{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.8,
  "global_step": 21,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "logps_train/chosen": -100.49485778808594,
      "logps_train/ref_chosen": -100.5,
      "logps_train/ref_rejected": -105.0,
      "logps_train/rejected": -104.80752563476562,
      "rewards_train/accuracies": 0.46875,
      "rewards_train/chosen": -0.020498279482126236,
      "rewards_train/margins": -0.021032828837633133,
      "rewards_train/rejected": 0.000534549355506897,
      "step": 0
    },
    {
      "epoch": 0,
      "logps_train/chosen": -89.90950012207031,
      "logps_train/ref_chosen": -90.0,
      "logps_train/ref_rejected": -101.0,
      "logps_train/rejected": -100.86872863769531,
      "rewards_train/accuracies": 0.5,
      "rewards_train/chosen": 0.0191262885928154,
      "rewards_train/margins": 0.0020734891295433044,
      "rewards_train/rejected": 0.017052799463272095,
      "step": 0
    },
    {
      "epoch": 0.13,
      "learning_rate": 5e-06,
      "loss": 0.7043,
      "step": 1
    },
    {
      "epoch": 0.13,
      "logps_train/chosen": -104.38166809082031,
      "logps_train/ref_chosen": -104.5,
      "logps_train/ref_rejected": -98.0,
      "logps_train/rejected": -97.89907836914062,
      "rewards_train/accuracies": 0.515625,
      "rewards_train/chosen": 0.03059956058859825,
      "rewards_train/margins": 0.004916800186038017,
      "rewards_train/rejected": 0.025682760402560234,
      "step": 1
    },
    {
      "epoch": 0.13,
      "logps_train/chosen": -94.88069152832031,
      "logps_train/ref_chosen": -95.0,
      "logps_train/ref_rejected": -95.0,
      "logps_train/rejected": -95.29293823242188,
      "rewards_train/accuracies": 0.546875,
      "rewards_train/chosen": -0.009438544511795044,
      "rewards_train/margins": -0.020688317716121674,
      "rewards_train/rejected": 0.01124977320432663,
      "step": 1
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.9692208514878445e-06,
      "loss": 0.7061,
      "step": 2
    },
    {
      "epoch": 0.27,
      "logps_train/chosen": -82.53303527832031,
      "logps_train/ref_chosen": -82.5,
      "logps_train/ref_rejected": -89.0,
      "logps_train/rejected": -88.99131774902344,
      "rewards_train/accuracies": 0.484375,
      "rewards_train/chosen": 0.006856412626802921,
      "rewards_train/margins": -0.001938466913998127,
      "rewards_train/rejected": 0.008794879540801048,
      "step": 2
    },
    {
      "epoch": 0.27,
      "logps_train/chosen": -100.22808837890625,
      "logps_train/ref_chosen": -100.0,
      "logps_train/ref_rejected": -106.0,
      "logps_train/rejected": -106.06494140625,
      "rewards_train/accuracies": 0.46875,
      "rewards_train/chosen": -0.033964600414037704,
      "rewards_train/margins": -0.008083481341600418,
      "rewards_train/rejected": -0.025881119072437286,
      "step": 2
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.8776412907378845e-06,
      "loss": 0.702,
      "step": 3
    },
    {
      "epoch": 0.4,
      "logps_train/chosen": -85.86089324951172,
      "logps_train/ref_chosen": -86.0,
      "logps_train/ref_rejected": -90.0,
      "logps_train/rejected": -89.91110229492188,
      "rewards_train/accuracies": 0.3125,
      "rewards_train/chosen": -0.007595188915729523,
      "rewards_train/margins": -0.07023129612207413,
      "rewards_train/rejected": 0.0626361072063446,
      "step": 3
    },
    {
      "epoch": 0.4,
      "logps_train/chosen": -92.3138427734375,
      "logps_train/ref_chosen": -92.0,
      "logps_train/ref_rejected": -77.5,
      "logps_train/rejected": -77.53357696533203,
      "rewards_train/accuracies": 0.453125,
      "rewards_train/chosen": -0.0021358560770750046,
      "rewards_train/margins": -0.025264138355851173,
      "rewards_train/rejected": 0.02312828227877617,
      "step": 3
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.72751631047092e-06,
      "loss": 0.7233,
      "step": 4
    },
    {
      "epoch": 0.53,
      "logps_train/chosen": -88.56388854980469,
      "logps_train/ref_chosen": -88.5,
      "logps_train/ref_rejected": -91.5,
      "logps_train/rejected": -91.46757507324219,
      "rewards_train/accuracies": 0.40625,
      "rewards_train/chosen": 0.0178141500800848,
      "rewards_train/margins": -0.01190277375280857,
      "rewards_train/rejected": 0.02971692383289337,
      "step": 4
    },
    {
      "epoch": 0.53,
      "logps_train/chosen": -104.78536224365234,
      "logps_train/ref_chosen": -105.0,
      "logps_train/ref_rejected": -104.5,
      "logps_train/rejected": -104.33482360839844,
      "rewards_train/accuracies": 0.59375,
      "rewards_train/chosen": 0.07069776952266693,
      "rewards_train/margins": 0.03803768381476402,
      "rewards_train/rejected": 0.03266008570790291,
      "step": 4
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.522542485937369e-06,
      "loss": 0.6957,
      "step": 5
    },
    {
      "epoch": 0.67,
      "logps_train/chosen": -109.24040985107422,
      "logps_train/ref_chosen": -109.0,
      "logps_train/ref_rejected": -99.5,
      "logps_train/rejected": -99.40049743652344,
      "rewards_train/accuracies": 0.59375,
      "rewards_train/chosen": 0.030185997486114502,
      "rewards_train/margins": 0.04000556468963623,
      "rewards_train/rejected": -0.009819567203521729,
      "step": 5
    },
    {
      "epoch": 0.67,
      "logps_train/chosen": -82.44449615478516,
      "logps_train/ref_chosen": -82.5,
      "logps_train/ref_rejected": -80.0,
      "logps_train/rejected": -79.84416198730469,
      "rewards_train/accuracies": 0.5625,
      "rewards_train/chosen": 0.07816548645496368,
      "rewards_train/margins": 0.038515977561473846,
      "rewards_train/rejected": 0.03964950889348984,
      "step": 5
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.267766952966369e-06,
      "loss": 0.6776,
      "step": 6
    },
    {
      "epoch": 0.8,
      "logps_train/chosen": -92.00798034667969,
      "logps_train/ref_chosen": -92.0,
      "logps_train/ref_rejected": -89.5,
      "logps_train/rejected": -89.4742660522461,
      "rewards_train/accuracies": 0.59375,
      "rewards_train/chosen": 0.10367429256439209,
      "rewards_train/margins": 0.02098412811756134,
      "rewards_train/rejected": 0.08269016444683075,
      "step": 6
    },
    {
      "epoch": 0.8,
      "logps_train/chosen": -88.19847106933594,
      "logps_train/ref_chosen": -88.5,
      "logps_train/ref_rejected": -94.0,
      "logps_train/rejected": -93.69485473632812,
      "rewards_train/accuracies": 0.53125,
      "rewards_train/chosen": 0.054087504744529724,
      "rewards_train/margins": 0.018747031688690186,
      "rewards_train/rejected": 0.03534047305583954,
      "step": 6
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.969463130731183e-06,
      "loss": 0.6911,
      "step": 7
    },
    {
      "epoch": 0.93,
      "logps_train/chosen": -106.92861938476562,
      "logps_train/ref_chosen": -107.0,
      "logps_train/ref_rejected": -104.0,
      "logps_train/rejected": -104.0421142578125,
      "rewards_train/accuracies": 0.859375,
      "rewards_train/chosen": 0.18815864622592926,
      "rewards_train/margins": 0.27952703833580017,
      "rewards_train/rejected": -0.09136839210987091,
      "step": 7
    },
    {
      "epoch": 0.93,
      "logps_train/chosen": -95.47774505615234,
      "logps_train/ref_chosen": -96.0,
      "logps_train/ref_rejected": -90.5,
      "logps_train/rejected": -90.31706237792969,
      "rewards_train/accuracies": 0.75,
      "rewards_train/chosen": 0.20924821496009827,
      "rewards_train/margins": 0.21952810510993004,
      "rewards_train/rejected": -0.010279890149831772,
      "step": 7
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.634976249348867e-06,
      "loss": 0.5866,
      "step": 8
    },
    {
      "epoch": 1.07,
      "logps_train/chosen": -103.60890197753906,
      "logps_train/ref_chosen": -104.0,
      "logps_train/ref_rejected": -91.0,
      "logps_train/rejected": -91.19517517089844,
      "rewards_train/accuracies": 0.78125,
      "rewards_train/chosen": 0.22508788108825684,
      "rewards_train/margins": 0.2580987438559532,
      "rewards_train/rejected": -0.03301086276769638,
      "step": 8
    },
    {
      "epoch": 1.07,
      "logps_train/chosen": -89.79802703857422,
      "logps_train/ref_chosen": -90.0,
      "logps_train/ref_rejected": -91.0,
      "logps_train/rejected": -91.28385925292969,
      "rewards_train/accuracies": 0.796875,
      "rewards_train/chosen": 0.18155044317245483,
      "rewards_train/margins": 0.24022718518972397,
      "rewards_train/rejected": -0.058676742017269135,
      "step": 8
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.272542485937369e-06,
      "loss": 0.5875,
      "step": 9
    },
    {
      "epoch": 1.2,
      "logps_train/chosen": -99.75357818603516,
      "logps_train/ref_chosen": -100.0,
      "logps_train/ref_rejected": -91.5,
      "logps_train/rejected": -91.54853820800781,
      "rewards_train/accuracies": 0.765625,
      "rewards_train/chosen": 0.2291664034128189,
      "rewards_train/margins": 0.28212443739175797,
      "rewards_train/rejected": -0.052958033978939056,
      "step": 9
    },
    {
      "epoch": 1.2,
      "logps_train/chosen": -97.27603149414062,
      "logps_train/ref_chosen": -98.0,
      "logps_train/ref_rejected": -95.5,
      "logps_train/rejected": -95.36297607421875,
      "rewards_train/accuracies": 0.828125,
      "rewards_train/chosen": 0.23210078477859497,
      "rewards_train/margins": 0.23072939738631248,
      "rewards_train/rejected": 0.001371387392282486,
      "step": 9
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.8910861626005774e-06,
      "loss": 0.583,
      "step": 10
    },
    {
      "epoch": 1.33,
      "logps_train/chosen": -99.11138916015625,
      "logps_train/ref_chosen": -99.5,
      "logps_train/ref_rejected": -97.0,
      "logps_train/rejected": -96.76425170898438,
      "rewards_train/accuracies": 0.765625,
      "rewards_train/chosen": 0.24630612134933472,
      "rewards_train/margins": 0.23292576521635056,
      "rewards_train/rejected": 0.013380356132984161,
      "step": 10
    },
    {
      "epoch": 1.33,
      "logps_train/chosen": -90.8442611694336,
      "logps_train/ref_chosen": -91.0,
      "logps_train/ref_rejected": -97.5,
      "logps_train/rejected": -97.66960144042969,
      "rewards_train/accuracies": 0.765625,
      "rewards_train/chosen": 0.26005905866622925,
      "rewards_train/margins": 0.2439107522368431,
      "rewards_train/rejected": 0.01614830642938614,
      "step": 10
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5e-06,
      "loss": 0.5913,
      "step": 11
    },
    {
      "epoch": 1.47,
      "logps_train/chosen": -96.24613952636719,
      "logps_train/ref_chosen": -97.0,
      "logps_train/ref_rejected": -99.0,
      "logps_train/rejected": -98.8271484375,
      "rewards_train/accuracies": 0.84375,
      "rewards_train/chosen": 0.23923271894454956,
      "rewards_train/margins": 0.3027096092700958,
      "rewards_train/rejected": -0.06347689032554626,
      "step": 11
    },
    {
      "epoch": 1.47,
      "logps_train/chosen": -80.69171142578125,
      "logps_train/ref_chosen": -81.0,
      "logps_train/ref_rejected": -90.0,
      "logps_train/rejected": -90.06707000732422,
      "rewards_train/accuracies": 0.796875,
      "rewards_train/chosen": 0.21196044981479645,
      "rewards_train/margins": 0.21380476094782352,
      "rewards_train/rejected": -0.0018443111330270767,
      "step": 11
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.1089138373994226e-06,
      "loss": 0.5821,
      "step": 12
    },
    {
      "epoch": 1.6,
      "logps_train/chosen": -89.44386291503906,
      "logps_train/ref_chosen": -90.0,
      "logps_train/ref_rejected": -96.0,
      "logps_train/rejected": -95.8780746459961,
      "rewards_train/accuracies": 0.796875,
      "rewards_train/chosen": 0.2612239420413971,
      "rewards_train/margins": 0.28380540758371353,
      "rewards_train/rejected": -0.022581465542316437,
      "step": 12
    },
    {
      "epoch": 1.6,
      "logps_train/chosen": -82.67738342285156,
      "logps_train/ref_chosen": -83.0,
      "logps_train/ref_rejected": -85.0,
      "logps_train/rejected": -85.28672790527344,
      "rewards_train/accuracies": 0.859375,
      "rewards_train/chosen": 0.2267371118068695,
      "rewards_train/margins": 0.2917319983243942,
      "rewards_train/rejected": -0.06499488651752472,
      "step": 12
    },
    {
      "epoch": 1.73,
      "learning_rate": 1.7274575140626318e-06,
      "loss": 0.5736,
      "step": 13
    },
    {
      "epoch": 1.73,
      "logps_train/chosen": -104.39497375488281,
      "logps_train/ref_chosen": -105.0,
      "logps_train/ref_rejected": -101.0,
      "logps_train/rejected": -101.02708435058594,
      "rewards_train/accuracies": 0.921875,
      "rewards_train/chosen": 0.34890174865722656,
      "rewards_train/margins": 0.4168848916888237,
      "rewards_train/rejected": -0.06798314303159714,
      "step": 13
    },
    {
      "epoch": 1.73,
      "logps_train/chosen": -89.95936584472656,
      "logps_train/ref_chosen": -91.0,
      "logps_train/ref_rejected": -85.0,
      "logps_train/rejected": -85.16304016113281,
      "rewards_train/accuracies": 0.875,
      "rewards_train/chosen": 0.283745139837265,
      "rewards_train/margins": 0.3208296597003937,
      "rewards_train/rejected": -0.03708451986312866,
      "step": 13
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.3650237506511333e-06,
      "loss": 0.5417,
      "step": 14
    },
    {
      "epoch": 1.87,
      "logps_train/chosen": -92.31253051757812,
      "logps_train/ref_chosen": -93.0,
      "logps_train/ref_rejected": -108.5,
      "logps_train/rejected": -108.66498565673828,
      "rewards_train/accuracies": 0.796875,
      "rewards_train/chosen": 0.2648744583129883,
      "rewards_train/margins": 0.3167903497815132,
      "rewards_train/rejected": -0.05191589146852493,
      "step": 14
    },
    {
      "epoch": 1.87,
      "logps_train/chosen": -93.20501708984375,
      "logps_train/ref_chosen": -94.0,
      "logps_train/ref_rejected": -93.0,
      "logps_train/rejected": -93.12667846679688,
      "rewards_train/accuracies": 0.890625,
      "rewards_train/chosen": 0.41396820545196533,
      "rewards_train/margins": 0.4813934415578842,
      "rewards_train/rejected": -0.06742523610591888,
      "step": 14
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.0305368692688175e-06,
      "loss": 0.5331,
      "step": 15
    },
    {
      "epoch": 2.0,
      "logps_train/chosen": -96.6799087524414,
      "logps_train/ref_chosen": -97.5,
      "logps_train/ref_rejected": -96.5,
      "logps_train/rejected": -96.50393676757812,
      "rewards_train/accuracies": 0.9375,
      "rewards_train/chosen": 0.3427841067314148,
      "rewards_train/margins": 0.3952302448451519,
      "rewards_train/rejected": -0.052446138113737106,
      "step": 15
    },
    {
      "epoch": 2.0,
      "logps_train/chosen": -92.8008041381836,
      "logps_train/ref_chosen": -93.5,
      "logps_train/ref_rejected": -106.0,
      "logps_train/rejected": -105.96940612792969,
      "rewards_train/accuracies": 0.828125,
      "rewards_train/chosen": 0.3604634702205658,
      "rewards_train/margins": 0.4462389573454857,
      "rewards_train/rejected": -0.08577548712491989,
      "step": 15
    },
    {
      "epoch": 2.13,
      "learning_rate": 7.322330470336314e-07,
      "loss": 0.5232,
      "step": 16
    },
    {
      "epoch": 2.13,
      "logps_train/chosen": -94.60572814941406,
      "logps_train/ref_chosen": -95.0,
      "logps_train/ref_rejected": -89.0,
      "logps_train/rejected": -89.16722869873047,
      "rewards_train/accuracies": 0.953125,
      "rewards_train/chosen": 0.3311697542667389,
      "rewards_train/margins": 0.4457896798849106,
      "rewards_train/rejected": -0.11461992561817169,
      "step": 16
    },
    {
      "epoch": 2.13,
      "logps_train/chosen": -93.5136489868164,
      "logps_train/ref_chosen": -94.0,
      "logps_train/ref_rejected": -88.5,
      "logps_train/rejected": -88.58706665039062,
      "rewards_train/accuracies": 0.875,
      "rewards_train/chosen": 0.31861382722854614,
      "rewards_train/margins": 0.35628755390644073,
      "rewards_train/rejected": -0.03767372667789459,
      "step": 16
    },
    {
      "epoch": 2.27,
      "learning_rate": 4.774575140626317e-07,
      "loss": 0.5274,
      "step": 17
    },
    {
      "epoch": 2.27,
      "logps_train/chosen": -95.30484771728516,
      "logps_train/ref_chosen": -96.0,
      "logps_train/ref_rejected": -96.0,
      "logps_train/rejected": -96.07192993164062,
      "rewards_train/accuracies": 0.921875,
      "rewards_train/chosen": 0.3683273196220398,
      "rewards_train/margins": 0.48193351179361343,
      "rewards_train/rejected": -0.11360619217157364,
      "step": 17
    },
    {
      "epoch": 2.27,
      "logps_train/chosen": -90.89393615722656,
      "logps_train/ref_chosen": -91.5,
      "logps_train/ref_rejected": -100.0,
      "logps_train/rejected": -100.20901489257812,
      "rewards_train/accuracies": 0.9375,
      "rewards_train/chosen": 0.34770655632019043,
      "rewards_train/margins": 0.4666781648993492,
      "rewards_train/rejected": -0.11897160857915878,
      "step": 17
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.7248368952908055e-07,
      "loss": 0.4996,
      "step": 18
    },
    {
      "epoch": 2.4,
      "logps_train/chosen": -101.5006103515625,
      "logps_train/ref_chosen": -102.0,
      "logps_train/ref_rejected": -99.0,
      "logps_train/rejected": -98.9917221069336,
      "rewards_train/accuracies": 0.921875,
      "rewards_train/chosen": 0.39098966121673584,
      "rewards_train/margins": 0.46887996047735214,
      "rewards_train/rejected": -0.0778902992606163,
      "step": 18
    },
    {
      "epoch": 2.4,
      "logps_train/chosen": -100.89315795898438,
      "logps_train/ref_chosen": -101.5,
      "logps_train/ref_rejected": -101.0,
      "logps_train/rejected": -101.20696258544922,
      "rewards_train/accuracies": 0.875,
      "rewards_train/chosen": 0.3883822560310364,
      "rewards_train/margins": 0.4976026713848114,
      "rewards_train/rejected": -0.10922041535377502,
      "step": 18
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.223587092621162e-07,
      "loss": 0.499,
      "step": 19
    },
    {
      "epoch": 2.53,
      "logps_train/chosen": -93.89166259765625,
      "logps_train/ref_chosen": -94.5,
      "logps_train/ref_rejected": -101.0,
      "logps_train/rejected": -101.55459594726562,
      "rewards_train/accuracies": 0.859375,
      "rewards_train/chosen": 0.33444538712501526,
      "rewards_train/margins": 0.45895563066005707,
      "rewards_train/rejected": -0.12451024353504181,
      "step": 19
    },
    {
      "epoch": 2.53,
      "logps_train/chosen": -100.19595336914062,
      "logps_train/ref_chosen": -101.0,
      "logps_train/ref_rejected": -95.5,
      "logps_train/rejected": -95.7244873046875,
      "rewards_train/accuracies": 0.921875,
      "rewards_train/chosen": 0.38786405324935913,
      "rewards_train/margins": 0.44200433045625687,
      "rewards_train/rejected": -0.054140277206897736,
      "step": 19
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.077914851215585e-08,
      "loss": 0.5114,
      "step": 20
    },
    {
      "epoch": 2.67,
      "logps_train/chosen": -75.63420867919922,
      "logps_train/ref_chosen": -76.0,
      "logps_train/ref_rejected": -78.0,
      "logps_train/rejected": -78.19580841064453,
      "rewards_train/accuracies": 0.828125,
      "rewards_train/chosen": 0.28807422518730164,
      "rewards_train/margins": 0.3581909090280533,
      "rewards_train/rejected": -0.07011668384075165,
      "step": 20
    },
    {
      "epoch": 2.67,
      "logps_train/chosen": -94.97181701660156,
      "logps_train/ref_chosen": -96.0,
      "logps_train/ref_rejected": -94.0,
      "logps_train/rejected": -94.27290344238281,
      "rewards_train/accuracies": 0.921875,
      "rewards_train/chosen": 0.38176849484443665,
      "rewards_train/margins": 0.46939007192850113,
      "rewards_train/rejected": -0.08762157708406448,
      "step": 20
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0,
      "loss": 0.5237,
      "step": 21
    },
    {
      "epoch": 2.8,
      "step": 21,
      "total_flos": 0.0,
      "train_loss": 0.5982543017183032,
      "train_runtime": 243.8656,
      "train_samples_per_second": 11.182,
      "train_steps_per_second": 0.086
    }
  ],
  "max_steps": 21,
  "num_train_epochs": 3,
  "total_flos": 0.0,
  "trial_name": null,
  "trial_params": null
}