zephyr-7b-IL_DPOBPBase-full / trainer_state.json
TTTXXX01's picture
Model save
735ad65 verified
raw
history blame
26.1 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.998691442030882,
"eval_steps": 500,
"global_step": 477,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.002093692750588851,
"grad_norm": 4.60553352196104,
"learning_rate": 1.0416666666666666e-08,
"logits/chosen": 5002.53564453125,
"logits/rejected": 4591.72021484375,
"logps/chosen": -265.6396789550781,
"logps/rejected": -206.22401428222656,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.02093692750588851,
"grad_norm": 3.906214184095967,
"learning_rate": 1.0416666666666667e-07,
"logits/chosen": 5977.31591796875,
"logits/rejected": 5015.41943359375,
"logps/chosen": -292.51190185546875,
"logps/rejected": -259.33819580078125,
"loss": 0.6931,
"rewards/accuracies": 0.4791666567325592,
"rewards/chosen": 0.0004885931266471744,
"rewards/margins": 0.0006340253748930991,
"rewards/rejected": -0.00014543221914209425,
"step": 10
},
{
"epoch": 0.04187385501177702,
"grad_norm": 4.408104522173234,
"learning_rate": 2.0833333333333333e-07,
"logits/chosen": 5877.7744140625,
"logits/rejected": 4856.02734375,
"logps/chosen": -290.4228820800781,
"logps/rejected": -249.76522827148438,
"loss": 0.6927,
"rewards/accuracies": 0.6000000238418579,
"rewards/chosen": 0.0017061525722965598,
"rewards/margins": 0.002024381887167692,
"rewards/rejected": -0.00031822946039028466,
"step": 20
},
{
"epoch": 0.06281078251766553,
"grad_norm": 4.1323992714408995,
"learning_rate": 3.1249999999999997e-07,
"logits/chosen": 5644.30419921875,
"logits/rejected": 4922.60302734375,
"logps/chosen": -270.80462646484375,
"logps/rejected": -238.2168426513672,
"loss": 0.6907,
"rewards/accuracies": 0.5874999761581421,
"rewards/chosen": 0.014328369870781898,
"rewards/margins": 0.008474691770970821,
"rewards/rejected": 0.00585367763414979,
"step": 30
},
{
"epoch": 0.08374771002355404,
"grad_norm": 3.9871215863103235,
"learning_rate": 4.1666666666666667e-07,
"logits/chosen": 5738.802734375,
"logits/rejected": 4992.0400390625,
"logps/chosen": -269.14434814453125,
"logps/rejected": -256.4861145019531,
"loss": 0.6855,
"rewards/accuracies": 0.6656249761581421,
"rewards/chosen": 0.048127662390470505,
"rewards/margins": 0.028449412435293198,
"rewards/rejected": 0.019678251817822456,
"step": 40
},
{
"epoch": 0.10468463752944256,
"grad_norm": 4.4563458428725395,
"learning_rate": 4.999731868769026e-07,
"logits/chosen": 6019.1103515625,
"logits/rejected": 5225.02392578125,
"logps/chosen": -276.1994323730469,
"logps/rejected": -271.67559814453125,
"loss": 0.679,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": 0.024182714521884918,
"rewards/margins": 0.0657457783818245,
"rewards/rejected": -0.04156306013464928,
"step": 50
},
{
"epoch": 0.12562156503533106,
"grad_norm": 5.034590628294388,
"learning_rate": 4.990353313429303e-07,
"logits/chosen": 5789.7138671875,
"logits/rejected": 4927.72412109375,
"logps/chosen": -283.95257568359375,
"logps/rejected": -268.9764709472656,
"loss": 0.6716,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": -0.055300962179899216,
"rewards/margins": 0.11967425048351288,
"rewards/rejected": -0.1749752014875412,
"step": 60
},
{
"epoch": 0.14655849254121958,
"grad_norm": 6.512738637492193,
"learning_rate": 4.967625656594781e-07,
"logits/chosen": 5590.95166015625,
"logits/rejected": 5145.8408203125,
"logps/chosen": -288.2838439941406,
"logps/rejected": -293.23541259765625,
"loss": 0.664,
"rewards/accuracies": 0.6625000238418579,
"rewards/chosen": -0.09975917637348175,
"rewards/margins": 0.16456229984760284,
"rewards/rejected": -0.2643214762210846,
"step": 70
},
{
"epoch": 0.16749542004710807,
"grad_norm": 6.2250099209512655,
"learning_rate": 4.93167072587771e-07,
"logits/chosen": 5744.78662109375,
"logits/rejected": 4929.46435546875,
"logps/chosen": -284.8975524902344,
"logps/rejected": -283.4288330078125,
"loss": 0.6577,
"rewards/accuracies": 0.6781250238418579,
"rewards/chosen": -0.08810611069202423,
"rewards/margins": 0.21545740962028503,
"rewards/rejected": -0.30356353521347046,
"step": 80
},
{
"epoch": 0.1884323475529966,
"grad_norm": 6.510058896775878,
"learning_rate": 4.882681251368548e-07,
"logits/chosen": 5883.4892578125,
"logits/rejected": 4958.2470703125,
"logps/chosen": -292.3514709472656,
"logps/rejected": -294.89471435546875,
"loss": 0.6536,
"rewards/accuracies": 0.684374988079071,
"rewards/chosen": -0.14643868803977966,
"rewards/margins": 0.2527722418308258,
"rewards/rejected": -0.39921098947525024,
"step": 90
},
{
"epoch": 0.2093692750588851,
"grad_norm": 9.163808135102661,
"learning_rate": 4.820919832540181e-07,
"logits/chosen": 5974.72802734375,
"logits/rejected": 5452.3408203125,
"logps/chosen": -311.05804443359375,
"logps/rejected": -347.23944091796875,
"loss": 0.6466,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.3118935227394104,
"rewards/margins": 0.41534680128097534,
"rewards/rejected": -0.727240264415741,
"step": 100
},
{
"epoch": 0.23030620256477363,
"grad_norm": 7.986681346223677,
"learning_rate": 4.7467175306295647e-07,
"logits/chosen": 6145.7939453125,
"logits/rejected": 4966.3818359375,
"logps/chosen": -318.4931640625,
"logps/rejected": -344.3719177246094,
"loss": 0.6384,
"rewards/accuracies": 0.7093750238418579,
"rewards/chosen": -0.4387056231498718,
"rewards/margins": 0.4162977337837219,
"rewards/rejected": -0.8550033569335938,
"step": 110
},
{
"epoch": 0.2512431300706621,
"grad_norm": 9.318478390808158,
"learning_rate": 4.6604720940421207e-07,
"logits/chosen": 5758.07373046875,
"logits/rejected": 5305.98193359375,
"logps/chosen": -309.2460632324219,
"logps/rejected": -346.93157958984375,
"loss": 0.6446,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.42067036032676697,
"rewards/margins": 0.3798871636390686,
"rewards/rejected": -0.8005574941635132,
"step": 120
},
{
"epoch": 0.2721800575765506,
"grad_norm": 12.596663584835428,
"learning_rate": 4.5626458262912735e-07,
"logits/chosen": 5819.57373046875,
"logits/rejected": 5167.3623046875,
"logps/chosen": -298.8730163574219,
"logps/rejected": -338.01397705078125,
"loss": 0.6418,
"rewards/accuracies": 0.675000011920929,
"rewards/chosen": -0.3035568296909332,
"rewards/margins": 0.3864966928958893,
"rewards/rejected": -0.6900535225868225,
"step": 130
},
{
"epoch": 0.29311698508243916,
"grad_norm": 8.250031153064022,
"learning_rate": 4.453763107901675e-07,
"logits/chosen": 5601.96484375,
"logits/rejected": 4937.931640625,
"logps/chosen": -306.513427734375,
"logps/rejected": -329.6960754394531,
"loss": 0.6421,
"rewards/accuracies": 0.699999988079071,
"rewards/chosen": -0.3460861146450043,
"rewards/margins": 0.4395461976528168,
"rewards/rejected": -0.785632312297821,
"step": 140
},
{
"epoch": 0.31405391258832765,
"grad_norm": 10.215191441356916,
"learning_rate": 4.3344075855595097e-07,
"logits/chosen": 6188.27392578125,
"logits/rejected": 5056.6064453125,
"logps/chosen": -325.03436279296875,
"logps/rejected": -337.5921630859375,
"loss": 0.6367,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.3591935336589813,
"rewards/margins": 0.5018205642700195,
"rewards/rejected": -0.8610140681266785,
"step": 150
},
{
"epoch": 0.33499084009421615,
"grad_norm": 11.588598413111276,
"learning_rate": 4.2052190435769554e-07,
"logits/chosen": 5836.8857421875,
"logits/rejected": 4833.65673828125,
"logps/chosen": -338.6114807128906,
"logps/rejected": -368.7120056152344,
"loss": 0.6254,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.5401726961135864,
"rewards/margins": 0.583164393901825,
"rewards/rejected": -1.1233372688293457,
"step": 160
},
{
"epoch": 0.3559277676001047,
"grad_norm": 16.573341127038365,
"learning_rate": 4.0668899744407567e-07,
"logits/chosen": 6162.7734375,
"logits/rejected": 5209.9150390625,
"logps/chosen": -329.6107482910156,
"logps/rejected": -334.8266906738281,
"loss": 0.6289,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.47295036911964417,
"rewards/margins": 0.46886569261550903,
"rewards/rejected": -0.941815972328186,
"step": 170
},
{
"epoch": 0.3768646951059932,
"grad_norm": 8.541127331717703,
"learning_rate": 3.920161866827889e-07,
"logits/chosen": 5801.9912109375,
"logits/rejected": 4906.3310546875,
"logps/chosen": -352.50616455078125,
"logps/rejected": -369.984375,
"loss": 0.6347,
"rewards/accuracies": 0.721875011920929,
"rewards/chosen": -0.6521086096763611,
"rewards/margins": 0.5479272603988647,
"rewards/rejected": -1.200035810470581,
"step": 180
},
{
"epoch": 0.39780162261188173,
"grad_norm": 10.691736891405624,
"learning_rate": 3.765821230985757e-07,
"logits/chosen": 5942.0029296875,
"logits/rejected": 4911.9306640625,
"logps/chosen": -313.849853515625,
"logps/rejected": -363.1335144042969,
"loss": 0.6223,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.44098320603370667,
"rewards/margins": 0.7115194201469421,
"rewards/rejected": -1.1525026559829712,
"step": 190
},
{
"epoch": 0.4187385501177702,
"grad_norm": 12.414499403080086,
"learning_rate": 3.604695382782159e-07,
"logits/chosen": 5570.83056640625,
"logits/rejected": 4781.8603515625,
"logps/chosen": -307.8031921386719,
"logps/rejected": -357.94049072265625,
"loss": 0.6348,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.5345231294631958,
"rewards/margins": 0.6162750720977783,
"rewards/rejected": -1.1507983207702637,
"step": 200
},
{
"epoch": 0.4396754776236587,
"grad_norm": 9.725871680307973,
"learning_rate": 3.4376480090239047e-07,
"logits/chosen": 6185.18505859375,
"logits/rejected": 5210.94775390625,
"logps/chosen": -344.7357177734375,
"logps/rejected": -370.8301696777344,
"loss": 0.6292,
"rewards/accuracies": 0.7406250238418579,
"rewards/chosen": -0.5796831846237183,
"rewards/margins": 0.5115828514099121,
"rewards/rejected": -1.0912659168243408,
"step": 210
},
{
"epoch": 0.46061240512954726,
"grad_norm": 10.26939339999947,
"learning_rate": 3.265574537815398e-07,
"logits/chosen": 5568.31298828125,
"logits/rejected": 5116.90087890625,
"logps/chosen": -316.9978942871094,
"logps/rejected": -368.52630615234375,
"loss": 0.6218,
"rewards/accuracies": 0.800000011920929,
"rewards/chosen": -0.45882177352905273,
"rewards/margins": 0.6230064630508423,
"rewards/rejected": -1.0818283557891846,
"step": 220
},
{
"epoch": 0.48154933263543576,
"grad_norm": 11.837301535764077,
"learning_rate": 3.0893973387735683e-07,
"logits/chosen": 5724.6162109375,
"logits/rejected": 4797.6650390625,
"logps/chosen": -357.8770446777344,
"logps/rejected": -405.49285888671875,
"loss": 0.6198,
"rewards/accuracies": 0.7718750238418579,
"rewards/chosen": -0.8960827589035034,
"rewards/margins": 0.7045291066169739,
"rewards/rejected": -1.600611925125122,
"step": 230
},
{
"epoch": 0.5024862601413242,
"grad_norm": 11.232499614895293,
"learning_rate": 2.910060778827554e-07,
"logits/chosen": 5925.3447265625,
"logits/rejected": 4943.4423828125,
"logps/chosen": -331.6304931640625,
"logps/rejected": -372.5296325683594,
"loss": 0.6256,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.6605360507965088,
"rewards/margins": 0.5669505000114441,
"rewards/rejected": -1.2274866104125977,
"step": 240
},
{
"epoch": 0.5234231876472127,
"grad_norm": 12.90825856853013,
"learning_rate": 2.7285261601056697e-07,
"logits/chosen": 6004.00732421875,
"logits/rejected": 5001.3466796875,
"logps/chosen": -353.7926330566406,
"logps/rejected": -393.0778503417969,
"loss": 0.6308,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.6784394383430481,
"rewards/margins": 0.6315867900848389,
"rewards/rejected": -1.3100261688232422,
"step": 250
},
{
"epoch": 0.5443601151531012,
"grad_norm": 13.478020063581878,
"learning_rate": 2.5457665670441937e-07,
"logits/chosen": 5818.59130859375,
"logits/rejected": 5201.59130859375,
"logps/chosen": -355.6083068847656,
"logps/rejected": -395.66754150390625,
"loss": 0.6232,
"rewards/accuracies": 0.78125,
"rewards/chosen": -0.6953917145729065,
"rewards/margins": 0.6504613757133484,
"rewards/rejected": -1.3458532094955444,
"step": 260
},
{
"epoch": 0.5652970426589898,
"grad_norm": 11.290710956607453,
"learning_rate": 2.3627616503391812e-07,
"logits/chosen": 5881.2919921875,
"logits/rejected": 5333.2744140625,
"logps/chosen": -335.8533935546875,
"logps/rejected": -390.1767883300781,
"loss": 0.6207,
"rewards/accuracies": 0.793749988079071,
"rewards/chosen": -0.612395167350769,
"rewards/margins": 0.6854075193405151,
"rewards/rejected": -1.2978026866912842,
"step": 270
},
{
"epoch": 0.5862339701648783,
"grad_norm": 11.562696281009446,
"learning_rate": 2.1804923757009882e-07,
"logits/chosen": 5932.57861328125,
"logits/rejected": 4862.0146484375,
"logps/chosen": -341.2466735839844,
"logps/rejected": -384.97857666015625,
"loss": 0.6207,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.6052099466323853,
"rewards/margins": 0.7767351269721985,
"rewards/rejected": -1.3819451332092285,
"step": 280
},
{
"epoch": 0.6071708976707668,
"grad_norm": 14.192713016769995,
"learning_rate": 1.9999357655598891e-07,
"logits/chosen": 6152.61669921875,
"logits/rejected": 5673.72900390625,
"logps/chosen": -372.6900329589844,
"logps/rejected": -436.62286376953125,
"loss": 0.6269,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.8125594258308411,
"rewards/margins": 0.5744532346725464,
"rewards/rejected": -1.3870127201080322,
"step": 290
},
{
"epoch": 0.6281078251766553,
"grad_norm": 13.845927183072918,
"learning_rate": 1.8220596619089573e-07,
"logits/chosen": 5640.8662109375,
"logits/rejected": 5246.69091796875,
"logps/chosen": -338.5563659667969,
"logps/rejected": -399.28900146484375,
"loss": 0.6223,
"rewards/accuracies": 0.734375,
"rewards/chosen": -0.6720638275146484,
"rewards/margins": 0.61163729429245,
"rewards/rejected": -1.2837010622024536,
"step": 300
},
{
"epoch": 0.6490447526825438,
"grad_norm": 11.103620407034061,
"learning_rate": 1.647817538357072e-07,
"logits/chosen": 6317.97607421875,
"logits/rejected": 5491.85498046875,
"logps/chosen": -355.3973388671875,
"logps/rejected": -393.3858947753906,
"loss": 0.6185,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.6456772089004517,
"rewards/margins": 0.5926147699356079,
"rewards/rejected": -1.2382919788360596,
"step": 310
},
{
"epoch": 0.6699816801884323,
"grad_norm": 12.452201295868143,
"learning_rate": 1.478143389201113e-07,
"logits/chosen": 5882.0078125,
"logits/rejected": 5039.4091796875,
"logps/chosen": -335.7853698730469,
"logps/rejected": -396.05712890625,
"loss": 0.6228,
"rewards/accuracies": 0.71875,
"rewards/chosen": -0.6392192840576172,
"rewards/margins": 0.723227858543396,
"rewards/rejected": -1.3624470233917236,
"step": 320
},
{
"epoch": 0.6909186076943209,
"grad_norm": 11.031080841746133,
"learning_rate": 1.3139467229135998e-07,
"logits/chosen": 5675.63623046875,
"logits/rejected": 5184.22607421875,
"logps/chosen": -334.7772521972656,
"logps/rejected": -389.0362243652344,
"loss": 0.6271,
"rewards/accuracies": 0.7562500238418579,
"rewards/chosen": -0.5799298286437988,
"rewards/margins": 0.6671385765075684,
"rewards/rejected": -1.2470682859420776,
"step": 330
},
{
"epoch": 0.7118555352002094,
"grad_norm": 10.777776010932476,
"learning_rate": 1.1561076868822755e-07,
"logits/chosen": 5497.7236328125,
"logits/rejected": 4727.431640625,
"logps/chosen": -319.00543212890625,
"logps/rejected": -367.11663818359375,
"loss": 0.6253,
"rewards/accuracies": 0.7281249761581421,
"rewards/chosen": -0.6143845915794373,
"rewards/margins": 0.6447553038597107,
"rewards/rejected": -1.2591397762298584,
"step": 340
},
{
"epoch": 0.7327924627060979,
"grad_norm": 13.171378321554494,
"learning_rate": 1.0054723495346482e-07,
"logits/chosen": 6181.7666015625,
"logits/rejected": 4961.35400390625,
"logps/chosen": -359.21905517578125,
"logps/rejected": -390.53350830078125,
"loss": 0.6232,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.5830456018447876,
"rewards/margins": 0.6389049291610718,
"rewards/rejected": -1.2219504117965698,
"step": 350
},
{
"epoch": 0.7537293902119864,
"grad_norm": 12.17139481883657,
"learning_rate": 8.628481651367875e-08,
"logits/chosen": 5867.166015625,
"logits/rejected": 5056.07568359375,
"logps/chosen": -332.9352722167969,
"logps/rejected": -381.3428955078125,
"loss": 0.6279,
"rewards/accuracies": 0.7281249761581421,
"rewards/chosen": -0.5521169900894165,
"rewards/margins": 0.678789496421814,
"rewards/rejected": -1.2309064865112305,
"step": 360
},
{
"epoch": 0.7746663177178749,
"grad_norm": 10.377060003916368,
"learning_rate": 7.289996455765748e-08,
"logits/chosen": 5769.4111328125,
"logits/rejected": 4844.27880859375,
"logps/chosen": -343.3131103515625,
"logps/rejected": -383.9873962402344,
"loss": 0.6145,
"rewards/accuracies": 0.765625,
"rewards/chosen": -0.5452004671096802,
"rewards/margins": 0.725028932094574,
"rewards/rejected": -1.270229458808899,
"step": 370
},
{
"epoch": 0.7956032452237635,
"grad_norm": 13.056548862753138,
"learning_rate": 6.046442623320145e-08,
"logits/chosen": 6045.7568359375,
"logits/rejected": 5162.32958984375,
"logps/chosen": -337.4881896972656,
"logps/rejected": -362.61529541015625,
"loss": 0.626,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.527867317199707,
"rewards/margins": 0.5897092223167419,
"rewards/rejected": -1.1175765991210938,
"step": 380
},
{
"epoch": 0.816540172729652,
"grad_norm": 9.602896358416052,
"learning_rate": 4.904486005914027e-08,
"logits/chosen": 6276.33837890625,
"logits/rejected": 5479.82177734375,
"logps/chosen": -344.8057861328125,
"logps/rejected": -400.14154052734375,
"loss": 0.6156,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.530563473701477,
"rewards/margins": 0.7230854034423828,
"rewards/rejected": -1.2536487579345703,
"step": 390
},
{
"epoch": 0.8374771002355405,
"grad_norm": 10.002000791834785,
"learning_rate": 3.8702478614051345e-08,
"logits/chosen": 5853.0087890625,
"logits/rejected": 5118.06494140625,
"logps/chosen": -347.6117248535156,
"logps/rejected": -372.6192932128906,
"loss": 0.6239,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": -0.5670760273933411,
"rewards/margins": 0.6005390882492065,
"rewards/rejected": -1.1676150560379028,
"step": 400
},
{
"epoch": 0.8584140277414289,
"grad_norm": 10.39957529568141,
"learning_rate": 2.9492720416985e-08,
"logits/chosen": 5829.76318359375,
"logits/rejected": 5050.9814453125,
"logps/chosen": -343.9368591308594,
"logps/rejected": -402.4644470214844,
"loss": 0.6171,
"rewards/accuracies": 0.715624988079071,
"rewards/chosen": -0.6206948757171631,
"rewards/margins": 0.592847466468811,
"rewards/rejected": -1.2135423421859741,
"step": 410
},
{
"epoch": 0.8793509552473174,
"grad_norm": 12.525714643168069,
"learning_rate": 2.1464952759020856e-08,
"logits/chosen": 6221.03662109375,
"logits/rejected": 5368.93994140625,
"logps/chosen": -350.761474609375,
"logps/rejected": -397.9172058105469,
"loss": 0.6175,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": -0.5878224968910217,
"rewards/margins": 0.6664667129516602,
"rewards/rejected": -1.254289150238037,
"step": 420
},
{
"epoch": 0.9002878827532059,
"grad_norm": 11.787131409532831,
"learning_rate": 1.4662207078575684e-08,
"logits/chosen": 6147.80322265625,
"logits/rejected": 4875.4560546875,
"logps/chosen": -347.69866943359375,
"logps/rejected": -387.70556640625,
"loss": 0.612,
"rewards/accuracies": 0.762499988079071,
"rewards/chosen": -0.6009602546691895,
"rewards/margins": 0.7015070915222168,
"rewards/rejected": -1.3024674654006958,
"step": 430
},
{
"epoch": 0.9212248102590945,
"grad_norm": 11.622178629621203,
"learning_rate": 9.12094829893642e-09,
"logits/chosen": 5975.74169921875,
"logits/rejected": 5603.3095703125,
"logps/chosen": -335.4075622558594,
"logps/rejected": -398.03759765625,
"loss": 0.619,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": -0.5686401724815369,
"rewards/margins": 0.6512762308120728,
"rewards/rejected": -1.2199163436889648,
"step": 440
},
{
"epoch": 0.942161737764983,
"grad_norm": 12.334438799788524,
"learning_rate": 4.8708793644441086e-09,
"logits/chosen": 6104.8115234375,
"logits/rejected": 5160.8291015625,
"logps/chosen": -347.5029602050781,
"logps/rejected": -385.18292236328125,
"loss": 0.6176,
"rewards/accuracies": 0.7749999761581421,
"rewards/chosen": -0.5716773271560669,
"rewards/margins": 0.6656473875045776,
"rewards/rejected": -1.237324595451355,
"step": 450
},
{
"epoch": 0.9630986652708715,
"grad_norm": 14.274087744528849,
"learning_rate": 1.9347820230782295e-09,
"logits/chosen": 6278.4931640625,
"logits/rejected": 5222.73828125,
"logps/chosen": -374.8132629394531,
"logps/rejected": -419.890625,
"loss": 0.6239,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": -0.6181422472000122,
"rewards/margins": 0.7236369848251343,
"rewards/rejected": -1.3417792320251465,
"step": 460
},
{
"epoch": 0.98403559277676,
"grad_norm": 9.711310883436695,
"learning_rate": 3.2839470889836627e-10,
"logits/chosen": 5856.66259765625,
"logits/rejected": 5124.48193359375,
"logps/chosen": -340.29803466796875,
"logps/rejected": -388.76165771484375,
"loss": 0.6131,
"rewards/accuracies": 0.7718750238418579,
"rewards/chosen": -0.5945884585380554,
"rewards/margins": 0.7149998545646667,
"rewards/rejected": -1.3095881938934326,
"step": 470
},
{
"epoch": 0.998691442030882,
"step": 477,
"total_flos": 0.0,
"train_loss": 0.6354130409048788,
"train_runtime": 12872.7493,
"train_samples_per_second": 4.749,
"train_steps_per_second": 0.037
}
],
"logging_steps": 10,
"max_steps": 477,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 1000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": false,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}