -2.8203928470611572,
"logps/chosen": -338.4170837402344,
"logps/rejected": -375.84893798828125,
"loss": 0.5257,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.182822346687317,
"rewards/margins": 0.7907478213310242,
"rewards/rejected": -1.9735702276229858,
"step": 170
},
{
"epoch": 0.8080330773774365,
"grad_norm": 5.011470232701305,
"learning_rate": 3.374069347993218e-07,
"logits/chosen": -2.6921019554138184,
"logits/rejected": -2.7788760662078857,
"logps/chosen": -329.470703125,
"logps/rejected": -467.9057922363281,
"loss": 0.5396,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.1266531944274902,
"rewards/margins": 0.8941323757171631,
"rewards/rejected": -2.0207855701446533,
"step": 171
},
{
"epoch": 0.8127584170112226,
"grad_norm": 3.7253097168704294,
"learning_rate": 3.356052770052119e-07,
"logits/chosen": -2.716782331466675,
"logits/rejected": -2.6556971073150635,
"logps/chosen": -309.70672607421875,
"logps/rejected": -426.3639221191406,
"loss": 0.5239,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.2285258769989014,
"rewards/margins": 0.8051817417144775,
"rewards/rejected": -2.033707618713379,
"step": 172
},
{
"epoch": 0.8127584170112226,
"eval_logits/chosen": -2.9802942276000977,
"eval_logits/rejected": -2.9883527755737305,
"eval_logps/chosen": -339.8587646484375,
"eval_logps/rejected": -423.5323791503906,
"eval_loss": 0.51226407289505,
"eval_rewards/accuracies": 0.6287878751754761,
"eval_rewards/chosen": -1.2862800359725952,
"eval_rewards/margins": 0.9495444893836975,
"eval_rewards/rejected": -2.2358245849609375,
"eval_runtime": 225.4382,
"eval_samples_per_second": 16.217,
"eval_steps_per_second": 0.293,
"step": 172
},
{
"epoch": 0.8174837566450088,
"grad_norm": 4.039346342004204,
"learning_rate": 3.337985685115926e-07,
"logits/chosen": -2.9982471466064453,
"logits/rejected": -2.934654712677002,
"logps/chosen": -345.1435546875,
"logps/rejected": -418.8468017578125,
"loss": 0.5227,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.2590099573135376,
"rewards/margins": 0.7078821659088135,
"rewards/rejected": -1.9668920040130615,
"step": 173
},
{
"epoch": 0.822209096278795,
"grad_norm": 3.355839857834348,
"learning_rate": 3.319869159140152e-07,
"logits/chosen": -2.412257432937622,
"logits/rejected": -2.5528626441955566,
"logps/chosen": -313.64422607421875,
"logps/rejected": -387.2737121582031,
"loss": 0.5154,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.2666703462600708,
"rewards/margins": 0.9918266534805298,
"rewards/rejected": -2.2584969997406006,
"step": 174
},
{
"epoch": 0.8269344359125812,
"grad_norm": 3.8883187343016528,
"learning_rate": 3.301704260997325e-07,
"logits/chosen": -2.835768222808838,
"logits/rejected": -2.861588716506958,
"logps/chosen": -310.4635314941406,
"logps/rejected": -418.094970703125,
"loss": 0.5263,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.149375557899475,
"rewards/margins": 1.2631374597549438,
"rewards/rejected": -2.41251277923584,
"step": 175
},
{
"epoch": 0.8316597755463674,
"grad_norm": 3.943606898054145,
"learning_rate": 3.283492062413925e-07,
"logits/chosen": -2.8773105144500732,
"logits/rejected": -2.919630527496338,
"logps/chosen": -315.5,
"logps/rejected": -414.37908935546875,
"loss": 0.5139,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.2905051708221436,
"rewards/margins": 1.0659555196762085,
"rewards/rejected": -2.3564605712890625,
"step": 176
},
{
"epoch": 0.8363851151801536,
"grad_norm": 4.394795298842495,
"learning_rate": 3.2652336379071506e-07,
"logits/chosen": -2.7635695934295654,
"logits/rejected": -2.8052563667297363,
"logps/chosen": -371.2409973144531,
"logps/rejected": -429.0415954589844,
"loss": 0.4828,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.2244281768798828,
"rewards/margins": 1.045049786567688,
"rewards/rejected": -2.2694778442382812,
"step": 177
},
{
"epoch": 0.8411104548139398,
"grad_norm": 3.6672724994810593,
"learning_rate": 3.246930064721523e-07,
"logits/chosen": -2.7490479946136475,
"logits/rejected": -2.6954996585845947,
"logps/chosen": -265.0568542480469,
"logps/rejected": -373.4683532714844,
"loss": 0.5134,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.1422216892242432,
"rewards/margins": 1.010473370552063,
"rewards/rejected": -2.1526949405670166,
"step": 178
},
{
"epoch": 0.8458357944477259,
"grad_norm": 3.438810171868027,
"learning_rate": 3.228582422765331e-07,
"logits/chosen": -2.661006212234497,
"logits/rejected": -2.756809711456299,
"logps/chosen": -363.7588806152344,
"logps/rejected": -392.8025207519531,
"loss": 0.5044,
"rewards/accuracies": 0.609375,
"rewards/chosen": -1.341508388519287,
"rewards/margins": 0.7005224227905273,
"rewards/rejected": -2.0420308113098145,
"step": 179
},
{
"epoch": 0.8505611340815121,
"grad_norm": 4.113182426182398,
"learning_rate": 3.2101917945469135e-07,
"logits/chosen": -2.700942277908325,
"logits/rejected": -2.82185435295105,
"logps/chosen": -364.240234375,
"logps/rejected": -429.81292724609375,
"loss": 0.4816,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.1254488229751587,
"rewards/margins": 0.9122301340103149,
"rewards/rejected": -2.0376789569854736,
"step": 180
},
{
"epoch": 0.8552864737152983,
"grad_norm": 4.022329679746118,
"learning_rate": 3.1917592651107927e-07,
"logits/chosen": -2.8973255157470703,
"logits/rejected": -2.794524908065796,
"logps/chosen": -352.9599914550781,
"logps/rejected": -422.71197509765625,
"loss": 0.5153,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.2179690599441528,
"rewards/margins": 0.7948740124702454,
"rewards/rejected": -2.012843132019043,
"step": 181
},
{
"epoch": 0.8600118133490845,
"grad_norm": 3.7009474178127353,
"learning_rate": 3.173285921973657e-07,
"logits/chosen": -2.793835401535034,
"logits/rejected": -2.7703464031219482,
"logps/chosen": -320.6808776855469,
"logps/rejected": -454.6962890625,
"loss": 0.512,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.1430678367614746,
"rewards/margins": 1.6093158721923828,
"rewards/rejected": -2.7523837089538574,
"step": 182
},
{
"epoch": 0.8647371529828707,
"grad_norm": 3.2328530891675182,
"learning_rate": 3.1547728550601983e-07,
"logits/chosen": -2.6808881759643555,
"logits/rejected": -2.676565170288086,
"logps/chosen": -301.3103332519531,
"logps/rejected": -402.32281494140625,
"loss": 0.5161,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.1162338256835938,
"rewards/margins": 1.0167958736419678,
"rewards/rejected": -2.1330299377441406,
"step": 183
},
{
"epoch": 0.8694624926166569,
"grad_norm": 3.899851002215307,
"learning_rate": 3.1362211566388057e-07,
"logits/chosen": -2.8450677394866943,
"logits/rejected": -2.836167097091675,
"logps/chosen": -341.335693359375,
"logps/rejected": -407.4953918457031,
"loss": 0.5422,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.2810035943984985,
"rewards/margins": 0.6536043882369995,
"rewards/rejected": -1.934607982635498,
"step": 184
},
{
"epoch": 0.874187832250443,
"grad_norm": 3.2712338921521518,
"learning_rate": 3.1176319212571204e-07,
"logits/chosen": -2.405541181564331,
"logits/rejected": -2.4331328868865967,
"logps/chosen": -288.65325927734375,
"logps/rejected": -364.7303466796875,
"loss": 0.5114,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.2161575555801392,
"rewards/margins": 0.9309273362159729,
"rewards/rejected": -2.147084951400757,
"step": 185
},
{
"epoch": 0.8789131718842291,
"grad_norm": 3.8818256203788852,
"learning_rate": 3.099006245677461e-07,
"logits/chosen": -2.4217336177825928,
"logits/rejected": -2.360973358154297,
"logps/chosen": -381.958740234375,
"logps/rejected": -487.387451171875,
"loss": 0.5393,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.2569398880004883,
"rewards/margins": 0.7359199523925781,
"rewards/rejected": -1.9928598403930664,
"step": 186
},
{
"epoch": 0.8836385115180153,
"grad_norm": 3.241850575925625,
"learning_rate": 3.0803452288121113e-07,
"logits/chosen": -2.6186816692352295,
"logits/rejected": -2.5035552978515625,
"logps/chosen": -356.3555603027344,
"logps/rejected": -538.8609619140625,
"loss": 0.4985,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.2594354152679443,
"rewards/margins": 1.667135238647461,
"rewards/rejected": -2.9265708923339844,
"step": 187
},
{
"epoch": 0.8883638511518015,
"grad_norm": 2.994070703824197,
"learning_rate": 3.0616499716584874e-07,
"logits/chosen": -2.936795473098755,
"logits/rejected": -2.855689525604248,
"logps/chosen": -339.5811767578125,
"logps/rejected": -430.01483154296875,
"loss": 0.5116,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.1258800029754639,
"rewards/margins": 0.6621576547622681,
"rewards/rejected": -1.7880375385284424,
"step": 188
},
{
"epoch": 0.8930891907855877,
"grad_norm": 3.4626212317934146,
"learning_rate": 3.042921577234177e-07,
"logits/chosen": -2.6947526931762695,
"logits/rejected": -2.8166098594665527,
"logps/chosen": -328.1815185546875,
"logps/rejected": -340.3358154296875,
"loss": 0.5048,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.0728776454925537,
"rewards/margins": 0.6414874792098999,
"rewards/rejected": -1.7143651247024536,
"step": 189
},
{
"epoch": 0.8978145304193739,
"grad_norm": 3.1647602539221182,
"learning_rate": 3.024161150511861e-07,
"logits/chosen": -2.9085636138916016,
"logits/rejected": -3.0228068828582764,
"logps/chosen": -323.0364074707031,
"logps/rejected": -358.6117248535156,
"loss": 0.5248,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.1753543615341187,
"rewards/margins": 0.9408324956893921,
"rewards/rejected": -2.1161868572235107,
"step": 190
},
{
"epoch": 0.90253987005316,
"grad_norm": 3.025256770279786,
"learning_rate": 3.0053697983541247e-07,
"logits/chosen": -2.545339345932007,
"logits/rejected": -2.5873234272003174,
"logps/chosen": -364.4934387207031,
"logps/rejected": -404.4205322265625,
"loss": 0.5114,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.3082777261734009,
"rewards/margins": 0.9447382688522339,
"rewards/rejected": -2.2530159950256348,
"step": 191
},
{
"epoch": 0.9072652096869462,
"grad_norm": 3.2107060614205682,
"learning_rate": 2.986548629448146e-07,
"logits/chosen": -2.5320749282836914,
"logits/rejected": -2.6858551502227783,
"logps/chosen": -357.6763916015625,
"logps/rejected": -418.71173095703125,
"loss": 0.5194,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.0752955675125122,
"rewards/margins": 1.176898717880249,
"rewards/rejected": -2.252194404602051,
"step": 192
},
{
"epoch": 0.9119905493207324,
"grad_norm": 4.290813644018959,
"learning_rate": 2.967698754240289e-07,
"logits/chosen": -2.6695892810821533,
"logits/rejected": -2.660761833190918,
"logps/chosen": -347.8155517578125,
"logps/rejected": -403.8181457519531,
"loss": 0.5543,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.207077145576477,
"rewards/margins": 0.7051151394844055,
"rewards/rejected": -1.9121922254562378,
"step": 193
},
{
"epoch": 0.9167158889545186,
"grad_norm": 3.656365041410898,
"learning_rate": 2.948821284870585e-07,
"logits/chosen": -3.009221315383911,
"logits/rejected": -2.9382755756378174,
"logps/chosen": -338.1603698730469,
"logps/rejected": -425.35931396484375,
"loss": 0.5007,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.3645328283309937,
"rewards/margins": 0.9410390257835388,
"rewards/rejected": -2.305572032928467,
"step": 194
},
{
"epoch": 0.9214412285883048,
"grad_norm": 3.166267065915439,
"learning_rate": 2.9299173351071176e-07,
"logits/chosen": -2.544590950012207,
"logits/rejected": -2.5580813884735107,
"logps/chosen": -375.34112548828125,
"logps/rejected": -437.5019836425781,
"loss": 0.5057,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.1270973682403564,
"rewards/margins": 0.8198693990707397,
"rewards/rejected": -1.9469666481018066,
"step": 195
},
{
"epoch": 0.926166568222091,
"grad_norm": 2.9416082082853805,
"learning_rate": 2.9109880202803097e-07,
"logits/chosen": -2.5898144245147705,
"logits/rejected": -2.590017795562744,
"logps/chosen": -325.3600769042969,
"logps/rejected": -417.5689392089844,
"loss": 0.5251,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.1922552585601807,
"rewards/margins": 1.2383495569229126,
"rewards/rejected": -2.4306044578552246,
"step": 196
},
{
"epoch": 0.9308919078558772,
"grad_norm": 4.344454107439186,
"learning_rate": 2.892034457217119e-07,
"logits/chosen": -2.793138027191162,
"logits/rejected": -2.6765213012695312,
"logps/chosen": -393.07080078125,
"logps/rejected": -508.9129638671875,
"loss": 0.5302,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.202684998512268,
"rewards/margins": 1.2748346328735352,
"rewards/rejected": -2.4775197505950928,
"step": 197
},
{
"epoch": 0.9356172474896634,
"grad_norm": 4.341995602773921,
"learning_rate": 2.8730577641751474e-07,
"logits/chosen": -2.5986645221710205,
"logits/rejected": -2.7119102478027344,
"logps/chosen": -301.242431640625,
"logps/rejected": -423.28619384765625,
"loss": 0.5145,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.1418884992599487,
"rewards/margins": 1.2161611318588257,
"rewards/rejected": -2.3580498695373535,
"step": 198
},
{
"epoch": 0.9403425871234495,
"grad_norm": 3.5092494052777536,
"learning_rate": 2.854059060776659e-07,
"logits/chosen": -2.4758505821228027,
"logits/rejected": -2.5398991107940674,
"logps/chosen": -311.9541015625,
"logps/rejected": -413.2162780761719,
"loss": 0.4867,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.084080457687378,
"rewards/margins": 1.512494683265686,
"rewards/rejected": -2.5965750217437744,
"step": 199
},
{
"epoch": 0.9450679267572357,
"grad_norm": 3.5187855585130294,
"learning_rate": 2.835039467942529e-07,
"logits/chosen": -2.5920052528381348,
"logits/rejected": -2.7105183601379395,
"logps/chosen": -340.66351318359375,
"logps/rejected": -416.5780944824219,
"loss": 0.5222,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.0971770286560059,
"rewards/margins": 0.9351829290390015,
"rewards/rejected": -2.032360315322876,
"step": 200
},
{
"epoch": 0.9497932663910219,
"grad_norm": 3.793717608121411,
"learning_rate": 2.8160001078261055e-07,
"logits/chosen": -2.498663902282715,
"logits/rejected": -2.4728269577026367,
"logps/chosen": -301.9998474121094,
"logps/rejected": -426.3241271972656,
"loss": 0.5251,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.031512975692749,
"rewards/margins": 1.083221197128296,
"rewards/rejected": -2.114734172821045,
"step": 201
},
{
"epoch": 0.954518606024808,
"grad_norm": 2.9737105768220244,
"learning_rate": 2.7969421037470033e-07,
"logits/chosen": -2.7715539932250977,
"logits/rejected": -2.7849650382995605,
"logps/chosen": -345.68963623046875,
"logps/rejected": -444.9283752441406,
"loss": 0.5044,
"rewards/accuracies": 0.609375,
"rewards/chosen": -1.262139081954956,
"rewards/margins": 0.9420502185821533,
"rewards/rejected": -2.2041893005371094,
"step": 202
},
{
"epoch": 0.9592439456585942,
"grad_norm": 3.4484596414693893,
"learning_rate": 2.777866580124829e-07,
"logits/chosen": -2.491079330444336,
"logits/rejected": -2.7043981552124023,
"logps/chosen": -395.48211669921875,
"logps/rejected": -463.1803894042969,
"loss": 0.5037,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.2879524230957031,
"rewards/margins": 0.926118791103363,
"rewards/rejected": -2.214071273803711,
"step": 203
},
{
"epoch": 0.9639692852923804,
"grad_norm": 3.553922161828863,
"learning_rate": 2.758774662412838e-07,
"logits/chosen": -2.5517563819885254,
"logits/rejected": -2.4807627201080322,
"logps/chosen": -305.34637451171875,
"logps/rejected": -486.2059631347656,
"loss": 0.5027,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.346076250076294,
"rewards/margins": 1.699803352355957,
"rewards/rejected": -3.04587984085083,
"step": 204
},
{
"epoch": 0.9686946249261665,
"grad_norm": 3.191519565656703,
"learning_rate": 2.739667477031538e-07,
"logits/chosen": -2.7191619873046875,
"logits/rejected": -2.7111451625823975,
"logps/chosen": -344.4493408203125,
"logps/rejected": -480.70135498046875,
"loss": 0.4982,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.2324879169464111,
"rewards/margins": 1.2888904809951782,
"rewards/rejected": -2.521378517150879,
"step": 205
},
{
"epoch": 0.9734199645599527,
"grad_norm": 4.672920315021906,
"learning_rate": 2.7205461513022233e-07,
"logits/chosen": -2.0949220657348633,
"logits/rejected": -2.1367533206939697,
"logps/chosen": -390.1219482421875,
"logps/rejected": -415.38037109375,
"loss": 0.4815,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.0024380683898926,
"rewards/margins": 0.9189928770065308,
"rewards/rejected": -1.9214308261871338,
"step": 206
},
{
"epoch": 0.9781453041937389,
"grad_norm": 3.5306370227740262,
"learning_rate": 2.70141181338047e-07,
"logits/chosen": -2.406205177307129,
"logits/rejected": -2.38879132270813,
"logps/chosen": -356.2658386230469,
"logps/rejected": -456.2667236328125,
"loss": 0.4946,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.220351219177246,
"rewards/margins": 1.0728812217712402,
"rewards/rejected": -2.2932324409484863,
"step": 207
},
{
"epoch": 0.9828706438275251,
"grad_norm": 3.479933152007338,
"learning_rate": 2.6822655921895693e-07,
"logits/chosen": -2.446387767791748,
"logits/rejected": -2.346810817718506,
"logps/chosen": -312.9122619628906,
"logps/rejected": -459.10040283203125,
"loss": 0.5015,
"rewards/accuracies": 0.859375,
"rewards/chosen": -1.142736792564392,
"rewards/margins": 1.2611982822418213,
"rewards/rejected": -2.403934955596924,
"step": 208
},
{
"epoch": 0.9875959834613113,
"grad_norm": 3.8539520740205018,
"learning_rate": 2.663108617353926e-07,
"logits/chosen": -2.7008585929870605,
"logits/rejected": -2.7585320472717285,
"logps/chosen": -410.00634765625,
"logps/rejected": -455.63812255859375,
"loss": 0.5103,
"rewards/accuracies": 0.609375,
"rewards/chosen": -1.2332756519317627,
"rewards/margins": 0.693227231502533,
"rewards/rejected": -1.9265029430389404,
"step": 209
},
{
"epoch": 0.9923213230950975,
"grad_norm": 3.845322423212567,
"learning_rate": 2.6439420191324064e-07,
"logits/chosen": -2.462545871734619,
"logits/rejected": -2.5116636753082275,
"logps/chosen": -305.0135803222656,
"logps/rejected": -405.6048583984375,
"loss": 0.4911,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.231694221496582,
"rewards/margins": 0.9429072141647339,
"rewards/rejected": -2.1746013164520264,
"step": 210
},
{
"epoch": 0.9970466627288836,
"grad_norm": 5.6022942181300905,
"learning_rate": 2.6247669283516556e-07,
"logits/chosen": -2.544132947921753,
"logits/rejected": -2.5758156776428223,
"logps/chosen": -366.85009765625,
"logps/rejected": -442.0340881347656,
"loss": 0.5449,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.4171723127365112,
"rewards/margins": 1.0115364789962769,
"rewards/rejected": -2.428708791732788,
"step": 211
},
{
"epoch": 1.0017720023626697,
"grad_norm": 3.186023679965871,
"learning_rate": 2.60558447633938e-07,
"logits/chosen": -2.244544506072998,
"logits/rejected": -2.3220551013946533,
"logps/chosen": -366.2366638183594,
"logps/rejected": -452.18707275390625,
"loss": 0.5058,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.4039456844329834,
"rewards/margins": 1.031445860862732,
"rewards/rejected": -2.435391426086426,
"step": 212
},
{
"epoch": 1.006497341996456,
"grad_norm": 3.6450488482030523,
"learning_rate": 2.5863957948575963e-07,
"logits/chosen": -2.3453848361968994,
"logits/rejected": -2.468796968460083,
"logps/chosen": -326.9305114746094,
"logps/rejected": -368.9020690917969,
"loss": 0.4954,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.1048694849014282,
"rewards/margins": 0.7470109462738037,
"rewards/rejected": -1.8518803119659424,
"step": 213
},
{
"epoch": 1.011222681630242,
"grad_norm": 4.137227043648897,
"learning_rate": 2.567202016035859e-07,
"logits/chosen": -2.5369303226470947,
"logits/rejected": -2.535897731781006,
"logps/chosen": -310.4713439941406,
"logps/rejected": -399.39398193359375,
"loss": 0.4987,
"rewards/accuracies": 0.59375,
"rewards/chosen": -1.2701810598373413,
"rewards/margins": 0.7798756957054138,
"rewards/rejected": -2.0500564575195312,
"step": 214
},
{
"epoch": 1.0159480212640284,
"grad_norm": 4.02450997349156,
"learning_rate": 2.5480042723044653e-07,
"logits/chosen": -2.567810535430908,
"logits/rejected": -2.575981378555298,
"logps/chosen": -350.37957763671875,
"logps/rejected": -428.45745849609375,
"loss": 0.4668,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.4323270320892334,
"rewards/margins": 1.0882251262664795,
"rewards/rejected": -2.520552396774292,
"step": 215
},
{
"epoch": 1.0159480212640284,
"eval_logits/chosen": -2.584289789199829,
"eval_logits/rejected": -2.5910158157348633,
"eval_logps/chosen": -361.1751708984375,
"eval_logps/rejected": -463.7195129394531,
"eval_loss": 0.49447911977767944,
"eval_rewards/accuracies": 0.6439393758773804,
"eval_rewards/chosen": -1.499444603919983,
"eval_rewards/margins": 1.138251781463623,
"eval_rewards/rejected": -2.6376962661743164,
"eval_runtime": 224.7997,
"eval_samples_per_second": 16.263,
"eval_steps_per_second": 0.294,
"step": 215
},
{
"epoch": 1.0206733608978145,
"grad_norm": 4.359766282431931,
"learning_rate": 2.5288036963276414e-07,
"logits/chosen": -2.4916322231292725,
"logits/rejected": -2.482870101928711,
"logps/chosen": -297.8340759277344,
"logps/rejected": -387.14703369140625,
"loss": 0.4968,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.1933519840240479,
"rewards/margins": 1.0262653827667236,
"rewards/rejected": -2.2196173667907715,
"step": 216
},
{
"epoch": 1.0253987005316008,
"grad_norm": 3.593941578167617,
"learning_rate": 2.509601420936717e-07,
"logits/chosen": -2.5306057929992676,
"logits/rejected": -2.440415382385254,
"logps/chosen": -341.2530212402344,
"logps/rejected": -464.18487548828125,
"loss": 0.488,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.4438025951385498,
"rewards/margins": 1.340946078300476,
"rewards/rejected": -2.7847485542297363,
"step": 217
},
{
"epoch": 1.0301240401653868,
"grad_norm": 3.664745354346814,
"learning_rate": 2.490398579063283e-07,
"logits/chosen": -2.6102001667022705,
"logits/rejected": -2.529940128326416,
"logps/chosen": -345.69677734375,
"logps/rejected": -484.93035888671875,
"loss": 0.5001,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.5673726797103882,
"rewards/margins": 1.314273476600647,
"rewards/rejected": -2.8816463947296143,
"step": 218
},
{
"epoch": 1.0348493797991731,
"grad_norm": 5.084728530421535,
"learning_rate": 2.4711963036723583e-07,
"logits/chosen": -2.3622119426727295,
"logits/rejected": -2.365530490875244,
"logps/chosen": -367.52984619140625,
"logps/rejected": -406.59375,
"loss": 0.4928,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.4226937294006348,
"rewards/margins": 0.49759745597839355,
"rewards/rejected": -1.9202911853790283,
"step": 219
},
{
"epoch": 1.0395747194329592,
"grad_norm": 3.6814125755312093,
"learning_rate": 2.451995727695535e-07,
"logits/chosen": -2.4771206378936768,
"logits/rejected": -2.5038909912109375,
"logps/chosen": -365.91815185546875,
"logps/rejected": -477.00347900390625,
"loss": 0.4837,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.5666453838348389,
"rewards/margins": 1.2746491432189941,
"rewards/rejected": -2.841294527053833,
"step": 220
},
{
"epoch": 1.0443000590667455,
"grad_norm": 4.348311431123137,
"learning_rate": 2.432797983964141e-07,
"logits/chosen": -2.514589786529541,
"logits/rejected": -2.460106134414673,
"logps/chosen": -358.013427734375,
"logps/rejected": -444.3507080078125,
"loss": 0.4922,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.4816346168518066,
"rewards/margins": 0.9881500601768494,
"rewards/rejected": -2.469784736633301,
"step": 221
},
{
"epoch": 1.0490253987005316,
"grad_norm": 4.474335447495181,
"learning_rate": 2.413604205142404e-07,
"logits/chosen": -2.2786381244659424,
"logits/rejected": -2.371419668197632,
"logps/chosen": -377.5779113769531,
"logps/rejected": -431.80316162109375,
"loss": 0.5267,
"rewards/accuracies": 0.59375,
"rewards/chosen": -1.8127973079681396,
"rewards/margins": 0.9888743162155151,
"rewards/rejected": -2.8016717433929443,
"step": 222
},
{
"epoch": 1.0537507383343179,
"grad_norm": 3.9171412475826557,
"learning_rate": 2.3944155236606196e-07,
"logits/chosen": -2.4430832862854004,
"logits/rejected": -2.409916639328003,
"logps/chosen": -344.40399169921875,
"logps/rejected": -461.02349853515625,
"loss": 0.4846,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.4063448905944824,
"rewards/margins": 1.1203367710113525,
"rewards/rejected": -2.526681661605835,
"step": 223
},
{
"epoch": 1.058476077968104,
"grad_norm": 5.279702970403167,
"learning_rate": 2.3752330716483444e-07,
"logits/chosen": -2.7835280895233154,
"logits/rejected": -2.7787797451019287,
"logps/chosen": -353.5722351074219,
"logps/rejected": -415.2908935546875,
"loss": 0.5057,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.6296898126602173,
"rewards/margins": 0.9214704632759094,
"rewards/rejected": -2.5511598587036133,
"step": 224
},
{
"epoch": 1.0632014176018902,
"grad_norm": 3.8586230520693157,
"learning_rate": 2.356057980867594e-07,
"logits/chosen": -2.547018527984619,
"logits/rejected": -2.5479323863983154,
"logps/chosen": -411.9010009765625,
"logps/rejected": -543.1708984375,
"loss": 0.4834,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.6013555526733398,
"rewards/margins": 1.6427891254425049,
"rewards/rejected": -3.244144916534424,
"step": 225
},
{
"epoch": 1.0679267572356763,
"grad_norm": 3.836310963297125,
"learning_rate": 2.3368913826460742e-07,
"logits/chosen": -2.622213840484619,
"logits/rejected": -2.699857473373413,
"logps/chosen": -383.7267150878906,
"logps/rejected": -406.25152587890625,
"loss": 0.5085,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.4025287628173828,
"rewards/margins": 0.56195068359375,
"rewards/rejected": -1.9644795656204224,
"step": 226
},
{
"epoch": 1.0726520968694624,
"grad_norm": 4.999193459222086,
"learning_rate": 2.3177344078104305e-07,
"logits/chosen": -2.6717772483825684,
"logits/rejected": -2.672889232635498,
"logps/chosen": -346.08544921875,
"logps/rejected": -417.90545654296875,
"loss": 0.4906,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.6554023027420044,
"rewards/margins": 0.8562977313995361,
"rewards/rejected": -2.51170015335083,
"step": 227
},
{
"epoch": 1.0773774365032487,
"grad_norm": 4.6410271396992595,
"learning_rate": 2.2985881866195304e-07,
"logits/chosen": -2.646639347076416,
"logits/rejected": -2.7165169715881348,
"logps/chosen": -355.2235107421875,
"logps/rejected": -395.57550048828125,
"loss": 0.5089,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.3660098314285278,
"rewards/margins": 0.7286862134933472,
"rewards/rejected": -2.094696044921875,
"step": 228
},
{
"epoch": 1.0821027761370348,
"grad_norm": 3.9004711748243412,
"learning_rate": 2.2794538486977765e-07,
"logits/chosen": -2.573826313018799,
"logits/rejected": -2.6808300018310547,
"logps/chosen": -355.0180969238281,
"logps/rejected": -473.90252685546875,
"loss": 0.4922,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.4073047637939453,
"rewards/margins": 1.1750006675720215,
"rewards/rejected": -2.582305431365967,
"step": 229
},
{
"epoch": 1.086828115770821,
"grad_norm": 3.6316010874552704,
"learning_rate": 2.2603325229684628e-07,
"logits/chosen": -2.4409735202789307,
"logits/rejected": -2.4068901538848877,
"logps/chosen": -306.1089172363281,
"logps/rejected": -447.4771423339844,
"loss": 0.4552,
"rewards/accuracies": 0.609375,
"rewards/chosen": -1.4156140089035034,
"rewards/margins": 1.0425467491149902,
"rewards/rejected": -2.458160877227783,
"step": 230
},
{
"epoch": 1.0915534554046071,
"grad_norm": 4.420363258356965,
"learning_rate": 2.2412253375871618e-07,
"logits/chosen": -2.681562662124634,
"logits/rejected": -2.5477468967437744,
"logps/chosen": -338.50775146484375,
"logps/rejected": -500.01568603515625,
"loss": 0.4909,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.4533276557922363,
"rewards/margins": 1.5830377340316772,
"rewards/rejected": -3.036365270614624,
"step": 231
},
{
"epoch": 1.0962787950383934,
"grad_norm": 3.3398932376367116,
"learning_rate": 2.2221334198751717e-07,
"logits/chosen": -2.566534996032715,
"logits/rejected": -2.663203716278076,
"logps/chosen": -423.2144470214844,
"logps/rejected": -525.0465087890625,
"loss": 0.4902,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.552110195159912,
"rewards/margins": 1.6297154426574707,
"rewards/rejected": -3.181825876235962,
"step": 232
},
{
"epoch": 1.1010041346721795,
"grad_norm": 3.258958216613353,
"learning_rate": 2.2030578962529964e-07,
"logits/chosen": -2.533161163330078,
"logits/rejected": -2.598869800567627,
"logps/chosen": -384.85443115234375,
"logps/rejected": -436.78302001953125,
"loss": 0.4943,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.1795471906661987,
"rewards/margins": 0.7562814354896545,
"rewards/rejected": -1.935828685760498,
"step": 233
},
{
"epoch": 1.1057294743059658,
"grad_norm": 4.3977570998277695,
"learning_rate": 2.1839998921738948e-07,
"logits/chosen": -2.682744026184082,
"logits/rejected": -2.8130013942718506,
"logps/chosen": -375.19012451171875,
"logps/rejected": -446.29486083984375,
"loss": 0.4962,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.338588833808899,
"rewards/margins": 0.9564355611801147,
"rewards/rejected": -2.2950241565704346,
"step": 234
},
{
"epoch": 1.1104548139397519,
"grad_norm": 3.711007822076339,
"learning_rate": 2.1649605320574715e-07,
"logits/chosen": -2.770697832107544,
"logits/rejected": -2.7958450317382812,
"logps/chosen": -263.7291259765625,
"logps/rejected": -410.407470703125,
"loss": 0.4984,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.2530757188796997,
"rewards/margins": 1.3193039894104004,
"rewards/rejected": -2.5723795890808105,
"step": 235
},
{
"epoch": 1.1151801535735382,
"grad_norm": 5.205159446883962,
"learning_rate": 2.1459409392233414e-07,
"logits/chosen": -2.509124279022217,
"logits/rejected": -2.4804513454437256,
"logps/chosen": -431.5054931640625,
"logps/rejected": -569.149169921875,
"loss": 0.4849,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.6995575428009033,
"rewards/margins": 1.6153841018676758,
"rewards/rejected": -3.3149421215057373,
"step": 236
},
{
"epoch": 1.1199054932073242,
"grad_norm": 3.6720237761578276,
"learning_rate": 2.1269422358248534e-07,
"logits/chosen": -2.2790334224700928,
"logits/rejected": -2.5470128059387207,
"logps/chosen": -368.87750244140625,
"logps/rejected": -383.09478759765625,
"loss": 0.4829,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.3113298416137695,
"rewards/margins": 0.9517258405685425,
"rewards/rejected": -2.2630558013916016,
"step": 237
},
{
"epoch": 1.1246308328411105,
"grad_norm": 4.38203569485804,
"learning_rate": 2.1079655427828804e-07,
"logits/chosen": -2.484546184539795,
"logits/rejected": -2.513195753097534,
"logps/chosen": -329.1628723144531,
"logps/rejected": -448.17327880859375,
"loss": 0.4783,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.3690838813781738,
"rewards/margins": 1.233577847480774,
"rewards/rejected": -2.602661609649658,
"step": 238
},
{
"epoch": 1.1293561724748966,
"grad_norm": 4.1915678394143665,
"learning_rate": 2.0890119797196904e-07,
"logits/chosen": -2.489327907562256,
"logits/rejected": -2.619502544403076,
"logps/chosen": -367.636474609375,
"logps/rejected": -442.066650390625,
"loss": 0.4619,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.260221004486084,
"rewards/margins": 1.034293293952942,
"rewards/rejected": -2.2945144176483154,
"step": 239
},
{
"epoch": 1.1340815121086827,
"grad_norm": 4.912500043289106,
"learning_rate": 2.0700826648928827e-07,
"logits/chosen": -2.445549964904785,
"logits/rejected": -2.4925429821014404,
"logps/chosen": -403.63580322265625,
"logps/rejected": -492.6475830078125,
"loss": 0.4837,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.4245916604995728,
"rewards/margins": 1.291830062866211,
"rewards/rejected": -2.7164220809936523,
"step": 240
},
{
"epoch": 1.138806851742469,
"grad_norm": 3.973403985509937,
"learning_rate": 2.0511787151294153e-07,
"logits/chosen": -2.6269099712371826,
"logits/rejected": -2.6633949279785156,
"logps/chosen": -362.7601318359375,
"logps/rejected": -465.685791015625,
"loss": 0.4772,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.3287506103515625,
"rewards/margins": 1.2157796621322632,
"rewards/rejected": -2.544530153274536,
"step": 241
},
{
"epoch": 1.1435321913762553,
"grad_norm": 3.706128513697955,
"learning_rate": 2.0323012457597113e-07,
"logits/chosen": -2.6716468334198,
"logits/rejected": -2.5537989139556885,
"logps/chosen": -313.87188720703125,
"logps/rejected": -456.5760498046875,
"loss": 0.4663,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.3276413679122925,
"rewards/margins": 1.286893367767334,
"rewards/rejected": -2.614534854888916,
"step": 242
},
{
"epoch": 1.1482575310100414,
"grad_norm": 4.436897843537744,
"learning_rate": 2.0134513705518544e-07,
"logits/chosen": -2.442168951034546,
"logits/rejected": -2.475062608718872,
"logps/chosen": -349.3672180175781,
"logps/rejected": -437.71075439453125,
"loss": 0.4738,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.4281671047210693,
"rewards/margins": 1.169339656829834,
"rewards/rejected": -2.5975069999694824,
"step": 243
},
{
"epoch": 1.1529828706438274,
"grad_norm": 3.739187250501708,
"learning_rate": 1.9946302016458754e-07,
"logits/chosen": -2.5069191455841064,
"logits/rejected": -2.4427642822265625,
"logps/chosen": -378.0785827636719,
"logps/rejected": -515.2130126953125,
"loss": 0.4537,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.5397615432739258,
"rewards/margins": 1.4216468334197998,
"rewards/rejected": -2.9614078998565674,
"step": 244
},
{
"epoch": 1.1577082102776137,
"grad_norm": 4.618667609463329,
"learning_rate": 1.975838849488139e-07,
"logits/chosen": -2.6840784549713135,
"logits/rejected": -2.608031988143921,
"logps/chosen": -353.92388916015625,
"logps/rejected": -443.1690368652344,
"loss": 0.4789,
"rewards/accuracies": 0.59375,
"rewards/chosen": -1.5656286478042603,
"rewards/margins": 0.7766789793968201,
"rewards/rejected": -2.3423075675964355,
"step": 245
},
{
"epoch": 1.1624335499113998,
"grad_norm": 3.85370799421685,
"learning_rate": 1.957078422765823e-07,
"logits/chosen": -2.512361764907837,
"logits/rejected": -2.664332389831543,
"logps/chosen": -423.3899230957031,
"logps/rejected": -493.901123046875,
"loss": 0.508,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.5689921379089355,
"rewards/margins": 1.2216757535934448,
"rewards/rejected": -2.790667772293091,
"step": 246
},
{
"epoch": 1.167158889545186,
"grad_norm": 4.071196958136995,
"learning_rate": 1.9383500283415127e-07,
"logits/chosen": -2.7708868980407715,
"logits/rejected": -2.928670644760132,
"logps/chosen": -413.68780517578125,
"logps/rejected": -419.13287353515625,
"loss": 0.4674,
"rewards/accuracies": 0.59375,
"rewards/chosen": -1.6208066940307617,
"rewards/margins": 0.8534724116325378,
"rewards/rejected": -2.4742789268493652,
"step": 247
},
{
"epoch": 1.1718842291789722,
"grad_norm": 4.6057615724394445,
"learning_rate": 1.9196547711878882e-07,
"logits/chosen": -2.7272331714630127,
"logits/rejected": -2.7651190757751465,
"logps/chosen": -391.68218994140625,
"logps/rejected": -536.7830810546875,
"loss": 0.4898,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.3025808334350586,
"rewards/margins": 1.699794888496399,
"rewards/rejected": -3.002375602722168,
"step": 248
},
{
"epoch": 1.1766095688127585,
"grad_norm": 4.356687184481579,
"learning_rate": 1.9009937543225393e-07,
"logits/chosen": -2.6314167976379395,
"logits/rejected": -2.5595688819885254,
"logps/chosen": -311.2470703125,
"logps/rejected": -458.4511413574219,
"loss": 0.4912,
"rewards/accuracies": 0.609375,
"rewards/chosen": -1.5766427516937256,
"rewards/margins": 1.1243988275527954,
"rewards/rejected": -2.7010414600372314,
"step": 249
},
{
"epoch": 1.1813349084465445,
"grad_norm": 3.8646816254377168,
"learning_rate": 1.8823680787428804e-07,
"logits/chosen": -2.628770589828491,
"logits/rejected": -2.6632299423217773,
"logps/chosen": -362.2929382324219,
"logps/rejected": -451.97686767578125,
"loss": 0.5168,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.444336175918579,
"rewards/margins": 1.026198387145996,
"rewards/rejected": -2.470534324645996,
"step": 250
},
{
"epoch": 1.1860602480803308,
"grad_norm": 4.972160856261013,
"learning_rate": 1.8637788433611946e-07,
"logits/chosen": -2.6531898975372314,
"logits/rejected": -2.6115176677703857,
"logps/chosen": -374.62554931640625,
"logps/rejected": -556.6736450195312,
"loss": 0.4812,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.495612382888794,
"rewards/margins": 1.505082368850708,
"rewards/rejected": -3.000694751739502,
"step": 251
},
{
"epoch": 1.190785587714117,
"grad_norm": 5.493325018181506,
"learning_rate": 1.8452271449398015e-07,
"logits/chosen": -2.6560559272766113,
"logits/rejected": -2.7018203735351562,
"logps/chosen": -352.3895568847656,
"logps/rejected": -441.25994873046875,
"loss": 0.4814,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.6333725452423096,
"rewards/margins": 1.1176269054412842,
"rewards/rejected": -2.7509994506835938,
"step": 252
},
{
"epoch": 1.1955109273479032,
"grad_norm": 4.859694388307905,
"learning_rate": 1.8267140780263424e-07,
"logits/chosen": -2.634824514389038,
"logits/rejected": -2.638603448867798,
"logps/chosen": -336.9228515625,
"logps/rejected": -454.23883056640625,
"loss": 0.4996,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.460708498954773,
"rewards/margins": 1.1815128326416016,
"rewards/rejected": -2.642221450805664,
"step": 253
},
{
"epoch": 1.2002362669816893,
"grad_norm": 4.23344077907717,
"learning_rate": 1.8082407348892076e-07,
"logits/chosen": -2.581425666809082,
"logits/rejected": -2.4411489963531494,
"logps/chosen": -360.4097595214844,
"logps/rejected": -511.65484619140625,
"loss": 0.4754,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.5460442304611206,
"rewards/margins": 1.2290416955947876,
"rewards/rejected": -2.775085687637329,
"step": 254
},
{
"epoch": 1.2049616066154756,
"grad_norm": 5.021641074779002,
"learning_rate": 1.7898082054530868e-07,
"logits/chosen": -2.5814576148986816,
"logits/rejected": -2.562331199645996,
"logps/chosen": -391.0801696777344,
"logps/rejected": -486.2853698730469,
"loss": 0.4886,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.5095934867858887,
"rewards/margins": 0.9785588979721069,
"rewards/rejected": -2.488152503967285,
"step": 255
},
{
"epoch": 1.2096869462492617,
"grad_norm": 4.032966746831722,
"learning_rate": 1.7714175772346683e-07,
"logits/chosen": -2.73325514793396,
"logits/rejected": -2.8232052326202393,
"logps/chosen": -361.27423095703125,
"logps/rejected": -461.78125,
"loss": 0.5027,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.4946091175079346,
"rewards/margins": 1.1646876335144043,
"rewards/rejected": -2.659296751022339,
"step": 256
},
{
"epoch": 1.2144122858830477,
"grad_norm": 4.348859426367674,
"learning_rate": 1.753069935278477e-07,
"logits/chosen": -2.5508053302764893,
"logits/rejected": -2.6324799060821533,
"logps/chosen": -349.6580810546875,
"logps/rejected": -436.2727966308594,
"loss": 0.4904,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.5449261665344238,
"rewards/margins": 1.1617025136947632,
"rewards/rejected": -2.7066287994384766,
"step": 257
},
{
"epoch": 1.219137625516834,
"grad_norm": 3.7399337348111246,
"learning_rate": 1.7347663620928494e-07,
"logits/chosen": -2.6396665573120117,
"logits/rejected": -2.686690330505371,
"logps/chosen": -373.62762451171875,
"logps/rejected": -456.91876220703125,
"loss": 0.4607,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.5388782024383545,
"rewards/margins": 1.0700188875198364,
"rewards/rejected": -2.6088972091674805,
"step": 258
},
{
"epoch": 1.219137625516834,
"eval_logits/chosen": -2.7951161861419678,
"eval_logits/rejected": -2.8025708198547363,
"eval_logps/chosen": -369.3280029296875,
"eval_logps/rejected": -488.8177490234375,
"eval_loss": 0.48161956667900085,
"eval_rewards/accuracies": 0.6401515007019043,
"eval_rewards/chosen": -1.5809730291366577,
"eval_rewards/margins": 1.307705044746399,
"eval_rewards/rejected": -2.8886778354644775,
"eval_runtime": 225.1752,
"eval_samples_per_second": 16.236,
"eval_steps_per_second": 0.293,
"step": 258
},
{
"epoch": 1.2238629651506203,
"grad_norm": 4.565788447191084,
"learning_rate": 1.7165079375860752e-07,
"logits/chosen": -2.7769393920898438,
"logits/rejected": -2.7542011737823486,
"logps/chosen": -327.9093322753906,
"logps/rejected": -441.0569763183594,
"loss": 0.5122,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.4275953769683838,
"rewards/margins": 1.174314022064209,
"rewards/rejected": -2.6019093990325928,
"step": 259
},
{
"epoch": 1.2285883047844064,
"grad_norm": 3.8161499063717206,
"learning_rate": 1.6982957390026748e-07,
"logits/chosen": -2.5881879329681396,
"logits/rejected": -2.5274972915649414,
"logps/chosen": -380.05670166015625,
"logps/rejected": -548.7481079101562,
"loss": 0.4618,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.7176233530044556,
"rewards/margins": 1.5351811647415161,
"rewards/rejected": -3.2528045177459717,
"step": 260
},
{
"epoch": 1.2333136444181925,
"grad_norm": 4.44964808900954,
"learning_rate": 1.680130840859848e-07,
"logits/chosen": -2.6286840438842773,
"logits/rejected": -2.606605291366577,
"logps/chosen": -325.35833740234375,
"logps/rejected": -435.7322082519531,
"loss": 0.4912,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.2944529056549072,
"rewards/margins": 1.2223491668701172,
"rewards/rejected": -2.5168020725250244,
"step": 261
},
{
"epoch": 1.2380389840519788,
"grad_norm": 4.264258297752704,
"learning_rate": 1.662014314884074e-07,
"logits/chosen": -2.6938886642456055,
"logits/rejected": -2.6410956382751465,
"logps/chosen": -341.89141845703125,
"logps/rejected": -533.8803100585938,
"loss": 0.4583,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.6077433824539185,
"rewards/margins": 1.743915319442749,
"rewards/rejected": -3.351658821105957,
"step": 262
},
{
"epoch": 1.2427643236857648,
"grad_norm": 3.8551209282619374,
"learning_rate": 1.64394722994788e-07,
"logits/chosen": -2.644559383392334,
"logits/rejected": -2.672788381576538,
"logps/chosen": -332.5156555175781,
"logps/rejected": -419.4230041503906,
"loss": 0.4707,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.4292283058166504,
"rewards/margins": 0.7848268151283264,
"rewards/rejected": -2.214055299758911,
"step": 263
},
{
"epoch": 1.2474896633195511,
"grad_norm": 5.862600550730873,
"learning_rate": 1.625930652006782e-07,
"logits/chosen": -2.806763172149658,
"logits/rejected": -2.818608283996582,
"logps/chosen": -332.0267639160156,
"logps/rejected": -421.3615417480469,
"loss": 0.5,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.3579407930374146,
"rewards/margins": 1.1036723852157593,
"rewards/rejected": -2.461613178253174,
"step": 264
},
{
"epoch": 1.2522150029533372,
"grad_norm": 4.952960168127134,
"learning_rate": 1.607965644036386e-07,
"logits/chosen": -2.397037982940674,
"logits/rejected": -2.370586633682251,
"logps/chosen": -380.63433837890625,
"logps/rejected": -567.65478515625,
"loss": 0.4619,
"rewards/accuracies": 0.796875,
"rewards/chosen": -1.3215548992156982,
"rewards/margins": 1.9972357749938965,
"rewards/rejected": -3.318790912628174,
"step": 265
},
{
"epoch": 1.2569403425871235,
"grad_norm": 4.451231917032084,
"learning_rate": 1.5900532659696786e-07,
"logits/chosen": -2.5172245502471924,
"logits/rejected": -2.549943208694458,
"logps/chosen": -310.1299743652344,
"logps/rejected": -405.5203857421875,
"loss": 0.493,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.271753191947937,
"rewards/margins": 0.9632126688957214,
"rewards/rejected": -2.2349658012390137,
"step": 266
},
{
"epoch": 1.2616656822209096,
"grad_norm": 3.9305163776837313,
"learning_rate": 1.5721945746344914e-07,
"logits/chosen": -2.5553438663482666,
"logits/rejected": -2.6015634536743164,
"logps/chosen": -336.09326171875,
"logps/rejected": -444.8936462402344,
"loss": 0.4871,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.417650580406189,
"rewards/margins": 1.3679817914962769,
"rewards/rejected": -2.785632371902466,
"step": 267
},
{
"epoch": 1.2663910218546959,
"grad_norm": 3.688073971085107,
"learning_rate": 1.5543906236911423e-07,
"logits/chosen": -2.798358917236328,
"logits/rejected": -2.803248167037964,
"logps/chosen": -307.88250732421875,
"logps/rejected": -455.02081298828125,
"loss": 0.4952,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.4341715574264526,
"rewards/margins": 1.0945826768875122,
"rewards/rejected": -2.528754234313965,
"step": 268
},
{
"epoch": 1.271116361488482,
"grad_norm": 4.072630703078056,
"learning_rate": 1.5366424635702773e-07,
"logits/chosen": -2.4071907997131348,
"logits/rejected": -2.469078540802002,
"logps/chosen": -344.1185302734375,
"logps/rejected": -481.708740234375,
"loss": 0.4607,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.4194681644439697,
"rewards/margins": 1.5102123022079468,
"rewards/rejected": -2.929680109024048,
"step": 269
},
{
"epoch": 1.2758417011222682,
"grad_norm": 4.509575176085059,
"learning_rate": 1.5189511414108902e-07,
"logits/chosen": -2.6352696418762207,
"logits/rejected": -2.5769996643066406,
"logps/chosen": -299.04486083984375,
"logps/rejected": -428.2113952636719,
"loss": 0.4548,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.261251449584961,
"rewards/margins": 1.3529876470565796,
"rewards/rejected": -2.61423921585083,
"step": 270
},
{
"epoch": 1.2805670407560543,
"grad_norm": 4.337503774611465,
"learning_rate": 1.5013177009985412e-07,
"logits/chosen": -2.492708683013916,
"logits/rejected": -2.5158631801605225,
"logps/chosen": -369.5946960449219,
"logps/rejected": -539.3471069335938,
"loss": 0.4839,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.4575221538543701,
"rewards/margins": 1.7974143028259277,
"rewards/rejected": -3.2549362182617188,
"step": 271
},
{
"epoch": 1.2852923803898406,
"grad_norm": 3.7985527417770797,
"learning_rate": 1.4837431827037786e-07,
"logits/chosen": -2.588874101638794,
"logits/rejected": -2.6791810989379883,
"logps/chosen": -399.30633544921875,
"logps/rejected": -391.2076416015625,
"loss": 0.4848,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.1733232736587524,
"rewards/margins": 0.8876461982727051,
"rewards/rejected": -2.060969591140747,
"step": 272
},
{
"epoch": 1.2900177200236267,
"grad_norm": 4.177927201443006,
"learning_rate": 1.466228623420751e-07,
"logits/chosen": -2.498131036758423,
"logits/rejected": -2.5749406814575195,
"logps/chosen": -323.9309997558594,
"logps/rejected": -450.7756042480469,
"loss": 0.4591,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.2552162408828735,
"rewards/margins": 1.4370646476745605,
"rewards/rejected": -2.6922807693481445,
"step": 273
},
{
"epoch": 1.2947430596574128,
"grad_norm": 3.6814005701875523,
"learning_rate": 1.448775056506036e-07,
"logits/chosen": -2.469701051712036,
"logits/rejected": -2.5528130531311035,
"logps/chosen": -382.1181335449219,
"logps/rejected": -487.5738525390625,
"loss": 0.4747,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.516263723373413,
"rewards/margins": 1.4078060388565063,
"rewards/rejected": -2.92406964302063,
"step": 274
},
{
"epoch": 1.299468399291199,
"grad_norm": 3.7200492832656766,
"learning_rate": 1.4313835117176692e-07,
"logits/chosen": -2.9147932529449463,
"logits/rejected": -2.988351821899414,
"logps/chosen": -376.6600341796875,
"logps/rejected": -456.061279296875,
"loss": 0.4718,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.3018884658813477,
"rewards/margins": 1.2087831497192383,
"rewards/rejected": -2.510671615600586,
"step": 275
},
{
"epoch": 1.3041937389249854,
"grad_norm": 4.704523030577493,
"learning_rate": 1.4140550151543872e-07,
"logits/chosen": -2.5561208724975586,
"logits/rejected": -2.669656276702881,
"logps/chosen": -391.99273681640625,
"logps/rejected": -470.0566711425781,
"loss": 0.4779,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.384698510169983,
"rewards/margins": 1.2832698822021484,
"rewards/rejected": -2.667968511581421,
"step": 276
},
{
"epoch": 1.3089190785587714,
"grad_norm": 3.890211089107488,
"learning_rate": 1.3967905891950936e-07,
"logits/chosen": -2.525979995727539,
"logits/rejected": -2.5187900066375732,
"logps/chosen": -329.0636291503906,
"logps/rejected": -518.5650024414062,
"loss": 0.4978,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.3041456937789917,
"rewards/margins": 1.7594897747039795,
"rewards/rejected": -3.0636353492736816,
"step": 277
},
{
"epoch": 1.3136444181925575,
"grad_norm": 3.900954884610721,
"learning_rate": 1.3795912524385322e-07,
"logits/chosen": -2.6802122592926025,
"logits/rejected": -2.7386527061462402,
"logps/chosen": -394.46905517578125,
"logps/rejected": -530.927978515625,
"loss": 0.4819,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.5945687294006348,
"rewards/margins": 1.4686897993087769,
"rewards/rejected": -3.063258647918701,
"step": 278
},
{
"epoch": 1.3183697578263438,
"grad_norm": 5.012829563655198,
"learning_rate": 1.3624580196431952e-07,
"logits/chosen": -2.735568046569824,
"logits/rejected": -2.7510178089141846,
"logps/chosen": -375.1429748535156,
"logps/rejected": -484.94390869140625,
"loss": 0.4769,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.396562099456787,
"rewards/margins": 1.301504135131836,
"rewards/rejected": -2.698065996170044,
"step": 279
},
{
"epoch": 1.3230950974601299,
"grad_norm": 6.486088065803327,
"learning_rate": 1.3453919016674483e-07,
"logits/chosen": -2.5972790718078613,
"logits/rejected": -2.6828713417053223,
"logps/chosen": -317.81280517578125,
"logps/rejected": -378.3294982910156,
"loss": 0.4867,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.1421467065811157,
"rewards/margins": 1.074657917022705,
"rewards/rejected": -2.2168045043945312,
"step": 280
},
{
"epoch": 1.3278204370939162,
"grad_norm": 4.08450025143174,
"learning_rate": 1.328393905409892e-07,
"logits/chosen": -2.4976108074188232,
"logits/rejected": -2.5105791091918945,
"logps/chosen": -394.51141357421875,
"logps/rejected": -482.7172546386719,
"loss": 0.4878,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.521199107170105,
"rewards/margins": 1.0865689516067505,
"rewards/rejected": -2.6077680587768555,
"step": 281
},
{
"epoch": 1.3325457767277022,
"grad_norm": 4.246089354170224,
"learning_rate": 1.3114650337499578e-07,
"logits/chosen": -2.629361152648926,
"logits/rejected": -2.595665693283081,
"logps/chosen": -334.8639831542969,
"logps/rejected": -423.8031921386719,
"loss": 0.4994,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.5956919193267822,
"rewards/margins": 0.8941479921340942,
"rewards/rejected": -2.489840030670166,
"step": 282
},
{
"epoch": 1.3372711163614885,
"grad_norm": 3.6260185771038103,
"learning_rate": 1.2946062854887314e-07,
"logits/chosen": -2.430432081222534,
"logits/rejected": -2.3944997787475586,
"logps/chosen": -371.2943115234375,
"logps/rejected": -522.4259033203125,
"loss": 0.4836,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.5247899293899536,
"rewards/margins": 1.4179781675338745,
"rewards/rejected": -2.942767858505249,
"step": 283
},
{
"epoch": 1.3419964559952746,
"grad_norm": 4.374381091295728,
"learning_rate": 1.2778186552900316e-07,
"logits/chosen": -2.7117838859558105,
"logits/rejected": -2.761711597442627,
"logps/chosen": -400.6455078125,
"logps/rejected": -503.24371337890625,
"loss": 0.484,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.717864990234375,
"rewards/margins": 1.5052025318145752,
"rewards/rejected": -3.2230677604675293,
"step": 284
},
{
"epoch": 1.346721795629061,
"grad_norm": 3.8541867456367847,
"learning_rate": 1.261103133621718e-07,
"logits/chosen": -2.594248056411743,
"logits/rejected": -2.603362798690796,
"logps/chosen": -360.4008483886719,
"logps/rejected": -533.9642944335938,
"loss": 0.4763,
"rewards/accuracies": 0.546875,
"rewards/chosen": -1.626147747039795,
"rewards/margins": 1.3955036401748657,
"rewards/rejected": -3.021651268005371,
"step": 285
},
{
"epoch": 1.351447135262847,
"grad_norm": 3.8033607627593815,
"learning_rate": 1.2444607066972583e-07,
"logits/chosen": -2.385476589202881,
"logits/rejected": -2.4700021743774414,
"logps/chosen": -379.0484924316406,
"logps/rejected": -446.36932373046875,
"loss": 0.4707,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.5443971157073975,
"rewards/margins": 1.0206142663955688,
"rewards/rejected": -2.5650112628936768,
"step": 286
},
{
"epoch": 1.356172474896633,
"grad_norm": 3.9322741636171017,
"learning_rate": 1.227892356417542e-07,
"logits/chosen": -2.8771088123321533,
"logits/rejected": -2.838731050491333,
"logps/chosen": -366.8982849121094,
"logps/rejected": -519.6290893554688,
"loss": 0.4657,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.7020255327224731,
"rewards/margins": 1.5927929878234863,
"rewards/rejected": -3.29481840133667,
"step": 287
},
{
"epoch": 1.3608978145304194,
"grad_norm": 4.441774997254426,
"learning_rate": 1.211399060312943e-07,
"logits/chosen": -2.6161060333251953,
"logits/rejected": -2.6837587356567383,
"logps/chosen": -333.47747802734375,
"logps/rejected": -396.8973388671875,
"loss": 0.4805,
"rewards/accuracies": 0.5625,
"rewards/chosen": -1.4345372915267944,
"rewards/margins": 0.6499552726745605,
"rewards/rejected": -2.0844926834106445,
"step": 288
},
{
"epoch": 1.3656231541642057,
"grad_norm": 4.102742239919904,
"learning_rate": 1.1949817914856539e-07,
"logits/chosen": -2.6814827919006348,
"logits/rejected": -2.6281232833862305,
"logps/chosen": -309.97998046875,
"logps/rejected": -483.1641540527344,
"loss": 0.4876,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.347307801246643,
"rewards/margins": 1.4000307321548462,
"rewards/rejected": -2.7473385334014893,
"step": 289
},
{
"epoch": 1.3703484937979917,
"grad_norm": 3.510044851118614,
"learning_rate": 1.1786415185522644e-07,
"logits/chosen": -2.4141433238983154,
"logits/rejected": -2.440483331680298,
"logps/chosen": -371.58050537109375,
"logps/rejected": -485.5816345214844,
"loss": 0.462,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.6337575912475586,
"rewards/margins": 1.3851597309112549,
"rewards/rejected": -3.0189173221588135,
"step": 290
},
{
"epoch": 1.3750738334317778,
"grad_norm": 4.439828210645555,
"learning_rate": 1.1623792055866182e-07,
"logits/chosen": -2.9460198879241943,
"logits/rejected": -2.792397975921631,
"logps/chosen": -306.8702392578125,
"logps/rejected": -499.4562072753906,
"loss": 0.4817,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.5415101051330566,
"rewards/margins": 1.6944992542266846,
"rewards/rejected": -3.2360095977783203,
"step": 291
},
{
"epoch": 1.379799173065564,
"grad_norm": 3.808563735975941,
"learning_rate": 1.1461958120629345e-07,
"logits/chosen": -2.601799488067627,
"logits/rejected": -2.6055119037628174,
"logps/chosen": -349.54119873046875,
"logps/rejected": -453.0423278808594,
"loss": 0.4919,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.38273286819458,
"rewards/margins": 1.1617132425308228,
"rewards/rejected": -2.5444459915161133,
"step": 292
},
{
"epoch": 1.3845245126993504,
"grad_norm": 3.614960145049863,
"learning_rate": 1.1300922927991912e-07,
"logits/chosen": -2.3846492767333984,
"logits/rejected": -2.35745906829834,
"logps/chosen": -377.6761169433594,
"logps/rejected": -496.96405029296875,
"loss": 0.4831,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.5498136281967163,
"rewards/margins": 1.1908408403396606,
"rewards/rejected": -2.740654468536377,
"step": 293
},
{
"epoch": 1.3892498523331365,
"grad_norm": 5.27301677074656,
"learning_rate": 1.1140695979008017e-07,
"logits/chosen": -2.359983205795288,
"logits/rejected": -2.346726894378662,
"logps/chosen": -323.6959533691406,
"logps/rejected": -452.9242248535156,
"loss": 0.4725,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.4340472221374512,
"rewards/margins": 1.28817880153656,
"rewards/rejected": -2.722226142883301,
"step": 294
},
{
"epoch": 1.3939751919669225,
"grad_norm": 3.7242744476436704,
"learning_rate": 1.0981286727045483e-07,
"logits/chosen": -2.4720327854156494,
"logits/rejected": -2.353053331375122,
"logps/chosen": -352.6996765136719,
"logps/rejected": -499.5018615722656,
"loss": 0.476,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.4100433588027954,
"rewards/margins": 1.4266828298568726,
"rewards/rejected": -2.836726188659668,
"step": 295
},
{
"epoch": 1.3987005316007088,
"grad_norm": 4.955539561562781,
"learning_rate": 1.0822704577228131e-07,
"logits/chosen": -2.642940044403076,
"logits/rejected": -2.639770746231079,
"logps/chosen": -338.60595703125,
"logps/rejected": -491.56805419921875,
"loss": 0.4635,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.4772026538848877,
"rewards/margins": 1.517985224723816,
"rewards/rejected": -2.995187759399414,
"step": 296
},
{
"epoch": 1.403425871234495,
"grad_norm": 4.549161233388332,
"learning_rate": 1.0664958885880901e-07,
"logits/chosen": -2.5309808254241943,
"logits/rejected": -2.6360573768615723,
"logps/chosen": -335.5814514160156,
"logps/rejected": -437.315673828125,
"loss": 0.4888,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.466670036315918,
"rewards/margins": 1.2143834829330444,
"rewards/rejected": -2.681053638458252,
"step": 297
},
{
"epoch": 1.4081512108682812,
"grad_norm": 3.7991551857385533,
"learning_rate": 1.0508058959977756e-07,
"logits/chosen": -2.673210382461548,
"logits/rejected": -2.5948569774627686,
"logps/chosen": -333.4312744140625,
"logps/rejected": -518.317626953125,
"loss": 0.47,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.2860051393508911,
"rewards/margins": 1.857041358947754,
"rewards/rejected": -3.1430463790893555,
"step": 298
},
{
"epoch": 1.4128765505020673,
"grad_norm": 4.382508044742225,
"learning_rate": 1.0352014056592653e-07,
"logits/chosen": -2.746319055557251,
"logits/rejected": -2.8318300247192383,
"logps/chosen": -361.51220703125,
"logps/rejected": -441.3756103515625,
"loss": 0.492,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.3820509910583496,
"rewards/margins": 1.1350202560424805,
"rewards/rejected": -2.517071485519409,
"step": 299
},
{
"epoch": 1.4176018901358536,
"grad_norm": 4.674467616661016,
"learning_rate": 1.0196833382353303e-07,
"logits/chosen": -2.731412887573242,
"logits/rejected": -2.6908507347106934,
"logps/chosen": -322.404052734375,
"logps/rejected": -462.2546081542969,
"loss": 0.4942,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.4098780155181885,
"rewards/margins": 1.386682391166687,
"rewards/rejected": -2.796560287475586,
"step": 300
},
{
"epoch": 1.4223272297696397,
"grad_norm": 4.067329027842824,
"learning_rate": 1.0042526092898049e-07,
"logits/chosen": -2.8876852989196777,
"logits/rejected": -2.7597179412841797,
"logps/chosen": -328.04437255859375,
"logps/rejected": -431.107666015625,
"loss": 0.5068,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.4123650789260864,
"rewards/margins": 1.0988068580627441,
"rewards/rejected": -2.511171817779541,
"step": 301
},
{
"epoch": 1.4223272297696397,
"eval_logits/chosen": -2.751302719116211,
"eval_logits/rejected": -2.7585830688476562,
"eval_logps/chosen": -369.27899169921875,
"eval_logps/rejected": -500.55902099609375,
"eval_loss": 0.47641003131866455,
"eval_rewards/accuracies": 0.6401515007019043,
"eval_rewards/chosen": -1.5804827213287354,
"eval_rewards/margins": 1.4256082773208618,
"eval_rewards/rejected": -3.0060908794403076,
"eval_runtime": 225.4536,
"eval_samples_per_second": 16.216,
"eval_steps_per_second": 0.293,
"step": 301
},
{
"epoch": 1.427052569403426,
"grad_norm": 4.636151437447421,
"learning_rate": 9.889101292335625e-08,
"logits/chosen": -2.6484196186065674,
"logits/rejected": -2.6770687103271484,
"logps/chosen": -402.7945251464844,
"logps/rejected": -453.43646240234375,
"loss": 0.4782,
"rewards/accuracies": 0.796875,
"rewards/chosen": -1.1831938028335571,
"rewards/margins": 1.1924808025360107,
"rewards/rejected": -2.3756744861602783,
"step": 302
},
{
"epoch": 1.431777909037212,
"grad_norm": 4.004369409074385,
"learning_rate": 9.736568032708068e-08,
"logits/chosen": -2.5602633953094482,
"logits/rejected": -2.6748807430267334,
"logps/chosen": -366.6900634765625,
"logps/rejected": -485.0727844238281,
"loss": 0.482,
"rewards/accuracies": 0.578125,
"rewards/chosen": -1.4285987615585327,
"rewards/margins": 1.416813611984253,
"rewards/rejected": -2.845412254333496,
"step": 303
},
{
"epoch": 1.436503248670998,
"grad_norm": 4.878881366827008,
"learning_rate": 9.584935313456596e-08,
"logits/chosen": -2.231307029724121,
"logits/rejected": -2.287929058074951,
"logps/chosen": -352.46282958984375,
"logps/rejected": -453.6793212890625,
"loss": 0.4942,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.5504112243652344,
"rewards/margins": 1.3251781463623047,
"rewards/rejected": -2.875589609146118,
"step": 304
},
{
"epoch": 1.4412285883047844,
"grad_norm": 4.241730784056723,
"learning_rate": 9.4342120808907e-08,
"logits/chosen": -2.703420639038086,
"logits/rejected": -2.6591320037841797,
"logps/chosen": -383.8643493652344,
"logps/rejected": -594.2821655273438,
"loss": 0.4792,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.5567654371261597,
"rewards/margins": 2.0447824001312256,
"rewards/rejected": -3.6015477180480957,
"step": 305
},
{
"epoch": 1.4459539279385707,
"grad_norm": 3.8843212431707284,
"learning_rate": 9.284407227660249e-08,
"logits/chosen": -2.8023083209991455,
"logits/rejected": -2.7946949005126953,
"logps/chosen": -337.86090087890625,
"logps/rejected": -434.48260498046875,
"loss": 0.4894,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.2694463729858398,
"rewards/margins": 1.1360807418823242,
"rewards/rejected": -2.405527114868164,
"step": 306
},
{
"epoch": 1.4506792675723568,
"grad_norm": 3.978483022316159,
"learning_rate": 9.13552959223089e-08,
"logits/chosen": -2.676576852798462,
"logits/rejected": -2.5387959480285645,
"logps/chosen": -313.46954345703125,
"logps/rejected": -421.6120300292969,
"loss": 0.4765,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.3958847522735596,
"rewards/margins": 0.8875546455383301,
"rewards/rejected": -2.2834396362304688,
"step": 307
},
{
"epoch": 1.4554046072061428,
"grad_norm": 4.395780053942796,
"learning_rate": 8.987587958362516e-08,
"logits/chosen": -2.856872320175171,
"logits/rejected": -2.8188376426696777,
"logps/chosen": -371.1302795410156,
"logps/rejected": -465.9201354980469,
"loss": 0.4681,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.4001535177230835,
"rewards/margins": 1.085311770439148,
"rewards/rejected": -2.4854652881622314,
"step": 308
},
{
"epoch": 1.4601299468399291,
"grad_norm": 3.8656297909933772,
"learning_rate": 8.840591054591096e-08,
"logits/chosen": -2.5090444087982178,
"logits/rejected": -2.6375505924224854,
"logps/chosen": -410.7342529296875,
"logps/rejected": -467.5154724121094,
"loss": 0.4832,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.2629905939102173,
"rewards/margins": 1.3948893547058105,
"rewards/rejected": -2.6578800678253174,
"step": 309
},
{
"epoch": 1.4648552864737154,
"grad_norm": 4.334534835752672,
"learning_rate": 8.694547553713618e-08,
"logits/chosen": -2.759681224822998,
"logits/rejected": -2.73026180267334,
"logps/chosen": -355.33624267578125,
"logps/rejected": -529.0985107421875,
"loss": 0.4896,
"rewards/accuracies": 0.796875,
"rewards/chosen": -1.521816372871399,
"rewards/margins": 1.6343505382537842,
"rewards/rejected": -3.1561670303344727,
"step": 310
},
{
"epoch": 1.4695806261075015,
"grad_norm": 4.853059543604455,
"learning_rate": 8.54946607227644e-08,
"logits/chosen": -2.4591803550720215,
"logits/rejected": -2.5936403274536133,
"logps/chosen": -391.37298583984375,
"logps/rejected": -434.3221435546875,
"loss": 0.4857,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.5846033096313477,
"rewards/margins": 1.0046393871307373,
"rewards/rejected": -2.589242696762085,
"step": 311
},
{
"epoch": 1.4743059657412876,
"grad_norm": 5.028900388618828,
"learning_rate": 8.405355170066925e-08,
"logits/chosen": -2.7181220054626465,
"logits/rejected": -2.590919017791748,
"logps/chosen": -380.94818115234375,
"logps/rejected": -548.143798828125,
"loss": 0.4851,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.6279135942459106,
"rewards/margins": 1.5961647033691406,
"rewards/rejected": -3.2240781784057617,
"step": 312
},
{
"epoch": 1.4790313053750739,
"grad_norm": 4.261294839829192,
"learning_rate": 8.262223349608366e-08,
"logits/chosen": -2.7533867359161377,
"logits/rejected": -2.8102259635925293,
"logps/chosen": -394.7510681152344,
"logps/rejected": -499.4160461425781,
"loss": 0.4965,
"rewards/accuracies": 0.546875,
"rewards/chosen": -1.5835039615631104,
"rewards/margins": 1.1018283367156982,
"rewards/rejected": -2.6853325366973877,
"step": 313
},
{
"epoch": 1.48375664500886,
"grad_norm": 4.418632236406977,
"learning_rate": 8.120079055658402e-08,
"logits/chosen": -2.642446279525757,
"logits/rejected": -2.6767466068267822,
"logps/chosen": -323.5433349609375,
"logps/rejected": -469.7967834472656,
"loss": 0.475,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.2628949880599976,
"rewards/margins": 1.5510131120681763,
"rewards/rejected": -2.813908100128174,
"step": 314
},
{
"epoch": 1.4884819846426462,
"grad_norm": 4.312754452868253,
"learning_rate": 7.978930674710719e-08,
"logits/chosen": -2.4338035583496094,
"logits/rejected": -2.444002151489258,
"logps/chosen": -374.0926513671875,
"logps/rejected": -496.3292236328125,
"loss": 0.4875,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.539766788482666,
"rewards/margins": 1.5390400886535645,
"rewards/rejected": -3.0788071155548096,
"step": 315
},
{
"epoch": 1.4932073242764323,
"grad_norm": 4.036827603807639,
"learning_rate": 7.838786534500269e-08,
"logits/chosen": -2.7101027965545654,
"logits/rejected": -2.7315640449523926,
"logps/chosen": -367.41107177734375,
"logps/rejected": -477.13916015625,
"loss": 0.4933,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.501757025718689,
"rewards/margins": 1.17995023727417,
"rewards/rejected": -2.6817073822021484,
"step": 316
},
{
"epoch": 1.4979326639102186,
"grad_norm": 4.1034959013822645,
"learning_rate": 7.699654903511971e-08,
"logits/chosen": -2.4980247020721436,
"logits/rejected": -2.569985866546631,
"logps/chosen": -306.99346923828125,
"logps/rejected": -443.2658996582031,
"loss": 0.4761,
"rewards/accuracies": 0.828125,
"rewards/chosen": -1.0971927642822266,
"rewards/margins": 1.3962390422821045,
"rewards/rejected": -2.49343204498291,
"step": 317
},
{
"epoch": 1.5026580035440047,
"grad_norm": 3.8908709553914083,
"learning_rate": 7.561543990492803e-08,
"logits/chosen": -2.5545809268951416,
"logits/rejected": -2.7232065200805664,
"logps/chosen": -382.45367431640625,
"logps/rejected": -496.840087890625,
"loss": 0.4696,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.5967504978179932,
"rewards/margins": 1.5658732652664185,
"rewards/rejected": -3.162623643875122,
"step": 318
},
{
"epoch": 1.507383343177791,
"grad_norm": 4.1097836221394095,
"learning_rate": 7.424461943967555e-08,
"logits/chosen": -2.5873563289642334,
"logits/rejected": -2.7303099632263184,
"logps/chosen": -394.63916015625,
"logps/rejected": -551.1517333984375,
"loss": 0.4793,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.5619065761566162,
"rewards/margins": 1.4349513053894043,
"rewards/rejected": -2.9968576431274414,
"step": 319
},
{
"epoch": 1.512108682811577,
"grad_norm": 3.8818233399377373,
"learning_rate": 7.288416851758016e-08,
"logits/chosen": -2.638657808303833,
"logits/rejected": -2.547767162322998,
"logps/chosen": -392.5679626464844,
"logps/rejected": -627.8018188476562,
"loss": 0.4782,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.5840284824371338,
"rewards/margins": 2.276487112045288,
"rewards/rejected": -3.860515594482422,
"step": 320
},
{
"epoch": 1.5168340224453631,
"grad_norm": 4.298412046444391,
"learning_rate": 7.153416740505814e-08,
"logits/chosen": -2.473698854446411,
"logits/rejected": -2.580441951751709,
"logps/chosen": -402.406982421875,
"logps/rejected": -486.8596496582031,
"loss": 0.4701,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.5713368654251099,
"rewards/margins": 1.1730860471725464,
"rewards/rejected": -2.7444231510162354,
"step": 321
},
{
"epoch": 1.5215593620791494,
"grad_norm": 3.844411445149988,
"learning_rate": 7.01946957519886e-08,
"logits/chosen": -2.414036273956299,
"logits/rejected": -2.5014243125915527,
"logps/chosen": -408.16455078125,
"logps/rejected": -474.1005859375,
"loss": 0.4766,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.5195822715759277,
"rewards/margins": 1.162341594696045,
"rewards/rejected": -2.6819238662719727,
"step": 322
},
{
"epoch": 1.5262847017129357,
"grad_norm": 4.081324399203426,
"learning_rate": 6.88658325870138e-08,
"logits/chosen": -2.6832616329193115,
"logits/rejected": -2.7357096672058105,
"logps/chosen": -385.5028076171875,
"logps/rejected": -469.4779052734375,
"loss": 0.4665,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.399859070777893,
"rewards/margins": 1.3238856792449951,
"rewards/rejected": -2.7237446308135986,
"step": 323
},
{
"epoch": 1.5310100413467218,
"grad_norm": 5.228883712742776,
"learning_rate": 6.754765631287695e-08,
"logits/chosen": -2.544619560241699,
"logits/rejected": -2.655355453491211,
"logps/chosen": -355.68328857421875,
"logps/rejected": -468.37738037109375,
"loss": 0.4799,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.6572238206863403,
"rewards/margins": 1.6221368312835693,
"rewards/rejected": -3.27936053276062,
"step": 324
},
{
"epoch": 1.5357353809805079,
"grad_norm": 3.85359128112212,
"learning_rate": 6.62402447017959e-08,
"logits/chosen": -2.255566358566284,
"logits/rejected": -2.2861790657043457,
"logps/chosen": -374.5622863769531,
"logps/rejected": -525.822265625,
"loss": 0.4804,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.5297114849090576,
"rewards/margins": 1.5574190616607666,
"rewards/rejected": -3.087130546569824,
"step": 325
},
{
"epoch": 1.5404607206142942,
"grad_norm": 3.9384759163947733,
"learning_rate": 6.494367489087488e-08,
"logits/chosen": -2.310734987258911,
"logits/rejected": -2.331479072570801,
"logps/chosen": -353.3757629394531,
"logps/rejected": -456.4471740722656,
"loss": 0.4973,
"rewards/accuracies": 0.796875,
"rewards/chosen": -1.2544013261795044,
"rewards/margins": 0.9933812618255615,
"rewards/rejected": -2.2477827072143555,
"step": 326
},
{
"epoch": 1.5451860602480805,
"grad_norm": 4.707537372396568,
"learning_rate": 6.365802337755364e-08,
"logits/chosen": -2.5162229537963867,
"logits/rejected": -2.5781514644622803,
"logps/chosen": -359.474365234375,
"logps/rejected": -459.92413330078125,
"loss": 0.4737,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.3576124906539917,
"rewards/margins": 1.2539989948272705,
"rewards/rejected": -2.6116113662719727,
"step": 327
},
{
"epoch": 1.5499113998818665,
"grad_norm": 3.5347435278013024,
"learning_rate": 6.238336601509364e-08,
"logits/chosen": -2.4307329654693604,
"logits/rejected": -2.413248300552368,
"logps/chosen": -346.82537841796875,
"logps/rejected": -487.83111572265625,
"loss": 0.4487,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.5922647714614868,
"rewards/margins": 1.7097985744476318,
"rewards/rejected": -3.302063465118408,
"step": 328
},
{
"epoch": 1.5546367395156526,
"grad_norm": 4.5192043219385925,
"learning_rate": 6.111977800810316e-08,
"logits/chosen": -2.4796946048736572,
"logits/rejected": -2.3877992630004883,
"logps/chosen": -332.8540344238281,
"logps/rejected": -508.25360107421875,
"loss": 0.4836,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.7624458074569702,
"rewards/margins": 1.4423408508300781,
"rewards/rejected": -3.204786777496338,
"step": 329
},
{
"epoch": 1.559362079149439,
"grad_norm": 4.386475280759837,
"learning_rate": 5.986733390809993e-08,
"logits/chosen": -2.400326728820801,
"logits/rejected": -2.2906858921051025,
"logps/chosen": -386.3865966796875,
"logps/rejected": -556.1158447265625,
"loss": 0.4623,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.8006043434143066,
"rewards/margins": 1.7274423837661743,
"rewards/rejected": -3.5280466079711914,
"step": 330
},
{
"epoch": 1.564087418783225,
"grad_norm": 4.101101033266814,
"learning_rate": 5.862610760911257e-08,
"logits/chosen": -2.4113216400146484,
"logits/rejected": -2.3908936977386475,
"logps/chosen": -364.8887939453125,
"logps/rejected": -459.69854736328125,
"loss": 0.4372,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.4839426279067993,
"rewards/margins": 1.3043051958084106,
"rewards/rejected": -2.78824782371521,
"step": 331
},
{
"epoch": 1.5688127584170113,
"grad_norm": 4.330222141211974,
"learning_rate": 5.739617234332131e-08,
"logits/chosen": -2.6859869956970215,
"logits/rejected": -2.7122979164123535,
"logps/chosen": -405.88018798828125,
"logps/rejected": -452.80926513671875,
"loss": 0.4648,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.5999912023544312,
"rewards/margins": 0.9445231556892395,
"rewards/rejected": -2.5445144176483154,
"step": 332
},
{
"epoch": 1.5735380980507974,
"grad_norm": 5.485683540439284,
"learning_rate": 5.6177600676736656e-08,
"logits/chosen": -2.5637035369873047,
"logits/rejected": -2.5589380264282227,
"logps/chosen": -393.2574462890625,
"logps/rejected": -529.729248046875,
"loss": 0.5169,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.7628885507583618,
"rewards/margins": 1.4583940505981445,
"rewards/rejected": -3.221282482147217,
"step": 333
},
{
"epoch": 1.5782634376845834,
"grad_norm": 3.9160914653232983,
"learning_rate": 5.4970464504918654e-08,
"logits/chosen": -2.7670090198516846,
"logits/rejected": -2.7049059867858887,
"logps/chosen": -369.9678039550781,
"logps/rejected": -494.07196044921875,
"loss": 0.4565,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.4675428867340088,
"rewards/margins": 1.4338531494140625,
"rewards/rejected": -2.901395797729492,
"step": 334
},
{
"epoch": 1.5829887773183697,
"grad_norm": 3.637994981997343,
"learning_rate": 5.37748350487344e-08,
"logits/chosen": -2.6616098880767822,
"logits/rejected": -2.6261179447174072,
"logps/chosen": -340.9493408203125,
"logps/rejected": -505.6307067871094,
"loss": 0.4528,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.4446711540222168,
"rewards/margins": 1.6602320671081543,
"rewards/rejected": -3.10490345954895,
"step": 335
},
{
"epoch": 1.587714116952156,
"grad_norm": 3.742063853340133,
"learning_rate": 5.2590782850156667e-08,
"logits/chosen": -2.755837917327881,
"logits/rejected": -2.6713008880615234,
"logps/chosen": -421.58154296875,
"logps/rejected": -625.4422607421875,
"loss": 0.4608,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.701414704322815,
"rewards/margins": 1.9440244436264038,
"rewards/rejected": -3.6454391479492188,
"step": 336
},
{
"epoch": 1.592439456585942,
"grad_norm": 3.5742676774235154,
"learning_rate": 5.14183777681014e-08,
"logits/chosen": -2.466548442840576,
"logits/rejected": -2.5580849647521973,
"logps/chosen": -414.5927734375,
"logps/rejected": -506.8092956542969,
"loss": 0.5037,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.518671989440918,
"rewards/margins": 1.1154457330703735,
"rewards/rejected": -2.634117603302002,
"step": 337
},
{
"epoch": 1.5971647962197282,
"grad_norm": 3.6653171076626205,
"learning_rate": 5.0257688974306436e-08,
"logits/chosen": -2.8738627433776855,
"logits/rejected": -3.015981435775757,
"logps/chosen": -381.5254211425781,
"logps/rejected": -457.140869140625,
"loss": 0.4883,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.607730507850647,
"rewards/margins": 1.3426533937454224,
"rewards/rejected": -2.9503836631774902,
"step": 338
},
{
"epoch": 1.6018901358535145,
"grad_norm": 3.837456752855929,
"learning_rate": 4.910878494925008e-08,
"logits/chosen": -2.6002864837646484,
"logits/rejected": -2.6144702434539795,
"logps/chosen": -402.1292724609375,
"logps/rejected": -593.8958740234375,
"loss": 0.4436,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.5465564727783203,
"rewards/margins": 1.739166498184204,
"rewards/rejected": -3.2857229709625244,
"step": 339
},
{
"epoch": 1.6066154754873008,
"grad_norm": 3.9821598438730565,
"learning_rate": 4.7971733478111094e-08,
"logits/chosen": -2.5669634342193604,
"logits/rejected": -2.5983939170837402,
"logps/chosen": -384.938720703125,
"logps/rejected": -557.4461669921875,
"loss": 0.471,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.5989502668380737,
"rewards/margins": 1.8718353509902954,
"rewards/rejected": -3.4707858562469482,
"step": 340
},
{
"epoch": 1.6113408151210868,
"grad_norm": 4.1505854551229735,
"learning_rate": 4.684660164676896e-08,
"logits/chosen": -2.4149627685546875,
"logits/rejected": -2.3984737396240234,
"logps/chosen": -367.2955322265625,
"logps/rejected": -546.0213623046875,
"loss": 0.4849,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.5166963338851929,
"rewards/margins": 1.747159719467163,
"rewards/rejected": -3.2638559341430664,
"step": 341
},
{
"epoch": 1.616066154754873,
"grad_norm": 4.195319716535671,
"learning_rate": 4.5733455837846325e-08,
"logits/chosen": -2.6863296031951904,
"logits/rejected": -2.740741014480591,
"logps/chosen": -388.5705261230469,
"logps/rejected": -527.72265625,
"loss": 0.4803,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.659536600112915,
"rewards/margins": 1.501134991645813,
"rewards/rejected": -3.1606712341308594,
"step": 342
},
{
"epoch": 1.6207914943886592,
"grad_norm": 4.414273407573815,
"learning_rate": 4.4632361726791914e-08,
"logits/chosen": -2.6036033630371094,
"logits/rejected": -2.6751937866210938,
"logps/chosen": -401.4815979003906,
"logps/rejected": -471.8222961425781,
"loss": 0.5018,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.435232162475586,
"rewards/margins": 1.3340187072753906,
"rewards/rejected": -2.7692506313323975,
"step": 343
},
{
"epoch": 1.6255168340224455,
"grad_norm": 5.041321789664628,
"learning_rate": 4.354338427800619e-08,
"logits/chosen": -2.6457765102386475,
"logits/rejected": -2.5664923191070557,
"logps/chosen": -325.753662109375,
"logps/rejected": -515.0615844726562,
"loss": 0.4724,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.4971141815185547,
"rewards/margins": 1.5831940174102783,
"rewards/rejected": -3.080307960510254,
"step": 344
},
{
"epoch": 1.6255168340224455,
"eval_logits/chosen": -2.623720645904541,
"eval_logits/rejected": -2.6295571327209473,
"eval_logps/chosen": -379.5492858886719,
"eval_logps/rejected": -517.3630981445312,
"eval_loss": 0.47301965951919556,
"eval_rewards/accuracies": 0.6382575631141663,
"eval_rewards/chosen": -1.683185338973999,
"eval_rewards/margins": 1.4909465312957764,
"eval_rewards/rejected": -3.1741318702697754,
"eval_runtime": 225.158,
"eval_samples_per_second": 16.237,
"eval_steps_per_second": 0.293,
"step": 344
},
{
"epoch": 1.6302421736562316,
"grad_norm": 3.9177181536561396,
"learning_rate": 4.246658774100803e-08,
"logits/chosen": -2.5313777923583984,
"logits/rejected": -2.6277084350585938,
"logps/chosen": -407.18621826171875,
"logps/rejected": -496.5007629394531,
"loss": 0.4807,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.6058672666549683,
"rewards/margins": 1.1191009283065796,
"rewards/rejected": -2.724968194961548,
"step": 345
},
{
"epoch": 1.6349675132900177,
"grad_norm": 3.841031231021936,
"learning_rate": 4.140203564664421e-08,
"logits/chosen": -2.6209938526153564,
"logits/rejected": -2.650219678878784,
"logps/chosen": -361.2805480957031,
"logps/rejected": -478.2210998535156,
"loss": 0.4595,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.4036047458648682,
"rewards/margins": 1.4318450689315796,
"rewards/rejected": -2.8354499340057373,
"step": 346
},
{
"epoch": 1.6396928529238037,
"grad_norm": 3.522512280432432,
"learning_rate": 4.0349790803341274e-08,
"logits/chosen": -2.8207521438598633,
"logits/rejected": -2.7012057304382324,
"logps/chosen": -348.5565490722656,
"logps/rejected": -503.2954406738281,
"loss": 0.4711,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.5178475379943848,
"rewards/margins": 1.5804749727249146,
"rewards/rejected": -3.0983223915100098,
"step": 347
},
{
"epoch": 1.64441819255759,
"grad_norm": 4.0111825885134245,
"learning_rate": 3.930991529339936e-08,
"logits/chosen": -2.629011631011963,
"logits/rejected": -2.583953380584717,
"logps/chosen": -376.8917236328125,
"logps/rejected": -623.347412109375,
"loss": 0.4686,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.57673978805542,
"rewards/margins": 2.396878719329834,
"rewards/rejected": -3.973618268966675,
"step": 348
},
{
"epoch": 1.6491435321913763,
"grad_norm": 4.471413252688273,
"learning_rate": 3.828247046932992e-08,
"logits/chosen": -2.6523804664611816,
"logits/rejected": -2.6967380046844482,
"logps/chosen": -339.6617126464844,
"logps/rejected": -441.6883239746094,
"loss": 0.4632,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.344817876815796,
"rewards/margins": 1.1860837936401367,
"rewards/rejected": -2.5309014320373535,
"step": 349
},
{
"epoch": 1.6538688718251624,
"grad_norm": 3.9015948159102902,
"learning_rate": 3.7267516950235525e-08,
"logits/chosen": -2.590344190597534,
"logits/rejected": -2.6862494945526123,
"logps/chosen": -347.7746887207031,
"logps/rejected": -500.335693359375,
"loss": 0.4407,
"rewards/accuracies": 0.828125,
"rewards/chosen": -1.4618185758590698,
"rewards/margins": 1.6277704238891602,
"rewards/rejected": -3.0895891189575195,
"step": 350
},
{
"epoch": 1.6585942114589485,
"grad_norm": 4.642097450354513,
"learning_rate": 3.62651146182334e-08,
"logits/chosen": -2.656710624694824,
"logits/rejected": -2.673964738845825,
"logps/chosen": -364.386962890625,
"logps/rejected": -458.63348388671875,
"loss": 0.5201,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.533918023109436,
"rewards/margins": 1.1320774555206299,
"rewards/rejected": -2.6659955978393555,
"step": 351
},
{
"epoch": 1.6633195510927348,
"grad_norm": 4.219548915754237,
"learning_rate": 3.527532261492272e-08,
"logits/chosen": -2.572221517562866,
"logits/rejected": -2.526576042175293,
"logps/chosen": -370.37396240234375,
"logps/rejected": -487.1659240722656,
"loss": 0.4737,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.4427790641784668,
"rewards/margins": 1.223225712776184,
"rewards/rejected": -2.6660046577453613,
"step": 352
},
{
"epoch": 1.668044890726521,
"grad_norm": 4.167493075518676,
"learning_rate": 3.4298199337894685e-08,
"logits/chosen": -2.6304097175598145,
"logits/rejected": -2.6287808418273926,
"logps/chosen": -387.1392822265625,
"logps/rejected": -581.6126708984375,
"loss": 0.4896,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.8162624835968018,
"rewards/margins": 2.150383234024048,
"rewards/rejected": -3.9666457176208496,
"step": 353
},
{
"epoch": 1.6727702303603071,
"grad_norm": 5.897960463157494,
"learning_rate": 3.333380243728773e-08,
"logits/chosen": -2.4372665882110596,
"logits/rejected": -2.5481488704681396,
"logps/chosen": -380.5074157714844,
"logps/rejected": -474.3066711425781,
"loss": 0.4799,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.3929729461669922,
"rewards/margins": 1.5269252061843872,
"rewards/rejected": -2.919898271560669,
"step": 354
},
{
"epoch": 1.6774955699940932,
"grad_norm": 4.6734891602402,
"learning_rate": 3.238218881238558e-08,
"logits/chosen": -2.68146014213562,
"logits/rejected": -2.764531135559082,
"logps/chosen": -385.89788818359375,
"logps/rejected": -453.99371337890625,
"loss": 0.4799,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.5357441902160645,
"rewards/margins": 1.310141921043396,
"rewards/rejected": -2.845885992050171,
"step": 355
},
{
"epoch": 1.6822209096278795,
"grad_norm": 3.9989351330319503,
"learning_rate": 3.1443414608260526e-08,
"logits/chosen": -2.7942676544189453,
"logits/rejected": -2.7728328704833984,
"logps/chosen": -417.6398010253906,
"logps/rejected": -534.0235595703125,
"loss": 0.4725,
"rewards/accuracies": 0.796875,
"rewards/chosen": -1.648948073387146,
"rewards/margins": 1.4638010263442993,
"rewards/rejected": -3.1127490997314453,
"step": 356
},
{
"epoch": 1.6869462492616658,
"grad_norm": 4.114134264165304,
"learning_rate": 3.0517535212460946e-08,
"logits/chosen": -2.6477482318878174,
"logits/rejected": -2.7860238552093506,
"logps/chosen": -470.87103271484375,
"logps/rejected": -513.6318969726562,
"loss": 0.469,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.568166732788086,
"rewards/margins": 0.9764127731323242,
"rewards/rejected": -2.5445797443389893,
"step": 357
},
{
"epoch": 1.6916715888954519,
"grad_norm": 4.425834666691968,
"learning_rate": 2.960460525174313e-08,
"logits/chosen": -2.8916306495666504,
"logits/rejected": -2.853088855743408,
"logps/chosen": -339.81610107421875,
"logps/rejected": -483.9287109375,
"loss": 0.472,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.479756236076355,
"rewards/margins": 1.5092005729675293,
"rewards/rejected": -2.9889566898345947,
"step": 358
},
{
"epoch": 1.696396928529238,
"grad_norm": 4.184257952905371,
"learning_rate": 2.8704678588848535e-08,
"logits/chosen": -2.52712345123291,
"logits/rejected": -2.5050594806671143,
"logps/chosen": -368.9576416015625,
"logps/rejected": -528.0908813476562,
"loss": 0.4573,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.523732304573059,
"rewards/margins": 1.6147029399871826,
"rewards/rejected": -3.1384353637695312,
"step": 359
},
{
"epoch": 1.7011222681630243,
"grad_norm": 4.172727890906571,
"learning_rate": 2.781780831932595e-08,
"logits/chosen": -2.6111361980438232,
"logits/rejected": -2.634065628051758,
"logps/chosen": -437.898681640625,
"logps/rejected": -492.3134460449219,
"loss": 0.4917,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.5102522373199463,
"rewards/margins": 0.8861820697784424,
"rewards/rejected": -2.3964343070983887,
"step": 360
},
{
"epoch": 1.7058476077968105,
"grad_norm": 3.6954976996048376,
"learning_rate": 2.6944046768398565e-08,
"logits/chosen": -2.5292959213256836,
"logits/rejected": -2.5336508750915527,
"logps/chosen": -348.8612060546875,
"logps/rejected": -474.258056640625,
"loss": 0.4838,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.3767406940460205,
"rewards/margins": 1.3499197959899902,
"rewards/rejected": -2.7266602516174316,
"step": 361
},
{
"epoch": 1.7105729474305966,
"grad_norm": 4.557110158231638,
"learning_rate": 2.608344548787722e-08,
"logits/chosen": -2.493603229522705,
"logits/rejected": -2.6656622886657715,
"logps/chosen": -443.8504638671875,
"logps/rejected": -544.0814819335938,
"loss": 0.4946,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.7300480604171753,
"rewards/margins": 1.5911731719970703,
"rewards/rejected": -3.321221113204956,
"step": 362
},
{
"epoch": 1.7152982870643827,
"grad_norm": 4.456300726365723,
"learning_rate": 2.523605525311842e-08,
"logits/chosen": -2.5634899139404297,
"logits/rejected": -2.5102734565734863,
"logps/chosen": -380.3689270019531,
"logps/rejected": -480.8353271484375,
"loss": 0.457,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.3825726509094238,
"rewards/margins": 0.8929464221000671,
"rewards/rejected": -2.2755191326141357,
"step": 363
},
{
"epoch": 1.7200236266981688,
"grad_norm": 4.384534987558066,
"learning_rate": 2.440192606002889e-08,
"logits/chosen": -2.7241060733795166,
"logits/rejected": -2.700328826904297,
"logps/chosen": -400.6265869140625,
"logps/rejected": -509.753173828125,
"loss": 0.4544,
"rewards/accuracies": 0.609375,
"rewards/chosen": -1.817798137664795,
"rewards/margins": 1.4839262962341309,
"rewards/rejected": -3.3017241954803467,
"step": 364
},
{
"epoch": 1.724748966331955,
"grad_norm": 4.665836153801188,
"learning_rate": 2.3581107122115723e-08,
"logits/chosen": -2.7754364013671875,
"logits/rejected": -2.7966537475585938,
"logps/chosen": -398.6808166503906,
"logps/rejected": -486.95263671875,
"loss": 0.4908,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.9782556295394897,
"rewards/margins": 1.371333360671997,
"rewards/rejected": -3.3495888710021973,
"step": 365
},
{
"epoch": 1.7294743059657414,
"grad_norm": 4.147614147053133,
"learning_rate": 2.2773646867582763e-08,
"logits/chosen": -2.626425266265869,
"logits/rejected": -2.57529354095459,
"logps/chosen": -398.737548828125,
"logps/rejected": -586.5382690429688,
"loss": 0.488,
"rewards/accuracies": 0.734375,
"rewards/chosen": -1.5953752994537354,
"rewards/margins": 1.6417709589004517,
"rewards/rejected": -3.2371463775634766,
"step": 366
},
{
"epoch": 1.7341996455995274,
"grad_norm": 4.0997514353208775,
"learning_rate": 2.19795929364735e-08,
"logits/chosen": -2.473259449005127,
"logits/rejected": -2.6184892654418945,
"logps/chosen": -344.3788757324219,
"logps/rejected": -420.8124694824219,
"loss": 0.4917,
"rewards/accuracies": 0.625,
"rewards/chosen": -1.3588237762451172,
"rewards/margins": 1.1373924016952515,
"rewards/rejected": -2.496216058731079,
"step": 367
},
{
"epoch": 1.7389249852333135,
"grad_norm": 4.374774389574994,
"learning_rate": 2.119899217785995e-08,
"logits/chosen": -2.467965841293335,
"logits/rejected": -2.5698554515838623,
"logps/chosen": -387.9084777832031,
"logps/rejected": -457.1981201171875,
"loss": 0.4716,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.393808364868164,
"rewards/margins": 1.2519505023956299,
"rewards/rejected": -2.645759105682373,
"step": 368
},
{
"epoch": 1.7436503248670998,
"grad_norm": 4.049649530225524,
"learning_rate": 2.0431890647079093e-08,
"logits/chosen": -2.407700538635254,
"logits/rejected": -2.458270788192749,
"logps/chosen": -414.94561767578125,
"logps/rejected": -544.7018432617188,
"loss": 0.4652,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.8057122230529785,
"rewards/margins": 1.333924412727356,
"rewards/rejected": -3.139636754989624,
"step": 369
},
{
"epoch": 1.748375664500886,
"grad_norm": 4.671069391094125,
"learning_rate": 1.967833360301513e-08,
"logits/chosen": -2.637674331665039,
"logits/rejected": -2.722353458404541,
"logps/chosen": -357.71142578125,
"logps/rejected": -436.8525390625,
"loss": 0.4661,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.5406451225280762,
"rewards/margins": 1.1408922672271729,
"rewards/rejected": -2.681537389755249,
"step": 370
},
{
"epoch": 1.7531010041346722,
"grad_norm": 4.49257028234618,
"learning_rate": 1.8938365505429544e-08,
"logits/chosen": -2.710331678390503,
"logits/rejected": -2.7880280017852783,
"logps/chosen": -386.74945068359375,
"logps/rejected": -489.5121154785156,
"loss": 0.4942,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.6727674007415771,
"rewards/margins": 1.33446204662323,
"rewards/rejected": -3.0072293281555176,
"step": 371
},
{
"epoch": 1.7578263437684583,
"grad_norm": 3.5465325285307228,
"learning_rate": 1.8212030012337704e-08,
"logits/chosen": -2.757737874984741,
"logits/rejected": -2.687244415283203,
"logps/chosen": -360.2991943359375,
"logps/rejected": -512.8646240234375,
"loss": 0.4577,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.4896515607833862,
"rewards/margins": 1.5841938257217407,
"rewards/rejected": -3.0738449096679688,
"step": 372
},
{
"epoch": 1.7625516834022446,
"grad_norm": 4.417035193749527,
"learning_rate": 1.7499369977433453e-08,
"logits/chosen": -2.6367974281311035,
"logits/rejected": -2.677651882171631,
"logps/chosen": -362.3944396972656,
"logps/rejected": -460.9113464355469,
"loss": 0.4586,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.5859841108322144,
"rewards/margins": 1.0494407415390015,
"rewards/rejected": -2.6354243755340576,
"step": 373
},
{
"epoch": 1.7672770230360308,
"grad_norm": 4.392962342030237,
"learning_rate": 1.680042744756016e-08,
"logits/chosen": -2.9928336143493652,
"logits/rejected": -2.9165008068084717,
"logps/chosen": -366.8034973144531,
"logps/rejected": -550.0221557617188,
"loss": 0.4769,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.5658485889434814,
"rewards/margins": 1.7793083190917969,
"rewards/rejected": -3.3451569080352783,
"step": 374
},
{
"epoch": 1.772002362669817,
"grad_norm": 4.856077218434121,
"learning_rate": 1.611524366023062e-08,
"logits/chosen": -2.722025156021118,
"logits/rejected": -2.8585572242736816,
"logps/chosen": -347.9723815917969,
"logps/rejected": -446.63604736328125,
"loss": 0.4742,
"rewards/accuracies": 0.59375,
"rewards/chosen": -1.4827804565429688,
"rewards/margins": 1.0786786079406738,
"rewards/rejected": -2.5614588260650635,
"step": 375
},
{
"epoch": 1.776727702303603,
"grad_norm": 3.752916160722341,
"learning_rate": 1.544385904119344e-08,
"logits/chosen": -2.913771867752075,
"logits/rejected": -2.971179962158203,
"logps/chosen": -360.3616638183594,
"logps/rejected": -423.8080749511719,
"loss": 0.4586,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.482804298400879,
"rewards/margins": 1.2081830501556396,
"rewards/rejected": -2.6909875869750977,
"step": 376
},
{
"epoch": 1.7814530419373893,
"grad_norm": 4.12751400895733,
"learning_rate": 1.4786313202048456e-08,
"logits/chosen": -2.6907248497009277,
"logits/rejected": -2.816317319869995,
"logps/chosen": -403.627685546875,
"logps/rejected": -487.9366760253906,
"loss": 0.4929,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.6239163875579834,
"rewards/margins": 1.5215915441513062,
"rewards/rejected": -3.1455078125,
"step": 377
},
{
"epoch": 1.7861783815711756,
"grad_norm": 4.590826584263257,
"learning_rate": 1.4142644937909203e-08,
"logits/chosen": -2.6118569374084473,
"logits/rejected": -2.6019818782806396,
"logps/chosen": -341.8358154296875,
"logps/rejected": -419.81829833984375,
"loss": 0.484,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.3214576244354248,
"rewards/margins": 0.8901782631874084,
"rewards/rejected": -2.2116360664367676,
"step": 378
},
{
"epoch": 1.7909037212049617,
"grad_norm": 4.146292704965862,
"learning_rate": 1.351289222511426e-08,
"logits/chosen": -2.610994577407837,
"logits/rejected": -2.78933048248291,
"logps/chosen": -369.8974609375,
"logps/rejected": -480.7877197265625,
"loss": 0.4694,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.5018396377563477,
"rewards/margins": 1.5635069608688354,
"rewards/rejected": -3.0653464794158936,
"step": 379
},
{
"epoch": 1.7956290608387477,
"grad_norm": 5.1734179405150895,
"learning_rate": 1.2897092218986716e-08,
"logits/chosen": -2.523732900619507,
"logits/rejected": -2.6424248218536377,
"logps/chosen": -425.2537536621094,
"logps/rejected": -560.3030395507812,
"loss": 0.4854,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.5777177810668945,
"rewards/margins": 1.5381261110305786,
"rewards/rejected": -3.1158437728881836,
"step": 380
},
{
"epoch": 1.8003544004725338,
"grad_norm": 3.865564251130276,
"learning_rate": 1.2295281251641698e-08,
"logits/chosen": -2.653510570526123,
"logits/rejected": -2.6763200759887695,
"logps/chosen": -392.48577880859375,
"logps/rejected": -508.2447509765625,
"loss": 0.4676,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.5738812685012817,
"rewards/margins": 1.344365119934082,
"rewards/rejected": -2.918246269226074,
"step": 381
},
{
"epoch": 1.80507974010632,
"grad_norm": 3.9007660136930062,
"learning_rate": 1.1707494829843207e-08,
"logits/chosen": -2.4932150840759277,
"logits/rejected": -2.525545835494995,
"logps/chosen": -393.6082763671875,
"logps/rejected": -494.412109375,
"loss": 0.4637,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.6010535955429077,
"rewards/margins": 1.5001673698425293,
"rewards/rejected": -3.1012210845947266,
"step": 382
},
{
"epoch": 1.8098050797401064,
"grad_norm": 3.789687566531843,
"learning_rate": 1.1133767632908798e-08,
"logits/chosen": -2.773787021636963,
"logits/rejected": -2.7511181831359863,
"logps/chosen": -374.2195739746094,
"logps/rejected": -590.141357421875,
"loss": 0.5006,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.784470558166504,
"rewards/margins": 1.8250298500061035,
"rewards/rejected": -3.6095001697540283,
"step": 383
},
{
"epoch": 1.8145304193738925,
"grad_norm": 4.27781176287506,
"learning_rate": 1.0574133510663747e-08,
"logits/chosen": -2.4717490673065186,
"logits/rejected": -2.580211639404297,
"logps/chosen": -404.38250732421875,
"logps/rejected": -501.7881774902344,
"loss": 0.4858,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.4478615522384644,
"rewards/margins": 1.3333170413970947,
"rewards/rejected": -2.7811787128448486,
"step": 384
},
{
"epoch": 1.8192557590076786,
"grad_norm": 4.660154426544148,
"learning_rate": 1.0028625481443981e-08,
"logits/chosen": -2.6154394149780273,
"logits/rejected": -2.6079351902008057,
"logps/chosen": -332.8676452636719,
"logps/rejected": -475.0517272949219,
"loss": 0.4503,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.4882500171661377,
"rewards/margins": 1.4799097776412964,
"rewards/rejected": -2.9681599140167236,
"step": 385
},
{
"epoch": 1.8239810986414648,
"grad_norm": 4.32107981908328,
"learning_rate": 9.497275730147774e-09,
"logits/chosen": -2.57356333732605,
"logits/rejected": -2.566416025161743,
"logps/chosen": -400.1445007324219,
"logps/rejected": -566.9680786132812,
"loss": 0.4524,
"rewards/accuracies": 0.796875,
"rewards/chosen": -1.724095344543457,
"rewards/margins": 1.8485801219940186,
"rewards/rejected": -3.5726757049560547,
"step": 386
},
{
"epoch": 1.8287064382752511,
"grad_norm": 4.382882347588404,
"learning_rate": 8.980115606337046e-09,
"logits/chosen": -2.744180202484131,
"logits/rejected": -2.6595263481140137,
"logps/chosen": -318.4762268066406,
"logps/rejected": -484.2186584472656,
"loss": 0.4836,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.4589486122131348,
"rewards/margins": 1.453284502029419,
"rewards/rejected": -2.912233352661133,
"step": 387
},
{
"epoch": 1.8287064382752511,
"eval_logits/chosen": -2.637449264526367,
"eval_logits/rejected": -2.6433615684509277,
"eval_logps/chosen": -379.1832275390625,
"eval_logps/rejected": -518.951416015625,
"eval_loss": 0.4717705249786377,
"eval_rewards/accuracies": 0.6420454382896423,
"eval_rewards/chosen": -1.679525375366211,
"eval_rewards/margins": 1.5104897022247314,
"eval_rewards/rejected": -3.1900153160095215,
"eval_runtime": 226.5578,
"eval_samples_per_second": 16.137,
"eval_steps_per_second": 0.291,
"step": 387
},
{
"epoch": 1.8334317779090372,
"grad_norm": 4.767778942870907,
"learning_rate": 8.47717562238756e-09,
"logits/chosen": -2.504225254058838,
"logits/rejected": -2.5545616149902344,
"logps/chosen": -355.3597717285156,
"logps/rejected": -519.2000732421875,
"loss": 0.4663,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.5064764022827148,
"rewards/margins": 1.655312418937683,
"rewards/rejected": -3.1617889404296875,
"step": 388
},
{
"epoch": 1.8381571175428233,
"grad_norm": 4.59007333084006,
"learning_rate": 7.988485451688815e-09,
"logits/chosen": -2.8325204849243164,
"logits/rejected": -2.8123779296875,
"logps/chosen": -341.4407043457031,
"logps/rejected": -501.8046875,
"loss": 0.4958,
"rewards/accuracies": 0.609375,
"rewards/chosen": -1.6452869176864624,
"rewards/margins": 1.5238325595855713,
"rewards/rejected": -3.1691195964813232,
"step": 389
},
{
"epoch": 1.8428824571766096,
"grad_norm": 4.836517410103638,
"learning_rate": 7.514073926893432e-09,
"logits/chosen": -2.441648006439209,
"logits/rejected": -2.528268814086914,
"logps/chosen": -386.49493408203125,
"logps/rejected": -454.25616455078125,
"loss": 0.5151,
"rewards/accuracies": 0.671875,
"rewards/chosen": -1.5899585485458374,
"rewards/margins": 0.9383600950241089,
"rewards/rejected": -2.5283186435699463,
"step": 390
},
{
"epoch": 1.8476077968103959,
"grad_norm": 4.204493754201481,
"learning_rate": 7.053969038215674e-09,
"logits/chosen": -2.7649574279785156,
"logits/rejected": -2.6198995113372803,
"logps/chosen": -405.90283203125,
"logps/rejected": -581.5423583984375,
"loss": 0.4766,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.7321686744689941,
"rewards/margins": 1.7127902507781982,
"rewards/rejected": -3.4449586868286133,
"step": 391
},
{
"epoch": 1.852333136444182,
"grad_norm": 3.805689590990249,
"learning_rate": 6.608197931780496e-09,
"logits/chosen": -2.5429623126983643,
"logits/rejected": -2.4480888843536377,
"logps/chosen": -373.41375732421875,
"logps/rejected": -588.4783935546875,
"loss": 0.4787,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.4610140323638916,
"rewards/margins": 1.9005025625228882,
"rewards/rejected": -3.3615164756774902,
"step": 392
},
{
"epoch": 1.857058476077968,
"grad_norm": 4.640569841501701,
"learning_rate": 6.176786908021453e-09,
"logits/chosen": -2.661363363265991,
"logits/rejected": -2.734570026397705,
"logps/chosen": -411.58306884765625,
"logps/rejected": -510.5803527832031,
"loss": 0.475,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.6014535427093506,
"rewards/margins": 1.3125067949295044,
"rewards/rejected": -2.9139604568481445,
"step": 393
},
{
"epoch": 1.8617838157117543,
"grad_norm": 4.140728148044369,
"learning_rate": 5.759761420129322e-09,
"logits/chosen": -2.901158332824707,
"logits/rejected": -2.942783832550049,
"logps/chosen": -331.9306945800781,
"logps/rejected": -466.704833984375,
"loss": 0.4703,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.4073690176010132,
"rewards/margins": 1.4787429571151733,
"rewards/rejected": -2.8861119747161865,
"step": 394
},
{
"epoch": 1.8665091553455406,
"grad_norm": 3.736333802301315,
"learning_rate": 5.357146072550278e-09,
"logits/chosen": -2.4809322357177734,
"logits/rejected": -2.499831438064575,
"logps/chosen": -393.06109619140625,
"logps/rejected": -467.2709655761719,
"loss": 0.4632,
"rewards/accuracies": 0.59375,
"rewards/chosen": -1.567178726196289,
"rewards/margins": 0.7932885885238647,
"rewards/rejected": -2.3604674339294434,
"step": 395
},
{
"epoch": 1.8712344949793267,
"grad_norm": 4.263089946197189,
"learning_rate": 4.968964619534138e-09,
"logits/chosen": -2.6070809364318848,
"logits/rejected": -2.5771572589874268,
"logps/chosen": -346.75042724609375,
"logps/rejected": -519.9202880859375,
"loss": 0.443,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.525521993637085,
"rewards/margins": 1.564245581626892,
"rewards/rejected": -3.0897674560546875,
"step": 396
},
{
"epoch": 1.8759598346131128,
"grad_norm": 4.104385303806953,
"learning_rate": 4.595239963733011e-09,
"logits/chosen": -2.834834575653076,
"logits/rejected": -2.8623130321502686,
"logps/chosen": -387.9482727050781,
"logps/rejected": -544.7717895507812,
"loss": 0.427,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.8901216983795166,
"rewards/margins": 1.8027396202087402,
"rewards/rejected": -3.6928610801696777,
"step": 397
},
{
"epoch": 1.8806851742468988,
"grad_norm": 3.9166519315043407,
"learning_rate": 4.2359941548499035e-09,
"logits/chosen": -2.438992977142334,
"logits/rejected": -2.4675240516662598,
"logps/chosen": -377.0653076171875,
"logps/rejected": -579.7406616210938,
"loss": 0.4621,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.7994550466537476,
"rewards/margins": 1.7710639238357544,
"rewards/rejected": -3.570518970489502,
"step": 398
},
{
"epoch": 1.8854105138806851,
"grad_norm": 3.980027730791039,
"learning_rate": 3.891248388337847e-09,
"logits/chosen": -2.461378574371338,
"logits/rejected": -2.3893392086029053,
"logps/chosen": -391.1559753417969,
"logps/rejected": -487.46112060546875,
"loss": 0.4763,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.5593876838684082,
"rewards/margins": 1.065595030784607,
"rewards/rejected": -2.6249828338623047,
"step": 399
},
{
"epoch": 1.8901358535144714,
"grad_norm": 4.768587791597909,
"learning_rate": 3.5610230041494828e-09,
"logits/chosen": -2.3206558227539062,
"logits/rejected": -2.3519835472106934,
"logps/chosen": -392.8856506347656,
"logps/rejected": -517.7825317382812,
"loss": 0.4687,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.5414066314697266,
"rewards/margins": 1.4560898542404175,
"rewards/rejected": -2.9974963665008545,
"step": 400
},
{
"epoch": 1.8948611931482575,
"grad_norm": 4.155264459662159,
"learning_rate": 3.2453374855367366e-09,
"logits/chosen": -2.830385684967041,
"logits/rejected": -2.893541097640991,
"logps/chosen": -356.2238464355469,
"logps/rejected": -443.7274169921875,
"loss": 0.4612,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.463138461112976,
"rewards/margins": 1.1921138763427734,
"rewards/rejected": -2.65525221824646,
"step": 401
},
{
"epoch": 1.8995865327820436,
"grad_norm": 4.0526022624158955,
"learning_rate": 2.9442104579016356e-09,
"logits/chosen": -2.313565731048584,
"logits/rejected": -2.4396870136260986,
"logps/chosen": -441.73736572265625,
"logps/rejected": -458.2696533203125,
"loss": 0.4724,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.3574057817459106,
"rewards/margins": 0.8890299797058105,
"rewards/rejected": -2.2464358806610107,
"step": 402
},
{
"epoch": 1.9043118724158299,
"grad_norm": 3.860540669652651,
"learning_rate": 2.657659687697156e-09,
"logits/chosen": -2.6873679161071777,
"logits/rejected": -2.549762725830078,
"logps/chosen": -317.7322082519531,
"logps/rejected": -429.63787841796875,
"loss": 0.4784,
"rewards/accuracies": 0.640625,
"rewards/chosen": -1.5817471742630005,
"rewards/margins": 1.0039292573928833,
"rewards/rejected": -2.5856761932373047,
"step": 403
},
{
"epoch": 1.9090372120496162,
"grad_norm": 4.53787076168368,
"learning_rate": 2.385702081379143e-09,
"logits/chosen": -2.3574860095977783,
"logits/rejected": -2.427652359008789,
"logps/chosen": -431.2988586425781,
"logps/rejected": -567.0403442382812,
"loss": 0.4604,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.6895103454589844,
"rewards/margins": 1.765432357788086,
"rewards/rejected": -3.4549427032470703,
"step": 404
},
{
"epoch": 1.9137625516834023,
"grad_norm": 3.735971197718604,
"learning_rate": 2.1283536844087513e-09,
"logits/chosen": -2.5645933151245117,
"logits/rejected": -2.5676231384277344,
"logps/chosen": -362.2431335449219,
"logps/rejected": -531.4500122070312,
"loss": 0.4731,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.5010064840316772,
"rewards/margins": 1.6250488758087158,
"rewards/rejected": -3.1260552406311035,
"step": 405
},
{
"epoch": 1.9184878913171883,
"grad_norm": 3.613782865337248,
"learning_rate": 1.885629680305867e-09,
"logits/chosen": -2.569301128387451,
"logits/rejected": -2.568390130996704,
"logps/chosen": -384.6716613769531,
"logps/rejected": -521.9403076171875,
"loss": 0.4467,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.6642777919769287,
"rewards/margins": 1.619908332824707,
"rewards/rejected": -3.284186363220215,
"step": 406
},
{
"epoch": 1.9232132309509746,
"grad_norm": 4.461343319483535,
"learning_rate": 1.6575443897531294e-09,
"logits/chosen": -2.5552725791931152,
"logits/rejected": -2.436856985092163,
"logps/chosen": -367.128662109375,
"logps/rejected": -545.8114624023438,
"loss": 0.4754,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.6492211818695068,
"rewards/margins": 1.592346429824829,
"rewards/rejected": -3.241567373275757,
"step": 407
},
{
"epoch": 1.927938570584761,
"grad_norm": 4.454271459686257,
"learning_rate": 1.4441112697511638e-09,
"logits/chosen": -2.6292550563812256,
"logits/rejected": -2.6084656715393066,
"logps/chosen": -421.9637756347656,
"logps/rejected": -506.38897705078125,
"loss": 0.484,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.6983362436294556,
"rewards/margins": 1.19191312789917,
"rewards/rejected": -2.890249252319336,
"step": 408
},
{
"epoch": 1.932663910218547,
"grad_norm": 4.002197751374577,
"learning_rate": 1.2453429128245762e-09,
"logits/chosen": -2.5047454833984375,
"logits/rejected": -2.451660633087158,
"logps/chosen": -431.7708740234375,
"logps/rejected": -614.2373657226562,
"loss": 0.471,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.9678634405136108,
"rewards/margins": 1.840644359588623,
"rewards/rejected": -3.8085079193115234,
"step": 409
},
{
"epoch": 1.937389249852333,
"grad_norm": 4.283657038062509,
"learning_rate": 1.061251046278938e-09,
"logits/chosen": -2.5947906970977783,
"logits/rejected": -2.700439214706421,
"logps/chosen": -392.77850341796875,
"logps/rejected": -458.53570556640625,
"loss": 0.4785,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.4087834358215332,
"rewards/margins": 1.301395297050476,
"rewards/rejected": -2.7101786136627197,
"step": 410
},
{
"epoch": 1.9421145894861194,
"grad_norm": 4.320389688019286,
"learning_rate": 8.918465315088941e-10,
"logits/chosen": -2.529167413711548,
"logits/rejected": -2.595038652420044,
"logps/chosen": -436.0292663574219,
"logps/rejected": -522.2467041015625,
"loss": 0.4691,
"rewards/accuracies": 0.65625,
"rewards/chosen": -1.7954521179199219,
"rewards/margins": 1.5213781595230103,
"rewards/rejected": -3.3168301582336426,
"step": 411
},
{
"epoch": 1.9468399291199054,
"grad_norm": 3.917488632599818,
"learning_rate": 7.371393633574252e-10,
"logits/chosen": -2.508873224258423,
"logits/rejected": -2.5457823276519775,
"logps/chosen": -384.64923095703125,
"logps/rejected": -475.94134521484375,
"loss": 0.469,
"rewards/accuracies": 0.828125,
"rewards/chosen": -1.4391266107559204,
"rewards/margins": 1.211023211479187,
"rewards/rejected": -2.6501498222351074,
"step": 412
},
{
"epoch": 1.9515652687536917,
"grad_norm": 3.9351124104785584,
"learning_rate": 5.971386695260705e-10,
"logits/chosen": -2.465684652328491,
"logits/rejected": -2.5066022872924805,
"logps/chosen": -385.0665283203125,
"logps/rejected": -479.13214111328125,
"loss": 0.4643,
"rewards/accuracies": 0.765625,
"rewards/chosen": -1.4245293140411377,
"rewards/margins": 1.249526858329773,
"rewards/rejected": -2.6740562915802,
"step": 413
},
{
"epoch": 1.9562906083874778,
"grad_norm": 3.704041812352823,
"learning_rate": 4.718527100364134e-10,
"logits/chosen": -2.6545798778533936,
"logits/rejected": -2.6624419689178467,
"logps/chosen": -366.58184814453125,
"logps/rejected": -510.66436767578125,
"loss": 0.4482,
"rewards/accuracies": 0.78125,
"rewards/chosen": -1.5549781322479248,
"rewards/margins": 1.5164021253585815,
"rewards/rejected": -3.071380376815796,
"step": 414
},
{
"epoch": 1.9610159480212639,
"grad_norm": 4.499954109459605,
"learning_rate": 3.6128887674272133e-10,
"logits/chosen": -2.5801494121551514,
"logits/rejected": -2.551297187805176,
"logps/chosen": -360.39935302734375,
"logps/rejected": -539.8038330078125,
"loss": 0.4529,
"rewards/accuracies": 0.75,
"rewards/chosen": -1.6478443145751953,
"rewards/margins": 1.7761938571929932,
"rewards/rejected": -3.4240384101867676,
"step": 415
},
{
"epoch": 1.9657412876550502,
"grad_norm": 4.539840176257443,
"learning_rate": 2.6545369289587836e-10,
"logits/chosen": -2.506206750869751,
"logits/rejected": -2.729609489440918,
"logps/chosen": -423.645751953125,
"logps/rejected": -465.3453674316406,
"loss": 0.5018,
"rewards/accuracies": 0.8125,
"rewards/chosen": -1.2937572002410889,
"rewards/margins": 1.3675696849822998,
"rewards/rejected": -2.6613268852233887,
"step": 416
},
{
"epoch": 1.9704666272888365,
"grad_norm": 4.977216825161169,
"learning_rate": 1.843528127584981e-10,
"logits/chosen": -2.5777127742767334,
"logits/rejected": -2.611386775970459,
"logps/chosen": -396.95892333984375,
"logps/rejected": -500.09375,
"loss": 0.4951,
"rewards/accuracies": 0.609375,
"rewards/chosen": -1.7503533363342285,
"rewards/margins": 1.2793786525726318,
"rewards/rejected": -3.0297319889068604,
"step": 417
},
{
"epoch": 1.9751919669226226,
"grad_norm": 4.452268262277202,
"learning_rate": 1.17991021271302e-10,
"logits/chosen": -2.4709465503692627,
"logits/rejected": -2.399839401245117,
"logps/chosen": -379.194091796875,
"logps/rejected": -520.7940063476562,
"loss": 0.4696,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.5808072090148926,
"rewards/margins": 1.5406217575073242,
"rewards/rejected": -3.121428966522217,
"step": 418
},
{
"epoch": 1.9799173065564086,
"grad_norm": 4.060313383389935,
"learning_rate": 6.637223377078949e-11,
"logits/chosen": -2.822049856185913,
"logits/rejected": -2.800950765609741,
"logps/chosen": -309.0566711425781,
"logps/rejected": -446.6823425292969,
"loss": 0.4561,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.39582097530365,
"rewards/margins": 1.2124607563018799,
"rewards/rejected": -2.6082818508148193,
"step": 419
},
{
"epoch": 1.984642646190195,
"grad_norm": 4.291839752698122,
"learning_rate": 2.949949575833943e-11,
"logits/chosen": -2.625284433364868,
"logits/rejected": -2.6156272888183594,
"logps/chosen": -342.9141845703125,
"logps/rejected": -463.14312744140625,
"loss": 0.5061,
"rewards/accuracies": 0.71875,
"rewards/chosen": -1.5005218982696533,
"rewards/margins": 1.2724413871765137,
"rewards/rejected": -2.772963047027588,
"step": 420
},
{
"epoch": 1.9893679858239812,
"grad_norm": 4.75166895203396,
"learning_rate": 7.374982720326217e-12,
"logits/chosen": -2.7700271606445312,
"logits/rejected": -2.7563564777374268,
"logps/chosen": -397.4788818359375,
"logps/rejected": -582.6298828125,
"loss": 0.4718,
"rewards/accuracies": 0.703125,
"rewards/chosen": -1.6806347370147705,
"rewards/margins": 1.8375290632247925,
"rewards/rejected": -3.5181639194488525,
"step": 421
},
{
"epoch": 1.9940933254577673,
"grad_norm": 4.42823795907204,
"learning_rate": 0.0,
"logits/chosen": -2.568351984024048,
"logits/rejected": -2.728876829147339,
"logps/chosen": -369.7898254394531,
"logps/rejected": -406.9199523925781,
"loss": 0.4865,
"rewards/accuracies": 0.6875,
"rewards/chosen": -1.3815803527832031,
"rewards/margins": 0.7950088977813721,
"rewards/rejected": -2.176589250564575,
"step": 422
},
{
"epoch": 1.9940933254577673,
"step": 422,
"total_flos": 0.0,
"train_loss": 0.5345145690638872,
"train_runtime": 33183.7631,
"train_samples_per_second": 5.711,
"train_steps_per_second": 0.013
}
],
"logging_steps": 1,
"max_steps": 422,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 43,
"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": 8,
"trial_name": null,
"trial_params": null
}