trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9984301412872841,
  "eval_steps": 100,
  "global_step": 477,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.0416666666666666e-08,
      "logits/chosen": -1.375,
      "logits/rejected": -1.21875,
      "logps/chosen": -404.0,
      "logps/rejected": -374.0,
      "loss": 0.6914,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.0416666666666667e-07,
      "logits/chosen": -1.1875,
      "logits/rejected": -1.15625,
      "logps/chosen": -394.0,
      "logps/rejected": -334.0,
      "loss": 0.6944,
      "rewards/accuracies": 0.34375,
      "rewards/chosen": -0.0024261474609375,
      "rewards/margins": -0.004364013671875,
      "rewards/rejected": 0.001922607421875,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.0833333333333333e-07,
      "logits/chosen": -1.2109375,
      "logits/rejected": -1.171875,
      "logps/chosen": -412.0,
      "logps/rejected": -342.0,
      "loss": 0.6931,
      "rewards/accuracies": 0.47187501192092896,
      "rewards/chosen": 0.000743865966796875,
      "rewards/margins": 0.003692626953125,
      "rewards/rejected": -0.0029449462890625,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.1249999999999997e-07,
      "logits/chosen": -1.2109375,
      "logits/rejected": -1.2265625,
      "logps/chosen": -432.0,
      "logps/rejected": -372.0,
      "loss": 0.6923,
      "rewards/accuracies": 0.45625001192092896,
      "rewards/chosen": 0.002197265625,
      "rewards/margins": 0.003204345703125,
      "rewards/rejected": -0.0009918212890625,
      "step": 30
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.1666666666666667e-07,
      "logits/chosen": -1.2421875,
      "logits/rejected": -1.21875,
      "logps/chosen": -444.0,
      "logps/rejected": -368.0,
      "loss": 0.6933,
      "rewards/accuracies": 0.46875,
      "rewards/chosen": -0.0019073486328125,
      "rewards/margins": 9.357929229736328e-06,
      "rewards/rejected": -0.00191497802734375,
      "step": 40
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.999731868769026e-07,
      "logits/chosen": -1.1953125,
      "logits/rejected": -1.15625,
      "logps/chosen": -432.0,
      "logps/rejected": -384.0,
      "loss": 0.6918,
      "rewards/accuracies": 0.4906249940395355,
      "rewards/chosen": 0.000812530517578125,
      "rewards/margins": 0.00439453125,
      "rewards/rejected": -0.0035858154296875,
      "step": 50
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.990353313429303e-07,
      "logits/chosen": -1.1484375,
      "logits/rejected": -1.078125,
      "logps/chosen": -400.0,
      "logps/rejected": -308.0,
      "loss": 0.6915,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": 0.0038299560546875,
      "rewards/margins": 0.00457763671875,
      "rewards/rejected": -0.000759124755859375,
      "step": 60
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.967625656594781e-07,
      "logits/chosen": -1.203125,
      "logits/rejected": -1.203125,
      "logps/chosen": -406.0,
      "logps/rejected": -336.0,
      "loss": 0.6907,
      "rewards/accuracies": 0.4625000059604645,
      "rewards/chosen": -0.00010776519775390625,
      "rewards/margins": 0.002593994140625,
      "rewards/rejected": -0.002685546875,
      "step": 70
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.93167072587771e-07,
      "logits/chosen": -1.2265625,
      "logits/rejected": -1.2265625,
      "logps/chosen": -416.0,
      "logps/rejected": -394.0,
      "loss": 0.6907,
      "rewards/accuracies": 0.45625001192092896,
      "rewards/chosen": -0.000720977783203125,
      "rewards/margins": 0.00144195556640625,
      "rewards/rejected": -0.002166748046875,
      "step": 80
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.882681251368548e-07,
      "logits/chosen": -1.171875,
      "logits/rejected": -1.1796875,
      "logps/chosen": -382.0,
      "logps/rejected": -324.0,
      "loss": 0.6871,
      "rewards/accuracies": 0.546875,
      "rewards/chosen": 0.0035858154296875,
      "rewards/margins": 0.01318359375,
      "rewards/rejected": -0.0096435546875,
      "step": 90
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.820919832540181e-07,
      "logits/chosen": -1.28125,
      "logits/rejected": -1.171875,
      "logps/chosen": -420.0,
      "logps/rejected": -372.0,
      "loss": 0.6867,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.00103759765625,
      "rewards/margins": 0.015625,
      "rewards/rejected": -0.0167236328125,
      "step": 100
    },
    {
      "epoch": 0.21,
      "eval_logits/chosen": -1.2421875,
      "eval_logits/rejected": -1.1953125,
      "eval_logps/chosen": -424.0,
      "eval_logps/rejected": -368.0,
      "eval_loss": 0.6852812767028809,
      "eval_rewards/accuracies": 0.5555555820465088,
      "eval_rewards/chosen": -0.005645751953125,
      "eval_rewards/margins": 0.0216064453125,
      "eval_rewards/rejected": -0.027099609375,
      "eval_runtime": 51.3914,
      "eval_samples_per_second": 38.917,
      "eval_steps_per_second": 1.226,
      "step": 100
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.7467175306295647e-07,
      "logits/chosen": -1.234375,
      "logits/rejected": -1.21875,
      "logps/chosen": -448.0,
      "logps/rejected": -366.0,
      "loss": 0.6836,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.0045166015625,
      "rewards/margins": 0.0247802734375,
      "rewards/rejected": -0.029296875,
      "step": 110
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.6604720940421207e-07,
      "logits/chosen": -1.203125,
      "logits/rejected": -1.140625,
      "logps/chosen": -410.0,
      "logps/rejected": -332.0,
      "loss": 0.6802,
      "rewards/accuracies": 0.578125,
      "rewards/chosen": -0.0155029296875,
      "rewards/margins": 0.025390625,
      "rewards/rejected": -0.040771484375,
      "step": 120
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.5626458262912735e-07,
      "logits/chosen": -1.265625,
      "logits/rejected": -1.2265625,
      "logps/chosen": -404.0,
      "logps/rejected": -380.0,
      "loss": 0.6806,
      "rewards/accuracies": 0.5718749761581421,
      "rewards/chosen": -0.04638671875,
      "rewards/margins": 0.0257568359375,
      "rewards/rejected": -0.072265625,
      "step": 130
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.453763107901675e-07,
      "logits/chosen": -1.265625,
      "logits/rejected": -1.2578125,
      "logps/chosen": -454.0,
      "logps/rejected": -388.0,
      "loss": 0.6776,
      "rewards/accuracies": 0.574999988079071,
      "rewards/chosen": -0.0654296875,
      "rewards/margins": 0.037841796875,
      "rewards/rejected": -0.10302734375,
      "step": 140
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.3344075855595097e-07,
      "logits/chosen": -1.2421875,
      "logits/rejected": -1.234375,
      "logps/chosen": -420.0,
      "logps/rejected": -356.0,
      "loss": 0.6716,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.09423828125,
      "rewards/margins": 0.049560546875,
      "rewards/rejected": -0.1435546875,
      "step": 150
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.2052190435769554e-07,
      "logits/chosen": -1.28125,
      "logits/rejected": -1.234375,
      "logps/chosen": -422.0,
      "logps/rejected": -388.0,
      "loss": 0.6672,
      "rewards/accuracies": 0.6000000238418579,
      "rewards/chosen": -0.12353515625,
      "rewards/margins": 0.061767578125,
      "rewards/rejected": -0.185546875,
      "step": 160
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.0668899744407567e-07,
      "logits/chosen": -1.28125,
      "logits/rejected": -1.2265625,
      "logps/chosen": -390.0,
      "logps/rejected": -336.0,
      "loss": 0.674,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.1630859375,
      "rewards/margins": 0.037353515625,
      "rewards/rejected": -0.2001953125,
      "step": 170
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.920161866827889e-07,
      "logits/chosen": -1.3125,
      "logits/rejected": -1.296875,
      "logps/chosen": -426.0,
      "logps/rejected": -382.0,
      "loss": 0.6659,
      "rewards/accuracies": 0.574999988079071,
      "rewards/chosen": -0.181640625,
      "rewards/margins": 0.06298828125,
      "rewards/rejected": -0.244140625,
      "step": 180
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.765821230985757e-07,
      "logits/chosen": -1.3203125,
      "logits/rejected": -1.265625,
      "logps/chosen": -410.0,
      "logps/rejected": -380.0,
      "loss": 0.6594,
      "rewards/accuracies": 0.659375011920929,
      "rewards/chosen": -0.19140625,
      "rewards/margins": 0.09814453125,
      "rewards/rejected": -0.2890625,
      "step": 190
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.604695382782159e-07,
      "logits/chosen": -1.3671875,
      "logits/rejected": -1.265625,
      "logps/chosen": -438.0,
      "logps/rejected": -374.0,
      "loss": 0.6632,
      "rewards/accuracies": 0.621874988079071,
      "rewards/chosen": -0.21875,
      "rewards/margins": 0.07421875,
      "rewards/rejected": -0.29296875,
      "step": 200
    },
    {
      "epoch": 0.42,
      "eval_logits/chosen": -1.3671875,
      "eval_logits/rejected": -1.3203125,
      "eval_logps/chosen": -446.0,
      "eval_logps/rejected": -396.0,
      "eval_loss": 0.6625468730926514,
      "eval_rewards/accuracies": 0.625,
      "eval_rewards/chosen": -0.228515625,
      "eval_rewards/margins": 0.09130859375,
      "eval_rewards/rejected": -0.3203125,
      "eval_runtime": 52.3612,
      "eval_samples_per_second": 38.196,
      "eval_steps_per_second": 1.203,
      "step": 200
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.4376480090239047e-07,
      "logits/chosen": -1.4296875,
      "logits/rejected": -1.359375,
      "logps/chosen": -442.0,
      "logps/rejected": -388.0,
      "loss": 0.6604,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.2421875,
      "rewards/margins": 0.0888671875,
      "rewards/rejected": -0.33203125,
      "step": 210
    },
    {
      "epoch": 0.46,
      "learning_rate": 3.265574537815398e-07,
      "logits/chosen": -1.296875,
      "logits/rejected": -1.28125,
      "logps/chosen": -406.0,
      "logps/rejected": -374.0,
      "loss": 0.6588,
      "rewards/accuracies": 0.590624988079071,
      "rewards/chosen": -0.271484375,
      "rewards/margins": 0.0673828125,
      "rewards/rejected": -0.33984375,
      "step": 220
    },
    {
      "epoch": 0.48,
      "learning_rate": 3.0893973387735683e-07,
      "logits/chosen": -1.34375,
      "logits/rejected": -1.296875,
      "logps/chosen": -406.0,
      "logps/rejected": -372.0,
      "loss": 0.6516,
      "rewards/accuracies": 0.6000000238418579,
      "rewards/chosen": -0.26171875,
      "rewards/margins": 0.1064453125,
      "rewards/rejected": -0.369140625,
      "step": 230
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.910060778827554e-07,
      "logits/chosen": -1.3515625,
      "logits/rejected": -1.328125,
      "logps/chosen": -420.0,
      "logps/rejected": -368.0,
      "loss": 0.6635,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.302734375,
      "rewards/margins": 0.09619140625,
      "rewards/rejected": -0.400390625,
      "step": 240
    },
    {
      "epoch": 0.52,
      "learning_rate": 2.7285261601056697e-07,
      "logits/chosen": -1.3828125,
      "logits/rejected": -1.328125,
      "logps/chosen": -426.0,
      "logps/rejected": -384.0,
      "loss": 0.6546,
      "rewards/accuracies": 0.550000011920929,
      "rewards/chosen": -0.30859375,
      "rewards/margins": 0.10693359375,
      "rewards/rejected": -0.416015625,
      "step": 250
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.5457665670441937e-07,
      "logits/chosen": -1.3515625,
      "logits/rejected": -1.2890625,
      "logps/chosen": -432.0,
      "logps/rejected": -384.0,
      "loss": 0.6521,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": -0.337890625,
      "rewards/margins": 0.142578125,
      "rewards/rejected": -0.48046875,
      "step": 260
    },
    {
      "epoch": 0.57,
      "learning_rate": 2.3627616503391812e-07,
      "logits/chosen": -1.3828125,
      "logits/rejected": -1.390625,
      "logps/chosen": -454.0,
      "logps/rejected": -426.0,
      "loss": 0.6522,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.34375,
      "rewards/margins": 0.11181640625,
      "rewards/rejected": -0.45703125,
      "step": 270
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.1804923757009882e-07,
      "logits/chosen": -1.296875,
      "logits/rejected": -1.3046875,
      "logps/chosen": -426.0,
      "logps/rejected": -372.0,
      "loss": 0.6545,
      "rewards/accuracies": 0.606249988079071,
      "rewards/chosen": -0.35546875,
      "rewards/margins": 0.134765625,
      "rewards/rejected": -0.48828125,
      "step": 280
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.9999357655598891e-07,
      "logits/chosen": -1.3515625,
      "logits/rejected": -1.3671875,
      "logps/chosen": -436.0,
      "logps/rejected": -388.0,
      "loss": 0.6546,
      "rewards/accuracies": 0.6000000238418579,
      "rewards/chosen": -0.37890625,
      "rewards/margins": 0.109375,
      "rewards/rejected": -0.48828125,
      "step": 290
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8220596619089573e-07,
      "logits/chosen": -1.375,
      "logits/rejected": -1.3046875,
      "logps/chosen": -474.0,
      "logps/rejected": -442.0,
      "loss": 0.6493,
      "rewards/accuracies": 0.6312500238418579,
      "rewards/chosen": -0.33984375,
      "rewards/margins": 0.12109375,
      "rewards/rejected": -0.458984375,
      "step": 300
    },
    {
      "epoch": 0.63,
      "eval_logits/chosen": -1.3984375,
      "eval_logits/rejected": -1.3515625,
      "eval_logps/chosen": -460.0,
      "eval_logps/rejected": -412.0,
      "eval_loss": 0.653249979019165,
      "eval_rewards/accuracies": 0.625,
      "eval_rewards/chosen": -0.35546875,
      "eval_rewards/margins": 0.134765625,
      "eval_rewards/rejected": -0.490234375,
      "eval_runtime": 52.3802,
      "eval_samples_per_second": 38.182,
      "eval_steps_per_second": 1.203,
      "step": 300
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.647817538357072e-07,
      "logits/chosen": -1.390625,
      "logits/rejected": -1.3203125,
      "logps/chosen": -432.0,
      "logps/rejected": -366.0,
      "loss": 0.6467,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": -0.353515625,
      "rewards/margins": 0.125,
      "rewards/rejected": -0.478515625,
      "step": 310
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.478143389201113e-07,
      "logits/chosen": -1.296875,
      "logits/rejected": -1.328125,
      "logps/chosen": -416.0,
      "logps/rejected": -386.0,
      "loss": 0.6499,
      "rewards/accuracies": 0.6343749761581421,
      "rewards/chosen": -0.3515625,
      "rewards/margins": 0.11865234375,
      "rewards/rejected": -0.47265625,
      "step": 320
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.3139467229135998e-07,
      "logits/chosen": -1.4296875,
      "logits/rejected": -1.390625,
      "logps/chosen": -476.0,
      "logps/rejected": -422.0,
      "loss": 0.6409,
      "rewards/accuracies": 0.6343749761581421,
      "rewards/chosen": -0.373046875,
      "rewards/margins": 0.1669921875,
      "rewards/rejected": -0.5390625,
      "step": 330
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.1561076868822755e-07,
      "logits/chosen": -1.3671875,
      "logits/rejected": -1.328125,
      "logps/chosen": -428.0,
      "logps/rejected": -404.0,
      "loss": 0.6505,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.384765625,
      "rewards/margins": 0.10400390625,
      "rewards/rejected": -0.48828125,
      "step": 340
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.0054723495346482e-07,
      "logits/chosen": -1.359375,
      "logits/rejected": -1.2890625,
      "logps/chosen": -420.0,
      "logps/rejected": -388.0,
      "loss": 0.6417,
      "rewards/accuracies": 0.6312500238418579,
      "rewards/chosen": -0.37890625,
      "rewards/margins": 0.1455078125,
      "rewards/rejected": -0.5234375,
      "step": 350
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.628481651367875e-08,
      "logits/chosen": -1.3828125,
      "logits/rejected": -1.3515625,
      "logps/chosen": -476.0,
      "logps/rejected": -400.0,
      "loss": 0.6398,
      "rewards/accuracies": 0.6312500238418579,
      "rewards/chosen": -0.37890625,
      "rewards/margins": 0.1396484375,
      "rewards/rejected": -0.51953125,
      "step": 360
    },
    {
      "epoch": 0.77,
      "learning_rate": 7.289996455765748e-08,
      "logits/chosen": -1.421875,
      "logits/rejected": -1.3984375,
      "logps/chosen": -492.0,
      "logps/rejected": -404.0,
      "loss": 0.6472,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.376953125,
      "rewards/margins": 0.138671875,
      "rewards/rejected": -0.515625,
      "step": 370
    },
    {
      "epoch": 0.8,
      "learning_rate": 6.046442623320145e-08,
      "logits/chosen": -1.4140625,
      "logits/rejected": -1.3984375,
      "logps/chosen": -448.0,
      "logps/rejected": -416.0,
      "loss": 0.6386,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.39453125,
      "rewards/margins": 0.126953125,
      "rewards/rejected": -0.51953125,
      "step": 380
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.904486005914027e-08,
      "logits/chosen": -1.3515625,
      "logits/rejected": -1.3515625,
      "logps/chosen": -442.0,
      "logps/rejected": -416.0,
      "loss": 0.6437,
      "rewards/accuracies": 0.6312500238418579,
      "rewards/chosen": -0.39453125,
      "rewards/margins": 0.126953125,
      "rewards/rejected": -0.5234375,
      "step": 390
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.8702478614051345e-08,
      "logits/chosen": -1.375,
      "logits/rejected": -1.34375,
      "logps/chosen": -492.0,
      "logps/rejected": -426.0,
      "loss": 0.6462,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.392578125,
      "rewards/margins": 0.1318359375,
      "rewards/rejected": -0.5234375,
      "step": 400
    },
    {
      "epoch": 0.84,
      "eval_logits/chosen": -1.421875,
      "eval_logits/rejected": -1.375,
      "eval_logps/chosen": -464.0,
      "eval_logps/rejected": -420.0,
      "eval_loss": 0.6495468616485596,
      "eval_rewards/accuracies": 0.6388888955116272,
      "eval_rewards/chosen": -0.400390625,
      "eval_rewards/margins": 0.1552734375,
      "eval_rewards/rejected": -0.5546875,
      "eval_runtime": 51.9691,
      "eval_samples_per_second": 38.484,
      "eval_steps_per_second": 1.212,
      "step": 400
    },
    {
      "epoch": 0.86,
      "learning_rate": 2.9492720416985e-08,
      "logits/chosen": -1.328125,
      "logits/rejected": -1.3046875,
      "logps/chosen": -428.0,
      "logps/rejected": -374.0,
      "loss": 0.651,
      "rewards/accuracies": 0.6343749761581421,
      "rewards/chosen": -0.400390625,
      "rewards/margins": 0.130859375,
      "rewards/rejected": -0.53125,
      "step": 410
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.1464952759020856e-08,
      "logits/chosen": -1.40625,
      "logits/rejected": -1.375,
      "logps/chosen": -432.0,
      "logps/rejected": -408.0,
      "loss": 0.6437,
      "rewards/accuracies": 0.6187499761581421,
      "rewards/chosen": -0.396484375,
      "rewards/margins": 0.12109375,
      "rewards/rejected": -0.51953125,
      "step": 420
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.4662207078575684e-08,
      "logits/chosen": -1.375,
      "logits/rejected": -1.359375,
      "logps/chosen": -458.0,
      "logps/rejected": -404.0,
      "loss": 0.6478,
      "rewards/accuracies": 0.643750011920929,
      "rewards/chosen": -0.380859375,
      "rewards/margins": 0.181640625,
      "rewards/rejected": -0.5625,
      "step": 430
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.12094829893642e-09,
      "logits/chosen": -1.421875,
      "logits/rejected": -1.390625,
      "logps/chosen": -420.0,
      "logps/rejected": -392.0,
      "loss": 0.6468,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.390625,
      "rewards/margins": 0.138671875,
      "rewards/rejected": -0.52734375,
      "step": 440
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.8708793644441086e-09,
      "logits/chosen": -1.4296875,
      "logits/rejected": -1.359375,
      "logps/chosen": -452.0,
      "logps/rejected": -426.0,
      "loss": 0.6587,
      "rewards/accuracies": 0.5843750238418579,
      "rewards/chosen": -0.40234375,
      "rewards/margins": 0.08447265625,
      "rewards/rejected": -0.48828125,
      "step": 450
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.9347820230782295e-09,
      "logits/chosen": -1.359375,
      "logits/rejected": -1.3359375,
      "logps/chosen": -426.0,
      "logps/rejected": -390.0,
      "loss": 0.6394,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.40234375,
      "rewards/margins": 0.140625,
      "rewards/rejected": -0.54296875,
      "step": 460
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.2839470889836627e-10,
      "logits/chosen": -1.421875,
      "logits/rejected": -1.375,
      "logps/chosen": -470.0,
      "logps/rejected": -422.0,
      "loss": 0.6303,
      "rewards/accuracies": 0.690625011920929,
      "rewards/chosen": -0.36328125,
      "rewards/margins": 0.19921875,
      "rewards/rejected": -0.55859375,
      "step": 470
    },
    {
      "epoch": 1.0,
      "step": 477,
      "total_flos": 0.0,
      "train_loss": 0.6625334221861897,
      "train_runtime": 3684.7873,
      "train_samples_per_second": 16.591,
      "train_steps_per_second": 0.129
    }
  ],
  "logging_steps": 10,
  "max_steps": 477,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}