{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 100, "global_step": 37, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "debug/policy_chosen_logits": -3.41206693649292, "debug/policy_chosen_logps": -143.21307373046875, "debug/policy_rejected_logits": -3.2551214694976807, "debug/policy_rejected_logps": -150.2328338623047, "debug/reference_chosen_logps": -143.21307373046875, "debug/reference_rejected_logps": -150.2328338623047, "epoch": 0.02702702702702703, "grad_norm": 3.5140838495338116, "learning_rate": 1e-06, "logits/chosen": -3.41206693649292, "logits/rejected": -3.2551214694976807, "logps/chosen": -143.21307373046875, "logps/rejected": -150.2328338623047, "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": -3.4049649238586426, "debug/policy_chosen_logps": -143.7670135498047, "debug/policy_rejected_logits": -3.4008069038391113, "debug/policy_rejected_logps": -138.0754852294922, "debug/reference_chosen_logps": -143.82278442382812, "debug/reference_rejected_logps": -138.1428985595703, "epoch": 0.05405405405405406, "grad_norm": 3.526231895119079, "learning_rate": 1e-06, "logits/chosen": -3.4049649238586426, "logits/rejected": -3.4008069038391113, "logps/chosen": -143.7670135498047, "logps/rejected": -138.0754852294922, "loss": 0.4994, "rewards/accuracies": 0.375, "rewards/chosen": 0.0005577945848926902, "rewards/margins": -0.00011638639261946082, "rewards/rejected": 0.000674180977512151, "step": 2 }, { "debug/policy_chosen_logits": -3.2883780002593994, "debug/policy_chosen_logps": -143.35385131835938, "debug/policy_rejected_logits": -3.2103564739227295, "debug/policy_rejected_logps": -139.85372924804688, "debug/reference_chosen_logps": -143.38311767578125, "debug/reference_rejected_logps": -139.98326110839844, "epoch": 0.08108108108108109, "grad_norm": 3.618148106626628, "learning_rate": 1e-06, "logits/chosen": -3.2883780002593994, "logits/rejected": -3.2103564739227295, "logps/chosen": -143.35385131835938, "logps/rejected": -139.85372924804688, "loss": 0.5006, "rewards/accuracies": 0.25, "rewards/chosen": 0.00029258732683956623, "rewards/margins": -0.001002683537080884, "rewards/rejected": 0.0012952708639204502, "step": 3 }, { "debug/policy_chosen_logits": -3.294776678085327, "debug/policy_chosen_logps": -132.79519653320312, "debug/policy_rejected_logits": -3.4150912761688232, "debug/policy_rejected_logps": -138.08180236816406, "debug/reference_chosen_logps": -132.62167358398438, "debug/reference_rejected_logps": -137.927978515625, "epoch": 0.10810810810810811, "grad_norm": 3.5093502617713077, "learning_rate": 1e-06, "logits/chosen": -3.294776678085327, "logits/rejected": -3.4150912761688232, "logps/chosen": -132.79519653320312, "logps/rejected": -138.08180236816406, "loss": 0.4994, "rewards/accuracies": 0.5, "rewards/chosen": -0.0017352199647575617, "rewards/margins": -0.0001968955621123314, "rewards/rejected": -0.0015383241698145866, "step": 4 }, { "debug/policy_chosen_logits": -3.343012809753418, "debug/policy_chosen_logps": -126.95188903808594, "debug/policy_rejected_logits": -3.3124771118164062, "debug/policy_rejected_logps": -129.9374237060547, "debug/reference_chosen_logps": -126.82910919189453, "debug/reference_rejected_logps": -129.91493225097656, "epoch": 0.13513513513513514, "grad_norm": 3.250245901043351, "learning_rate": 1e-06, "logits/chosen": -3.343012809753418, "logits/rejected": -3.3124771118164062, "logps/chosen": -126.95188903808594, "logps/rejected": -129.9374237060547, "loss": 0.5009, "rewards/accuracies": 0.625, "rewards/chosen": -0.001227750675752759, "rewards/margins": -0.001002778997644782, "rewards/rejected": -0.00022497176541946828, "step": 5 }, { "debug/policy_chosen_logits": -3.2374277114868164, "debug/policy_chosen_logps": -154.26406860351562, "debug/policy_rejected_logits": -3.2978382110595703, "debug/policy_rejected_logps": -143.40101623535156, "debug/reference_chosen_logps": -153.9961395263672, "debug/reference_rejected_logps": -143.3165283203125, "epoch": 0.16216216216216217, "grad_norm": 3.439711062354016, "learning_rate": 1e-06, "logits/chosen": -3.2374277114868164, "logits/rejected": -3.2978382110595703, "logps/chosen": -154.26406860351562, "logps/rejected": -143.40101623535156, "loss": 0.5009, "rewards/accuracies": 0.375, "rewards/chosen": -0.002679176162928343, "rewards/margins": -0.0018340873066335917, "rewards/rejected": -0.000845088972710073, "step": 6 }, { "debug/policy_chosen_logits": -3.2252252101898193, "debug/policy_chosen_logps": -162.3474578857422, "debug/policy_rejected_logits": -3.230537176132202, "debug/policy_rejected_logps": -147.3297576904297, "debug/reference_chosen_logps": -162.26889038085938, "debug/reference_rejected_logps": -147.57199096679688, "epoch": 0.1891891891891892, "grad_norm": 3.5128248367150468, "learning_rate": 1e-06, "logits/chosen": -3.2252252101898193, "logits/rejected": -3.230537176132202, "logps/chosen": -162.3474578857422, "logps/rejected": -147.3297576904297, "loss": 0.5005, "rewards/accuracies": 0.25, "rewards/chosen": -0.0007855510921217501, "rewards/margins": -0.003207969479262829, "rewards/rejected": 0.0024224184453487396, "step": 7 }, { "debug/policy_chosen_logits": -3.3006181716918945, "debug/policy_chosen_logps": -149.18630981445312, "debug/policy_rejected_logits": -3.3146488666534424, "debug/policy_rejected_logps": -147.72348022460938, "debug/reference_chosen_logps": -149.20960998535156, "debug/reference_rejected_logps": -147.54893493652344, "epoch": 0.21621621621621623, "grad_norm": 3.499360817762659, "learning_rate": 1e-06, "logits/chosen": -3.3006181716918945, "logits/rejected": -3.3146488666534424, "logps/chosen": -149.18630981445312, "logps/rejected": -147.72348022460938, "loss": 0.4992, "rewards/accuracies": 0.75, "rewards/chosen": 0.000233030179515481, "rewards/margins": 0.001978444866836071, "rewards/rejected": -0.00174541468732059, "step": 8 }, { "debug/policy_chosen_logits": -3.3503048419952393, "debug/policy_chosen_logps": -150.99197387695312, "debug/policy_rejected_logits": -3.2822837829589844, "debug/policy_rejected_logps": -145.26345825195312, "debug/reference_chosen_logps": -151.06939697265625, "debug/reference_rejected_logps": -145.52816772460938, "epoch": 0.24324324324324326, "grad_norm": 3.4264003876223215, "learning_rate": 1e-06, "logits/chosen": -3.3503048419952393, "logits/rejected": -3.2822837829589844, "logps/chosen": -150.99197387695312, "logps/rejected": -145.26345825195312, "loss": 0.4987, "rewards/accuracies": 0.375, "rewards/chosen": 0.0007742022862657905, "rewards/margins": -0.0018729494186118245, "rewards/rejected": 0.002647151704877615, "step": 9 }, { "debug/policy_chosen_logits": -3.304420232772827, "debug/policy_chosen_logps": -133.21044921875, "debug/policy_rejected_logits": -3.412870168685913, "debug/policy_rejected_logps": -125.90888977050781, "debug/reference_chosen_logps": -133.26504516601562, "debug/reference_rejected_logps": -125.95523834228516, "epoch": 0.2702702702702703, "grad_norm": 3.3820605783151074, "learning_rate": 1e-06, "logits/chosen": -3.304420232772827, "logits/rejected": -3.412870168685913, "logps/chosen": -133.21044921875, "logps/rejected": -125.90888977050781, "loss": 0.5005, "rewards/accuracies": 0.5, "rewards/chosen": 0.0005457591032609344, "rewards/margins": 8.22543297545053e-05, "rewards/rejected": 0.0004635047516785562, "step": 10 }, { "debug/policy_chosen_logits": -3.3585329055786133, "debug/policy_chosen_logps": -139.380859375, "debug/policy_rejected_logits": -3.3599538803100586, "debug/policy_rejected_logps": -132.3596649169922, "debug/reference_chosen_logps": -139.71542358398438, "debug/reference_rejected_logps": -132.61981201171875, "epoch": 0.2972972972972973, "grad_norm": 3.288218575224273, "learning_rate": 1e-06, "logits/chosen": -3.3585329055786133, "logits/rejected": -3.3599538803100586, "logps/chosen": -139.380859375, "logps/rejected": -132.3596649169922, "loss": 0.4999, "rewards/accuracies": 0.5, "rewards/chosen": 0.0033456420060247183, "rewards/margins": 0.0007441901252605021, "rewards/rejected": 0.0026014517061412334, "step": 11 }, { "debug/policy_chosen_logits": -3.2882096767425537, "debug/policy_chosen_logps": -156.53334045410156, "debug/policy_rejected_logits": -3.2960500717163086, "debug/policy_rejected_logps": -143.13995361328125, "debug/reference_chosen_logps": -156.6325225830078, "debug/reference_rejected_logps": -142.83905029296875, "epoch": 0.32432432432432434, "grad_norm": 3.813316903837083, "learning_rate": 1e-06, "logits/chosen": -3.2882096767425537, "logits/rejected": -3.2960500717163086, "logps/chosen": -156.53334045410156, "logps/rejected": -143.13995361328125, "loss": 0.4989, "rewards/accuracies": 0.625, "rewards/chosen": 0.0009917640127241611, "rewards/margins": 0.004000749439001083, "rewards/rejected": -0.0030089854262769222, "step": 12 }, { "debug/policy_chosen_logits": -3.2771499156951904, "debug/policy_chosen_logps": -133.85403442382812, "debug/policy_rejected_logits": -3.3577096462249756, "debug/policy_rejected_logps": -145.81051635742188, "debug/reference_chosen_logps": -134.21636962890625, "debug/reference_rejected_logps": -145.71588134765625, "epoch": 0.35135135135135137, "grad_norm": 3.5238547026932254, "learning_rate": 1e-06, "logits/chosen": -3.2771499156951904, "logits/rejected": -3.3577096462249756, "logps/chosen": -133.85403442382812, "logps/rejected": -145.81051635742188, "loss": 0.5004, "rewards/accuracies": 0.75, "rewards/chosen": 0.003623266238719225, "rewards/margins": 0.004569797310978174, "rewards/rejected": -0.0009465311886742711, "step": 13 }, { "debug/policy_chosen_logits": -3.4804084300994873, "debug/policy_chosen_logps": -165.59515380859375, "debug/policy_rejected_logits": -3.5207834243774414, "debug/policy_rejected_logps": -139.4119873046875, "debug/reference_chosen_logps": -165.61618041992188, "debug/reference_rejected_logps": -139.546142578125, "epoch": 0.3783783783783784, "grad_norm": 3.271576173715924, "learning_rate": 1e-06, "logits/chosen": -3.4804084300994873, "logits/rejected": -3.5207834243774414, "logps/chosen": -165.59515380859375, "logps/rejected": -139.4119873046875, "loss": 0.4997, "rewards/accuracies": 0.625, "rewards/chosen": 0.00021032325457781553, "rewards/margins": -0.0011312393471598625, "rewards/rejected": 0.0013415624853223562, "step": 14 }, { "debug/policy_chosen_logits": -3.371809244155884, "debug/policy_chosen_logps": -138.2847900390625, "debug/policy_rejected_logits": -3.2725796699523926, "debug/policy_rejected_logps": -146.83644104003906, "debug/reference_chosen_logps": -138.04966735839844, "debug/reference_rejected_logps": -146.4534454345703, "epoch": 0.40540540540540543, "grad_norm": 3.5549880331732084, "learning_rate": 1e-06, "logits/chosen": -3.371809244155884, "logits/rejected": -3.2725796699523926, "logps/chosen": -138.2847900390625, "logps/rejected": -146.83644104003906, "loss": 0.4978, "rewards/accuracies": 0.5, "rewards/chosen": -0.002351150382310152, "rewards/margins": 0.0014788818079978228, "rewards/rejected": -0.0038300324231386185, "step": 15 }, { "debug/policy_chosen_logits": -3.2811360359191895, "debug/policy_chosen_logps": -145.2406005859375, "debug/policy_rejected_logits": -3.2666900157928467, "debug/policy_rejected_logps": -136.08499145507812, "debug/reference_chosen_logps": -145.38519287109375, "debug/reference_rejected_logps": -136.1790313720703, "epoch": 0.43243243243243246, "grad_norm": 3.390241199656903, "learning_rate": 1e-06, "logits/chosen": -3.2811360359191895, "logits/rejected": -3.2666900157928467, "logps/chosen": -145.2406005859375, "logps/rejected": -136.08499145507812, "loss": 0.4992, "rewards/accuracies": 0.375, "rewards/chosen": 0.0014458559453487396, "rewards/margins": 0.0005054186331108212, "rewards/rejected": 0.0009404371376149356, "step": 16 }, { "debug/policy_chosen_logits": -3.4209792613983154, "debug/policy_chosen_logps": -154.63943481445312, "debug/policy_rejected_logits": -3.3486948013305664, "debug/policy_rejected_logps": -133.35781860351562, "debug/reference_chosen_logps": -154.70895385742188, "debug/reference_rejected_logps": -133.4116668701172, "epoch": 0.4594594594594595, "grad_norm": 3.6742049248327437, "learning_rate": 1e-06, "logits/chosen": -3.4209792613983154, "logits/rejected": -3.3486948013305664, "logps/chosen": -154.63943481445312, "logps/rejected": -133.35781860351562, "loss": 0.4989, "rewards/accuracies": 0.5, "rewards/chosen": 0.0006950664683245122, "rewards/margins": 0.00015656478353776038, "rewards/rejected": 0.0005385016556829214, "step": 17 }, { "debug/policy_chosen_logits": -3.298342704772949, "debug/policy_chosen_logps": -146.52044677734375, "debug/policy_rejected_logits": -3.3201894760131836, "debug/policy_rejected_logps": -156.97808837890625, "debug/reference_chosen_logps": -146.34840393066406, "debug/reference_rejected_logps": -156.84378051757812, "epoch": 0.4864864864864865, "grad_norm": 3.5194962148324103, "learning_rate": 1e-06, "logits/chosen": -3.298342704772949, "logits/rejected": -3.3201894760131836, "logps/chosen": -146.52044677734375, "logps/rejected": -156.97808837890625, "loss": 0.4973, "rewards/accuracies": 0.5, "rewards/chosen": -0.0017202282324433327, "rewards/margins": -0.0003771304036490619, "rewards/rejected": -0.001343097654171288, "step": 18 }, { "debug/policy_chosen_logits": -3.3400371074676514, "debug/policy_chosen_logps": -132.01437377929688, "debug/policy_rejected_logits": -3.36023211479187, "debug/policy_rejected_logps": -148.31887817382812, "debug/reference_chosen_logps": -131.88931274414062, "debug/reference_rejected_logps": -148.13345336914062, "epoch": 0.5135135135135135, "grad_norm": 3.61313418435521, "learning_rate": 1e-06, "logits/chosen": -3.3400371074676514, "logits/rejected": -3.36023211479187, "logps/chosen": -132.01437377929688, "logps/rejected": -148.31887817382812, "loss": 0.498, "rewards/accuracies": 0.5, "rewards/chosen": -0.00125067715998739, "rewards/margins": 0.0006034375401213765, "rewards/rejected": -0.001854114467278123, "step": 19 }, { "debug/policy_chosen_logits": -3.2641985416412354, "debug/policy_chosen_logps": -134.37350463867188, "debug/policy_rejected_logits": -3.255669355392456, "debug/policy_rejected_logps": -124.11454772949219, "debug/reference_chosen_logps": -134.53549194335938, "debug/reference_rejected_logps": -123.25413513183594, "epoch": 0.5405405405405406, "grad_norm": 3.4240940627433663, "learning_rate": 1e-06, "logits/chosen": -3.2641985416412354, "logits/rejected": -3.255669355392456, "logps/chosen": -134.37350463867188, "logps/rejected": -124.11454772949219, "loss": 0.4969, "rewards/accuracies": 0.5, "rewards/chosen": 0.0016198731027543545, "rewards/margins": 0.010224045254290104, "rewards/rejected": -0.008604173548519611, "step": 20 }, { "debug/policy_chosen_logits": -3.389690637588501, "debug/policy_chosen_logps": -144.95230102539062, "debug/policy_rejected_logits": -3.3237528800964355, "debug/policy_rejected_logps": -140.4608917236328, "debug/reference_chosen_logps": -145.19155883789062, "debug/reference_rejected_logps": -139.9624481201172, "epoch": 0.5675675675675675, "grad_norm": 3.5636895297483804, "learning_rate": 1e-06, "logits/chosen": -3.389690637588501, "logits/rejected": -3.3237528800964355, "logps/chosen": -144.95230102539062, "logps/rejected": -140.4608917236328, "loss": 0.4972, "rewards/accuracies": 0.375, "rewards/chosen": 0.0023925015702843666, "rewards/margins": 0.007377022877335548, "rewards/rejected": -0.004984522238373756, "step": 21 }, { "debug/policy_chosen_logits": -3.2673821449279785, "debug/policy_chosen_logps": -159.7883758544922, "debug/policy_rejected_logits": -3.2593047618865967, "debug/policy_rejected_logps": -141.72947692871094, "debug/reference_chosen_logps": -159.80300903320312, "debug/reference_rejected_logps": -142.28695678710938, "epoch": 0.5945945945945946, "grad_norm": 3.8014985351330988, "learning_rate": 1e-06, "logits/chosen": -3.2673821449279785, "logits/rejected": -3.2593047618865967, "logps/chosen": -159.7883758544922, "logps/rejected": -141.72947692871094, "loss": 0.499, "rewards/accuracies": 0.125, "rewards/chosen": 0.00014627468772232533, "rewards/margins": -0.005428638309240341, "rewards/rejected": 0.00557491322979331, "step": 22 }, { "debug/policy_chosen_logits": -3.3076963424682617, "debug/policy_chosen_logps": -147.4810791015625, "debug/policy_rejected_logits": -3.3063511848449707, "debug/policy_rejected_logps": -155.9129638671875, "debug/reference_chosen_logps": -147.6890869140625, "debug/reference_rejected_logps": -155.93359375, "epoch": 0.6216216216216216, "grad_norm": 3.5943677168048143, "learning_rate": 1e-06, "logits/chosen": -3.3076963424682617, "logits/rejected": -3.3063511848449707, "logps/chosen": -147.4810791015625, "logps/rejected": -155.9129638671875, "loss": 0.499, "rewards/accuracies": 0.625, "rewards/chosen": 0.0020801923237740993, "rewards/margins": 0.001873721950687468, "rewards/rejected": 0.00020647048950195312, "step": 23 }, { "debug/policy_chosen_logits": -3.3162076473236084, "debug/policy_chosen_logps": -146.12327575683594, "debug/policy_rejected_logits": -3.2808167934417725, "debug/policy_rejected_logps": -150.74459838867188, "debug/reference_chosen_logps": -146.20852661132812, "debug/reference_rejected_logps": -151.0212860107422, "epoch": 0.6486486486486487, "grad_norm": 3.490255067904084, "learning_rate": 1e-06, "logits/chosen": -3.3162076473236084, "logits/rejected": -3.2808167934417725, "logps/chosen": -146.12327575683594, "logps/rejected": -150.74459838867188, "loss": 0.4979, "rewards/accuracies": 0.75, "rewards/chosen": 0.0008525657467544079, "rewards/margins": -0.0019143482204526663, "rewards/rejected": 0.002766914200037718, "step": 24 }, { "debug/policy_chosen_logits": -3.341583490371704, "debug/policy_chosen_logps": -145.74295043945312, "debug/policy_rejected_logits": -3.2622642517089844, "debug/policy_rejected_logps": -148.95095825195312, "debug/reference_chosen_logps": -145.28700256347656, "debug/reference_rejected_logps": -148.38320922851562, "epoch": 0.6756756756756757, "grad_norm": 3.4930826080691433, "learning_rate": 1e-06, "logits/chosen": -3.341583490371704, "logits/rejected": -3.2622642517089844, "logps/chosen": -145.74295043945312, "logps/rejected": -148.95095825195312, "loss": 0.4992, "rewards/accuracies": 0.375, "rewards/chosen": -0.0045594689436256886, "rewards/margins": 0.001117925625294447, "rewards/rejected": -0.0056773945689201355, "step": 25 }, { "debug/policy_chosen_logits": -3.3710758686065674, "debug/policy_chosen_logps": -134.76580810546875, "debug/policy_rejected_logits": -3.268131971359253, "debug/policy_rejected_logps": -164.87716674804688, "debug/reference_chosen_logps": -134.9357452392578, "debug/reference_rejected_logps": -164.06057739257812, "epoch": 0.7027027027027027, "grad_norm": 3.5324806343379795, "learning_rate": 1e-06, "logits/chosen": -3.3710758686065674, "logits/rejected": -3.268131971359253, "logps/chosen": -134.76580810546875, "logps/rejected": -164.87716674804688, "loss": 0.4961, "rewards/accuracies": 0.875, "rewards/chosen": 0.0016993426252156496, "rewards/margins": 0.009865359403192997, "rewards/rejected": -0.008166017010807991, "step": 26 }, { "debug/policy_chosen_logits": -3.276531934738159, "debug/policy_chosen_logps": -128.13336181640625, "debug/policy_rejected_logits": -3.2201032638549805, "debug/policy_rejected_logps": -148.01954650878906, "debug/reference_chosen_logps": -127.79987335205078, "debug/reference_rejected_logps": -148.602783203125, "epoch": 0.7297297297297297, "grad_norm": 3.5504923092772165, "learning_rate": 1e-06, "logits/chosen": -3.276531934738159, "logits/rejected": -3.2201032638549805, "logps/chosen": -128.13336181640625, "logps/rejected": -148.01954650878906, "loss": 0.4978, "rewards/accuracies": 0.25, "rewards/chosen": -0.0033348274882882833, "rewards/margins": -0.009167090058326721, "rewards/rejected": 0.005832261871546507, "step": 27 }, { "debug/policy_chosen_logits": -3.3670849800109863, "debug/policy_chosen_logps": -142.864013671875, "debug/policy_rejected_logits": -3.314009428024292, "debug/policy_rejected_logps": -159.29888916015625, "debug/reference_chosen_logps": -143.54417419433594, "debug/reference_rejected_logps": -159.19894409179688, "epoch": 0.7567567567567568, "grad_norm": 3.6333481085660533, "learning_rate": 1e-06, "logits/chosen": -3.3670849800109863, "logits/rejected": -3.314009428024292, "logps/chosen": -142.864013671875, "logps/rejected": -159.29888916015625, "loss": 0.4946, "rewards/accuracies": 0.875, "rewards/chosen": 0.006801585666835308, "rewards/margins": 0.007801036350429058, "rewards/rejected": -0.0009994508000090718, "step": 28 }, { "debug/policy_chosen_logits": -3.4510252475738525, "debug/policy_chosen_logps": -144.7528839111328, "debug/policy_rejected_logits": -3.4992246627807617, "debug/policy_rejected_logps": -129.56163024902344, "debug/reference_chosen_logps": -144.27435302734375, "debug/reference_rejected_logps": -128.7982177734375, "epoch": 0.7837837837837838, "grad_norm": 3.519707420288352, "learning_rate": 1e-06, "logits/chosen": -3.4510252475738525, "logits/rejected": -3.4992246627807617, "logps/chosen": -144.7528839111328, "logps/rejected": -129.56163024902344, "loss": 0.4917, "rewards/accuracies": 0.625, "rewards/chosen": -0.0047853561118245125, "rewards/margins": 0.0028487201780080795, "rewards/rejected": -0.007634077221155167, "step": 29 }, { "debug/policy_chosen_logits": -3.442349910736084, "debug/policy_chosen_logps": -129.01779174804688, "debug/policy_rejected_logits": -3.355781078338623, "debug/policy_rejected_logps": -125.97911071777344, "debug/reference_chosen_logps": -129.1351318359375, "debug/reference_rejected_logps": -125.61131286621094, "epoch": 0.8108108108108109, "grad_norm": 3.436282122029211, "learning_rate": 1e-06, "logits/chosen": -3.442349910736084, "logits/rejected": -3.355781078338623, "logps/chosen": -129.01779174804688, "logps/rejected": -125.97911071777344, "loss": 0.4957, "rewards/accuracies": 0.375, "rewards/chosen": 0.0011734389699995518, "rewards/margins": 0.004851359874010086, "rewards/rejected": -0.003677921136841178, "step": 30 }, { "debug/policy_chosen_logits": -3.46421480178833, "debug/policy_chosen_logps": -148.1017608642578, "debug/policy_rejected_logits": -3.4284045696258545, "debug/policy_rejected_logps": -155.14401245117188, "debug/reference_chosen_logps": -147.857666015625, "debug/reference_rejected_logps": -153.7999267578125, "epoch": 0.8378378378378378, "grad_norm": 3.573380219675295, "learning_rate": 1e-06, "logits/chosen": -3.46421480178833, "logits/rejected": -3.4284045696258545, "logps/chosen": -148.1017608642578, "logps/rejected": -155.14401245117188, "loss": 0.4994, "rewards/accuracies": 0.5, "rewards/chosen": -0.002440805546939373, "rewards/margins": 0.010999973863363266, "rewards/rejected": -0.013440780341625214, "step": 31 }, { "debug/policy_chosen_logits": -3.296452522277832, "debug/policy_chosen_logps": -146.0049285888672, "debug/policy_rejected_logits": -3.3844947814941406, "debug/policy_rejected_logps": -145.01739501953125, "debug/reference_chosen_logps": -146.03688049316406, "debug/reference_rejected_logps": -144.96478271484375, "epoch": 0.8648648648648649, "grad_norm": 3.7844677501283552, "learning_rate": 1e-06, "logits/chosen": -3.296452522277832, "logits/rejected": -3.3844947814941406, "logps/chosen": -146.0049285888672, "logps/rejected": -145.01739501953125, "loss": 0.4983, "rewards/accuracies": 0.625, "rewards/chosen": 0.0003195476019755006, "rewards/margins": 0.0008457758231088519, "rewards/rejected": -0.000526227755472064, "step": 32 }, { "debug/policy_chosen_logits": -3.393280506134033, "debug/policy_chosen_logps": -135.49090576171875, "debug/policy_rejected_logits": -3.4107859134674072, "debug/policy_rejected_logps": -126.17267608642578, "debug/reference_chosen_logps": -135.9659423828125, "debug/reference_rejected_logps": -123.26311492919922, "epoch": 0.8918918918918919, "grad_norm": 3.6987210522943323, "learning_rate": 1e-06, "logits/chosen": -3.393280506134033, "logits/rejected": -3.4107859134674072, "logps/chosen": -135.49090576171875, "logps/rejected": -126.17267608642578, "loss": 0.4982, "rewards/accuracies": 0.5, "rewards/chosen": 0.00475044222548604, "rewards/margins": 0.03384600579738617, "rewards/rejected": -0.029095562174916267, "step": 33 }, { "debug/policy_chosen_logits": -3.3232836723327637, "debug/policy_chosen_logps": -159.23611450195312, "debug/policy_rejected_logits": -3.3686249256134033, "debug/policy_rejected_logps": -165.2159423828125, "debug/reference_chosen_logps": -158.15670776367188, "debug/reference_rejected_logps": -164.94976806640625, "epoch": 0.918918918918919, "grad_norm": 3.553957133112115, "learning_rate": 1e-06, "logits/chosen": -3.3232836723327637, "logits/rejected": -3.3686249256134033, "logps/chosen": -159.23611450195312, "logps/rejected": -165.2159423828125, "loss": 0.4998, "rewards/accuracies": 0.25, "rewards/chosen": -0.010794105939567089, "rewards/margins": -0.008132400922477245, "rewards/rejected": -0.0026617045514285564, "step": 34 }, { "debug/policy_chosen_logits": -3.3675332069396973, "debug/policy_chosen_logps": -129.2323455810547, "debug/policy_rejected_logits": -3.39815092086792, "debug/policy_rejected_logps": -157.40768432617188, "debug/reference_chosen_logps": -129.46514892578125, "debug/reference_rejected_logps": -157.58755493164062, "epoch": 0.9459459459459459, "grad_norm": 3.6687608874400897, "learning_rate": 1e-06, "logits/chosen": -3.3675332069396973, "logits/rejected": -3.39815092086792, "logps/chosen": -129.2323455810547, "logps/rejected": -157.40768432617188, "loss": 0.4972, "rewards/accuracies": 0.625, "rewards/chosen": 0.0023280431050807238, "rewards/margins": 0.0005296231247484684, "rewards/rejected": 0.0017984198639169335, "step": 35 }, { "debug/policy_chosen_logits": -3.2530903816223145, "debug/policy_chosen_logps": -139.84768676757812, "debug/policy_rejected_logits": -3.380523204803467, "debug/policy_rejected_logps": -127.18656921386719, "debug/reference_chosen_logps": -140.35733032226562, "debug/reference_rejected_logps": -127.09129333496094, "epoch": 0.972972972972973, "grad_norm": 3.747493235351115, "learning_rate": 1e-06, "logits/chosen": -3.2530903816223145, "logits/rejected": -3.380523204803467, "logps/chosen": -139.84768676757812, "logps/rejected": -127.18656921386719, "loss": 0.4949, "rewards/accuracies": 0.75, "rewards/chosen": 0.005096264183521271, "rewards/margins": 0.006049060728400946, "rewards/rejected": -0.0009527970105409622, "step": 36 }, { "debug/policy_chosen_logits": -3.1672980785369873, "debug/policy_chosen_logps": -165.81317138671875, "debug/policy_rejected_logits": -3.1891133785247803, "debug/policy_rejected_logps": -135.89328002929688, "debug/reference_chosen_logps": -164.97256469726562, "debug/reference_rejected_logps": -135.76358032226562, "epoch": 1.0, "grad_norm": 3.5111470282705084, "learning_rate": 1e-06, "logits/chosen": -3.1672980785369873, "logits/rejected": -3.1891133785247803, "logps/chosen": -165.81317138671875, "logps/rejected": -135.89328002929688, "loss": 0.4809, "rewards/accuracies": 0.375, "rewards/chosen": -0.00840616226196289, "rewards/margins": -0.007109222002327442, "rewards/rejected": -0.0012969397939741611, "step": 37 }, { "epoch": 1.0, "step": 37, "total_flos": 0.0, "train_loss": 0.4979530750094233, "train_runtime": 139.1789, "train_samples_per_second": 16.849, "train_steps_per_second": 0.266 } ], "logging_steps": 1, "max_steps": 37, "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 }