sigmoid_ds_chat_rmsprop_iter3 / trainer_state.json
yiran-wang3's picture
End of training
bd4aa30 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 100,
"global_step": 42,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"debug/policy_chosen_logits": 0.09027537703514099,
"debug/policy_chosen_logps": -162.730224609375,
"debug/policy_rejected_logits": 0.5158556699752808,
"debug/policy_rejected_logps": -184.16571044921875,
"debug/reference_chosen_logps": -162.730224609375,
"debug/reference_rejected_logps": -184.16571044921875,
"epoch": 0.023809523809523808,
"grad_norm": 4.685966665550777,
"learning_rate": 5e-07,
"logits/chosen": 0.09027537703514099,
"logits/rejected": 0.5158556699752808,
"logps/chosen": -162.730224609375,
"logps/rejected": -184.16571044921875,
"loss": 0.6931,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"debug/policy_chosen_logits": 0.7074397802352905,
"debug/policy_chosen_logps": -150.46807861328125,
"debug/policy_rejected_logits": 0.43174317479133606,
"debug/policy_rejected_logps": -140.48440551757812,
"debug/reference_chosen_logps": -153.92564392089844,
"debug/reference_rejected_logps": -142.85406494140625,
"epoch": 0.047619047619047616,
"grad_norm": 15.044621835270732,
"learning_rate": 5e-07,
"logits/chosen": 0.7074397802352905,
"logits/rejected": 0.43174317479133606,
"logps/chosen": -150.46807861328125,
"logps/rejected": -140.48440551757812,
"loss": 0.6973,
"rewards/accuracies": 0.625,
"rewards/chosen": 0.034575510770082474,
"rewards/margins": 0.010879031382501125,
"rewards/rejected": 0.023696478456258774,
"step": 2
},
{
"debug/policy_chosen_logits": 0.4280781149864197,
"debug/policy_chosen_logps": -153.80137634277344,
"debug/policy_rejected_logits": 1.081570029258728,
"debug/policy_rejected_logps": -173.27056884765625,
"debug/reference_chosen_logps": -151.54473876953125,
"debug/reference_rejected_logps": -169.95703125,
"epoch": 0.07142857142857142,
"grad_norm": 4.933146485934281,
"learning_rate": 5e-07,
"logits/chosen": 0.4280781149864197,
"logits/rejected": 1.081570029258728,
"logps/chosen": -153.80137634277344,
"logps/rejected": -173.27056884765625,
"loss": 0.6923,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.022566460072994232,
"rewards/margins": 0.010568867437541485,
"rewards/rejected": -0.03313532844185829,
"step": 3
},
{
"debug/policy_chosen_logits": 0.49108535051345825,
"debug/policy_chosen_logps": -155.66534423828125,
"debug/policy_rejected_logits": 0.4322296977043152,
"debug/policy_rejected_logps": -149.7783203125,
"debug/reference_chosen_logps": -153.9282989501953,
"debug/reference_rejected_logps": -148.75108337402344,
"epoch": 0.09523809523809523,
"grad_norm": 6.198349828531256,
"learning_rate": 5e-07,
"logits/chosen": 0.49108535051345825,
"logits/rejected": 0.4322296977043152,
"logps/chosen": -155.66534423828125,
"logps/rejected": -149.7783203125,
"loss": 0.6899,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.0173704344779253,
"rewards/margins": -0.0070981215685606,
"rewards/rejected": -0.010272311978042126,
"step": 4
},
{
"debug/policy_chosen_logits": 1.0142210721969604,
"debug/policy_chosen_logps": -158.026611328125,
"debug/policy_rejected_logits": 1.0418132543563843,
"debug/policy_rejected_logps": -199.23785400390625,
"debug/reference_chosen_logps": -155.4060516357422,
"debug/reference_rejected_logps": -194.74618530273438,
"epoch": 0.11904761904761904,
"grad_norm": 4.004423522491588,
"learning_rate": 5e-07,
"logits/chosen": 1.0142210721969604,
"logits/rejected": 1.0418132543563843,
"logps/chosen": -158.026611328125,
"logps/rejected": -199.23785400390625,
"loss": 0.6916,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.026205480098724365,
"rewards/margins": 0.018711339682340622,
"rewards/rejected": -0.04491681978106499,
"step": 5
},
{
"debug/policy_chosen_logits": 0.803213894367218,
"debug/policy_chosen_logps": -156.5592041015625,
"debug/policy_rejected_logits": 0.9607799649238586,
"debug/policy_rejected_logps": -173.58987426757812,
"debug/reference_chosen_logps": -146.985595703125,
"debug/reference_rejected_logps": -163.15786743164062,
"epoch": 0.14285714285714285,
"grad_norm": 10.059361542630269,
"learning_rate": 5e-07,
"logits/chosen": 0.803213894367218,
"logits/rejected": 0.9607799649238586,
"logps/chosen": -156.5592041015625,
"logps/rejected": -173.58987426757812,
"loss": 0.6869,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.09573620557785034,
"rewards/margins": 0.008583765476942062,
"rewards/rejected": -0.1043199747800827,
"step": 6
},
{
"debug/policy_chosen_logits": 0.5479093194007874,
"debug/policy_chosen_logps": -176.60946655273438,
"debug/policy_rejected_logits": 0.11254727840423584,
"debug/policy_rejected_logps": -175.84767150878906,
"debug/reference_chosen_logps": -174.020751953125,
"debug/reference_rejected_logps": -170.24949645996094,
"epoch": 0.16666666666666666,
"grad_norm": 20.07889018294013,
"learning_rate": 5e-07,
"logits/chosen": 0.5479093194007874,
"logits/rejected": 0.11254727840423584,
"logps/chosen": -176.60946655273438,
"logps/rejected": -175.84767150878906,
"loss": 0.6892,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.025887146592140198,
"rewards/margins": 0.030094660818576813,
"rewards/rejected": -0.05598180741071701,
"step": 7
},
{
"debug/policy_chosen_logits": 0.5916139483451843,
"debug/policy_chosen_logps": -164.50384521484375,
"debug/policy_rejected_logits": 0.5801162123680115,
"debug/policy_rejected_logps": -156.9475860595703,
"debug/reference_chosen_logps": -153.96173095703125,
"debug/reference_rejected_logps": -147.9217071533203,
"epoch": 0.19047619047619047,
"grad_norm": 4.5921375475915776,
"learning_rate": 5e-07,
"logits/chosen": 0.5916139483451843,
"logits/rejected": 0.5801162123680115,
"logps/chosen": -164.50384521484375,
"logps/rejected": -156.9475860595703,
"loss": 0.69,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.10542111098766327,
"rewards/margins": -0.015162268653512001,
"rewards/rejected": -0.09025884419679642,
"step": 8
},
{
"debug/policy_chosen_logits": 0.2481817752122879,
"debug/policy_chosen_logps": -169.9642333984375,
"debug/policy_rejected_logits": 0.48347601294517517,
"debug/policy_rejected_logps": -190.4657440185547,
"debug/reference_chosen_logps": -162.74264526367188,
"debug/reference_rejected_logps": -181.60940551757812,
"epoch": 0.21428571428571427,
"grad_norm": 5.023010276429253,
"learning_rate": 5e-07,
"logits/chosen": 0.2481817752122879,
"logits/rejected": 0.48347601294517517,
"logps/chosen": -169.9642333984375,
"logps/rejected": -190.4657440185547,
"loss": 0.6877,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.07221580296754837,
"rewards/margins": 0.01634763740003109,
"rewards/rejected": -0.08856344223022461,
"step": 9
},
{
"debug/policy_chosen_logits": 0.2575492560863495,
"debug/policy_chosen_logps": -184.9705810546875,
"debug/policy_rejected_logits": 0.5005592703819275,
"debug/policy_rejected_logps": -172.09518432617188,
"debug/reference_chosen_logps": -172.8156280517578,
"debug/reference_rejected_logps": -160.72515869140625,
"epoch": 0.23809523809523808,
"grad_norm": 8.326836844842857,
"learning_rate": 5e-07,
"logits/chosen": 0.2575492560863495,
"logits/rejected": 0.5005592703819275,
"logps/chosen": -184.9705810546875,
"logps/rejected": -172.09518432617188,
"loss": 0.6811,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.12154942750930786,
"rewards/margins": -0.00784902460873127,
"rewards/rejected": -0.11370040476322174,
"step": 10
},
{
"debug/policy_chosen_logits": 0.02329457737505436,
"debug/policy_chosen_logps": -151.5535888671875,
"debug/policy_rejected_logits": 0.5039985179901123,
"debug/policy_rejected_logps": -164.93890380859375,
"debug/reference_chosen_logps": -145.47381591796875,
"debug/reference_rejected_logps": -155.04107666015625,
"epoch": 0.2619047619047619,
"grad_norm": 8.956479762651878,
"learning_rate": 5e-07,
"logits/chosen": 0.02329457737505436,
"logits/rejected": 0.5039985179901123,
"logps/chosen": -151.5535888671875,
"logps/rejected": -164.93890380859375,
"loss": 0.6868,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.06079769879579544,
"rewards/margins": 0.038180749863386154,
"rewards/rejected": -0.0989784449338913,
"step": 11
},
{
"debug/policy_chosen_logits": 0.47942259907722473,
"debug/policy_chosen_logps": -159.60877990722656,
"debug/policy_rejected_logits": 0.5704939365386963,
"debug/policy_rejected_logps": -154.61744689941406,
"debug/reference_chosen_logps": -147.24301147460938,
"debug/reference_rejected_logps": -141.2715301513672,
"epoch": 0.2857142857142857,
"grad_norm": 11.992421788281984,
"learning_rate": 5e-07,
"logits/chosen": 0.47942259907722473,
"logits/rejected": 0.5704939365386963,
"logps/chosen": -159.60877990722656,
"logps/rejected": -154.61744689941406,
"loss": 0.6853,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.12365761399269104,
"rewards/margins": 0.009801514446735382,
"rewards/rejected": -0.13345913589000702,
"step": 12
},
{
"debug/policy_chosen_logits": -0.20997528731822968,
"debug/policy_chosen_logps": -142.75146484375,
"debug/policy_rejected_logits": 0.3726802468299866,
"debug/policy_rejected_logps": -175.70962524414062,
"debug/reference_chosen_logps": -134.545166015625,
"debug/reference_rejected_logps": -164.0076141357422,
"epoch": 0.30952380952380953,
"grad_norm": 5.358671233182435,
"learning_rate": 5e-07,
"logits/chosen": -0.20997528731822968,
"logits/rejected": 0.3726802468299866,
"logps/chosen": -142.75146484375,
"logps/rejected": -175.70962524414062,
"loss": 0.689,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.08206304907798767,
"rewards/margins": 0.0349571518599987,
"rewards/rejected": -0.11702020466327667,
"step": 13
},
{
"debug/policy_chosen_logits": 0.20442210137844086,
"debug/policy_chosen_logps": -177.51991271972656,
"debug/policy_rejected_logits": 0.34923601150512695,
"debug/policy_rejected_logps": -161.62881469726562,
"debug/reference_chosen_logps": -164.69485473632812,
"debug/reference_rejected_logps": -150.70733642578125,
"epoch": 0.3333333333333333,
"grad_norm": 6.471200581198782,
"learning_rate": 5e-07,
"logits/chosen": 0.20442210137844086,
"logits/rejected": 0.34923601150512695,
"logps/chosen": -177.51991271972656,
"logps/rejected": -161.62881469726562,
"loss": 0.6899,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.12825068831443787,
"rewards/margins": -0.019035786390304565,
"rewards/rejected": -0.1092148944735527,
"step": 14
},
{
"debug/policy_chosen_logits": 0.20042140781879425,
"debug/policy_chosen_logps": -198.75933837890625,
"debug/policy_rejected_logits": 0.24180738627910614,
"debug/policy_rejected_logps": -177.14825439453125,
"debug/reference_chosen_logps": -178.46697998046875,
"debug/reference_rejected_logps": -158.2596435546875,
"epoch": 0.35714285714285715,
"grad_norm": 15.202276429910315,
"learning_rate": 5e-07,
"logits/chosen": 0.20042140781879425,
"logits/rejected": 0.24180738627910614,
"logps/chosen": -198.75933837890625,
"logps/rejected": -177.14825439453125,
"loss": 0.6823,
"rewards/accuracies": 0.25,
"rewards/chosen": -0.2029237002134323,
"rewards/margins": -0.014037556946277618,
"rewards/rejected": -0.1888861358165741,
"step": 15
},
{
"debug/policy_chosen_logits": 0.3575197160243988,
"debug/policy_chosen_logps": -159.78720092773438,
"debug/policy_rejected_logits": 0.6836833953857422,
"debug/policy_rejected_logps": -167.8487548828125,
"debug/reference_chosen_logps": -148.35433959960938,
"debug/reference_rejected_logps": -153.20465087890625,
"epoch": 0.38095238095238093,
"grad_norm": 5.011772584899409,
"learning_rate": 5e-07,
"logits/chosen": 0.3575197160243988,
"logits/rejected": 0.6836833953857422,
"logps/chosen": -159.78720092773438,
"logps/rejected": -167.8487548828125,
"loss": 0.6846,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.11432872712612152,
"rewards/margins": 0.0321124903857708,
"rewards/rejected": -0.14644122123718262,
"step": 16
},
{
"debug/policy_chosen_logits": 0.3418176472187042,
"debug/policy_chosen_logps": -153.46934509277344,
"debug/policy_rejected_logits": 0.33436495065689087,
"debug/policy_rejected_logps": -164.7410888671875,
"debug/reference_chosen_logps": -145.3973388671875,
"debug/reference_rejected_logps": -149.6763458251953,
"epoch": 0.40476190476190477,
"grad_norm": 20.39016346970483,
"learning_rate": 5e-07,
"logits/chosen": 0.3418176472187042,
"logits/rejected": 0.33436495065689087,
"logps/chosen": -153.46934509277344,
"logps/rejected": -164.7410888671875,
"loss": 0.6823,
"rewards/accuracies": 0.875,
"rewards/chosen": -0.08072000741958618,
"rewards/margins": 0.06992734968662262,
"rewards/rejected": -0.1506473571062088,
"step": 17
},
{
"debug/policy_chosen_logits": 0.21861077845096588,
"debug/policy_chosen_logps": -184.25680541992188,
"debug/policy_rejected_logits": 0.3139030635356903,
"debug/policy_rejected_logps": -186.92042541503906,
"debug/reference_chosen_logps": -167.13250732421875,
"debug/reference_rejected_logps": -167.22145080566406,
"epoch": 0.42857142857142855,
"grad_norm": 9.604003163834712,
"learning_rate": 5e-07,
"logits/chosen": 0.21861077845096588,
"logits/rejected": 0.3139030635356903,
"logps/chosen": -184.25680541992188,
"logps/rejected": -186.92042541503906,
"loss": 0.6871,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.17124298214912415,
"rewards/margins": 0.025746773928403854,
"rewards/rejected": -0.1969897449016571,
"step": 18
},
{
"debug/policy_chosen_logits": 0.4868623614311218,
"debug/policy_chosen_logps": -164.40745544433594,
"debug/policy_rejected_logits": 0.407953143119812,
"debug/policy_rejected_logps": -192.24801635742188,
"debug/reference_chosen_logps": -149.22422790527344,
"debug/reference_rejected_logps": -176.274658203125,
"epoch": 0.4523809523809524,
"grad_norm": 5.467573091749328,
"learning_rate": 5e-07,
"logits/chosen": 0.4868623614311218,
"logits/rejected": 0.407953143119812,
"logps/chosen": -164.40745544433594,
"logps/rejected": -192.24801635742188,
"loss": 0.6788,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.1518322378396988,
"rewards/margins": 0.007901255041360855,
"rewards/rejected": -0.15973350405693054,
"step": 19
},
{
"debug/policy_chosen_logits": 0.059103213250637054,
"debug/policy_chosen_logps": -174.8370361328125,
"debug/policy_rejected_logits": 0.5786897540092468,
"debug/policy_rejected_logps": -199.56698608398438,
"debug/reference_chosen_logps": -161.7840118408203,
"debug/reference_rejected_logps": -185.17050170898438,
"epoch": 0.47619047619047616,
"grad_norm": 5.234500875119642,
"learning_rate": 5e-07,
"logits/chosen": 0.059103213250637054,
"logits/rejected": 0.5786897540092468,
"logps/chosen": -174.8370361328125,
"logps/rejected": -199.56698608398438,
"loss": 0.6848,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.13053017854690552,
"rewards/margins": 0.013434587977826595,
"rewards/rejected": -0.1439647674560547,
"step": 20
},
{
"debug/policy_chosen_logits": 0.5743213891983032,
"debug/policy_chosen_logps": -178.57594299316406,
"debug/policy_rejected_logits": 0.20286375284194946,
"debug/policy_rejected_logps": -179.84762573242188,
"debug/reference_chosen_logps": -158.3350067138672,
"debug/reference_rejected_logps": -160.45053100585938,
"epoch": 0.5,
"grad_norm": 6.102878645835385,
"learning_rate": 5e-07,
"logits/chosen": 0.5743213891983032,
"logits/rejected": 0.20286375284194946,
"logps/chosen": -178.57594299316406,
"logps/rejected": -179.84762573242188,
"loss": 0.6845,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.20240947604179382,
"rewards/margins": -0.008438415825366974,
"rewards/rejected": -0.19397103786468506,
"step": 21
},
{
"debug/policy_chosen_logits": 0.3122093677520752,
"debug/policy_chosen_logps": -164.51431274414062,
"debug/policy_rejected_logits": 0.3530707061290741,
"debug/policy_rejected_logps": -176.93601989746094,
"debug/reference_chosen_logps": -147.92042541503906,
"debug/reference_rejected_logps": -156.74851989746094,
"epoch": 0.5238095238095238,
"grad_norm": 6.315550057559611,
"learning_rate": 5e-07,
"logits/chosen": 0.3122093677520752,
"logits/rejected": 0.3530707061290741,
"logps/chosen": -164.51431274414062,
"logps/rejected": -176.93601989746094,
"loss": 0.6815,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.16593879461288452,
"rewards/margins": 0.03593616560101509,
"rewards/rejected": -0.20187495648860931,
"step": 22
},
{
"debug/policy_chosen_logits": -0.044268831610679626,
"debug/policy_chosen_logps": -157.81039428710938,
"debug/policy_rejected_logits": 0.3559872508049011,
"debug/policy_rejected_logps": -147.35525512695312,
"debug/reference_chosen_logps": -150.7872314453125,
"debug/reference_rejected_logps": -138.82229614257812,
"epoch": 0.5476190476190477,
"grad_norm": 23.620495593781406,
"learning_rate": 5e-07,
"logits/chosen": -0.044268831610679626,
"logits/rejected": 0.3559872508049011,
"logps/chosen": -157.81039428710938,
"logps/rejected": -147.35525512695312,
"loss": 0.6842,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.07023164629936218,
"rewards/margins": 0.015097856521606445,
"rewards/rejected": -0.08532950282096863,
"step": 23
},
{
"debug/policy_chosen_logits": 0.018966468051075935,
"debug/policy_chosen_logps": -159.8534698486328,
"debug/policy_rejected_logits": 0.1428254395723343,
"debug/policy_rejected_logps": -164.7286834716797,
"debug/reference_chosen_logps": -147.0450439453125,
"debug/reference_rejected_logps": -149.85382080078125,
"epoch": 0.5714285714285714,
"grad_norm": 4.677349017880479,
"learning_rate": 5e-07,
"logits/chosen": 0.018966468051075935,
"logits/rejected": 0.1428254395723343,
"logps/chosen": -159.8534698486328,
"logps/rejected": -164.7286834716797,
"loss": 0.6819,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1280841827392578,
"rewards/margins": 0.02066453918814659,
"rewards/rejected": -0.1487487256526947,
"step": 24
},
{
"debug/policy_chosen_logits": 0.2853807806968689,
"debug/policy_chosen_logps": -176.42611694335938,
"debug/policy_rejected_logits": 0.13074414432048798,
"debug/policy_rejected_logps": -155.15415954589844,
"debug/reference_chosen_logps": -158.4669647216797,
"debug/reference_rejected_logps": -136.29364013671875,
"epoch": 0.5952380952380952,
"grad_norm": 11.079802876278416,
"learning_rate": 5e-07,
"logits/chosen": 0.2853807806968689,
"logits/rejected": 0.13074414432048798,
"logps/chosen": -176.42611694335938,
"logps/rejected": -155.15415954589844,
"loss": 0.6821,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.179591566324234,
"rewards/margins": 0.00901371892541647,
"rewards/rejected": -0.18860529363155365,
"step": 25
},
{
"debug/policy_chosen_logits": 0.40262606739997864,
"debug/policy_chosen_logps": -153.15701293945312,
"debug/policy_rejected_logits": 0.7936873435974121,
"debug/policy_rejected_logps": -169.62689208984375,
"debug/reference_chosen_logps": -140.70889282226562,
"debug/reference_rejected_logps": -152.8561248779297,
"epoch": 0.6190476190476191,
"grad_norm": 5.32694876840012,
"learning_rate": 5e-07,
"logits/chosen": 0.40262606739997864,
"logits/rejected": 0.7936873435974121,
"logps/chosen": -153.15701293945312,
"logps/rejected": -169.62689208984375,
"loss": 0.6733,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.12448111921548843,
"rewards/margins": 0.04322664067149162,
"rewards/rejected": -0.16770777106285095,
"step": 26
},
{
"debug/policy_chosen_logits": 0.4326091408729553,
"debug/policy_chosen_logps": -157.39126586914062,
"debug/policy_rejected_logits": 0.607225239276886,
"debug/policy_rejected_logps": -188.34918212890625,
"debug/reference_chosen_logps": -145.33380126953125,
"debug/reference_rejected_logps": -162.48890686035156,
"epoch": 0.6428571428571429,
"grad_norm": 4.700163204340666,
"learning_rate": 5e-07,
"logits/chosen": 0.4326091408729553,
"logits/rejected": 0.607225239276886,
"logps/chosen": -157.39126586914062,
"logps/rejected": -188.34918212890625,
"loss": 0.6642,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.12057456374168396,
"rewards/margins": 0.13802826404571533,
"rewards/rejected": -0.2586028277873993,
"step": 27
},
{
"debug/policy_chosen_logits": 0.031958021223545074,
"debug/policy_chosen_logps": -159.67169189453125,
"debug/policy_rejected_logits": 0.1899116486310959,
"debug/policy_rejected_logps": -187.28189086914062,
"debug/reference_chosen_logps": -141.6376953125,
"debug/reference_rejected_logps": -162.19659423828125,
"epoch": 0.6666666666666666,
"grad_norm": 6.569512152499291,
"learning_rate": 5e-07,
"logits/chosen": 0.031958021223545074,
"logits/rejected": 0.1899116486310959,
"logps/chosen": -159.67169189453125,
"logps/rejected": -187.28189086914062,
"loss": 0.6824,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.180339977145195,
"rewards/margins": 0.07051312178373337,
"rewards/rejected": -0.2508530914783478,
"step": 28
},
{
"debug/policy_chosen_logits": 0.10809233784675598,
"debug/policy_chosen_logps": -176.20724487304688,
"debug/policy_rejected_logits": 0.41801854968070984,
"debug/policy_rejected_logps": -197.15541076660156,
"debug/reference_chosen_logps": -158.1036834716797,
"debug/reference_rejected_logps": -176.26634216308594,
"epoch": 0.6904761904761905,
"grad_norm": 6.541965417685165,
"learning_rate": 5e-07,
"logits/chosen": 0.10809233784675598,
"logits/rejected": 0.41801854968070984,
"logps/chosen": -176.20724487304688,
"logps/rejected": -197.15541076660156,
"loss": 0.6681,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.181035578250885,
"rewards/margins": 0.02785516157746315,
"rewards/rejected": -0.20889073610305786,
"step": 29
},
{
"debug/policy_chosen_logits": -0.15116974711418152,
"debug/policy_chosen_logps": -164.75247192382812,
"debug/policy_rejected_logits": 0.12009341269731522,
"debug/policy_rejected_logps": -175.12867736816406,
"debug/reference_chosen_logps": -146.2644805908203,
"debug/reference_rejected_logps": -149.75460815429688,
"epoch": 0.7142857142857143,
"grad_norm": 4.855645858354973,
"learning_rate": 5e-07,
"logits/chosen": -0.15116974711418152,
"logits/rejected": 0.12009341269731522,
"logps/chosen": -164.75247192382812,
"logps/rejected": -175.12867736816406,
"loss": 0.6786,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.18488001823425293,
"rewards/margins": 0.06886060535907745,
"rewards/rejected": -0.2537406086921692,
"step": 30
},
{
"debug/policy_chosen_logits": 0.17149780690670013,
"debug/policy_chosen_logps": -188.273193359375,
"debug/policy_rejected_logits": 0.3072517216205597,
"debug/policy_rejected_logps": -187.52651977539062,
"debug/reference_chosen_logps": -165.44805908203125,
"debug/reference_rejected_logps": -158.7125244140625,
"epoch": 0.7380952380952381,
"grad_norm": 6.097067478609718,
"learning_rate": 5e-07,
"logits/chosen": 0.17149780690670013,
"logits/rejected": 0.3072517216205597,
"logps/chosen": -188.273193359375,
"logps/rejected": -187.52651977539062,
"loss": 0.6829,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.22825142741203308,
"rewards/margins": 0.059888482093811035,
"rewards/rejected": -0.2881399095058441,
"step": 31
},
{
"debug/policy_chosen_logits": 0.46266281604766846,
"debug/policy_chosen_logps": -169.596435546875,
"debug/policy_rejected_logits": 0.4477306604385376,
"debug/policy_rejected_logps": -173.21981811523438,
"debug/reference_chosen_logps": -152.52273559570312,
"debug/reference_rejected_logps": -151.03050231933594,
"epoch": 0.7619047619047619,
"grad_norm": 10.520633954833,
"learning_rate": 5e-07,
"logits/chosen": 0.46266281604766846,
"logits/rejected": 0.4477306604385376,
"logps/chosen": -169.596435546875,
"logps/rejected": -173.21981811523438,
"loss": 0.6878,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.17073702812194824,
"rewards/margins": 0.05115606635808945,
"rewards/rejected": -0.2218931019306183,
"step": 32
},
{
"debug/policy_chosen_logits": 0.7084560394287109,
"debug/policy_chosen_logps": -161.38735961914062,
"debug/policy_rejected_logits": 0.6357196569442749,
"debug/policy_rejected_logps": -164.42718505859375,
"debug/reference_chosen_logps": -154.86412048339844,
"debug/reference_rejected_logps": -157.79238891601562,
"epoch": 0.7857142857142857,
"grad_norm": 15.939572697711974,
"learning_rate": 5e-07,
"logits/chosen": 0.7084560394287109,
"logits/rejected": 0.6357196569442749,
"logps/chosen": -161.38735961914062,
"logps/rejected": -164.42718505859375,
"loss": 0.6803,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.06523235142230988,
"rewards/margins": 0.001115655992180109,
"rewards/rejected": -0.06634800881147385,
"step": 33
},
{
"debug/policy_chosen_logits": 0.12151144444942474,
"debug/policy_chosen_logps": -161.50515747070312,
"debug/policy_rejected_logits": 1.0420786142349243,
"debug/policy_rejected_logps": -164.21615600585938,
"debug/reference_chosen_logps": -150.59689331054688,
"debug/reference_rejected_logps": -152.6244354248047,
"epoch": 0.8095238095238095,
"grad_norm": 9.868780868712001,
"learning_rate": 5e-07,
"logits/chosen": 0.12151144444942474,
"logits/rejected": 1.0420786142349243,
"logps/chosen": -161.50515747070312,
"logps/rejected": -164.21615600585938,
"loss": 0.6786,
"rewards/accuracies": 0.375,
"rewards/chosen": -0.10908253490924835,
"rewards/margins": 0.006834707222878933,
"rewards/rejected": -0.11591724306344986,
"step": 34
},
{
"debug/policy_chosen_logits": 0.027947237715125084,
"debug/policy_chosen_logps": -157.77456665039062,
"debug/policy_rejected_logits": 0.31475889682769775,
"debug/policy_rejected_logps": -159.06002807617188,
"debug/reference_chosen_logps": -136.69996643066406,
"debug/reference_rejected_logps": -138.58349609375,
"epoch": 0.8333333333333334,
"grad_norm": 8.177811902764494,
"learning_rate": 5e-07,
"logits/chosen": 0.027947237715125084,
"logits/rejected": 0.31475889682769775,
"logps/chosen": -157.77456665039062,
"logps/rejected": -159.06002807617188,
"loss": 0.6874,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.21074604988098145,
"rewards/margins": -0.005980661138892174,
"rewards/rejected": -0.20476537942886353,
"step": 35
},
{
"debug/policy_chosen_logits": 0.2593013346195221,
"debug/policy_chosen_logps": -165.31509399414062,
"debug/policy_rejected_logits": -0.04938528686761856,
"debug/policy_rejected_logps": -168.42660522460938,
"debug/reference_chosen_logps": -145.93374633789062,
"debug/reference_rejected_logps": -145.29168701171875,
"epoch": 0.8571428571428571,
"grad_norm": 12.91232630873052,
"learning_rate": 5e-07,
"logits/chosen": 0.2593013346195221,
"logits/rejected": -0.04938528686761856,
"logps/chosen": -165.31509399414062,
"logps/rejected": -168.42660522460938,
"loss": 0.6712,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.19381342828273773,
"rewards/margins": 0.03753571957349777,
"rewards/rejected": -0.2313491404056549,
"step": 36
},
{
"debug/policy_chosen_logits": 1.0012390613555908,
"debug/policy_chosen_logps": -165.09347534179688,
"debug/policy_rejected_logits": 1.0178093910217285,
"debug/policy_rejected_logps": -171.16152954101562,
"debug/reference_chosen_logps": -150.7286834716797,
"debug/reference_rejected_logps": -150.47354125976562,
"epoch": 0.8809523809523809,
"grad_norm": 4.747576829430422,
"learning_rate": 5e-07,
"logits/chosen": 1.0012390613555908,
"logits/rejected": 1.0178093910217285,
"logps/chosen": -165.09347534179688,
"logps/rejected": -171.16152954101562,
"loss": 0.6849,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.1436479538679123,
"rewards/margins": 0.06323190778493881,
"rewards/rejected": -0.2068798542022705,
"step": 37
},
{
"debug/policy_chosen_logits": -0.11002390831708908,
"debug/policy_chosen_logps": -173.82029724121094,
"debug/policy_rejected_logits": 0.446510910987854,
"debug/policy_rejected_logps": -212.50643920898438,
"debug/reference_chosen_logps": -150.2817840576172,
"debug/reference_rejected_logps": -185.846923828125,
"epoch": 0.9047619047619048,
"grad_norm": 5.476549746954031,
"learning_rate": 5e-07,
"logits/chosen": -0.11002390831708908,
"logits/rejected": 0.446510910987854,
"logps/chosen": -173.82029724121094,
"logps/rejected": -212.50643920898438,
"loss": 0.6724,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.23538516461849213,
"rewards/margins": 0.031209895387291908,
"rewards/rejected": -0.2665950655937195,
"step": 38
},
{
"debug/policy_chosen_logits": -0.20052273571491241,
"debug/policy_chosen_logps": -163.867431640625,
"debug/policy_rejected_logits": 0.5886087417602539,
"debug/policy_rejected_logps": -185.58941650390625,
"debug/reference_chosen_logps": -141.18801879882812,
"debug/reference_rejected_logps": -159.76058959960938,
"epoch": 0.9285714285714286,
"grad_norm": 12.5233373844755,
"learning_rate": 5e-07,
"logits/chosen": -0.20052273571491241,
"logits/rejected": 0.5886087417602539,
"logps/chosen": -163.867431640625,
"logps/rejected": -185.58941650390625,
"loss": 0.6685,
"rewards/accuracies": 0.5,
"rewards/chosen": -0.22679391503334045,
"rewards/margins": 0.0314943790435791,
"rewards/rejected": -0.25828829407691956,
"step": 39
},
{
"debug/policy_chosen_logits": 0.25073912739753723,
"debug/policy_chosen_logps": -157.51876831054688,
"debug/policy_rejected_logits": 0.30381596088409424,
"debug/policy_rejected_logps": -176.589599609375,
"debug/reference_chosen_logps": -138.98110961914062,
"debug/reference_rejected_logps": -154.03880310058594,
"epoch": 0.9523809523809523,
"grad_norm": 5.529655141820795,
"learning_rate": 5e-07,
"logits/chosen": 0.25073912739753723,
"logits/rejected": 0.30381596088409424,
"logps/chosen": -157.51876831054688,
"logps/rejected": -176.589599609375,
"loss": 0.6694,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.1853766143321991,
"rewards/margins": 0.040131378918886185,
"rewards/rejected": -0.22550798952579498,
"step": 40
},
{
"debug/policy_chosen_logits": 0.23757268488407135,
"debug/policy_chosen_logps": -167.54112243652344,
"debug/policy_rejected_logits": 0.40399065613746643,
"debug/policy_rejected_logps": -191.4041748046875,
"debug/reference_chosen_logps": -145.42752075195312,
"debug/reference_rejected_logps": -164.51507568359375,
"epoch": 0.9761904761904762,
"grad_norm": 5.959237498635192,
"learning_rate": 5e-07,
"logits/chosen": 0.23757268488407135,
"logits/rejected": 0.40399065613746643,
"logps/chosen": -167.54112243652344,
"logps/rejected": -191.4041748046875,
"loss": 0.6706,
"rewards/accuracies": 0.75,
"rewards/chosen": -0.22113589942455292,
"rewards/margins": 0.047754913568496704,
"rewards/rejected": -0.2688907980918884,
"step": 41
},
{
"debug/policy_chosen_logits": 0.029359659180045128,
"debug/policy_chosen_logps": -178.32794189453125,
"debug/policy_rejected_logits": 0.20693586766719818,
"debug/policy_rejected_logps": -185.86007690429688,
"debug/reference_chosen_logps": -151.25294494628906,
"debug/reference_rejected_logps": -157.61886596679688,
"epoch": 1.0,
"grad_norm": 15.740573011032756,
"learning_rate": 5e-07,
"logits/chosen": 0.029359659180045128,
"logits/rejected": 0.20693586766719818,
"logps/chosen": -178.32794189453125,
"logps/rejected": -185.86007690429688,
"loss": 0.6925,
"rewards/accuracies": 0.625,
"rewards/chosen": -0.27075010538101196,
"rewards/margins": 0.01166202500462532,
"rewards/rejected": -0.2824121117591858,
"step": 42
},
{
"epoch": 1.0,
"step": 42,
"total_flos": 0.0,
"train_loss": 0.6828021520660037,
"train_runtime": 390.3052,
"train_samples_per_second": 6.784,
"train_steps_per_second": 0.108
}
],
"logging_steps": 1,
"max_steps": 42,
"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
}