{
  "best_global_step": 437,
  "best_metric": 0.35658061504364014,
  "best_model_checkpoint": "/NHNHOME/WORKSPACE/26msit001_A/mnpo/revision_qwen3_8b/full_iter1/train/ronpo_k_only_s42_nhn-b200-q3-seq-07101430-konly/checkpoint-437",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 437,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02288329519450801,
      "grad_norm": 507.4701728682287,
      "learning_rate": 1.0227272727272728e-07,
      "logits/chosen": -0.79296875,
      "logits/rejected": -0.82421875,
      "logps/chosen": -15.875,
      "logps/rejected": -15.8125,
      "loss": 0.7230703234672546,
      "rewards/accuracies": 0.5375000238418579,
      "rewards/chosen": -153.16549682617188,
      "rewards/margins": 1.1008796691894531,
      "rewards/rejected": -154.26638793945312,
      "ronpo/logit": 0.11004491150379181,
      "ronpo/residual": 0.09169502556324005,
      "ronpo/residual_abs": 0.4952625334262848,
      "ronpo/target": 0.018349889665842056,
      "ronpo/target_abs": 0.26459527015686035,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 10
    },
    {
      "epoch": 0.04576659038901602,
      "grad_norm": 336.7714258925431,
      "learning_rate": 2.159090909090909e-07,
      "logits/chosen": -0.82421875,
      "logits/rejected": -0.82421875,
      "logps/chosen": -15.875,
      "logps/rejected": -15.9375,
      "loss": 0.6306931376457214,
      "rewards/accuracies": 0.612500011920929,
      "rewards/chosen": -153.71337890625,
      "rewards/margins": 1.7559814453125,
      "rewards/rejected": -155.4693603515625,
      "ronpo/logit": 0.1755521446466446,
      "ronpo/residual": 0.1825995147228241,
      "ronpo/residual_abs": 0.4917503297328949,
      "ronpo/target": -0.0070473672822117805,
      "ronpo/target_abs": 0.2585894465446472,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 20
    },
    {
      "epoch": 0.06864988558352403,
      "grad_norm": 179.93401054763783,
      "learning_rate": 3.295454545454545e-07,
      "logits/chosen": -0.83984375,
      "logits/rejected": -0.8203125,
      "logps/chosen": -15.8125,
      "logps/rejected": -15.875,
      "loss": 0.7665756940841675,
      "rewards/accuracies": 0.574999988079071,
      "rewards/chosen": -152.2275390625,
      "rewards/margins": 2.1743698120117188,
      "rewards/rejected": -154.4019012451172,
      "ronpo/logit": 0.21744287014007568,
      "ronpo/residual": 0.21943974494934082,
      "ronpo/residual_abs": 0.5555931925773621,
      "ronpo/target": -0.0019968985579907894,
      "ronpo/target_abs": 0.23704366385936737,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 30
    },
    {
      "epoch": 0.09153318077803203,
      "grad_norm": 188.32242349357378,
      "learning_rate": 4.4318181818181817e-07,
      "logits/chosen": -0.8515625,
      "logits/rejected": -0.84375,
      "logps/chosen": -15.75,
      "logps/rejected": -15.8125,
      "loss": 0.5709580183029175,
      "rewards/accuracies": 0.574999988079071,
      "rewards/chosen": -152.11019897460938,
      "rewards/margins": 1.608306884765625,
      "rewards/rejected": -153.718505859375,
      "ronpo/logit": 0.16085514426231384,
      "ronpo/residual": 0.1310609132051468,
      "ronpo/residual_abs": 0.4952216148376465,
      "ronpo/target": 0.02979421615600586,
      "ronpo/target_abs": 0.24758920073509216,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 40
    },
    {
      "epoch": 0.11441647597254005,
      "grad_norm": 178.24231576952798,
      "learning_rate": 4.998003327445819e-07,
      "logits/chosen": -0.83984375,
      "logits/rejected": -0.86328125,
      "logps/chosen": -15.625,
      "logps/rejected": -15.6875,
      "loss": 0.49001631140708923,
      "rewards/accuracies": 0.643750011920929,
      "rewards/chosen": -151.10308837890625,
      "rewards/margins": 1.7249259948730469,
      "rewards/rejected": -152.8280029296875,
      "ronpo/logit": 0.17246504127979279,
      "ronpo/residual": 0.14314237236976624,
      "ronpo/residual_abs": 0.4705863893032074,
      "ronpo/target": 0.02932267263531685,
      "ronpo/target_abs": 0.24458365142345428,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 50
    },
    {
      "epoch": 0.13729977116704806,
      "grad_norm": 189.3688920695095,
      "learning_rate": 4.98204907808407e-07,
      "logits/chosen": -0.9296875,
      "logits/rejected": -0.93359375,
      "logps/chosen": -15.1875,
      "logps/rejected": -15.1875,
      "loss": 1.0039470195770264,
      "rewards/accuracies": 0.6000000238418579,
      "rewards/chosen": -146.05941772460938,
      "rewards/margins": 1.8618888854980469,
      "rewards/rejected": -147.9213104248047,
      "ronpo/logit": 0.1862778663635254,
      "ronpo/residual": 0.19359330832958221,
      "ronpo/residual_abs": 0.5423773527145386,
      "ronpo/target": -0.007315444760024548,
      "ronpo/target_abs": 0.25545692443847656,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 60
    },
    {
      "epoch": 0.16018306636155608,
      "grad_norm": 221.19383431375573,
      "learning_rate": 4.95024247597132e-07,
      "logits/chosen": -1.0078125,
      "logits/rejected": -1.0078125,
      "logps/chosen": -14.8125,
      "logps/rejected": -14.8125,
      "loss": 0.5034725069999695,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -142.9036102294922,
      "rewards/margins": 1.5878067016601562,
      "rewards/rejected": -144.49142456054688,
      "ronpo/logit": 0.15876074135303497,
      "ronpo/residual": 0.16019420325756073,
      "ronpo/residual_abs": 0.4973166882991791,
      "ronpo/target": -0.00143346784170717,
      "ronpo/target_abs": 0.2591482996940613,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 70
    },
    {
      "epoch": 0.18306636155606407,
      "grad_norm": 166.72773225533953,
      "learning_rate": 4.902786663535712e-07,
      "logits/chosen": -1.0078125,
      "logits/rejected": -1.0234375,
      "logps/chosen": -14.4375,
      "logps/rejected": -14.375,
      "loss": 0.5039559602737427,
      "rewards/accuracies": 0.5375000238418579,
      "rewards/chosen": -139.4639129638672,
      "rewards/margins": 0.5908203125,
      "rewards/rejected": -140.0547332763672,
      "ronpo/logit": 0.05916024371981621,
      "ronpo/residual": 0.07047247886657715,
      "ronpo/residual_abs": 0.45131802558898926,
      "ronpo/target": -0.011312246322631836,
      "ronpo/target_abs": 0.26929640769958496,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 80
    },
    {
      "epoch": 0.20594965675057209,
      "grad_norm": 224.52627431026863,
      "learning_rate": 4.8399847315926e-07,
      "logits/chosen": -1.015625,
      "logits/rejected": -1.0234375,
      "logps/chosen": -14.1875,
      "logps/rejected": -14.125,
      "loss": 1.0758483409881592,
      "rewards/accuracies": 0.5874999761581421,
      "rewards/chosen": -134.73431396484375,
      "rewards/margins": 2.2587051391601562,
      "rewards/rejected": -136.99301147460938,
      "ronpo/logit": 0.225923091173172,
      "ronpo/residual": 0.21176382899284363,
      "ronpo/residual_abs": 0.5388052463531494,
      "ronpo/target": 0.0141592500731349,
      "ronpo/target_abs": 0.24621061980724335,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 90
    },
    {
      "epoch": 0.2288329519450801,
      "grad_norm": 193.7153670572699,
      "learning_rate": 4.762237783563906e-07,
      "logits/chosen": -0.99609375,
      "logits/rejected": -1.0078125,
      "logps/chosen": -13.875,
      "logps/rejected": -13.8125,
      "loss": 0.507349967956543,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -133.3758544921875,
      "rewards/margins": 1.0252304077148438,
      "rewards/rejected": -134.40109252929688,
      "ronpo/logit": 0.10251834243535995,
      "ronpo/residual": 0.09443376958370209,
      "ronpo/residual_abs": 0.43109574913978577,
      "ronpo/target": 0.008084583096206188,
      "ronpo/target_abs": 0.2518351674079895,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 100
    },
    {
      "epoch": 0.2517162471395881,
      "grad_norm": 162.81045220693468,
      "learning_rate": 4.670042373710523e-07,
      "logits/chosen": -1.0078125,
      "logits/rejected": -1.0,
      "logps/chosen": -13.6875,
      "logps/rejected": -13.625,
      "loss": 1.1557821035385132,
      "rewards/accuracies": 0.581250011920929,
      "rewards/chosen": -130.3563232421875,
      "rewards/margins": 2.2956924438476562,
      "rewards/rejected": -132.65200805664062,
      "ronpo/logit": 0.22950632870197296,
      "ronpo/residual": 0.2178325206041336,
      "ronpo/residual_abs": 0.5105395913124084,
      "ronpo/target": 0.011673808097839355,
      "ronpo/target_abs": 0.24373772740364075,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 110
    },
    {
      "epoch": 0.2745995423340961,
      "grad_norm": 188.78038929481818,
      "learning_rate": 4.5639873357392157e-07,
      "logits/chosen": -0.97265625,
      "logits/rejected": -0.95703125,
      "logps/chosen": -13.375,
      "logps/rejected": -13.4375,
      "loss": 0.47604840993881226,
      "rewards/accuracies": 0.612500011920929,
      "rewards/chosen": -128.6849365234375,
      "rewards/margins": 1.71466064453125,
      "rewards/rejected": -130.3996124267578,
      "ronpo/logit": 0.17151877284049988,
      "ronpo/residual": 0.1699337661266327,
      "ronpo/residual_abs": 0.4403519630432129,
      "ronpo/target": 0.0015850067138671875,
      "ronpo/target_abs": 0.25699958205223083,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 120
    },
    {
      "epoch": 0.2974828375286041,
      "grad_norm": 34.783846491386534,
      "learning_rate": 4.444750022039099e-07,
      "logits/chosen": -0.92578125,
      "logits/rejected": -0.9453125,
      "logps/chosen": -13.375,
      "logps/rejected": -13.375,
      "loss": 0.4631749093532562,
      "rewards/accuracies": 0.606249988079071,
      "rewards/chosen": -128.41290283203125,
      "rewards/margins": 1.4823760986328125,
      "rewards/rejected": -129.89527893066406,
      "ronpo/logit": 0.1482272893190384,
      "ronpo/residual": 0.13759438693523407,
      "ronpo/residual_abs": 0.41783371567726135,
      "ronpo/target": 0.010632896795868874,
      "ronpo/target_abs": 0.2536512315273285,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 130
    },
    {
      "epoch": 0.32036613272311215,
      "grad_norm": 66.75154575107938,
      "learning_rate": 4.313091977566937e-07,
      "logits/chosen": -0.90625,
      "logits/rejected": -0.921875,
      "logps/chosen": -13.3125,
      "logps/rejected": -13.25,
      "loss": 0.7874235510826111,
      "rewards/accuracies": 0.612500011920929,
      "rewards/chosen": -126.31758880615234,
      "rewards/margins": 2.0760345458984375,
      "rewards/rejected": -128.3936309814453,
      "ronpo/logit": 0.20748524367809296,
      "ronpo/residual": 0.20403257012367249,
      "ronpo/residual_abs": 0.4981064796447754,
      "ronpo/target": 0.00345268240198493,
      "ronpo/target_abs": 0.2723880708217621,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 140
    },
    {
      "epoch": 0.34324942791762014,
      "grad_norm": 192.9229979776684,
      "learning_rate": 4.169854076011292e-07,
      "logits/chosen": -0.86328125,
      "logits/rejected": -0.87109375,
      "logps/chosen": -13.25,
      "logps/rejected": -13.3125,
      "loss": 0.2791599631309509,
      "rewards/accuracies": 0.612500011920929,
      "rewards/chosen": -127.51309967041016,
      "rewards/margins": 1.3698883056640625,
      "rewards/rejected": -128.8829803466797,
      "ronpo/logit": 0.1369936764240265,
      "ronpo/residual": 0.14234663546085358,
      "ronpo/residual_abs": 0.39549511671066284,
      "ronpo/target": -0.005352973937988281,
      "ronpo/target_abs": 0.276210218667984,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 150
    },
    {
      "epoch": 0.36613272311212813,
      "grad_norm": 34.9536433242253,
      "learning_rate": 4.0159511492998744e-07,
      "logits/chosen": -0.84765625,
      "logits/rejected": -0.8515625,
      "logps/chosen": -13.1875,
      "logps/rejected": -13.125,
      "loss": 0.3614216148853302,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -126.5057144165039,
      "rewards/margins": 1.162130355834961,
      "rewards/rejected": -127.66786193847656,
      "ronpo/logit": 0.11622162908315659,
      "ronpo/residual": 0.15131323039531708,
      "ronpo/residual_abs": 0.3710349202156067,
      "ronpo/target": -0.0350915901362896,
      "ronpo/target_abs": 0.25014859437942505,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 160
    },
    {
      "epoch": 0.3890160183066362,
      "grad_norm": 42.22385969745458,
      "learning_rate": 3.8523661447503574e-07,
      "logits/chosen": -0.83203125,
      "logits/rejected": -0.828125,
      "logps/chosen": -13.3125,
      "logps/rejected": -13.25,
      "loss": 0.21069881319999695,
      "rewards/accuracies": 0.606249988079071,
      "rewards/chosen": -127.7530517578125,
      "rewards/margins": 0.88427734375,
      "rewards/rejected": -128.6373291015625,
      "ronpo/logit": 0.08846603333950043,
      "ronpo/residual": 0.06841827929019928,
      "ronpo/residual_abs": 0.35050949454307556,
      "ronpo/target": 0.020047759637236595,
      "ronpo/target_abs": 0.2535446286201477,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 170
    },
    {
      "epoch": 0.41189931350114417,
      "grad_norm": 72.12426453546038,
      "learning_rate": 3.6801438471817826e-07,
      "logits/chosen": -0.77734375,
      "logits/rejected": -0.7578125,
      "logps/chosen": -13.25,
      "logps/rejected": -13.25,
      "loss": 0.5257893800735474,
      "rewards/accuracies": 0.6000000238418579,
      "rewards/chosen": -127.363037109375,
      "rewards/margins": 1.278564453125,
      "rewards/rejected": -128.6416015625,
      "ronpo/logit": 0.12785287201404572,
      "ronpo/residual": 0.14572112262248993,
      "ronpo/residual_abs": 0.3929472267627716,
      "ronpo/target": -0.01786825619637966,
      "ronpo/target_abs": 0.272230327129364,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 180
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 48.4203371563711,
      "learning_rate": 3.5003842060821546e-07,
      "logits/chosen": -0.70703125,
      "logits/rejected": -0.703125,
      "logps/chosen": -13.375,
      "logps/rejected": -13.3125,
      "loss": 0.4416274428367615,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": -128.242431640625,
      "rewards/margins": 1.3032684326171875,
      "rewards/rejected": -129.5457000732422,
      "ronpo/logit": 0.13027086853981018,
      "ronpo/residual": 0.13186077773571014,
      "ronpo/residual_abs": 0.40120625495910645,
      "ronpo/target": -0.0015898942947387695,
      "ronpo/target_abs": 0.26229482889175415,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 190
    },
    {
      "epoch": 0.4576659038901602,
      "grad_norm": 32.19528363036125,
      "learning_rate": 3.3142353104503053e-07,
      "logits/chosen": -0.6328125,
      "logits/rejected": -0.6328125,
      "logps/chosen": -13.4375,
      "logps/rejected": -13.5,
      "loss": 0.4198226034641266,
      "rewards/accuracies": 0.574999988079071,
      "rewards/chosen": -128.51138305664062,
      "rewards/margins": 1.9074668884277344,
      "rewards/rejected": -130.41885375976562,
      "ronpo/logit": 0.19083575904369354,
      "ronpo/residual": 0.2115732729434967,
      "ronpo/residual_abs": 0.41171345114707947,
      "ronpo/target": -0.020737504586577415,
      "ronpo/target_abs": 0.24886098504066467,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 200
    },
    {
      "epoch": 0.4805491990846682,
      "grad_norm": 57.99776711568966,
      "learning_rate": 3.122886056180284e-07,
      "logits/chosen": -0.6015625,
      "logits/rejected": -0.59375,
      "logps/chosen": -13.5,
      "logps/rejected": -13.5,
      "loss": 0.5668808221817017,
      "rewards/accuracies": 0.6187499761581421,
      "rewards/chosen": -128.95379638671875,
      "rewards/margins": 2.156829833984375,
      "rewards/rejected": -131.11062622070312,
      "ronpo/logit": 0.21571657061576843,
      "ronpo/residual": 0.2658722400665283,
      "ronpo/residual_abs": 0.4744180738925934,
      "ronpo/target": -0.0501556396484375,
      "ronpo/target_abs": 0.26495361328125,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 210
    },
    {
      "epoch": 0.5034324942791762,
      "grad_norm": 66.5748797349874,
      "learning_rate": 2.927558552820251e-07,
      "logits/chosen": -0.57421875,
      "logits/rejected": -0.56640625,
      "logps/chosen": -13.5625,
      "logps/rejected": -13.5625,
      "loss": 0.8348878622055054,
      "rewards/accuracies": 0.5687500238418579,
      "rewards/chosen": -129.84304809570312,
      "rewards/margins": 1.874420166015625,
      "rewards/rejected": -131.71746826171875,
      "ronpo/logit": 0.18745450675487518,
      "ronpo/residual": 0.18469519913196564,
      "ronpo/residual_abs": 0.4471713602542877,
      "ronpo/target": 0.002759313676506281,
      "ronpo/target_abs": 0.27416548132896423,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 220
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 59.64590279351556,
      "learning_rate": 2.729500318202379e-07,
      "logits/chosen": -0.5625,
      "logits/rejected": -0.5703125,
      "logps/chosen": -13.625,
      "logps/rejected": -13.5625,
      "loss": 1.0018385648727417,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": -129.21832275390625,
      "rewards/margins": 2.456756591796875,
      "rewards/rejected": -131.67507934570312,
      "ronpo/logit": 0.24569907784461975,
      "ronpo/residual": 0.2502497136592865,
      "ronpo/residual_abs": 0.5137172937393188,
      "ronpo/target": -0.004550647921860218,
      "ronpo/target_abs": 0.25031614303588867,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 230
    },
    {
      "epoch": 0.5491990846681922,
      "grad_norm": 53.78348871569346,
      "learning_rate": 2.5299763107951076e-07,
      "logits/chosen": -0.5859375,
      "logits/rejected": -0.58203125,
      "logps/chosen": -13.625,
      "logps/rejected": -13.625,
      "loss": 0.45406943559646606,
      "rewards/accuracies": 0.668749988079071,
      "rewards/chosen": -129.8865966796875,
      "rewards/margins": 1.919342041015625,
      "rewards/rejected": -131.80593872070312,
      "ronpo/logit": 0.19197313487529755,
      "ronpo/residual": 0.2080519199371338,
      "ronpo/residual_abs": 0.4232204556465149,
      "ronpo/target": -0.016078775748610497,
      "ronpo/target_abs": 0.2563047707080841,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 240
    },
    {
      "epoch": 0.5720823798627003,
      "grad_norm": 314.6485896661544,
      "learning_rate": 2.3302608506655286e-07,
      "logits/chosen": -0.55859375,
      "logits/rejected": -0.5625,
      "logps/chosen": -13.5,
      "logps/rejected": -13.5625,
      "loss": 0.3197762072086334,
      "rewards/accuracies": 0.5249999761581421,
      "rewards/chosen": -130.083740234375,
      "rewards/margins": 1.003997802734375,
      "rewards/rejected": -131.08773803710938,
      "ronpo/logit": 0.10035202652215958,
      "ronpo/residual": 0.09121748805046082,
      "ronpo/residual_abs": 0.37010660767555237,
      "ronpo/target": 0.009134531021118164,
      "ronpo/target_abs": 0.26355957984924316,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 250
    },
    {
      "epoch": 0.5949656750572082,
      "grad_norm": 68.55328552753093,
      "learning_rate": 2.131629480651116e-07,
      "logits/chosen": -0.55078125,
      "logits/rejected": -0.55078125,
      "logps/chosen": -13.625,
      "logps/rejected": -13.5625,
      "loss": 0.2595209777355194,
      "rewards/accuracies": 0.6812499761581421,
      "rewards/chosen": -130.5323486328125,
      "rewards/margins": 1.3940658569335938,
      "rewards/rejected": -131.92642211914062,
      "ronpo/logit": 0.13951639831066132,
      "ronpo/residual": 0.16778139770030975,
      "ronpo/residual_abs": 0.3852796256542206,
      "ronpo/target": -0.028264999389648438,
      "ronpo/target_abs": 0.2798442840576172,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 260
    },
    {
      "epoch": 0.6178489702517163,
      "grad_norm": 90.54243633746498,
      "learning_rate": 1.935350819721849e-07,
      "logits/chosen": -0.51953125,
      "logits/rejected": -0.5234375,
      "logps/chosen": -13.625,
      "logps/rejected": -13.625,
      "loss": 0.41456565260887146,
      "rewards/accuracies": 0.5874999761581421,
      "rewards/chosen": -130.73831176757812,
      "rewards/margins": 1.3446502685546875,
      "rewards/rejected": -132.0829620361328,
      "ronpo/logit": 0.134442999958992,
      "ronpo/residual": 0.09629812836647034,
      "ronpo/residual_abs": 0.38442301750183105,
      "ronpo/target": 0.038144875317811966,
      "ronpo/target_abs": 0.23953285813331604,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 270
    },
    {
      "epoch": 0.6407322654462243,
      "grad_norm": 91.49025424779988,
      "learning_rate": 1.742678460563715e-07,
      "logits/chosen": -0.498046875,
      "logits/rejected": -0.5078125,
      "logps/chosen": -13.6875,
      "logps/rejected": -13.6875,
      "loss": 0.7765046954154968,
      "rewards/accuracies": 0.6499999761581421,
      "rewards/chosen": -130.361572265625,
      "rewards/margins": 2.2234344482421875,
      "rewards/rejected": -132.5850067138672,
      "ronpo/logit": 0.22241516411304474,
      "ronpo/residual": 0.19911065697669983,
      "ronpo/residual_abs": 0.44603413343429565,
      "ronpo/target": 0.023304533213377,
      "ronpo/target_abs": 0.26094645261764526,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 280
    },
    {
      "epoch": 0.6636155606407322,
      "grad_norm": 120.63542765790571,
      "learning_rate": 1.5548429631320804e-07,
      "logits/chosen": -0.50390625,
      "logits/rejected": -0.50390625,
      "logps/chosen": -13.6875,
      "logps/rejected": -13.6875,
      "loss": 0.6691421270370483,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -130.86163330078125,
      "rewards/margins": 1.9347991943359375,
      "rewards/rejected": -132.7964324951172,
      "ronpo/logit": 0.1935712993144989,
      "ronpo/residual": 0.19426050782203674,
      "ronpo/residual_abs": 0.464620977640152,
      "ronpo/target": -0.0006892204401083291,
      "ronpo/target_abs": 0.25403547286987305,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 290
    },
    {
      "epoch": 0.6864988558352403,
      "grad_norm": 55.729121734986606,
      "learning_rate": 1.373043995310524e-07,
      "logits/chosen": -0.490234375,
      "logits/rejected": -0.482421875,
      "logps/chosen": -13.6875,
      "logps/rejected": -13.625,
      "loss": 0.6174883842468262,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": -130.33990478515625,
      "rewards/margins": 2.029083251953125,
      "rewards/rejected": -132.36898803710938,
      "ronpo/logit": 0.20279976725578308,
      "ronpo/residual": 0.20034602284431458,
      "ronpo/residual_abs": 0.43287530541419983,
      "ronpo/target": 0.0024537325371056795,
      "ronpo/target_abs": 0.24373027682304382,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 300
    },
    {
      "epoch": 0.7093821510297483,
      "grad_norm": 31.586368551140392,
      "learning_rate": 1.1984426708711804e-07,
      "logits/chosen": -0.50390625,
      "logits/rejected": -0.49609375,
      "logps/chosen": -13.5625,
      "logps/rejected": -13.5625,
      "loss": 0.26410332322120667,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": -130.807373046875,
      "rewards/margins": 1.055145263671875,
      "rewards/rejected": -131.86251831054688,
      "ronpo/logit": 0.10549856722354889,
      "ronpo/residual": 0.14673687517642975,
      "ronpo/residual_abs": 0.3895283341407776,
      "ronpo/target": -0.04123830795288086,
      "ronpo/target_abs": 0.2849213480949402,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 310
    },
    {
      "epoch": 0.7322654462242563,
      "grad_norm": 41.977465977771885,
      "learning_rate": 1.0321541336725026e-07,
      "logits/chosen": -0.498046875,
      "logits/rejected": -0.5,
      "logps/chosen": -13.625,
      "logps/rejected": -13.625,
      "loss": 0.39825934171676636,
      "rewards/accuracies": 0.675000011920929,
      "rewards/chosen": -130.84564208984375,
      "rewards/margins": 1.891153335571289,
      "rewards/rejected": -132.73678588867188,
      "ronpo/logit": 0.1891597956418991,
      "ronpo/residual": 0.19329676032066345,
      "ronpo/residual_abs": 0.38093996047973633,
      "ronpo/target": -0.004136943724006414,
      "ronpo/target_abs": 0.2447056770324707,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 320
    },
    {
      "epoch": 0.7551487414187643,
      "grad_norm": 97.11696106748332,
      "learning_rate": 8.75240435457705e-08,
      "logits/chosen": -0.5,
      "logits/rejected": -0.494140625,
      "logps/chosen": -13.625,
      "logps/rejected": -13.625,
      "loss": 0.40726518630981445,
      "rewards/accuracies": 0.606249988079071,
      "rewards/chosen": -130.62158203125,
      "rewards/margins": 1.356597900390625,
      "rewards/rejected": -131.97817993164062,
      "ronpo/logit": 0.1355902999639511,
      "ronpo/residual": 0.13591110706329346,
      "ronpo/residual_abs": 0.4075256884098053,
      "ronpo/target": -0.0003208100679330528,
      "ronpo/target_abs": 0.2613794803619385,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 330
    },
    {
      "epoch": 0.7780320366132724,
      "grad_norm": 16.556229911254658,
      "learning_rate": 7.287037527419604e-08,
      "logits/chosen": -0.51171875,
      "logits/rejected": -0.5078125,
      "logps/chosen": -13.6875,
      "logps/rejected": -13.75,
      "loss": 1.1543911695480347,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -130.10595703125,
      "rewards/margins": 2.9382247924804688,
      "rewards/rejected": -133.044189453125,
      "ronpo/logit": 0.2938784062862396,
      "ronpo/residual": 0.28921040892601013,
      "ronpo/residual_abs": 0.4960741102695465,
      "ronpo/target": 0.004667997360229492,
      "ronpo/target_abs": 0.26037946343421936,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 340
    },
    {
      "epoch": 0.8009153318077803,
      "grad_norm": 49.66277733374995,
      "learning_rate": 5.934799861107506e-08,
      "logits/chosen": -0.490234375,
      "logits/rejected": -0.50390625,
      "logps/chosen": -13.5625,
      "logps/rejected": -13.5625,
      "loss": 0.2971951365470886,
      "rewards/accuracies": 0.550000011920929,
      "rewards/chosen": -130.4481201171875,
      "rewards/margins": 0.9669418334960938,
      "rewards/rejected": -131.41506958007812,
      "ronpo/logit": 0.09669926017522812,
      "ronpo/residual": 0.08134034276008606,
      "ronpo/residual_abs": 0.3665750324726105,
      "ronpo/target": 0.015358924865722656,
      "ronpo/target_abs": 0.2653478682041168,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 350
    },
    {
      "epoch": 0.8237986270022883,
      "grad_norm": 31.948876313907636,
      "learning_rate": 4.7043278280936406e-08,
      "logits/chosen": -0.51171875,
      "logits/rejected": -0.51171875,
      "logps/chosen": -13.5,
      "logps/rejected": -13.5625,
      "loss": 0.27478647232055664,
      "rewards/accuracies": 0.675000011920929,
      "rewards/chosen": -130.006591796875,
      "rewards/margins": 1.4485015869140625,
      "rewards/rejected": -131.45510864257812,
      "ronpo/logit": 0.14487479627132416,
      "ronpo/residual": 0.18561749160289764,
      "ronpo/residual_abs": 0.3878316283226013,
      "ronpo/target": -0.04074268415570259,
      "ronpo/target_abs": 0.23417606949806213,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 360
    },
    {
      "epoch": 0.8466819221967964,
      "grad_norm": 69.93387749352912,
      "learning_rate": 3.603480208000781e-08,
      "logits/chosen": -0.48828125,
      "logits/rejected": -0.484375,
      "logps/chosen": -13.625,
      "logps/rejected": -13.625,
      "loss": 0.4619961380958557,
      "rewards/accuracies": 0.6000000238418579,
      "rewards/chosen": -130.478759765625,
      "rewards/margins": 1.3962898254394531,
      "rewards/rejected": -131.8750457763672,
      "ronpo/logit": 0.13958875834941864,
      "ronpo/residual": 0.1094854325056076,
      "ronpo/residual_abs": 0.4327429234981537,
      "ronpo/target": 0.030103301629424095,
      "ronpo/target_abs": 0.2981323301792145,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 370
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 82.61470418012827,
      "learning_rate": 2.6392878951625258e-08,
      "logits/chosen": -0.470703125,
      "logits/rejected": -0.486328125,
      "logps/chosen": -13.625,
      "logps/rejected": -13.625,
      "loss": 0.6330326795578003,
      "rewards/accuracies": 0.606249988079071,
      "rewards/chosen": -131.38787841796875,
      "rewards/margins": 0.8808364868164062,
      "rewards/rejected": -132.2687225341797,
      "ronpo/logit": 0.08816994726657867,
      "ronpo/residual": 0.10562714189291,
      "ronpo/residual_abs": 0.4062245786190033,
      "ronpo/target": -0.017457198351621628,
      "ronpo/target_abs": 0.28133049607276917,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 380
    },
    {
      "epoch": 0.8924485125858124,
      "grad_norm": 29.76910596480733,
      "learning_rate": 1.8179089937025548e-08,
      "logits/chosen": -0.50390625,
      "logits/rejected": -0.50390625,
      "logps/chosen": -13.5625,
      "logps/rejected": -13.5,
      "loss": 0.20862670242786407,
      "rewards/accuracies": 0.5562499761581421,
      "rewards/chosen": -130.48080444335938,
      "rewards/margins": 0.68048095703125,
      "rewards/rejected": -131.16128540039062,
      "ronpo/logit": 0.06805794686079025,
      "ronpo/residual": 0.07399886846542358,
      "ronpo/residual_abs": 0.34743252396583557,
      "ronpo/target": -0.005940929055213928,
      "ronpo/target_abs": 0.2420380562543869,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 390
    },
    {
      "epoch": 0.9153318077803204,
      "grad_norm": 66.83030388275304,
      "learning_rate": 1.1445894869509865e-08,
      "logits/chosen": -0.5,
      "logits/rejected": -0.5,
      "logps/chosen": -13.5625,
      "logps/rejected": -13.5625,
      "loss": 0.617999255657196,
      "rewards/accuracies": 0.550000011920929,
      "rewards/chosen": -130.7327423095703,
      "rewards/margins": 1.04730224609375,
      "rewards/rejected": -131.78004455566406,
      "ronpo/logit": 0.10468678176403046,
      "ronpo/residual": 0.07548566162586212,
      "ronpo/residual_abs": 0.38671615719795227,
      "ronpo/target": 0.029201125726103783,
      "ronpo/target_abs": 0.2588233947753906,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 400
    },
    {
      "epoch": 0.9382151029748284,
      "grad_norm": 22.540093716859946,
      "learning_rate": 6.2362973239446785e-09,
      "logits/chosen": -0.50390625,
      "logits/rejected": -0.50390625,
      "logps/chosen": -13.6875,
      "logps/rejected": -13.6875,
      "loss": 0.688459038734436,
      "rewards/accuracies": 0.675000011920929,
      "rewards/chosen": -129.91021728515625,
      "rewards/margins": 2.6958179473876953,
      "rewards/rejected": -132.6060333251953,
      "ronpo/logit": 0.2695961594581604,
      "ronpo/residual": 0.24798472225666046,
      "ronpo/residual_abs": 0.48190993070602417,
      "ronpo/target": 0.021611403673887253,
      "ronpo/target_abs": 0.27506598830223083,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 410
    },
    {
      "epoch": 0.9610983981693364,
      "grad_norm": 31.14712184561151,
      "learning_rate": 2.583569961500276e-09,
      "logits/chosen": -0.48046875,
      "logits/rejected": -0.478515625,
      "logps/chosen": -13.6875,
      "logps/rejected": -13.625,
      "loss": 0.5411345362663269,
      "rewards/accuracies": 0.606249988079071,
      "rewards/chosen": -129.72723388671875,
      "rewards/margins": 1.803375244140625,
      "rewards/rejected": -131.53060913085938,
      "ronpo/logit": 0.18028709292411804,
      "ronpo/residual": 0.17998413741588593,
      "ronpo/residual_abs": 0.43158477544784546,
      "ronpo/target": 0.0003029584768228233,
      "ronpo/target_abs": 0.2634728252887726,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 420
    },
    {
      "epoch": 0.9839816933638444,
      "grad_norm": 37.17749009315313,
      "learning_rate": 5.110420237955382e-10,
      "logits/chosen": -0.515625,
      "logits/rejected": -0.5078125,
      "logps/chosen": -13.625,
      "logps/rejected": -13.625,
      "loss": 0.559510350227356,
      "rewards/accuracies": 0.675000011920929,
      "rewards/chosen": -129.8929443359375,
      "rewards/margins": 2.22955322265625,
      "rewards/rejected": -132.12249755859375,
      "ronpo/logit": 0.22305068373680115,
      "ronpo/residual": 0.23500566184520721,
      "ronpo/residual_abs": 0.4577879309654236,
      "ronpo/target": -0.011954975314438343,
      "ronpo/target_abs": 0.25632962584495544,
      "ronpo/weight": 1.0,
      "ronpo/weight_max": 1.0,
      "step": 430
    },
    {
      "epoch": 1.0,
      "eval_logits/chosen": -0.4921875,
      "eval_logits/rejected": -0.5,
      "eval_logps/chosen": -13.5,
      "eval_logps/rejected": -13.5,
      "eval_loss": 0.35658061504364014,
      "eval_rewards/accuracies": 0.5947136282920837,
      "eval_rewards/chosen": -130.1977996826172,
      "eval_rewards/margins": 0.8975045680999756,
      "eval_rewards/rejected": -131.0952911376953,
      "eval_ronpo/logit": 0.08975037932395935,
      "eval_ronpo/residual": 0.07896483689546585,
      "eval_ronpo/residual_abs": 0.3304201364517212,
      "eval_ronpo/target": 0.010785548947751522,
      "eval_ronpo/target_abs": 0.26927900314331055,
      "eval_ronpo/weight": 1.0,
      "eval_ronpo/weight_max": 1.0,
      "eval_runtime": 181.9721,
      "eval_samples_per_second": 9.958,
      "eval_steps_per_second": 1.247,
      "step": 437
    },
    {
      "epoch": 1.0,
      "step": 437,
      "total_flos": 0.0,
      "train_loss": 0.563431295704787,
      "train_runtime": 9963.7635,
      "train_samples_per_second": 5.614,
      "train_steps_per_second": 0.044
    }
  ],
  "logging_steps": 10,
  "max_steps": 437,
  "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": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}