{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 100,
  "global_step": 385,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 52.25,
      "learning_rate": 1.282051282051282e-07,
      "logits/chosen": 88.18099975585938,
      "logits/rejected": 88.25153350830078,
      "logps/chosen": -29.073104858398438,
      "logps/rejected": -26.25731658935547,
      "loss": 1.0,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.03,
      "grad_norm": 43.5,
      "learning_rate": 1.282051282051282e-06,
      "logits/chosen": 81.09413146972656,
      "logits/rejected": 80.79685974121094,
      "logps/chosen": -34.14472961425781,
      "logps/rejected": -33.07267761230469,
      "loss": 0.9155,
      "rewards/accuracies": 0.5555555820465088,
      "rewards/chosen": 0.038799185305833817,
      "rewards/margins": 0.08450882881879807,
      "rewards/rejected": -0.04570963978767395,
      "step": 10
    },
    {
      "epoch": 0.05,
      "grad_norm": 50.5,
      "learning_rate": 2.564102564102564e-06,
      "logits/chosen": 80.67982482910156,
      "logits/rejected": 80.56922149658203,
      "logps/chosen": -33.638526916503906,
      "logps/rejected": -30.80965805053711,
      "loss": 0.9668,
      "rewards/accuracies": 0.5,
      "rewards/chosen": 0.020552771165966988,
      "rewards/margins": 0.03342083841562271,
      "rewards/rejected": -0.012868070974946022,
      "step": 20
    },
    {
      "epoch": 0.08,
      "grad_norm": 48.25,
      "learning_rate": 3.846153846153847e-06,
      "logits/chosen": 82.50788879394531,
      "logits/rejected": 82.54085540771484,
      "logps/chosen": -33.867698669433594,
      "logps/rejected": -31.21173667907715,
      "loss": 1.0398,
      "rewards/accuracies": 0.44999998807907104,
      "rewards/chosen": 0.05804871767759323,
      "rewards/margins": -0.026789050549268723,
      "rewards/rejected": 0.08483776450157166,
      "step": 30
    },
    {
      "epoch": 0.1,
      "grad_norm": 37.0,
      "learning_rate": 4.999896948438434e-06,
      "logits/chosen": 81.12205505371094,
      "logits/rejected": 81.11637878417969,
      "logps/chosen": -32.64350128173828,
      "logps/rejected": -33.15576934814453,
      "loss": 0.9203,
      "rewards/accuracies": 0.5249999761581421,
      "rewards/chosen": 0.21222472190856934,
      "rewards/margins": 0.1271229237318039,
      "rewards/rejected": 0.08510179817676544,
      "step": 40
    },
    {
      "epoch": 0.13,
      "grad_norm": 28.125,
      "learning_rate": 4.987541037542187e-06,
      "logits/chosen": 78.77560424804688,
      "logits/rejected": 78.78538513183594,
      "logps/chosen": -30.54385757446289,
      "logps/rejected": -30.8847599029541,
      "loss": 0.8737,
      "rewards/accuracies": 0.5375000238418579,
      "rewards/chosen": 0.26455581188201904,
      "rewards/margins": 0.18914423882961273,
      "rewards/rejected": 0.07541157305240631,
      "step": 50
    },
    {
      "epoch": 0.16,
      "grad_norm": 40.25,
      "learning_rate": 4.954691471941119e-06,
      "logits/chosen": 83.37484741210938,
      "logits/rejected": 83.43515014648438,
      "logps/chosen": -30.861572265625,
      "logps/rejected": -29.3466796875,
      "loss": 1.0221,
      "rewards/accuracies": 0.4749999940395355,
      "rewards/chosen": 0.12543414533138275,
      "rewards/margins": 0.027898848056793213,
      "rewards/rejected": 0.09753530472517014,
      "step": 60
    },
    {
      "epoch": 0.18,
      "grad_norm": 51.25,
      "learning_rate": 4.901618883413549e-06,
      "logits/chosen": 84.05386352539062,
      "logits/rejected": 84.0824966430664,
      "logps/chosen": -30.508914947509766,
      "logps/rejected": -33.01020431518555,
      "loss": 0.9697,
      "rewards/accuracies": 0.5249999761581421,
      "rewards/chosen": 0.04804382845759392,
      "rewards/margins": 0.0370425209403038,
      "rewards/rejected": 0.011001313105225563,
      "step": 70
    },
    {
      "epoch": 0.21,
      "grad_norm": 40.25,
      "learning_rate": 4.828760511501322e-06,
      "logits/chosen": 81.74610900878906,
      "logits/rejected": 81.73453521728516,
      "logps/chosen": -31.33428382873535,
      "logps/rejected": -30.90468978881836,
      "loss": 0.9004,
      "rewards/accuracies": 0.5375000238418579,
      "rewards/chosen": 0.06826337426900864,
      "rewards/margins": 0.12708182632923126,
      "rewards/rejected": -0.05881844088435173,
      "step": 80
    },
    {
      "epoch": 0.23,
      "grad_norm": 47.75,
      "learning_rate": 4.7367166013034295e-06,
      "logits/chosen": 78.39300537109375,
      "logits/rejected": 78.36354064941406,
      "logps/chosen": -32.50352478027344,
      "logps/rejected": -31.204748153686523,
      "loss": 0.8878,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": 0.05386130139231682,
      "rewards/margins": 0.1551724374294281,
      "rewards/rejected": -0.10131114721298218,
      "step": 90
    },
    {
      "epoch": 0.26,
      "grad_norm": 48.5,
      "learning_rate": 4.626245458345211e-06,
      "logits/chosen": 83.5033950805664,
      "logits/rejected": 83.53245544433594,
      "logps/chosen": -34.1702880859375,
      "logps/rejected": -31.961788177490234,
      "loss": 0.8779,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": 0.05388103798031807,
      "rewards/margins": 0.13837869465351105,
      "rewards/rejected": -0.08449766039848328,
      "step": 100
    },
    {
      "epoch": 0.26,
      "eval_logits/chosen": 98.76129913330078,
      "eval_logits/rejected": 98.75178527832031,
      "eval_logps/chosen": -32.43000030517578,
      "eval_logps/rejected": -36.12626647949219,
      "eval_loss": 0.9410777688026428,
      "eval_rewards/accuracies": 0.5460963845252991,
      "eval_rewards/chosen": 0.005270163994282484,
      "eval_rewards/margins": 0.06916692852973938,
      "eval_rewards/rejected": -0.06389675289392471,
      "eval_runtime": 104.1853,
      "eval_samples_per_second": 3.292,
      "eval_steps_per_second": 0.413,
      "step": 100
    },
    {
      "epoch": 0.29,
      "grad_norm": 53.5,
      "learning_rate": 4.498257201263691e-06,
      "logits/chosen": 83.67477416992188,
      "logits/rejected": 83.56143951416016,
      "logps/chosen": -32.46035385131836,
      "logps/rejected": -32.912376403808594,
      "loss": 0.752,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.22498121857643127,
      "rewards/margins": 0.3275759816169739,
      "rewards/rejected": -0.102594755589962,
      "step": 110
    },
    {
      "epoch": 0.31,
      "grad_norm": 52.25,
      "learning_rate": 4.353806263777678e-06,
      "logits/chosen": 83.85575103759766,
      "logits/rejected": 83.95751953125,
      "logps/chosen": -28.3017520904541,
      "logps/rejected": -35.58953094482422,
      "loss": 0.7604,
      "rewards/accuracies": 0.6499999761581421,
      "rewards/chosen": 0.2511330246925354,
      "rewards/margins": 0.30328577756881714,
      "rewards/rejected": -0.05215279012918472,
      "step": 120
    },
    {
      "epoch": 0.34,
      "grad_norm": 38.0,
      "learning_rate": 4.1940827077152755e-06,
      "logits/chosen": 81.05609130859375,
      "logits/rejected": 81.07179260253906,
      "logps/chosen": -30.556743621826172,
      "logps/rejected": -32.228946685791016,
      "loss": 0.7882,
      "rewards/accuracies": 0.612500011920929,
      "rewards/chosen": 0.14024314284324646,
      "rewards/margins": 0.25916972756385803,
      "rewards/rejected": -0.11892654746770859,
      "step": 130
    },
    {
      "epoch": 0.36,
      "grad_norm": 45.25,
      "learning_rate": 4.0204024186666215e-06,
      "logits/chosen": 82.19850158691406,
      "logits/rejected": 82.19775390625,
      "logps/chosen": -26.9793701171875,
      "logps/rejected": -32.99781036376953,
      "loss": 0.7152,
      "rewards/accuracies": 0.699999988079071,
      "rewards/chosen": 0.18718235194683075,
      "rewards/margins": 0.4169556498527527,
      "rewards/rejected": -0.2297733575105667,
      "step": 140
    },
    {
      "epoch": 0.39,
      "grad_norm": 38.0,
      "learning_rate": 3.834196265035119e-06,
      "logits/chosen": 80.65988159179688,
      "logits/rejected": 80.63349914550781,
      "logps/chosen": -28.97601890563965,
      "logps/rejected": -33.183013916015625,
      "loss": 0.6687,
      "rewards/accuracies": 0.699999988079071,
      "rewards/chosen": 0.16523592174053192,
      "rewards/margins": 0.4118415415287018,
      "rewards/rejected": -0.24660558998584747,
      "step": 150
    },
    {
      "epoch": 0.42,
      "grad_norm": 52.25,
      "learning_rate": 3.636998309800573e-06,
      "logits/chosen": 82.44218444824219,
      "logits/rejected": 82.45413970947266,
      "logps/chosen": -33.828277587890625,
      "logps/rejected": -30.54928207397461,
      "loss": 0.7433,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": 0.12338707596063614,
      "rewards/margins": 0.36208492517471313,
      "rewards/rejected": -0.238697811961174,
      "step": 160
    },
    {
      "epoch": 0.44,
      "grad_norm": 40.25,
      "learning_rate": 3.4304331721118078e-06,
      "logits/chosen": 83.18556213378906,
      "logits/rejected": 83.12220764160156,
      "logps/chosen": -30.9090518951416,
      "logps/rejected": -32.738372802734375,
      "loss": 0.7036,
      "rewards/accuracies": 0.675000011920929,
      "rewards/chosen": 0.14009535312652588,
      "rewards/margins": 0.4274210035800934,
      "rewards/rejected": -0.28732559084892273,
      "step": 170
    },
    {
      "epoch": 0.47,
      "grad_norm": 28.125,
      "learning_rate": 3.2162026428305436e-06,
      "logits/chosen": 80.74728393554688,
      "logits/rejected": 80.72408294677734,
      "logps/chosen": -30.6306095123291,
      "logps/rejected": -31.813491821289062,
      "loss": 0.6778,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.2266170233488083,
      "rewards/margins": 0.48913127183914185,
      "rewards/rejected": -0.2625141739845276,
      "step": 180
    },
    {
      "epoch": 0.49,
      "grad_norm": 23.5,
      "learning_rate": 2.996071664294641e-06,
      "logits/chosen": 82.42842102050781,
      "logits/rejected": 82.41558837890625,
      "logps/chosen": -30.537174224853516,
      "logps/rejected": -30.8133602142334,
      "loss": 0.8564,
      "rewards/accuracies": 0.48750001192092896,
      "rewards/chosen": 0.09169743955135345,
      "rewards/margins": 0.22983141243457794,
      "rewards/rejected": -0.13813400268554688,
      "step": 190
    },
    {
      "epoch": 0.52,
      "grad_norm": 33.5,
      "learning_rate": 2.7718537898066833e-06,
      "logits/chosen": 77.84940338134766,
      "logits/rejected": 77.80615234375,
      "logps/chosen": -33.916507720947266,
      "logps/rejected": -32.88044738769531,
      "loss": 0.683,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.33449333906173706,
      "rewards/margins": 0.4893106520175934,
      "rewards/rejected": -0.1548173427581787,
      "step": 200
    },
    {
      "epoch": 0.52,
      "eval_logits/chosen": 98.61212158203125,
      "eval_logits/rejected": 98.5889663696289,
      "eval_logps/chosen": -32.7288703918457,
      "eval_logps/rejected": -36.450355529785156,
      "eval_loss": 0.9409860372543335,
      "eval_rewards/accuracies": 0.5568937063217163,
      "eval_rewards/chosen": -0.11428017914295197,
      "eval_rewards/margins": 0.07925106585025787,
      "eval_rewards/rejected": -0.19353125989437103,
      "eval_runtime": 104.0333,
      "eval_samples_per_second": 3.297,
      "eval_steps_per_second": 0.413,
      "step": 200
    },
    {
      "epoch": 0.55,
      "grad_norm": 59.0,
      "learning_rate": 2.5453962426402006e-06,
      "logits/chosen": 80.46003723144531,
      "logits/rejected": 80.36625671386719,
      "logps/chosen": -33.2611083984375,
      "logps/rejected": -35.46075439453125,
      "loss": 0.6831,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": 0.27384519577026367,
      "rewards/margins": 0.46370941400527954,
      "rewards/rejected": -0.18986421823501587,
      "step": 210
    },
    {
      "epoch": 0.57,
      "grad_norm": 45.0,
      "learning_rate": 2.3185646976551794e-06,
      "logits/chosen": 82.58009338378906,
      "logits/rejected": 82.66780853271484,
      "logps/chosen": -31.060461044311523,
      "logps/rejected": -31.333576202392578,
      "loss": 0.6185,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": 0.35539740324020386,
      "rewards/margins": 0.5972464084625244,
      "rewards/rejected": -0.24184902012348175,
      "step": 220
    },
    {
      "epoch": 0.6,
      "grad_norm": 46.25,
      "learning_rate": 2.0932279108998323e-06,
      "logits/chosen": 79.74901580810547,
      "logits/rejected": 79.80045318603516,
      "logps/chosen": -32.357086181640625,
      "logps/rejected": -34.57195281982422,
      "loss": 0.7442,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": 0.17946358025074005,
      "rewards/margins": 0.403384268283844,
      "rewards/rejected": -0.22392070293426514,
      "step": 230
    },
    {
      "epoch": 0.62,
      "grad_norm": 47.75,
      "learning_rate": 1.8712423238279358e-06,
      "logits/chosen": 82.10381317138672,
      "logits/rejected": 82.40473937988281,
      "logps/chosen": -30.87957763671875,
      "logps/rejected": -32.0345344543457,
      "loss": 0.5885,
      "rewards/accuracies": 0.7749999761581421,
      "rewards/chosen": 0.3233951926231384,
      "rewards/margins": 0.5243898630142212,
      "rewards/rejected": -0.20099470019340515,
      "step": 240
    },
    {
      "epoch": 0.65,
      "grad_norm": 39.75,
      "learning_rate": 1.6544367689701824e-06,
      "logits/chosen": 80.81050109863281,
      "logits/rejected": 80.87081909179688,
      "logps/chosen": -27.13393783569336,
      "logps/rejected": -30.56355857849121,
      "loss": 0.6984,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": 0.18478286266326904,
      "rewards/margins": 0.44991129636764526,
      "rewards/rejected": -0.2651284337043762,
      "step": 250
    },
    {
      "epoch": 0.68,
      "grad_norm": 30.25,
      "learning_rate": 1.4445974030621963e-06,
      "logits/chosen": 78.12925720214844,
      "logits/rejected": 78.25952911376953,
      "logps/chosen": -30.566410064697266,
      "logps/rejected": -36.688663482666016,
      "loss": 0.5996,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": 0.3978345990180969,
      "rewards/margins": 0.6326489448547363,
      "rewards/rejected": -0.2348143756389618,
      "step": 260
    },
    {
      "epoch": 0.7,
      "grad_norm": 42.25,
      "learning_rate": 1.243452991757889e-06,
      "logits/chosen": 77.4239501953125,
      "logits/rejected": 77.45281219482422,
      "logps/chosen": -30.834758758544922,
      "logps/rejected": -31.846826553344727,
      "loss": 0.626,
      "rewards/accuracies": 0.824999988079071,
      "rewards/chosen": 0.37766537070274353,
      "rewards/margins": 0.5142279863357544,
      "rewards/rejected": -0.13656257092952728,
      "step": 270
    },
    {
      "epoch": 0.73,
      "grad_norm": 54.0,
      "learning_rate": 1.0526606671603523e-06,
      "logits/chosen": 80.0883560180664,
      "logits/rejected": 79.87921905517578,
      "logps/chosen": -31.279199600219727,
      "logps/rejected": -29.9936580657959,
      "loss": 0.7705,
      "rewards/accuracies": 0.699999988079071,
      "rewards/chosen": 0.2122771292924881,
      "rewards/margins": 0.3429415822029114,
      "rewards/rejected": -0.13066445291042328,
      "step": 280
    },
    {
      "epoch": 0.75,
      "grad_norm": 40.5,
      "learning_rate": 8.737922755071455e-07,
      "logits/chosen": 80.207763671875,
      "logits/rejected": 80.12342834472656,
      "logps/chosen": -33.07125473022461,
      "logps/rejected": -32.81315612792969,
      "loss": 0.5368,
      "rewards/accuracies": 0.75,
      "rewards/chosen": 0.41319671273231506,
      "rewards/margins": 0.7317113280296326,
      "rewards/rejected": -0.3185145854949951,
      "step": 290
    },
    {
      "epoch": 0.78,
      "grad_norm": 43.0,
      "learning_rate": 7.08321427484816e-07,
      "logits/chosen": 75.83465576171875,
      "logits/rejected": 75.93038177490234,
      "logps/chosen": -32.472633361816406,
      "logps/rejected": -29.376873016357422,
      "loss": 0.7349,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.37323057651519775,
      "rewards/margins": 0.48316845297813416,
      "rewards/rejected": -0.109937883913517,
      "step": 300
    },
    {
      "epoch": 0.78,
      "eval_logits/chosen": 98.62049102783203,
      "eval_logits/rejected": 98.59721374511719,
      "eval_logps/chosen": -32.674007415771484,
      "eval_logps/rejected": -36.29110336303711,
      "eval_loss": 0.9808743000030518,
      "eval_rewards/accuracies": 0.5157807469367981,
      "eval_rewards/chosen": -0.0923333466053009,
      "eval_rewards/margins": 0.03749755397439003,
      "eval_rewards/rejected": -0.12983089685440063,
      "eval_runtime": 103.8205,
      "eval_samples_per_second": 3.304,
      "eval_steps_per_second": 0.414,
      "step": 300
    },
    {
      "epoch": 0.81,
      "grad_norm": 40.0,
      "learning_rate": 5.576113578589035e-07,
      "logits/chosen": 82.97636413574219,
      "logits/rejected": 83.00100708007812,
      "logps/chosen": -30.166763305664062,
      "logps/rejected": -32.79009246826172,
      "loss": 0.6505,
      "rewards/accuracies": 0.800000011920929,
      "rewards/chosen": 0.278707891702652,
      "rewards/margins": 0.5148409008979797,
      "rewards/rejected": -0.23613300919532776,
      "step": 310
    },
    {
      "epoch": 0.83,
      "grad_norm": 32.5,
      "learning_rate": 4.229036944380913e-07,
      "logits/chosen": 80.45620727539062,
      "logits/rejected": 80.45777893066406,
      "logps/chosen": -30.465673446655273,
      "logps/rejected": -29.35947036743164,
      "loss": 0.5314,
      "rewards/accuracies": 0.762499988079071,
      "rewards/chosen": 0.45315417647361755,
      "rewards/margins": 0.6291457414627075,
      "rewards/rejected": -0.17599165439605713,
      "step": 320
    },
    {
      "epoch": 0.86,
      "grad_norm": 34.0,
      "learning_rate": 3.053082288996112e-07,
      "logits/chosen": 77.60248565673828,
      "logits/rejected": 77.65824890136719,
      "logps/chosen": -29.066116333007812,
      "logps/rejected": -33.186683654785156,
      "loss": 0.4902,
      "rewards/accuracies": 0.75,
      "rewards/chosen": 0.5145958065986633,
      "rewards/margins": 0.735322117805481,
      "rewards/rejected": -0.2207263708114624,
      "step": 330
    },
    {
      "epoch": 0.88,
      "grad_norm": 46.75,
      "learning_rate": 2.0579377374915805e-07,
      "logits/chosen": 81.93863677978516,
      "logits/rejected": 81.96864318847656,
      "logps/chosen": -32.32833480834961,
      "logps/rejected": -33.94022750854492,
      "loss": 0.6732,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": 0.35352635383605957,
      "rewards/margins": 0.5821414589881897,
      "rewards/rejected": -0.22861509025096893,
      "step": 340
    },
    {
      "epoch": 0.91,
      "grad_norm": 31.875,
      "learning_rate": 1.2518018074041684e-07,
      "logits/chosen": 80.9842529296875,
      "logits/rejected": 80.99238586425781,
      "logps/chosen": -32.631919860839844,
      "logps/rejected": -33.691749572753906,
      "loss": 0.5834,
      "rewards/accuracies": 0.7250000238418579,
      "rewards/chosen": 0.4047267436981201,
      "rewards/margins": 0.6447058320045471,
      "rewards/rejected": -0.23997914791107178,
      "step": 350
    },
    {
      "epoch": 0.94,
      "grad_norm": 29.125,
      "learning_rate": 6.41315865106129e-08,
      "logits/chosen": 82.46900177001953,
      "logits/rejected": 82.49642944335938,
      "logps/chosen": -28.48472023010254,
      "logps/rejected": -31.835073471069336,
      "loss": 0.6674,
      "rewards/accuracies": 0.6499999761581421,
      "rewards/chosen": 0.42699599266052246,
      "rewards/margins": 0.5010865926742554,
      "rewards/rejected": -0.0740906298160553,
      "step": 360
    },
    {
      "epoch": 0.96,
      "grad_norm": 36.0,
      "learning_rate": 2.3150941078050325e-08,
      "logits/chosen": 81.88428497314453,
      "logits/rejected": 81.90206909179688,
      "logps/chosen": -31.865198135375977,
      "logps/rejected": -35.7531623840332,
      "loss": 0.5901,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": 0.40438637137413025,
      "rewards/margins": 0.6619016528129578,
      "rewards/rejected": -0.2575153410434723,
      "step": 370
    },
    {
      "epoch": 0.99,
      "grad_norm": 36.0,
      "learning_rate": 2.575864278703266e-09,
      "logits/chosen": 75.7695083618164,
      "logits/rejected": 75.64456939697266,
      "logps/chosen": -29.696298599243164,
      "logps/rejected": -28.395023345947266,
      "loss": 0.6836,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": 0.35193198919296265,
      "rewards/margins": 0.4457107484340668,
      "rewards/rejected": -0.09377876669168472,
      "step": 380
    },
    {
      "epoch": 1.0,
      "step": 385,
      "total_flos": 0.0,
      "train_loss": 0.741242148659446,
      "train_runtime": 2553.0854,
      "train_samples_per_second": 1.206,
      "train_steps_per_second": 0.151
    }
  ],
  "logging_steps": 10,
  "max_steps": 385,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}