qwen1_chat_adamw_iter5 / trainer_state.json
yiran-wang3's picture
End of training
d5687bc verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 100,
"global_step": 41,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"debug/policy_chosen_logits": -1.0657888650894165,
"debug/policy_chosen_logps": -135.17166137695312,
"debug/policy_rejected_logits": -1.0082604885101318,
"debug/policy_rejected_logps": -164.67491149902344,
"debug/reference_chosen_logps": -135.17166137695312,
"debug/reference_rejected_logps": -164.67491149902344,
"epoch": 0.024390243902439025,
"grad_norm": 7.455705642028753,
"learning_rate": 1e-06,
"logits/chosen": -1.0657888650894165,
"logits/rejected": -1.0082604885101318,
"logps/chosen": -135.17166137695312,
"logps/rejected": -164.67491149902344,
"loss": 0.5,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"debug/policy_chosen_logits": -0.9316875338554382,
"debug/policy_chosen_logps": -168.0986785888672,
"debug/policy_rejected_logits": -0.9154044389724731,
"debug/policy_rejected_logps": -152.3076934814453,
"debug/reference_chosen_logps": -168.41038513183594,
"debug/reference_rejected_logps": -152.78643798828125,
"epoch": 0.04878048780487805,
"grad_norm": 7.29758593491183,
"learning_rate": 1e-06,
"logits/chosen": -0.9316875338554382,
"logits/rejected": -0.9154044389724731,
"logps/chosen": -168.0986785888672,
"logps/rejected": -152.3076934814453,
"loss": 0.4997,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.003117017913609743,
"rewards/margins": -0.001670417725108564,
"rewards/rejected": 0.004787435755133629,
"step": 2
},
{
"debug/policy_chosen_logits": -1.1543760299682617,
"debug/policy_chosen_logps": -162.00730895996094,
"debug/policy_rejected_logits": -0.9926377534866333,
"debug/policy_rejected_logps": -148.77609252929688,
"debug/reference_chosen_logps": -161.8112030029297,
"debug/reference_rejected_logps": -148.79986572265625,
"epoch": 0.07317073170731707,
"grad_norm": 7.862425079439745,
"learning_rate": 1e-06,
"logits/chosen": -1.1543760299682617,
"logits/rejected": -0.9926377534866333,
"logps/chosen": -162.00730895996094,
"logps/rejected": -148.77609252929688,
"loss": 0.4992,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.0019610594026744366,
"rewards/margins": -0.0021987915970385075,
"rewards/rejected": 0.00023773196153342724,
"step": 3
},
{
"debug/policy_chosen_logits": -0.9400157332420349,
"debug/policy_chosen_logps": -156.6750030517578,
"debug/policy_rejected_logits": -1.0246046781539917,
"debug/policy_rejected_logps": -140.80905151367188,
"debug/reference_chosen_logps": -156.9263153076172,
"debug/reference_rejected_logps": -141.17587280273438,
"epoch": 0.0975609756097561,
"grad_norm": 8.32764259608944,
"learning_rate": 1e-06,
"logits/chosen": -0.9400157332420349,
"logits/rejected": -1.0246046781539917,
"logps/chosen": -156.6750030517578,
"logps/rejected": -140.80905151367188,
"loss": 0.4991,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.0025131413713097572,
"rewards/margins": -0.0011551285861060023,
"rewards/rejected": 0.0036682700738310814,
"step": 4
},
{
"debug/policy_chosen_logits": -0.8135665059089661,
"debug/policy_chosen_logps": -180.49978637695312,
"debug/policy_rejected_logits": -0.966259241104126,
"debug/policy_rejected_logps": -180.0769500732422,
"debug/reference_chosen_logps": -180.33763122558594,
"debug/reference_rejected_logps": -180.13702392578125,
"epoch": 0.12195121951219512,
"grad_norm": 9.5687290716109,
"learning_rate": 1e-06,
"logits/chosen": -0.8135665059089661,
"logits/rejected": -0.966259241104126,
"logps/chosen": -180.49978637695312,
"logps/rejected": -180.0769500732422,
"loss": 0.5012,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.0016215991927310824,
"rewards/margins": -0.002222432754933834,
"rewards/rejected": 0.0006008340278640389,
"step": 5
},
{
"debug/policy_chosen_logits": -0.875511109828949,
"debug/policy_chosen_logps": -169.20863342285156,
"debug/policy_rejected_logits": -0.8739838600158691,
"debug/policy_rejected_logps": -161.75338745117188,
"debug/reference_chosen_logps": -168.50558471679688,
"debug/reference_rejected_logps": -160.78341674804688,
"epoch": 0.14634146341463414,
"grad_norm": 8.253308110293828,
"learning_rate": 1e-06,
"logits/chosen": -0.875511109828949,
"logits/rejected": -0.8739838600158691,
"logps/chosen": -169.20863342285156,
"logps/rejected": -161.75338745117188,
"loss": 0.5028,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.007030477747321129,
"rewards/margins": 0.0026691434904932976,
"rewards/rejected": -0.009699621237814426,
"step": 6
},
{
"debug/policy_chosen_logits": -0.9064666628837585,
"debug/policy_chosen_logps": -163.56883239746094,
"debug/policy_rejected_logits": -0.88660728931427,
"debug/policy_rejected_logps": -167.95022583007812,
"debug/reference_chosen_logps": -162.90780639648438,
"debug/reference_rejected_logps": -166.8964080810547,
"epoch": 0.17073170731707318,
"grad_norm": 9.148150186683857,
"learning_rate": 1e-06,
"logits/chosen": -0.9064666628837585,
"logits/rejected": -0.88660728931427,
"logps/chosen": -163.56883239746094,
"logps/rejected": -167.95022583007812,
"loss": 0.5021,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.0066101741977036,
"rewards/margins": 0.0039279647171497345,
"rewards/rejected": -0.010538139380514622,
"step": 7
},
{
"debug/policy_chosen_logits": -1.0406209230422974,
"debug/policy_chosen_logps": -145.71910095214844,
"debug/policy_rejected_logits": -0.9474231004714966,
"debug/policy_rejected_logps": -176.6788330078125,
"debug/reference_chosen_logps": -145.38890075683594,
"debug/reference_rejected_logps": -176.26510620117188,
"epoch": 0.1951219512195122,
"grad_norm": 7.873026337514701,
"learning_rate": 1e-06,
"logits/chosen": -1.0406209230422974,
"logits/rejected": -0.9474231004714966,
"logps/chosen": -145.71910095214844,
"logps/rejected": -176.6788330078125,
"loss": 0.4983,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.003301925491541624,
"rewards/margins": 0.0008353232406079769,
"rewards/rejected": -0.004137249197810888,
"step": 8
},
{
"debug/policy_chosen_logits": -0.7725314497947693,
"debug/policy_chosen_logps": -191.3946533203125,
"debug/policy_rejected_logits": -0.9446115493774414,
"debug/policy_rejected_logps": -157.00643920898438,
"debug/reference_chosen_logps": -191.58938598632812,
"debug/reference_rejected_logps": -157.259521484375,
"epoch": 0.21951219512195122,
"grad_norm": 8.113917191223486,
"learning_rate": 1e-06,
"logits/chosen": -0.7725314497947693,
"logits/rejected": -0.9446115493774414,
"logps/chosen": -191.3946533203125,
"logps/rejected": -157.00643920898438,
"loss": 0.4984,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.0019471454434096813,
"rewards/margins": -0.0005835914053022861,
"rewards/rejected": 0.00253073638305068,
"step": 9
},
{
"debug/policy_chosen_logits": -0.9539018273353577,
"debug/policy_chosen_logps": -156.33084106445312,
"debug/policy_rejected_logits": -0.8661950826644897,
"debug/policy_rejected_logps": -166.2339630126953,
"debug/reference_chosen_logps": -156.4985809326172,
"debug/reference_rejected_logps": -165.98721313476562,
"epoch": 0.24390243902439024,
"grad_norm": 8.579040246656625,
"learning_rate": 1e-06,
"logits/chosen": -0.9539018273353577,
"logits/rejected": -0.8661950826644897,
"logps/chosen": -156.33084106445312,
"logps/rejected": -166.2339630126953,
"loss": 0.502,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.001677446300163865,
"rewards/margins": 0.0041448017582297325,
"rewards/rejected": -0.002467355690896511,
"step": 10
},
{
"debug/policy_chosen_logits": -0.9540179371833801,
"debug/policy_chosen_logps": -182.48159790039062,
"debug/policy_rejected_logits": -0.9883304834365845,
"debug/policy_rejected_logps": -147.31826782226562,
"debug/reference_chosen_logps": -184.1331024169922,
"debug/reference_rejected_logps": -149.22958374023438,
"epoch": 0.2682926829268293,
"grad_norm": 8.658564154949419,
"learning_rate": 1e-06,
"logits/chosen": -0.9540179371833801,
"logits/rejected": -0.9883304834365845,
"logps/chosen": -182.48159790039062,
"logps/rejected": -147.31826782226562,
"loss": 0.4978,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.016515053808689117,
"rewards/margins": -0.0025980568025261164,
"rewards/rejected": 0.01911311037838459,
"step": 11
},
{
"debug/policy_chosen_logits": -0.9231241345405579,
"debug/policy_chosen_logps": -173.3875274658203,
"debug/policy_rejected_logits": -0.9746717810630798,
"debug/policy_rejected_logps": -186.23223876953125,
"debug/reference_chosen_logps": -173.79617309570312,
"debug/reference_rejected_logps": -185.99221801757812,
"epoch": 0.2926829268292683,
"grad_norm": 9.641106516800372,
"learning_rate": 1e-06,
"logits/chosen": -0.9231241345405579,
"logits/rejected": -0.9746717810630798,
"logps/chosen": -173.3875274658203,
"logps/rejected": -186.23223876953125,
"loss": 0.4994,
"rewards/accuracies": 0.75,
"rewards/chosen": 0.0040865708142519,
"rewards/margins": 0.006486768834292889,
"rewards/rejected": -0.0024001982528716326,
"step": 12
},
{
"debug/policy_chosen_logits": -0.8564068675041199,
"debug/policy_chosen_logps": -155.58351135253906,
"debug/policy_rejected_logits": -0.78647381067276,
"debug/policy_rejected_logps": -180.34735107421875,
"debug/reference_chosen_logps": -155.46743774414062,
"debug/reference_rejected_logps": -179.4715576171875,
"epoch": 0.3170731707317073,
"grad_norm": 7.778587819967809,
"learning_rate": 1e-06,
"logits/chosen": -0.8564068675041199,
"logits/rejected": -0.78647381067276,
"logps/chosen": -155.58351135253906,
"logps/rejected": -180.34735107421875,
"loss": 0.4997,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.0011608502827584743,
"rewards/margins": 0.007597027346491814,
"rewards/rejected": -0.008757877163589,
"step": 13
},
{
"debug/policy_chosen_logits": -0.9291447997093201,
"debug/policy_chosen_logps": -137.7774658203125,
"debug/policy_rejected_logits": -0.8161742687225342,
"debug/policy_rejected_logps": -154.59149169921875,
"debug/reference_chosen_logps": -136.43026733398438,
"debug/reference_rejected_logps": -152.96876525878906,
"epoch": 0.34146341463414637,
"grad_norm": 9.452087496130199,
"learning_rate": 1e-06,
"logits/chosen": -0.9291447997093201,
"logits/rejected": -0.8161742687225342,
"logps/chosen": -137.7774658203125,
"logps/rejected": -154.59149169921875,
"loss": 0.4991,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.013471984304487705,
"rewards/margins": 0.002755412831902504,
"rewards/rejected": -0.016227398067712784,
"step": 14
},
{
"debug/policy_chosen_logits": -0.8452920317649841,
"debug/policy_chosen_logps": -175.22425842285156,
"debug/policy_rejected_logits": -0.7951302528381348,
"debug/policy_rejected_logps": -175.16729736328125,
"debug/reference_chosen_logps": -174.3819122314453,
"debug/reference_rejected_logps": -172.59144592285156,
"epoch": 0.36585365853658536,
"grad_norm": 8.577281209186165,
"learning_rate": 1e-06,
"logits/chosen": -0.8452920317649841,
"logits/rejected": -0.7951302528381348,
"logps/chosen": -175.22425842285156,
"logps/rejected": -175.16729736328125,
"loss": 0.4999,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.008423404768109322,
"rewards/margins": 0.017334945499897003,
"rewards/rejected": -0.025758352130651474,
"step": 15
},
{
"debug/policy_chosen_logits": -0.8255157470703125,
"debug/policy_chosen_logps": -196.6995391845703,
"debug/policy_rejected_logits": -1.0168753862380981,
"debug/policy_rejected_logps": -144.28012084960938,
"debug/reference_chosen_logps": -194.49815368652344,
"debug/reference_rejected_logps": -141.6895751953125,
"epoch": 0.3902439024390244,
"grad_norm": 10.569580209063629,
"learning_rate": 1e-06,
"logits/chosen": -0.8255157470703125,
"logits/rejected": -1.0168753862380981,
"logps/chosen": -196.6995391845703,
"logps/rejected": -144.28012084960938,
"loss": 0.5055,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.02201385423541069,
"rewards/margins": 0.003891811240464449,
"rewards/rejected": -0.025905665010213852,
"step": 16
},
{
"debug/policy_chosen_logits": -0.8963515162467957,
"debug/policy_chosen_logps": -167.45254516601562,
"debug/policy_rejected_logits": -1.1090606451034546,
"debug/policy_rejected_logps": -132.12088012695312,
"debug/reference_chosen_logps": -165.4241943359375,
"debug/reference_rejected_logps": -131.6002655029297,
"epoch": 0.4146341463414634,
"grad_norm": 8.216685819750042,
"learning_rate": 1e-06,
"logits/chosen": -0.8963515162467957,
"logits/rejected": -1.1090606451034546,
"logps/chosen": -167.45254516601562,
"logps/rejected": -132.12088012695312,
"loss": 0.5023,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.020283488556742668,
"rewards/margins": -0.015077323652803898,
"rewards/rejected": -0.00520616490393877,
"step": 17
},
{
"debug/policy_chosen_logits": -1.1278187036514282,
"debug/policy_chosen_logps": -163.56838989257812,
"debug/policy_rejected_logits": -0.7523326277732849,
"debug/policy_rejected_logps": -194.7676239013672,
"debug/reference_chosen_logps": -164.0474395751953,
"debug/reference_rejected_logps": -195.21795654296875,
"epoch": 0.43902439024390244,
"grad_norm": 8.12497169240311,
"learning_rate": 1e-06,
"logits/chosen": -1.1278187036514282,
"logits/rejected": -0.7523326277732849,
"logps/chosen": -163.56838989257812,
"logps/rejected": -194.7676239013672,
"loss": 0.5046,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.004790563136339188,
"rewards/margins": 0.0002871984615921974,
"rewards/rejected": 0.00450336467474699,
"step": 18
},
{
"debug/policy_chosen_logits": -0.8814194202423096,
"debug/policy_chosen_logps": -187.4921417236328,
"debug/policy_rejected_logits": -0.9221204519271851,
"debug/policy_rejected_logps": -184.78768920898438,
"debug/reference_chosen_logps": -187.89466857910156,
"debug/reference_rejected_logps": -186.1348876953125,
"epoch": 0.4634146341463415,
"grad_norm": 8.289942810390798,
"learning_rate": 1e-06,
"logits/chosen": -0.8814194202423096,
"logits/rejected": -0.9221204519271851,
"logps/chosen": -187.4921417236328,
"logps/rejected": -184.78768920898438,
"loss": 0.4986,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.004025382921099663,
"rewards/margins": -0.009446544572710991,
"rewards/rejected": 0.013471927493810654,
"step": 19
},
{
"debug/policy_chosen_logits": -0.9697524905204773,
"debug/policy_chosen_logps": -142.56431579589844,
"debug/policy_rejected_logits": -0.7087419033050537,
"debug/policy_rejected_logps": -181.8319091796875,
"debug/reference_chosen_logps": -145.44419860839844,
"debug/reference_rejected_logps": -184.153564453125,
"epoch": 0.4878048780487805,
"grad_norm": 9.078292331479483,
"learning_rate": 1e-06,
"logits/chosen": -0.9697524905204773,
"logits/rejected": -0.7087419033050537,
"logps/chosen": -142.56431579589844,
"logps/rejected": -181.8319091796875,
"loss": 0.4974,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.028798799961805344,
"rewards/margins": 0.005582190118730068,
"rewards/rejected": 0.0232166089117527,
"step": 20
},
{
"debug/policy_chosen_logits": -0.8631553649902344,
"debug/policy_chosen_logps": -181.8375701904297,
"debug/policy_rejected_logits": -0.8101003170013428,
"debug/policy_rejected_logps": -204.69305419921875,
"debug/reference_chosen_logps": -181.6129150390625,
"debug/reference_rejected_logps": -206.70823669433594,
"epoch": 0.5121951219512195,
"grad_norm": 9.63889554364432,
"learning_rate": 1e-06,
"logits/chosen": -0.8631553649902344,
"logits/rejected": -0.8101003170013428,
"logps/chosen": -181.8375701904297,
"logps/rejected": -204.69305419921875,
"loss": 0.5031,
"rewards/accuracies": 0.125,
"rewards/chosen": -0.0022465987130999565,
"rewards/margins": -0.02239835634827614,
"rewards/rejected": 0.020151756703853607,
"step": 21
},
{
"debug/policy_chosen_logits": -0.900942862033844,
"debug/policy_chosen_logps": -156.7996826171875,
"debug/policy_rejected_logits": -0.7672395706176758,
"debug/policy_rejected_logps": -190.6102294921875,
"debug/reference_chosen_logps": -157.8141326904297,
"debug/reference_rejected_logps": -191.65597534179688,
"epoch": 0.5365853658536586,
"grad_norm": 8.86160952811184,
"learning_rate": 1e-06,
"logits/chosen": -0.900942862033844,
"logits/rejected": -0.7672395706176758,
"logps/chosen": -156.7996826171875,
"logps/rejected": -190.6102294921875,
"loss": 0.4989,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.010144614614546299,
"rewards/margins": -0.0003129197284579277,
"rewards/rejected": 0.010457534343004227,
"step": 22
},
{
"debug/policy_chosen_logits": -0.8332923054695129,
"debug/policy_chosen_logps": -155.85830688476562,
"debug/policy_rejected_logits": -0.7771956324577332,
"debug/policy_rejected_logps": -165.27749633789062,
"debug/reference_chosen_logps": -156.92462158203125,
"debug/reference_rejected_logps": -165.16729736328125,
"epoch": 0.5609756097560976,
"grad_norm": 8.143779892532098,
"learning_rate": 1e-06,
"logits/chosen": -0.8332923054695129,
"logits/rejected": -0.7771956324577332,
"logps/chosen": -155.85830688476562,
"logps/rejected": -165.27749633789062,
"loss": 0.5003,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.0106631089001894,
"rewards/margins": 0.011765326373279095,
"rewards/rejected": -0.001102218870073557,
"step": 23
},
{
"debug/policy_chosen_logits": -1.01643705368042,
"debug/policy_chosen_logps": -142.14419555664062,
"debug/policy_rejected_logits": -0.9811778664588928,
"debug/policy_rejected_logps": -161.79629516601562,
"debug/reference_chosen_logps": -141.1592559814453,
"debug/reference_rejected_logps": -159.3284149169922,
"epoch": 0.5853658536585366,
"grad_norm": 8.084865068688668,
"learning_rate": 1e-06,
"logits/chosen": -1.01643705368042,
"logits/rejected": -0.9811778664588928,
"logps/chosen": -142.14419555664062,
"logps/rejected": -161.79629516601562,
"loss": 0.5022,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.009849337860941887,
"rewards/margins": 0.014829404652118683,
"rewards/rejected": -0.02467874437570572,
"step": 24
},
{
"debug/policy_chosen_logits": -0.7919837236404419,
"debug/policy_chosen_logps": -162.75352478027344,
"debug/policy_rejected_logits": -0.96061772108078,
"debug/policy_rejected_logps": -169.0644073486328,
"debug/reference_chosen_logps": -161.2084503173828,
"debug/reference_rejected_logps": -167.456787109375,
"epoch": 0.6097560975609756,
"grad_norm": 9.111682216370516,
"learning_rate": 1e-06,
"logits/chosen": -0.7919837236404419,
"logits/rejected": -0.96061772108078,
"logps/chosen": -162.75352478027344,
"logps/rejected": -169.0644073486328,
"loss": 0.5016,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.015450754202902317,
"rewards/margins": 0.0006255432963371277,
"rewards/rejected": -0.01607629843056202,
"step": 25
},
{
"debug/policy_chosen_logits": -0.9698644876480103,
"debug/policy_chosen_logps": -165.7355194091797,
"debug/policy_rejected_logits": -0.7252393364906311,
"debug/policy_rejected_logps": -183.15887451171875,
"debug/reference_chosen_logps": -164.15481567382812,
"debug/reference_rejected_logps": -182.38124084472656,
"epoch": 0.6341463414634146,
"grad_norm": 10.115977003883836,
"learning_rate": 1e-06,
"logits/chosen": -0.9698644876480103,
"logits/rejected": -0.7252393364906311,
"logps/chosen": -165.7355194091797,
"logps/rejected": -183.15887451171875,
"loss": 0.502,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.015806876122951508,
"rewards/margins": -0.008030500262975693,
"rewards/rejected": -0.00777637492865324,
"step": 26
},
{
"debug/policy_chosen_logits": -0.7707141637802124,
"debug/policy_chosen_logps": -205.21841430664062,
"debug/policy_rejected_logits": -0.9271693229675293,
"debug/policy_rejected_logps": -168.64834594726562,
"debug/reference_chosen_logps": -205.48428344726562,
"debug/reference_rejected_logps": -168.75538635253906,
"epoch": 0.6585365853658537,
"grad_norm": 9.132022098455337,
"learning_rate": 1e-06,
"logits/chosen": -0.7707141637802124,
"logits/rejected": -0.9271693229675293,
"logps/chosen": -205.21841430664062,
"logps/rejected": -168.64834594726562,
"loss": 0.4964,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.002658710815012455,
"rewards/margins": 0.0015882402658462524,
"rewards/rejected": 0.0010704714804887772,
"step": 27
},
{
"debug/policy_chosen_logits": -0.9867036938667297,
"debug/policy_chosen_logps": -170.24920654296875,
"debug/policy_rejected_logits": -0.958678126335144,
"debug/policy_rejected_logps": -172.13462829589844,
"debug/reference_chosen_logps": -169.69461059570312,
"debug/reference_rejected_logps": -171.46791076660156,
"epoch": 0.6829268292682927,
"grad_norm": 8.165335264629883,
"learning_rate": 1e-06,
"logits/chosen": -0.9867036938667297,
"logits/rejected": -0.958678126335144,
"logps/chosen": -170.24920654296875,
"logps/rejected": -172.13462829589844,
"loss": 0.4934,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.005545825697481632,
"rewards/margins": 0.0011212928220629692,
"rewards/rejected": -0.006667118053883314,
"step": 28
},
{
"debug/policy_chosen_logits": -0.6699000000953674,
"debug/policy_chosen_logps": -189.43618774414062,
"debug/policy_rejected_logits": -0.8863887786865234,
"debug/policy_rejected_logps": -179.74417114257812,
"debug/reference_chosen_logps": -188.31427001953125,
"debug/reference_rejected_logps": -179.5847625732422,
"epoch": 0.7073170731707317,
"grad_norm": 8.679731931376267,
"learning_rate": 1e-06,
"logits/chosen": -0.6699000000953674,
"logits/rejected": -0.8863887786865234,
"logps/chosen": -189.43618774414062,
"logps/rejected": -179.74417114257812,
"loss": 0.5078,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.011219177395105362,
"rewards/margins": -0.009625071659684181,
"rewards/rejected": -0.0015941057354211807,
"step": 29
},
{
"debug/policy_chosen_logits": -0.9941346645355225,
"debug/policy_chosen_logps": -139.71783447265625,
"debug/policy_rejected_logits": -0.9757084846496582,
"debug/policy_rejected_logps": -147.81222534179688,
"debug/reference_chosen_logps": -140.21261596679688,
"debug/reference_rejected_logps": -150.31747436523438,
"epoch": 0.7317073170731707,
"grad_norm": 8.152161947975058,
"learning_rate": 1e-06,
"logits/chosen": -0.9941346645355225,
"logits/rejected": -0.9757084846496582,
"logps/chosen": -139.71783447265625,
"logps/rejected": -147.81222534179688,
"loss": 0.4989,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.0049479007720947266,
"rewards/margins": -0.020104561001062393,
"rewards/rejected": 0.02505246177315712,
"step": 30
},
{
"debug/policy_chosen_logits": -0.9950529932975769,
"debug/policy_chosen_logps": -167.46485900878906,
"debug/policy_rejected_logits": -0.9336649179458618,
"debug/policy_rejected_logps": -161.46160888671875,
"debug/reference_chosen_logps": -170.01150512695312,
"debug/reference_rejected_logps": -164.1259765625,
"epoch": 0.7560975609756098,
"grad_norm": 8.534988911622145,
"learning_rate": 1e-06,
"logits/chosen": -0.9950529932975769,
"logits/rejected": -0.9336649179458618,
"logps/chosen": -167.46485900878906,
"logps/rejected": -161.46160888671875,
"loss": 0.4947,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.02546636573970318,
"rewards/margins": -0.0011772820726037025,
"rewards/rejected": 0.026643646880984306,
"step": 31
},
{
"debug/policy_chosen_logits": -0.99274080991745,
"debug/policy_chosen_logps": -141.17715454101562,
"debug/policy_rejected_logits": -0.9643717408180237,
"debug/policy_rejected_logps": -163.40673828125,
"debug/reference_chosen_logps": -142.4697723388672,
"debug/reference_rejected_logps": -163.57907104492188,
"epoch": 0.7804878048780488,
"grad_norm": 8.868022634011657,
"learning_rate": 1e-06,
"logits/chosen": -0.99274080991745,
"logits/rejected": -0.9643717408180237,
"logps/chosen": -141.17715454101562,
"logps/rejected": -163.40673828125,
"loss": 0.4887,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.012926094233989716,
"rewards/margins": 0.01120278425514698,
"rewards/rejected": 0.0017233085818588734,
"step": 32
},
{
"debug/policy_chosen_logits": -0.7678695321083069,
"debug/policy_chosen_logps": -179.30178833007812,
"debug/policy_rejected_logits": -0.9433090686798096,
"debug/policy_rejected_logps": -166.5570068359375,
"debug/reference_chosen_logps": -178.72622680664062,
"debug/reference_rejected_logps": -164.87799072265625,
"epoch": 0.8048780487804879,
"grad_norm": 8.118291328693896,
"learning_rate": 1e-06,
"logits/chosen": -0.7678695321083069,
"logits/rejected": -0.9433090686798096,
"logps/chosen": -179.30178833007812,
"logps/rejected": -166.5570068359375,
"loss": 0.4949,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.005755499936640263,
"rewards/margins": 0.011034755036234856,
"rewards/rejected": -0.016790255904197693,
"step": 33
},
{
"debug/policy_chosen_logits": -0.9250268936157227,
"debug/policy_chosen_logps": -174.24758911132812,
"debug/policy_rejected_logits": -0.8358609676361084,
"debug/policy_rejected_logps": -207.80772399902344,
"debug/reference_chosen_logps": -174.22793579101562,
"debug/reference_rejected_logps": -206.686279296875,
"epoch": 0.8292682926829268,
"grad_norm": 9.100751206625086,
"learning_rate": 1e-06,
"logits/chosen": -0.9250268936157227,
"logits/rejected": -0.8358609676361084,
"logps/chosen": -174.24758911132812,
"logps/rejected": -207.80772399902344,
"loss": 0.5021,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.00019640009850263596,
"rewards/margins": 0.01101800799369812,
"rewards/rejected": -0.01121440902352333,
"step": 34
},
{
"debug/policy_chosen_logits": -0.9135526418685913,
"debug/policy_chosen_logps": -159.21298217773438,
"debug/policy_rejected_logits": -1.1572299003601074,
"debug/policy_rejected_logps": -147.13705444335938,
"debug/reference_chosen_logps": -159.4290771484375,
"debug/reference_rejected_logps": -145.97109985351562,
"epoch": 0.8536585365853658,
"grad_norm": 8.661910409548431,
"learning_rate": 1e-06,
"logits/chosen": -0.9135526418685913,
"logits/rejected": -1.1572299003601074,
"logps/chosen": -159.21298217773438,
"logps/rejected": -147.13705444335938,
"loss": 0.4925,
"rewards/accuracies": 0.5,
"rewards/chosen": 0.0021607973612844944,
"rewards/margins": 0.013820314779877663,
"rewards/rejected": -0.011659517884254456,
"step": 35
},
{
"debug/policy_chosen_logits": -0.706568717956543,
"debug/policy_chosen_logps": -185.29225158691406,
"debug/policy_rejected_logits": -0.772286593914032,
"debug/policy_rejected_logps": -181.357177734375,
"debug/reference_chosen_logps": -186.060302734375,
"debug/reference_rejected_logps": -181.81985473632812,
"epoch": 0.8780487804878049,
"grad_norm": 8.794546491268,
"learning_rate": 1e-06,
"logits/chosen": -0.706568717956543,
"logits/rejected": -0.772286593914032,
"logps/chosen": -185.29225158691406,
"logps/rejected": -181.357177734375,
"loss": 0.4963,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.007680453825742006,
"rewards/margins": 0.0030536362901329994,
"rewards/rejected": 0.004626817535609007,
"step": 36
},
{
"debug/policy_chosen_logits": -0.9353126883506775,
"debug/policy_chosen_logps": -148.95077514648438,
"debug/policy_rejected_logits": -0.8417949080467224,
"debug/policy_rejected_logps": -192.06149291992188,
"debug/reference_chosen_logps": -148.9204864501953,
"debug/reference_rejected_logps": -192.1736297607422,
"epoch": 0.9024390243902439,
"grad_norm": 9.147337941113607,
"learning_rate": 1e-06,
"logits/chosen": -0.9353126883506775,
"logits/rejected": -0.8417949080467224,
"logps/chosen": -148.95077514648438,
"logps/rejected": -192.06149291992188,
"loss": 0.4959,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.00030298251658678055,
"rewards/margins": -0.001424331683665514,
"rewards/rejected": 0.0011213491670787334,
"step": 37
},
{
"debug/policy_chosen_logits": -0.9579962491989136,
"debug/policy_chosen_logps": -145.99325561523438,
"debug/policy_rejected_logits": -0.771964430809021,
"debug/policy_rejected_logps": -181.51898193359375,
"debug/reference_chosen_logps": -146.1923828125,
"debug/reference_rejected_logps": -181.7994384765625,
"epoch": 0.926829268292683,
"grad_norm": 8.770906218078656,
"learning_rate": 1e-06,
"logits/chosen": -0.9579962491989136,
"logits/rejected": -0.771964430809021,
"logps/chosen": -145.99325561523438,
"logps/rejected": -181.51898193359375,
"loss": 0.5028,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.0019913101568818092,
"rewards/margins": -0.0008130739443004131,
"rewards/rejected": 0.0028043838683515787,
"step": 38
},
{
"debug/policy_chosen_logits": -0.7748340964317322,
"debug/policy_chosen_logps": -192.96409606933594,
"debug/policy_rejected_logits": -0.7723251581192017,
"debug/policy_rejected_logps": -193.10032653808594,
"debug/reference_chosen_logps": -194.00680541992188,
"debug/reference_rejected_logps": -195.18478393554688,
"epoch": 0.9512195121951219,
"grad_norm": 9.455831322622839,
"learning_rate": 1e-06,
"logits/chosen": -0.7748340964317322,
"logits/rejected": -0.7723251581192017,
"logps/chosen": -192.96409606933594,
"logps/rejected": -193.10032653808594,
"loss": 0.4989,
"rewards/accuracies": 0.375,
"rewards/chosen": 0.010427169501781464,
"rewards/margins": -0.010417431592941284,
"rewards/rejected": 0.020844601094722748,
"step": 39
},
{
"debug/policy_chosen_logits": -0.9706252813339233,
"debug/policy_chosen_logps": -156.927490234375,
"debug/policy_rejected_logits": -0.8014242649078369,
"debug/policy_rejected_logps": -197.02548217773438,
"debug/reference_chosen_logps": -155.9664764404297,
"debug/reference_rejected_logps": -197.78335571289062,
"epoch": 0.975609756097561,
"grad_norm": 9.78790232486535,
"learning_rate": 1e-06,
"logits/chosen": -0.9706252813339233,
"logits/rejected": -0.8014242649078369,
"logps/chosen": -156.927490234375,
"logps/rejected": -197.02548217773438,
"loss": 0.5018,
"rewards/accuracies": 0.25,
"rewards/chosen": -0.009610195644199848,
"rewards/margins": -0.017188798636198044,
"rewards/rejected": 0.007578602526336908,
"step": 40
},
{
"debug/policy_chosen_logits": -0.7953402400016785,
"debug/policy_chosen_logps": -170.52337646484375,
"debug/policy_rejected_logits": -0.8171042799949646,
"debug/policy_rejected_logps": -175.24819946289062,
"debug/reference_chosen_logps": -169.29534912109375,
"debug/reference_rejected_logps": -173.3860626220703,
"epoch": 1.0,
"grad_norm": 10.67828699749108,
"learning_rate": 1e-06,
"logits/chosen": -0.7953402400016785,
"logits/rejected": -0.8171042799949646,
"logps/chosen": -170.52337646484375,
"logps/rejected": -175.24819946289062,
"loss": 0.4925,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.01228022575378418,
"rewards/margins": 0.006341142114251852,
"rewards/rejected": -0.01862136647105217,
"step": 41
},
{
"epoch": 1.0,
"step": 41,
"total_flos": 0.0,
"train_loss": 0.49933575202779074,
"train_runtime": 145.5342,
"train_samples_per_second": 17.989,
"train_steps_per_second": 0.282
}
],
"logging_steps": 1,
"max_steps": 41,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}