{
  "best_metric": 14.005528450012207,
  "best_model_checkpoint": "./qwen2.5-0.5b/qwen2.5-0.5b-expo-L1EXPO-ES-0.1/checkpoint-750",
  "epoch": 3.5474728389230044,
  "eval_steps": 50,
  "global_step": 1250,
  "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": 36.885068816813735,
      "learning_rate": 2.840909090909091e-08,
      "logits": -1.359458565711975,
      "logps": -84.69721221923828,
      "loss": 0.0051,
      "objective": 0.0046141319908201694,
      "ranking_idealized": 0.6458333134651184,
      "ranking_idealized_expo": 0.5833333134651184,
      "ranking_simple": 0.5833333134651184,
      "regularize": 0.0046141319908201694,
      "step": 1,
      "wo_beta": 14.840873718261719
    },
    {
      "dpo_loss": 0.6924303770065308,
      "epoch": 0.14170996693434104,
      "grad_norm": 33.742835742644765,
      "learning_rate": 1.4204545454545458e-06,
      "logits": -1.4566550254821777,
      "logps": -84.23489379882812,
      "loss": 0.0448,
      "objective": 0.04280169680714607,
      "ranking_idealized": 0.608418345451355,
      "ranking_idealized_expo": 0.5229591727256775,
      "ranking_simple": 0.5221088528633118,
      "regularize": 0.04280169680714607,
      "step": 50,
      "wo_beta": 15.654285430908203
    },
    {
      "epoch": 0.14170996693434104,
      "eval_dpo_loss": 0.6935604810714722,
      "eval_logits": -1.4298962354660034,
      "eval_logps": -90.38883209228516,
      "eval_loss": 0.062233828008174896,
      "eval_objective": 0.062142737209796906,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5243270993232727,
      "eval_regularize": 0.062142737209796906,
      "eval_runtime": 308.7283,
      "eval_samples_per_second": 18.754,
      "eval_steps_per_second": 1.564,
      "eval_wo_beta": 16.076759338378906,
      "step": 50
    },
    {
      "dpo_loss": 0.7021370530128479,
      "epoch": 0.2834199338686821,
      "grad_norm": 31.729424259376046,
      "learning_rate": 2.8409090909090916e-06,
      "logits": -1.3946272134780884,
      "logps": -82.87468719482422,
      "loss": 0.1716,
      "objective": 0.17463459074497223,
      "ranking_idealized": 0.6016666889190674,
      "ranking_idealized_expo": 0.5141666531562805,
      "ranking_simple": 0.51541668176651,
      "regularize": 0.17463459074497223,
      "step": 100,
      "wo_beta": 15.276419639587402
    },
    {
      "epoch": 0.2834199338686821,
      "eval_dpo_loss": 0.6981683969497681,
      "eval_logits": -1.3597227334976196,
      "eval_logps": -88.76753234863281,
      "eval_loss": 0.1556352823972702,
      "eval_objective": 0.15588510036468506,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5274327397346497,
      "eval_regularize": 0.15588510036468506,
      "eval_runtime": 308.3116,
      "eval_samples_per_second": 18.78,
      "eval_steps_per_second": 1.567,
      "eval_wo_beta": 15.943617820739746,
      "step": 100
    },
    {
      "dpo_loss": 0.7128496170043945,
      "epoch": 0.42512990080302315,
      "grad_norm": 24.11205441144116,
      "learning_rate": 4.2613636363636365e-06,
      "logits": -1.251375675201416,
      "logps": -79.16511535644531,
      "loss": 0.2858,
      "objective": 0.2846659719944,
      "ranking_idealized": 0.6066666841506958,
      "ranking_idealized_expo": 0.5287500023841858,
      "ranking_simple": 0.5274999737739563,
      "regularize": 0.2846659719944,
      "step": 150,
      "wo_beta": 15.22218132019043
    },
    {
      "epoch": 0.42512990080302315,
      "eval_dpo_loss": 0.718317449092865,
      "eval_logits": -1.2545664310455322,
      "eval_logps": -79.50674438476562,
      "eval_loss": 0.2911944091320038,
      "eval_objective": 0.2922578752040863,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.522774338722229,
      "eval_regularize": 0.2922578752040863,
      "eval_runtime": 308.3872,
      "eval_samples_per_second": 18.775,
      "eval_steps_per_second": 1.566,
      "eval_wo_beta": 15.05699634552002,
      "step": 150
    },
    {
      "dpo_loss": 0.7332326173782349,
      "epoch": 0.5668398677373642,
      "grad_norm": 19.83023586651878,
      "learning_rate": 4.997168347957521e-06,
      "logits": -1.015448808670044,
      "logps": -76.77922058105469,
      "loss": 0.3544,
      "objective": 0.3552749752998352,
      "ranking_idealized": 0.5924999713897705,
      "ranking_idealized_expo": 0.5166666507720947,
      "ranking_simple": 0.5020833611488342,
      "regularize": 0.3552749752998352,
      "step": 200,
      "wo_beta": 15.356170654296875
    },
    {
      "epoch": 0.5668398677373642,
      "eval_dpo_loss": 0.7308588027954102,
      "eval_logits": -0.8432308435440063,
      "eval_logps": -83.84849548339844,
      "eval_loss": 0.3898463547229767,
      "eval_objective": 0.3890216052532196,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.522774338722229,
      "eval_regularize": 0.3890216052532196,
      "eval_runtime": 307.8908,
      "eval_samples_per_second": 18.805,
      "eval_steps_per_second": 1.569,
      "eval_wo_beta": 14.712230682373047,
      "step": 200
    },
    {
      "dpo_loss": 0.7182620763778687,
      "epoch": 0.7085498346717053,
      "grad_norm": 19.421703617823624,
      "learning_rate": 4.973122855144066e-06,
      "logits": -0.7580794095993042,
      "logps": -78.16065216064453,
      "loss": 0.375,
      "objective": 0.37717047333717346,
      "ranking_idealized": 0.5991666913032532,
      "ranking_idealized_expo": 0.5170833468437195,
      "ranking_simple": 0.5195833444595337,
      "regularize": 0.37717047333717346,
      "step": 250,
      "wo_beta": 15.653904914855957
    },
    {
      "epoch": 0.7085498346717053,
      "eval_dpo_loss": 0.735299289226532,
      "eval_logits": -0.673379123210907,
      "eval_logps": -81.28996276855469,
      "eval_loss": 0.4397831857204437,
      "eval_objective": 0.43750080466270447,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5243270993232727,
      "eval_regularize": 0.43750080466270447,
      "eval_runtime": 351.635,
      "eval_samples_per_second": 16.466,
      "eval_steps_per_second": 1.374,
      "eval_wo_beta": 14.472906112670898,
      "step": 250
    },
    {
      "dpo_loss": 0.72139972448349,
      "epoch": 0.8502598016060463,
      "grad_norm": 19.146332148877114,
      "learning_rate": 4.924776641419513e-06,
      "logits": -0.564231812953949,
      "logps": -79.54463195800781,
      "loss": 0.3592,
      "objective": 0.3580860495567322,
      "ranking_idealized": 0.5799999833106995,
      "ranking_idealized_expo": 0.4970833361148834,
      "ranking_simple": 0.5024999976158142,
      "regularize": 0.3580860495567322,
      "step": 300,
      "wo_beta": 15.114410400390625
    },
    {
      "epoch": 0.8502598016060463,
      "eval_dpo_loss": 0.734784722328186,
      "eval_logits": -0.5500932335853577,
      "eval_logps": -84.41443634033203,
      "eval_loss": 0.442239373922348,
      "eval_objective": 0.43877851963043213,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5232919454574585,
      "eval_regularize": 0.43877851963043213,
      "eval_runtime": 404.6503,
      "eval_samples_per_second": 14.309,
      "eval_steps_per_second": 1.194,
      "eval_wo_beta": 14.44029712677002,
      "step": 300
    },
    {
      "dpo_loss": 0.717363178730011,
      "epoch": 0.9919697685403873,
      "grad_norm": 17.11167089420277,
      "learning_rate": 4.8526047530778175e-06,
      "logits": -0.5000432133674622,
      "logps": -79.52243041992188,
      "loss": 0.3351,
      "objective": 0.33388689160346985,
      "ranking_idealized": 0.60916668176651,
      "ranking_idealized_expo": 0.5270833373069763,
      "ranking_simple": 0.5262500047683716,
      "regularize": 0.33388689160346985,
      "step": 350,
      "wo_beta": 15.228816986083984
    },
    {
      "epoch": 0.9919697685403873,
      "eval_dpo_loss": 0.735403299331665,
      "eval_logits": -0.5360206961631775,
      "eval_logps": -82.93754577636719,
      "eval_loss": 0.4675760865211487,
      "eval_objective": 0.4601868689060211,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5341615080833435,
      "eval_regularize": 0.4601868689060211,
      "eval_runtime": 308.044,
      "eval_samples_per_second": 18.796,
      "eval_steps_per_second": 1.568,
      "eval_wo_beta": 14.272198677062988,
      "step": 350
    },
    {
      "dpo_loss": 0.7120790481567383,
      "epoch": 1.1336797354747283,
      "grad_norm": 17.95944897166815,
      "learning_rate": 4.757316345716554e-06,
      "logits": -0.5945844054222107,
      "logps": -77.92745208740234,
      "loss": 0.3056,
      "objective": 0.3030960261821747,
      "ranking_idealized": 0.6087499856948853,
      "ranking_idealized_expo": 0.5337499976158142,
      "ranking_simple": 0.5316666960716248,
      "regularize": 0.3030960261821747,
      "step": 400,
      "wo_beta": 15.400132179260254
    },
    {
      "epoch": 1.1336797354747283,
      "eval_dpo_loss": 0.7469586730003357,
      "eval_logits": -0.5685753226280212,
      "eval_logps": -80.56059265136719,
      "eval_loss": 0.48422977328300476,
      "eval_objective": 0.48037421703338623,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5253623127937317,
      "eval_regularize": 0.48037421703338623,
      "eval_runtime": 314.7441,
      "eval_samples_per_second": 18.396,
      "eval_steps_per_second": 1.535,
      "eval_wo_beta": 14.281224250793457,
      "step": 400
    },
    {
      "dpo_loss": 0.7077716588973999,
      "epoch": 1.2753897024090695,
      "grad_norm": 15.643717424703908,
      "learning_rate": 4.639847716126855e-06,
      "logits": -0.5214452743530273,
      "logps": -78.54476928710938,
      "loss": 0.2932,
      "objective": 0.29641959071159363,
      "ranking_idealized": 0.5975000262260437,
      "ranking_idealized_expo": 0.5199999809265137,
      "ranking_simple": 0.5199999809265137,
      "regularize": 0.29641959071159363,
      "step": 450,
      "wo_beta": 15.791983604431152
    },
    {
      "epoch": 1.2753897024090695,
      "eval_dpo_loss": 0.7439451217651367,
      "eval_logits": -0.5565418004989624,
      "eval_logps": -83.62307739257812,
      "eval_loss": 0.48051390051841736,
      "eval_objective": 0.47548484802246094,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5279502868652344,
      "eval_regularize": 0.47548484802246094,
      "eval_runtime": 308.6602,
      "eval_samples_per_second": 18.758,
      "eval_steps_per_second": 1.565,
      "eval_wo_beta": 14.464012145996094,
      "step": 450
    },
    {
      "dpo_loss": 0.7124494314193726,
      "epoch": 1.4170996693434104,
      "grad_norm": 15.324248073271528,
      "learning_rate": 4.501353102310901e-06,
      "logits": -0.5437880754470825,
      "logps": -78.15093994140625,
      "loss": 0.2864,
      "objective": 0.2836955189704895,
      "ranking_idealized": 0.57833331823349,
      "ranking_idealized_expo": 0.4983333349227905,
      "ranking_simple": 0.5,
      "regularize": 0.2836955189704895,
      "step": 500,
      "wo_beta": 15.279319763183594
    },
    {
      "epoch": 1.4170996693434104,
      "eval_dpo_loss": 0.7510210275650024,
      "eval_logits": -0.6556914448738098,
      "eval_logps": -82.91778564453125,
      "eval_loss": 0.4964132010936737,
      "eval_objective": 0.4971453845500946,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5274327397346497,
      "eval_regularize": 0.4971453845500946,
      "eval_runtime": 307.59,
      "eval_samples_per_second": 18.824,
      "eval_steps_per_second": 1.57,
      "eval_wo_beta": 14.282269477844238,
      "step": 500
    },
    {
      "dpo_loss": 0.7097735404968262,
      "epoch": 1.5588096362777515,
      "grad_norm": 15.632233049837359,
      "learning_rate": 4.34319334202531e-06,
      "logits": -0.5551173686981201,
      "logps": -78.2597885131836,
      "loss": 0.2635,
      "objective": 0.264424592256546,
      "ranking_idealized": 0.5945833325386047,
      "ranking_idealized_expo": 0.5116666555404663,
      "ranking_simple": 0.512499988079071,
      "regularize": 0.264424592256546,
      "step": 550,
      "wo_beta": 15.108202934265137
    },
    {
      "epoch": 1.5588096362777515,
      "eval_dpo_loss": 0.7502700686454773,
      "eval_logits": -0.6183538436889648,
      "eval_logps": -81.16139221191406,
      "eval_loss": 0.502347469329834,
      "eval_objective": 0.5043270587921143,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.522774338722229,
      "eval_regularize": 0.5043270587921143,
      "eval_runtime": 307.9389,
      "eval_samples_per_second": 18.802,
      "eval_steps_per_second": 1.568,
      "eval_wo_beta": 14.063152313232422,
      "step": 550
    },
    {
      "dpo_loss": 0.708111047744751,
      "epoch": 1.7005196032120926,
      "grad_norm": 15.258212996967583,
      "learning_rate": 4.16692250129073e-06,
      "logits": -0.48317351937294006,
      "logps": -79.20243835449219,
      "loss": 0.2561,
      "objective": 0.2518368065357208,
      "ranking_idealized": 0.6004166603088379,
      "ranking_idealized_expo": 0.51583331823349,
      "ranking_simple": 0.5074999928474426,
      "regularize": 0.2518368065357208,
      "step": 600,
      "wo_beta": 15.102338790893555
    },
    {
      "epoch": 1.7005196032120926,
      "eval_dpo_loss": 0.7486839294433594,
      "eval_logits": -0.5805073380470276,
      "eval_logps": -84.7038803100586,
      "eval_loss": 0.49804311990737915,
      "eval_objective": 0.49642521142959595,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5232919454574585,
      "eval_regularize": 0.49642521142959595,
      "eval_runtime": 308.8037,
      "eval_samples_per_second": 18.75,
      "eval_steps_per_second": 1.564,
      "eval_wo_beta": 14.335240364074707,
      "step": 600
    },
    {
      "dpo_loss": 0.7057015299797058,
      "epoch": 1.8422295701464337,
      "grad_norm": 15.137988457806076,
      "learning_rate": 3.974272604254906e-06,
      "logits": -0.44393202662467957,
      "logps": -80.69384002685547,
      "loss": 0.2448,
      "objective": 0.24766255915164948,
      "ranking_idealized": 0.6058333516120911,
      "ranking_idealized_expo": 0.5295833349227905,
      "ranking_simple": 0.5350000262260437,
      "regularize": 0.24766255915164948,
      "step": 650,
      "wo_beta": 15.657441139221191
    },
    {
      "epoch": 1.8422295701464337,
      "eval_dpo_loss": 0.750303328037262,
      "eval_logits": -0.42741408944129944,
      "eval_logps": -83.46288299560547,
      "eval_loss": 0.5170512795448303,
      "eval_objective": 0.5191380381584167,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5232919454574585,
      "eval_regularize": 0.5191380381584167,
      "eval_runtime": 311.9724,
      "eval_samples_per_second": 18.559,
      "eval_steps_per_second": 1.548,
      "eval_wo_beta": 14.215264320373535,
      "step": 650
    },
    {
      "dpo_loss": 0.7065611481666565,
      "epoch": 1.9839395370807746,
      "grad_norm": 16.215629133943352,
      "learning_rate": 3.767136614452458e-06,
      "logits": -0.43040552735328674,
      "logps": -79.25086212158203,
      "loss": 0.2235,
      "objective": 0.22704358398914337,
      "ranking_idealized": 0.5954166650772095,
      "ranking_idealized_expo": 0.5129166841506958,
      "ranking_simple": 0.5083333253860474,
      "regularize": 0.22704358398914337,
      "step": 700,
      "wo_beta": 15.360807418823242
    },
    {
      "epoch": 1.9839395370807746,
      "eval_dpo_loss": 0.7482582330703735,
      "eval_logits": -0.5056679844856262,
      "eval_logps": -81.71964263916016,
      "eval_loss": 0.49625101685523987,
      "eval_objective": 0.4948585331439972,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5232919454574585,
      "eval_regularize": 0.4948585331439972,
      "eval_runtime": 312.5183,
      "eval_samples_per_second": 18.527,
      "eval_steps_per_second": 1.546,
      "eval_wo_beta": 14.20261001586914,
      "step": 700
    },
    {
      "dpo_loss": 0.7034626007080078,
      "epoch": 2.1256495040151155,
      "grad_norm": 14.439374560790109,
      "learning_rate": 3.547549834686222e-06,
      "logits": -0.4137415587902069,
      "logps": -78.55575561523438,
      "loss": 0.21,
      "objective": 0.20508375763893127,
      "ranking_idealized": 0.6066666841506958,
      "ranking_idealized_expo": 0.5133333206176758,
      "ranking_simple": 0.5099999904632568,
      "regularize": 0.20508375763893127,
      "step": 750,
      "wo_beta": 15.407984733581543
    },
    {
      "epoch": 2.1256495040151155,
      "eval_dpo_loss": 0.7511767148971558,
      "eval_logits": -0.47565215826034546,
      "eval_logps": -82.51917266845703,
      "eval_loss": 0.5233561396598816,
      "eval_objective": 0.5224636197090149,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5253623127937317,
      "eval_regularize": 0.5224636197090149,
      "eval_runtime": 307.6951,
      "eval_samples_per_second": 18.817,
      "eval_steps_per_second": 1.57,
      "eval_wo_beta": 14.005528450012207,
      "step": 750
    },
    {
      "dpo_loss": 0.6996860504150391,
      "epoch": 2.2673594709494567,
      "grad_norm": 14.763670008384054,
      "learning_rate": 3.3176699082935546e-06,
      "logits": -0.44839462637901306,
      "logps": -79.29713439941406,
      "loss": 0.1988,
      "objective": 0.19862671196460724,
      "ranking_idealized": 0.5941666960716248,
      "ranking_idealized_expo": 0.5129166841506958,
      "ranking_simple": 0.5116666555404663,
      "regularize": 0.19862671196460724,
      "step": 800,
      "wo_beta": 14.99027156829834
    },
    {
      "epoch": 2.2673594709494567,
      "eval_dpo_loss": 0.7496048808097839,
      "eval_logits": -0.5577788949012756,
      "eval_logps": -81.05644226074219,
      "eval_loss": 0.5139943361282349,
      "eval_objective": 0.511448085308075,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5295031070709229,
      "eval_regularize": 0.511448085308075,
      "eval_runtime": 308.0865,
      "eval_samples_per_second": 18.793,
      "eval_steps_per_second": 1.568,
      "eval_wo_beta": 14.102994918823242,
      "step": 800
    },
    {
      "dpo_loss": 0.7008050680160522,
      "epoch": 2.409069437883798,
      "grad_norm": 14.833532828768124,
      "learning_rate": 3.0797556183036582e-06,
      "logits": -0.4541783630847931,
      "logps": -78.78826141357422,
      "loss": 0.1845,
      "objective": 0.18498587608337402,
      "ranking_idealized": 0.5979166626930237,
      "ranking_idealized_expo": 0.5166666507720947,
      "ranking_simple": 0.5095833539962769,
      "regularize": 0.18498587608337402,
      "step": 850,
      "wo_beta": 15.201929092407227
    },
    {
      "epoch": 2.409069437883798,
      "eval_dpo_loss": 0.7516361474990845,
      "eval_logits": -0.512949526309967,
      "eval_logps": -82.63258361816406,
      "eval_loss": 0.520516574382782,
      "eval_objective": 0.518623948097229,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5310559272766113,
      "eval_regularize": 0.518623948097229,
      "eval_runtime": 309.5699,
      "eval_samples_per_second": 18.703,
      "eval_steps_per_second": 1.56,
      "eval_wo_beta": 14.151838302612305,
      "step": 850
    },
    {
      "dpo_loss": 0.7016371488571167,
      "epoch": 2.550779404818139,
      "grad_norm": 16.31271001796413,
      "learning_rate": 2.8361446928038298e-06,
      "logits": -0.42467382550239563,
      "logps": -79.11641693115234,
      "loss": 0.1741,
      "objective": 0.1748921126127243,
      "ranking_idealized": 0.5887500047683716,
      "ranking_idealized_expo": 0.518750011920929,
      "ranking_simple": 0.51583331823349,
      "regularize": 0.1748921126127243,
      "step": 900,
      "wo_beta": 15.531429290771484
    },
    {
      "epoch": 2.550779404818139,
      "eval_dpo_loss": 0.7507295608520508,
      "eval_logits": -0.4789924621582031,
      "eval_logps": -82.98091125488281,
      "eval_loss": 0.5132278800010681,
      "eval_objective": 0.5117725729942322,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.523809552192688,
      "eval_regularize": 0.5117725729942322,
      "eval_runtime": 311.1948,
      "eval_samples_per_second": 18.606,
      "eval_steps_per_second": 1.552,
      "eval_wo_beta": 14.245877265930176,
      "step": 900
    },
    {
      "dpo_loss": 0.6993770003318787,
      "epoch": 2.69248937175248,
      "grad_norm": 14.7867391875567,
      "learning_rate": 2.5892308345974517e-06,
      "logits": -0.4542914927005768,
      "logps": -79.9416732788086,
      "loss": 0.1659,
      "objective": 0.16475924849510193,
      "ranking_idealized": 0.5975000262260437,
      "ranking_idealized_expo": 0.5087500214576721,
      "ranking_simple": 0.5141666531562805,
      "regularize": 0.16475924849510193,
      "step": 950,
      "wo_beta": 15.521940231323242
    },
    {
      "epoch": 2.69248937175248,
      "eval_dpo_loss": 0.7500060796737671,
      "eval_logits": -0.4840329587459564,
      "eval_logps": -83.8330307006836,
      "eval_loss": 0.5188658237457275,
      "eval_objective": 0.5192957520484924,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.523809552192688,
      "eval_regularize": 0.5192957520484924,
      "eval_runtime": 310.1876,
      "eval_samples_per_second": 18.666,
      "eval_steps_per_second": 1.557,
      "eval_wo_beta": 14.302889823913574,
      "step": 950
    },
    {
      "dpo_loss": 0.7024207711219788,
      "epoch": 2.8341993386868207,
      "grad_norm": 14.42838893249385,
      "learning_rate": 2.341440200858589e-06,
      "logits": -0.4285065233707428,
      "logps": -78.83317565917969,
      "loss": 0.1539,
      "objective": 0.15508733689785004,
      "ranking_idealized": 0.6020833253860474,
      "ranking_idealized_expo": 0.5104166865348816,
      "ranking_simple": 0.5091666579246521,
      "regularize": 0.15508733689785004,
      "step": 1000,
      "wo_beta": 15.06278133392334
    },
    {
      "epoch": 2.8341993386868207,
      "eval_dpo_loss": 0.7498777508735657,
      "eval_logits": -0.46712055802345276,
      "eval_logps": -82.88314056396484,
      "eval_loss": 0.5136557817459106,
      "eval_objective": 0.5127004981040955,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5269151329994202,
      "eval_regularize": 0.5127004981040955,
      "eval_runtime": 308.4784,
      "eval_samples_per_second": 18.77,
      "eval_steps_per_second": 1.566,
      "eval_wo_beta": 14.192536354064941,
      "step": 1000
    },
    {
      "dpo_loss": 0.6952893733978271,
      "epoch": 2.9806329711856403,
      "grad_norm": 14.77922239882152,
      "learning_rate": 2.0952075638923656e-06,
      "logits": -0.4941651523113251,
      "logps": -79.95951080322266,
      "loss": 0.1445,
      "objective": 0.14514465630054474,
      "ranking_idealized": 0.6079166531562805,
      "ranking_idealized_expo": 0.5183333158493042,
      "ranking_simple": 0.5220833420753479,
      "regularize": 0.14514465630054474,
      "step": 1050,
      "wo_beta": 15.50859546661377
    },
    {
      "epoch": 2.9806329711856403,
      "eval_dpo_loss": 0.7478482127189636,
      "eval_logits": -0.5530552864074707,
      "eval_logps": -83.16773986816406,
      "eval_loss": 0.5116256475448608,
      "eval_objective": 0.511193573474884,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5248447060585022,
      "eval_regularize": 0.511193573474884,
      "eval_runtime": 310.4184,
      "eval_samples_per_second": 18.652,
      "eval_steps_per_second": 1.556,
      "eval_wo_beta": 14.214123725891113,
      "step": 1050
    },
    {
      "dpo_loss": 0.6958988308906555,
      "epoch": 3.122342938119981,
      "grad_norm": 14.047766487106658,
      "learning_rate": 1.852952387243698e-06,
      "logits": -0.4758566915988922,
      "logps": -80.04509735107422,
      "loss": 0.1261,
      "objective": 0.12395481020212173,
      "ranking_idealized": 0.6070833206176758,
      "ranking_idealized_expo": 0.5304166674613953,
      "ranking_simple": 0.5270833373069763,
      "regularize": 0.12395481020212173,
      "step": 1100,
      "wo_beta": 15.6008939743042
    },
    {
      "epoch": 3.122342938119981,
      "eval_dpo_loss": 0.7515185475349426,
      "eval_logits": -0.5487966537475586,
      "eval_logps": -83.59542846679688,
      "eval_loss": 0.515699565410614,
      "eval_objective": 0.5165062546730042,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5232919454574585,
      "eval_regularize": 0.5165062546730042,
      "eval_runtime": 309.3286,
      "eval_samples_per_second": 18.718,
      "eval_steps_per_second": 1.561,
      "eval_wo_beta": 14.178275108337402,
      "step": 1100
    },
    {
      "dpo_loss": 0.6931909322738647,
      "epoch": 3.264052905054322,
      "grad_norm": 14.963308385384513,
      "learning_rate": 1.617055052228768e-06,
      "logits": -0.4729629456996918,
      "logps": -79.84102630615234,
      "loss": 0.1146,
      "objective": 0.11260777711868286,
      "ranking_idealized": 0.5950000286102295,
      "ranking_idealized_expo": 0.5095833539962769,
      "ranking_simple": 0.5149999856948853,
      "regularize": 0.11260777711868286,
      "step": 1150,
      "wo_beta": 15.298945426940918
    },
    {
      "epoch": 3.264052905054322,
      "eval_dpo_loss": 0.7487252354621887,
      "eval_logits": -0.5372445583343506,
      "eval_logps": -83.42646789550781,
      "eval_loss": 0.5174793601036072,
      "eval_objective": 0.5160741806030273,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5263975262641907,
      "eval_regularize": 0.5160741806030273,
      "eval_runtime": 308.2096,
      "eval_samples_per_second": 18.786,
      "eval_steps_per_second": 1.567,
      "eval_wo_beta": 14.195608139038086,
      "step": 1150
    },
    {
      "dpo_loss": 0.6932617425918579,
      "epoch": 3.4057628719886632,
      "grad_norm": 14.929601093854455,
      "learning_rate": 1.3898334684855647e-06,
      "logits": -0.4731375575065613,
      "logps": -80.88990783691406,
      "loss": 0.1076,
      "objective": 0.10945354402065277,
      "ranking_idealized": 0.5933333039283752,
      "ranking_idealized_expo": 0.5083333253860474,
      "ranking_simple": 0.5070833563804626,
      "regularize": 0.10945354402065277,
      "step": 1200,
      "wo_beta": 15.574357032775879
    },
    {
      "epoch": 3.4057628719886632,
      "eval_dpo_loss": 0.7491946816444397,
      "eval_logits": -0.4946048855781555,
      "eval_logps": -83.99122619628906,
      "eval_loss": 0.5169116258621216,
      "eval_objective": 0.5159533619880676,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5274327397346497,
      "eval_regularize": 0.5159533619880676,
      "eval_runtime": 309.8199,
      "eval_samples_per_second": 18.688,
      "eval_steps_per_second": 1.559,
      "eval_wo_beta": 14.124122619628906,
      "step": 1200
    },
    {
      "dpo_loss": 0.6921232342720032,
      "epoch": 3.5474728389230044,
      "grad_norm": 14.184398569090499,
      "learning_rate": 1.1735202983664803e-06,
      "logits": -0.45394301414489746,
      "logps": -80.82902526855469,
      "loss": 0.0981,
      "objective": 0.09646416455507278,
      "ranking_idealized": 0.5962499976158142,
      "ranking_idealized_expo": 0.5166666507720947,
      "ranking_simple": 0.51541668176651,
      "regularize": 0.09646416455507278,
      "step": 1250,
      "wo_beta": 15.384990692138672
    },
    {
      "epoch": 3.5474728389230044,
      "eval_dpo_loss": 0.750022828578949,
      "eval_logits": -0.5087407231330872,
      "eval_logps": -83.37907409667969,
      "eval_loss": 0.5174669623374939,
      "eval_objective": 0.5184855461120605,
      "eval_ranking_idealized": 0.6030020713806152,
      "eval_ranking_idealized_expo": 0.5222567319869995,
      "eval_ranking_simple": 0.5310559272766113,
      "eval_regularize": 0.5184855461120605,
      "eval_runtime": 307.8564,
      "eval_samples_per_second": 18.807,
      "eval_steps_per_second": 1.569,
      "eval_wo_beta": 14.215774536132812,
      "step": 1250
    },
    {
      "epoch": 3.5474728389230044,
      "step": 1250,
      "total_flos": 0.0,
      "train_loss": 0.02363297004699707,
      "train_runtime": 6884.3268,
      "train_samples_per_second": 36.897,
      "train_steps_per_second": 0.256
    }
  ],
  "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
}