{
  "best_metric": 6.651080131530762,
  "best_model_checkpoint": "./qwen2.5-0.5b/qwen2.5-0.5b-expo-DPO-ES-1000/checkpoint-600",
  "epoch": 3.122342938119981,
  "eval_steps": 50,
  "global_step": 1100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "dpo_loss": 0.6931471824645996,
      "epoch": 0.002834199338686821,
      "grad_norm": 184425.11607762048,
      "learning_rate": 2.840909090909091e-08,
      "logits": -1.359458565711975,
      "logps": -84.69721221923828,
      "loss": 0.6931,
      "objective": 0.6931471824645996,
      "ranking_idealized": 0.5833333134651184,
      "ranking_idealized_expo": 0.5833333134651184,
      "ranking_simple": 0.5833333134651184,
      "regularize": 0.6931471824645996,
      "step": 1,
      "wo_beta": 5.271125316619873
    },
    {
      "dpo_loss": 164.56210327148438,
      "epoch": 0.14170996693434104,
      "grad_norm": 220470.08226005477,
      "learning_rate": 1.4204545454545458e-06,
      "logits": -1.4481998682022095,
      "logps": -84.27718353271484,
      "loss": 171.6127,
      "objective": 164.56210327148438,
      "ranking_idealized": 0.5216836929321289,
      "ranking_idealized_expo": 0.5216836929321289,
      "ranking_simple": 0.521258533000946,
      "regularize": 164.56210327148438,
      "step": 50,
      "wo_beta": 7.096639156341553
    },
    {
      "epoch": 0.14170996693434104,
      "eval_dpo_loss": 308.1877136230469,
      "eval_logits": -1.4575201272964478,
      "eval_logps": -90.79700469970703,
      "eval_loss": 309.735595703125,
      "eval_objective": 308.1877136230469,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5253623127937317,
      "eval_regularize": 308.1877136230469,
      "eval_runtime": 308.9992,
      "eval_samples_per_second": 18.738,
      "eval_steps_per_second": 1.563,
      "eval_wo_beta": 7.70471715927124,
      "step": 50
    },
    {
      "dpo_loss": 567.9904174804688,
      "epoch": 0.2834199338686821,
      "grad_norm": 197852.1649626301,
      "learning_rate": 2.8409090909090916e-06,
      "logits": -1.4372496604919434,
      "logps": -82.37940216064453,
      "loss": 582.3149,
      "objective": 567.9904174804688,
      "ranking_idealized": 0.5137500166893005,
      "ranking_idealized_expo": 0.5137500166893005,
      "ranking_simple": 0.5429166555404663,
      "regularize": 567.9904174804688,
      "step": 100,
      "wo_beta": 6.490437030792236
    },
    {
      "epoch": 0.2834199338686821,
      "eval_dpo_loss": 708.8132934570312,
      "eval_logits": -1.3877463340759277,
      "eval_logps": -88.8039779663086,
      "eval_loss": 733.8201904296875,
      "eval_objective": 708.8132934570312,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5284678936004639,
      "eval_regularize": 708.8132934570312,
      "eval_runtime": 308.1307,
      "eval_samples_per_second": 18.791,
      "eval_steps_per_second": 1.568,
      "eval_wo_beta": 7.477492332458496,
      "step": 100
    },
    {
      "dpo_loss": 988.060302734375,
      "epoch": 0.42512990080302315,
      "grad_norm": 157694.24240229145,
      "learning_rate": 4.2613636363636365e-06,
      "logits": -1.3607189655303955,
      "logps": -79.60610961914062,
      "loss": 1002.3808,
      "objective": 988.060302734375,
      "ranking_idealized": 0.527916669845581,
      "ranking_idealized_expo": 0.527916669845581,
      "ranking_simple": 0.5629166960716248,
      "regularize": 988.060302734375,
      "step": 150,
      "wo_beta": 6.3590989112854
    },
    {
      "epoch": 0.42512990080302315,
      "eval_dpo_loss": 1245.726318359375,
      "eval_logits": -1.3138219118118286,
      "eval_logps": -83.66063690185547,
      "eval_loss": 1283.0697021484375,
      "eval_objective": 1245.726318359375,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5310559272766113,
      "eval_regularize": 1245.726318359375,
      "eval_runtime": 308.3631,
      "eval_samples_per_second": 18.777,
      "eval_steps_per_second": 1.566,
      "eval_wo_beta": 7.363152503967285,
      "step": 150
    },
    {
      "dpo_loss": 1228.03515625,
      "epoch": 0.5668398677373642,
      "grad_norm": 135260.80669895583,
      "learning_rate": 4.997168347957521e-06,
      "logits": -1.2575763463974,
      "logps": -76.39550018310547,
      "loss": 1199.7266,
      "objective": 1228.03515625,
      "ranking_idealized": 0.51541668176651,
      "ranking_idealized_expo": 0.51541668176651,
      "ranking_simple": 0.5520833134651184,
      "regularize": 1228.03515625,
      "step": 200,
      "wo_beta": 6.235730171203613
    },
    {
      "epoch": 0.5668398677373642,
      "eval_dpo_loss": 1471.0286865234375,
      "eval_logits": -1.2583951950073242,
      "eval_logps": -79.82491302490234,
      "eval_loss": 1530.312255859375,
      "eval_objective": 1471.0286865234375,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.534679114818573,
      "eval_regularize": 1471.0286865234375,
      "eval_runtime": 311.5857,
      "eval_samples_per_second": 18.582,
      "eval_steps_per_second": 1.55,
      "eval_wo_beta": 7.23297643661499,
      "step": 200
    },
    {
      "dpo_loss": 1277.0172119140625,
      "epoch": 0.7085498346717053,
      "grad_norm": 126264.1949530397,
      "learning_rate": 4.973122855144066e-06,
      "logits": -1.0750112533569336,
      "logps": -75.37919616699219,
      "loss": 1311.3106,
      "objective": 1277.0172119140625,
      "ranking_idealized": 0.5162500143051147,
      "ranking_idealized_expo": 0.5162500143051147,
      "ranking_simple": 0.5754166841506958,
      "regularize": 1277.0172119140625,
      "step": 250,
      "wo_beta": 6.227660179138184
    },
    {
      "epoch": 0.7085498346717053,
      "eval_dpo_loss": 1842.360107421875,
      "eval_logits": -1.1798591613769531,
      "eval_logps": -78.57501220703125,
      "eval_loss": 1873.1123046875,
      "eval_objective": 1842.360107421875,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.534679114818573,
      "eval_regularize": 1842.360107421875,
      "eval_runtime": 308.1159,
      "eval_samples_per_second": 18.792,
      "eval_steps_per_second": 1.568,
      "eval_wo_beta": 7.204605579376221,
      "step": 250
    },
    {
      "dpo_loss": 1138.1712646484375,
      "epoch": 0.8502598016060463,
      "grad_norm": 91406.60871461965,
      "learning_rate": 4.924776641419513e-06,
      "logits": -1.083847999572754,
      "logps": -76.23174285888672,
      "loss": 1216.5524,
      "objective": 1138.1712646484375,
      "ranking_idealized": 0.4950000047683716,
      "ranking_idealized_expo": 0.4950000047683716,
      "ranking_simple": 0.5774999856948853,
      "regularize": 1138.1712646484375,
      "step": 300,
      "wo_beta": 6.088484287261963
    },
    {
      "epoch": 0.8502598016060463,
      "eval_dpo_loss": 1949.1083984375,
      "eval_logits": -1.0463101863861084,
      "eval_logps": -80.68748474121094,
      "eval_loss": 2001.6103515625,
      "eval_objective": 1949.1083984375,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.532608687877655,
      "eval_regularize": 1949.1083984375,
      "eval_runtime": 308.2132,
      "eval_samples_per_second": 18.786,
      "eval_steps_per_second": 1.567,
      "eval_wo_beta": 6.943759918212891,
      "step": 300
    },
    {
      "dpo_loss": 1116.613525390625,
      "epoch": 0.9919697685403873,
      "grad_norm": 107657.35203851353,
      "learning_rate": 4.8526047530778175e-06,
      "logits": -0.8322110176086426,
      "logps": -76.44075775146484,
      "loss": 1157.2415,
      "objective": 1116.613525390625,
      "ranking_idealized": 0.5254166722297668,
      "ranking_idealized_expo": 0.5254166722297668,
      "ranking_simple": 0.6041666865348816,
      "regularize": 1116.613525390625,
      "step": 350,
      "wo_beta": 5.487993240356445
    },
    {
      "epoch": 0.9919697685403873,
      "eval_dpo_loss": 1956.4012451171875,
      "eval_logits": -0.8782438635826111,
      "eval_logps": -79.74929809570312,
      "eval_loss": 2064.322021484375,
      "eval_objective": 1956.4012451171875,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5439958572387695,
      "eval_regularize": 1956.4012451171875,
      "eval_runtime": 308.4172,
      "eval_samples_per_second": 18.773,
      "eval_steps_per_second": 1.566,
      "eval_wo_beta": 7.016916751861572,
      "step": 350
    },
    {
      "dpo_loss": 730.6525268554688,
      "epoch": 1.1336797354747283,
      "grad_norm": 85954.8176907515,
      "learning_rate": 4.757316345716554e-06,
      "logits": -0.6696025133132935,
      "logps": -74.4837875366211,
      "loss": 721.9005,
      "objective": 730.6525268554688,
      "ranking_idealized": 0.5320833325386047,
      "ranking_idealized_expo": 0.5320833325386047,
      "ranking_simple": 0.6329166889190674,
      "regularize": 730.6525268554688,
      "step": 400,
      "wo_beta": 4.743136882781982
    },
    {
      "epoch": 1.1336797354747283,
      "eval_dpo_loss": 2228.881103515625,
      "eval_logits": -0.5703257918357849,
      "eval_logps": -80.20221710205078,
      "eval_loss": 2276.4189453125,
      "eval_objective": 2228.881103515625,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5403726696968079,
      "eval_regularize": 2228.881103515625,
      "eval_runtime": 308.9371,
      "eval_samples_per_second": 18.742,
      "eval_steps_per_second": 1.563,
      "eval_wo_beta": 7.247954845428467,
      "step": 400
    },
    {
      "dpo_loss": 807.7144775390625,
      "epoch": 1.2753897024090695,
      "grad_norm": 87253.50953833942,
      "learning_rate": 4.639847716126855e-06,
      "logits": -0.5064049959182739,
      "logps": -74.17053985595703,
      "loss": 779.6797,
      "objective": 807.7144775390625,
      "ranking_idealized": 0.5191666483879089,
      "ranking_idealized_expo": 0.5191666483879089,
      "ranking_simple": 0.6362500190734863,
      "regularize": 807.7144775390625,
      "step": 450,
      "wo_beta": 5.3411478996276855
    },
    {
      "epoch": 1.2753897024090695,
      "eval_dpo_loss": 2016.328125,
      "eval_logits": -0.709108293056488,
      "eval_logps": -78.40538787841797,
      "eval_loss": 2069.493896484375,
      "eval_objective": 2016.328125,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5367494821548462,
      "eval_regularize": 2016.328125,
      "eval_runtime": 308.0644,
      "eval_samples_per_second": 18.795,
      "eval_steps_per_second": 1.568,
      "eval_wo_beta": 6.824249744415283,
      "step": 450
    },
    {
      "dpo_loss": 769.7364501953125,
      "epoch": 1.4170996693434104,
      "grad_norm": 88303.87563999402,
      "learning_rate": 4.501353102310901e-06,
      "logits": -0.6304615139961243,
      "logps": -75.90121459960938,
      "loss": 788.48,
      "objective": 769.7364501953125,
      "ranking_idealized": 0.4970833361148834,
      "ranking_idealized_expo": 0.4970833361148834,
      "ranking_simple": 0.6183333396911621,
      "regularize": 769.7364501953125,
      "step": 500,
      "wo_beta": 5.177425861358643
    },
    {
      "epoch": 1.4170996693434104,
      "eval_dpo_loss": 2044.074462890625,
      "eval_logits": -0.6658820509910583,
      "eval_logps": -81.98267364501953,
      "eval_loss": 2120.1181640625,
      "eval_objective": 2044.074462890625,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5341615080833435,
      "eval_regularize": 2044.074462890625,
      "eval_runtime": 308.3125,
      "eval_samples_per_second": 18.78,
      "eval_steps_per_second": 1.567,
      "eval_wo_beta": 6.866703033447266,
      "step": 500
    },
    {
      "dpo_loss": 711.3790893554688,
      "epoch": 1.5588096362777515,
      "grad_norm": 81196.09184670768,
      "learning_rate": 4.34319334202531e-06,
      "logits": -0.6250764727592468,
      "logps": -76.77790832519531,
      "loss": 684.4246,
      "objective": 711.3790893554688,
      "ranking_idealized": 0.5104166865348816,
      "ranking_idealized_expo": 0.5104166865348816,
      "ranking_simple": 0.6399999856948853,
      "regularize": 711.3790893554688,
      "step": 550,
      "wo_beta": 4.714800834655762
    },
    {
      "epoch": 1.5588096362777515,
      "eval_dpo_loss": 2053.837158203125,
      "eval_logits": -0.6751150488853455,
      "eval_logps": -81.63760375976562,
      "eval_loss": 2148.157958984375,
      "eval_objective": 2053.837158203125,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5341615080833435,
      "eval_regularize": 2053.837158203125,
      "eval_runtime": 310.078,
      "eval_samples_per_second": 18.673,
      "eval_steps_per_second": 1.558,
      "eval_wo_beta": 6.790140151977539,
      "step": 550
    },
    {
      "dpo_loss": 692.1633911132812,
      "epoch": 1.7005196032120926,
      "grad_norm": 81536.59548809296,
      "learning_rate": 4.16692250129073e-06,
      "logits": -0.5091446042060852,
      "logps": -76.25757598876953,
      "loss": 708.5259,
      "objective": 692.1633911132812,
      "ranking_idealized": 0.5149999856948853,
      "ranking_idealized_expo": 0.5149999856948853,
      "ranking_simple": 0.628333330154419,
      "regularize": 692.1633911132812,
      "step": 600,
      "wo_beta": 4.881653785705566
    },
    {
      "epoch": 1.7005196032120926,
      "eval_dpo_loss": 2071.194580078125,
      "eval_logits": -0.5595079660415649,
      "eval_logps": -79.44622039794922,
      "eval_loss": 2179.60009765625,
      "eval_objective": 2071.194580078125,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5362318754196167,
      "eval_regularize": 2071.194580078125,
      "eval_runtime": 310.8858,
      "eval_samples_per_second": 18.624,
      "eval_steps_per_second": 1.554,
      "eval_wo_beta": 6.651080131530762,
      "step": 600
    },
    {
      "dpo_loss": 673.4611206054688,
      "epoch": 1.8422295701464337,
      "grad_norm": 86567.62941958332,
      "learning_rate": 3.974272604254906e-06,
      "logits": -0.48710063099861145,
      "logps": -76.06396484375,
      "loss": 690.9902,
      "objective": 673.4611206054688,
      "ranking_idealized": 0.527916669845581,
      "ranking_idealized_expo": 0.527916669845581,
      "ranking_simple": 0.6387500166893005,
      "regularize": 673.4611206054688,
      "step": 650,
      "wo_beta": 5.110637187957764
    },
    {
      "epoch": 1.8422295701464337,
      "eval_dpo_loss": 2158.488525390625,
      "eval_logits": -0.5552332997322083,
      "eval_logps": -80.51081085205078,
      "eval_loss": 2241.3740234375,
      "eval_objective": 2158.488525390625,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5414078831672668,
      "eval_regularize": 2158.488525390625,
      "eval_runtime": 308.5347,
      "eval_samples_per_second": 18.766,
      "eval_steps_per_second": 1.565,
      "eval_wo_beta": 6.7740159034729,
      "step": 650
    },
    {
      "dpo_loss": 586.7876586914062,
      "epoch": 1.9839395370807746,
      "grad_norm": 79003.551065143,
      "learning_rate": 3.767136614452458e-06,
      "logits": -0.3871801495552063,
      "logps": -76.267578125,
      "loss": 617.6108,
      "objective": 586.7876586914062,
      "ranking_idealized": 0.5108333230018616,
      "ranking_idealized_expo": 0.5108333230018616,
      "ranking_simple": 0.6416666507720947,
      "regularize": 586.7876586914062,
      "step": 700,
      "wo_beta": 4.860604286193848
    },
    {
      "epoch": 1.9839395370807746,
      "eval_dpo_loss": 2132.251708984375,
      "eval_logits": -0.5078864097595215,
      "eval_logps": -80.38253784179688,
      "eval_loss": 2230.511474609375,
      "eval_objective": 2132.251708984375,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5403726696968079,
      "eval_regularize": 2132.251708984375,
      "eval_runtime": 309.2277,
      "eval_samples_per_second": 18.724,
      "eval_steps_per_second": 1.562,
      "eval_wo_beta": 6.795376777648926,
      "step": 700
    },
    {
      "dpo_loss": 347.239990234375,
      "epoch": 2.1256495040151155,
      "grad_norm": 48670.94258805941,
      "learning_rate": 3.547549834686222e-06,
      "logits": -0.3425801694393158,
      "logps": -76.91646575927734,
      "loss": 343.0455,
      "objective": 347.239990234375,
      "ranking_idealized": 0.5112500190734863,
      "ranking_idealized_expo": 0.5112500190734863,
      "ranking_simple": 0.6691666841506958,
      "regularize": 347.239990234375,
      "step": 750,
      "wo_beta": 4.557406902313232
    },
    {
      "epoch": 2.1256495040151155,
      "eval_dpo_loss": 2123.3603515625,
      "eval_logits": -0.5397638082504272,
      "eval_logps": -81.35387420654297,
      "eval_loss": 2199.717529296875,
      "eval_objective": 2123.3603515625,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5429606437683105,
      "eval_regularize": 2123.3603515625,
      "eval_runtime": 308.3199,
      "eval_samples_per_second": 18.779,
      "eval_steps_per_second": 1.567,
      "eval_wo_beta": 6.757833003997803,
      "step": 750
    },
    {
      "dpo_loss": 317.3254089355469,
      "epoch": 2.2673594709494567,
      "grad_norm": 42462.56905768485,
      "learning_rate": 3.3176699082935546e-06,
      "logits": -0.47370415925979614,
      "logps": -76.58859252929688,
      "loss": 311.7518,
      "objective": 317.3254089355469,
      "ranking_idealized": 0.512499988079071,
      "ranking_idealized_expo": 0.512499988079071,
      "ranking_simple": 0.6679166555404663,
      "regularize": 317.3254089355469,
      "step": 800,
      "wo_beta": 4.156112194061279
    },
    {
      "epoch": 2.2673594709494567,
      "eval_dpo_loss": 2038.6656494140625,
      "eval_logits": -0.5496692657470703,
      "eval_logps": -80.27388763427734,
      "eval_loss": 2139.787109375,
      "eval_objective": 2038.6656494140625,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5377846956253052,
      "eval_regularize": 2038.6656494140625,
      "eval_runtime": 308.3672,
      "eval_samples_per_second": 18.776,
      "eval_steps_per_second": 1.566,
      "eval_wo_beta": 6.676809310913086,
      "step": 800
    },
    {
      "dpo_loss": 342.0184020996094,
      "epoch": 2.409069437883798,
      "grad_norm": 50353.55073679767,
      "learning_rate": 3.0797556183036582e-06,
      "logits": -0.47750580310821533,
      "logps": -78.6015853881836,
      "loss": 315.5968,
      "objective": 342.0184020996094,
      "ranking_idealized": 0.5133333206176758,
      "ranking_idealized_expo": 0.5133333206176758,
      "ranking_simple": 0.6566666960716248,
      "regularize": 342.0184020996094,
      "step": 850,
      "wo_beta": 4.543591499328613
    },
    {
      "epoch": 2.409069437883798,
      "eval_dpo_loss": 2184.711181640625,
      "eval_logits": -0.5281842350959778,
      "eval_logps": -83.38427734375,
      "eval_loss": 2249.320068359375,
      "eval_objective": 2184.711181640625,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5403726696968079,
      "eval_regularize": 2184.711181640625,
      "eval_runtime": 308.9348,
      "eval_samples_per_second": 18.742,
      "eval_steps_per_second": 1.563,
      "eval_wo_beta": 6.807182312011719,
      "step": 850
    },
    {
      "dpo_loss": 1037.1005859375,
      "epoch": 2.555503070382617,
      "grad_norm": 148584.23094187575,
      "learning_rate": 2.8361446928038298e-06,
      "logits": -0.11181627959012985,
      "logps": -87.09500885009766,
      "loss": 6263.3387,
      "objective": 6087.2216796875,
      "ranking_idealized": 0.5183333158493042,
      "ranking_idealized_expo": 0.5179166793823242,
      "ranking_simple": 0.5941666960716248,
      "regularize": 6087.2216796875,
      "step": 900,
      "wo_beta": 15.719075202941895
    },
    {
      "epoch": 2.555503070382617,
      "eval_dpo_loss": 3261.5380859375,
      "eval_logits": -0.10349252820014954,
      "eval_logps": -90.69556427001953,
      "eval_loss": 6365.919921875,
      "eval_objective": 6410.73828125,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5248447060585022,
      "eval_regularize": 6410.73828125,
      "eval_runtime": 309.2116,
      "eval_samples_per_second": 18.725,
      "eval_steps_per_second": 1.562,
      "eval_wo_beta": 14.391923904418945,
      "step": 900
    },
    {
      "dpo_loss": 1062.182861328125,
      "epoch": 2.697213037316958,
      "grad_norm": 142709.83683573626,
      "learning_rate": 2.5892308345974517e-06,
      "logits": -0.11892472952604294,
      "logps": -86.6119613647461,
      "loss": 4964.9731,
      "objective": 4921.0400390625,
      "ranking_idealized": 0.5045833587646484,
      "ranking_idealized_expo": 0.5045833587646484,
      "ranking_simple": 0.574999988079071,
      "regularize": 4921.0400390625,
      "step": 950,
      "wo_beta": 14.98097038269043
    },
    {
      "epoch": 2.697213037316958,
      "eval_dpo_loss": 3203.624267578125,
      "eval_logits": -0.15411485731601715,
      "eval_logps": -88.87255096435547,
      "eval_loss": 6126.68994140625,
      "eval_objective": 6172.97900390625,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5258799195289612,
      "eval_regularize": 6172.97900390625,
      "eval_runtime": 308.5864,
      "eval_samples_per_second": 18.763,
      "eval_steps_per_second": 1.565,
      "eval_wo_beta": 14.286750793457031,
      "step": 950
    },
    {
      "dpo_loss": 1058.12353515625,
      "epoch": 2.838923004251299,
      "grad_norm": 141320.36114480646,
      "learning_rate": 2.341440200858589e-06,
      "logits": -0.06014474108815193,
      "logps": -84.71208190917969,
      "loss": 4278.7487,
      "objective": 4290.322265625,
      "ranking_idealized": 0.5112500190734863,
      "ranking_idealized_expo": 0.5112500190734863,
      "ranking_simple": 0.559583306312561,
      "regularize": 4290.322265625,
      "step": 1000,
      "wo_beta": 14.737447738647461
    },
    {
      "epoch": 2.838923004251299,
      "eval_dpo_loss": 3153.050048828125,
      "eval_logits": -0.094576895236969,
      "eval_logps": -87.88904571533203,
      "eval_loss": 6092.5341796875,
      "eval_objective": 6127.7734375,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5243270993232727,
      "eval_regularize": 6127.7734375,
      "eval_runtime": 308.7636,
      "eval_samples_per_second": 18.752,
      "eval_steps_per_second": 1.564,
      "eval_wo_beta": 14.248814582824707,
      "step": 1000
    },
    {
      "dpo_loss": 969.5340576171875,
      "epoch": 2.9806329711856403,
      "grad_norm": 145217.5279628283,
      "learning_rate": 2.0952075638923656e-06,
      "logits": -0.07964936643838882,
      "logps": -84.64329528808594,
      "loss": 3830.4475,
      "objective": 3857.745361328125,
      "ranking_idealized": 0.5174999833106995,
      "ranking_idealized_expo": 0.5174999833106995,
      "ranking_simple": 0.5695833563804626,
      "regularize": 3857.745361328125,
      "step": 1050,
      "wo_beta": 15.0871000289917
    },
    {
      "epoch": 2.9806329711856403,
      "eval_dpo_loss": 3127.49560546875,
      "eval_logits": -0.13649722933769226,
      "eval_logps": -86.9566421508789,
      "eval_loss": 6040.49169921875,
      "eval_objective": 6063.345703125,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.5232919454574585,
      "eval_regularize": 6063.345703125,
      "eval_runtime": 308.411,
      "eval_samples_per_second": 18.774,
      "eval_steps_per_second": 1.566,
      "eval_wo_beta": 14.031007766723633,
      "step": 1050
    },
    {
      "dpo_loss": 784.6331787109375,
      "epoch": 3.122342938119981,
      "grad_norm": 148785.0779157092,
      "learning_rate": 1.852952387243698e-06,
      "logits": -0.06228747218847275,
      "logps": -85.15433502197266,
      "loss": 3079.7456,
      "objective": 3018.007080078125,
      "ranking_idealized": 0.528333306312561,
      "ranking_idealized_expo": 0.528333306312561,
      "ranking_simple": 0.559166669845581,
      "regularize": 3018.007080078125,
      "step": 1100,
      "wo_beta": 15.30273151397705
    },
    {
      "epoch": 3.122342938119981,
      "eval_dpo_loss": 3144.784423828125,
      "eval_logits": -0.12487590312957764,
      "eval_logps": -88.1663589477539,
      "eval_loss": 6014.59765625,
      "eval_objective": 6010.9189453125,
      "eval_ranking_idealized": 0.5212215185165405,
      "eval_ranking_idealized_expo": 0.5212215185165405,
      "eval_ranking_simple": 0.52173912525177,
      "eval_regularize": 6010.9189453125,
      "eval_runtime": 308.9136,
      "eval_samples_per_second": 18.743,
      "eval_steps_per_second": 1.564,
      "eval_wo_beta": 14.233796119689941,
      "step": 1100
    },
    {
      "epoch": 3.122342938119981,
      "step": 1100,
      "total_flos": 0.0,
      "train_loss": 1018.9660795454546,
      "train_runtime": 6860.7986,
      "train_samples_per_second": 37.023,
      "train_steps_per_second": 0.257
    }
  ],
  "logging_steps": 50,
  "max_steps": 1760,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "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": 4,
  "trial_name": null,
  "trial_params": null
}