{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 100, "global_step": 52, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "debug/policy_chosen_logits": -2.4203386306762695, "debug/policy_chosen_logps": -185.7370147705078, "debug/policy_rejected_logits": -2.3896992206573486, "debug/policy_rejected_logps": -193.2849578857422, "debug/reference_chosen_logps": -185.7370147705078, "debug/reference_rejected_logps": -193.2849578857422, "epoch": 0.019230769230769232, "grad_norm": 2.861107813250133, "learning_rate": 1e-06, "logits/chosen": -2.4203386306762695, "logits/rejected": -2.3896992206573486, "logps/chosen": -185.7370147705078, "logps/rejected": -193.2849578857422, "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": -2.425750494003296, "debug/policy_chosen_logps": -179.03013610839844, "debug/policy_rejected_logits": -2.2569093704223633, "debug/policy_rejected_logps": -169.09234619140625, "debug/reference_chosen_logps": -178.87767028808594, "debug/reference_rejected_logps": -169.06729125976562, "epoch": 0.038461538461538464, "grad_norm": 3.758529521252219, "learning_rate": 1e-06, "logits/chosen": -2.425750494003296, "logits/rejected": -2.2569093704223633, "logps/chosen": -179.03013610839844, "logps/rejected": -169.09234619140625, "loss": 0.4996, "rewards/accuracies": 0.375, "rewards/chosen": -0.0015246201073750854, "rewards/margins": -0.0012739944504573941, "rewards/rejected": -0.0002506256278138608, "step": 2 }, { "debug/policy_chosen_logits": -2.4937379360198975, "debug/policy_chosen_logps": -194.95635986328125, "debug/policy_rejected_logits": -2.375201463699341, "debug/policy_rejected_logps": -179.0343017578125, "debug/reference_chosen_logps": -194.90005493164062, "debug/reference_rejected_logps": -179.09872436523438, "epoch": 0.057692307692307696, "grad_norm": 3.0965832459483087, "learning_rate": 1e-06, "logits/chosen": -2.4937379360198975, "logits/rejected": -2.375201463699341, "logps/chosen": -194.95635986328125, "logps/rejected": -179.0343017578125, "loss": 0.4994, "rewards/accuracies": 0.625, "rewards/chosen": -0.0005632019019685686, "rewards/margins": -0.0012073325924575329, "rewards/rejected": 0.0006441307486966252, "step": 3 }, { "debug/policy_chosen_logits": -2.4720897674560547, "debug/policy_chosen_logps": -188.27117919921875, "debug/policy_rejected_logits": -2.5355641841888428, "debug/policy_rejected_logps": -188.49911499023438, "debug/reference_chosen_logps": -188.7332763671875, "debug/reference_rejected_logps": -188.25445556640625, "epoch": 0.07692307692307693, "grad_norm": 3.5718687734725973, "learning_rate": 1e-06, "logits/chosen": -2.4720897674560547, "logits/rejected": -2.5355641841888428, "logps/chosen": -188.27117919921875, "logps/rejected": -188.49911499023438, "loss": 0.4969, "rewards/accuracies": 0.875, "rewards/chosen": 0.0046209716238081455, "rewards/margins": 0.007067641708999872, "rewards/rejected": -0.002446670550853014, "step": 4 }, { "debug/policy_chosen_logits": -2.379190683364868, "debug/policy_chosen_logps": -174.59059143066406, "debug/policy_rejected_logits": -2.5373787879943848, "debug/policy_rejected_logps": -175.47076416015625, "debug/reference_chosen_logps": -174.71871948242188, "debug/reference_rejected_logps": -175.51339721679688, "epoch": 0.09615384615384616, "grad_norm": 3.242762657272401, "learning_rate": 1e-06, "logits/chosen": -2.379190683364868, "logits/rejected": -2.5373787879943848, "logps/chosen": -174.59059143066406, "logps/rejected": -175.47076416015625, "loss": 0.4962, "rewards/accuracies": 0.5, "rewards/chosen": 0.0012813375797122717, "rewards/margins": 0.0008550643688067794, "rewards/rejected": 0.00042627332732081413, "step": 5 }, { "debug/policy_chosen_logits": -2.401658296585083, "debug/policy_chosen_logps": -183.67599487304688, "debug/policy_rejected_logits": -2.3361270427703857, "debug/policy_rejected_logps": -192.68539428710938, "debug/reference_chosen_logps": -183.98712158203125, "debug/reference_rejected_logps": -192.7581787109375, "epoch": 0.11538461538461539, "grad_norm": 2.5048963174015153, "learning_rate": 1e-06, "logits/chosen": -2.401658296585083, "logits/rejected": -2.3361270427703857, "logps/chosen": -183.67599487304688, "logps/rejected": -192.68539428710938, "loss": 0.4963, "rewards/accuracies": 0.75, "rewards/chosen": 0.003111190628260374, "rewards/margins": 0.0023833648301661015, "rewards/rejected": 0.0007278253324329853, "step": 6 }, { "debug/policy_chosen_logits": -2.4647703170776367, "debug/policy_chosen_logps": -196.40322875976562, "debug/policy_rejected_logits": -2.3510549068450928, "debug/policy_rejected_logps": -188.60353088378906, "debug/reference_chosen_logps": -196.56094360351562, "debug/reference_rejected_logps": -187.82266235351562, "epoch": 0.1346153846153846, "grad_norm": 2.954448778538458, "learning_rate": 1e-06, "logits/chosen": -2.4647703170776367, "logits/rejected": -2.3510549068450928, "logps/chosen": -196.40322875976562, "logps/rejected": -188.60353088378906, "loss": 0.4906, "rewards/accuracies": 0.875, "rewards/chosen": 0.0015772818587720394, "rewards/margins": 0.009385967627167702, "rewards/rejected": -0.007808685302734375, "step": 7 }, { "debug/policy_chosen_logits": -2.376976728439331, "debug/policy_chosen_logps": -186.48193359375, "debug/policy_rejected_logits": -2.4907095432281494, "debug/policy_rejected_logps": -184.52247619628906, "debug/reference_chosen_logps": -186.63832092285156, "debug/reference_rejected_logps": -184.3030548095703, "epoch": 0.15384615384615385, "grad_norm": 2.7660154133539856, "learning_rate": 1e-06, "logits/chosen": -2.376976728439331, "logits/rejected": -2.4907095432281494, "logps/chosen": -186.48193359375, "logps/rejected": -184.52247619628906, "loss": 0.4912, "rewards/accuracies": 0.375, "rewards/chosen": 0.0015639301855117083, "rewards/margins": 0.003758162958547473, "rewards/rejected": -0.0021942327730357647, "step": 8 }, { "debug/policy_chosen_logits": -2.4716763496398926, "debug/policy_chosen_logps": -184.51742553710938, "debug/policy_rejected_logits": -2.392301082611084, "debug/policy_rejected_logps": -163.52381896972656, "debug/reference_chosen_logps": -185.8481903076172, "debug/reference_rejected_logps": -162.64688110351562, "epoch": 0.17307692307692307, "grad_norm": 3.19925551824819, "learning_rate": 1e-06, "logits/chosen": -2.4716763496398926, "logits/rejected": -2.392301082611084, "logps/chosen": -184.51742553710938, "logps/rejected": -163.52381896972656, "loss": 0.4819, "rewards/accuracies": 0.875, "rewards/chosen": 0.013307590037584305, "rewards/margins": 0.022076930850744247, "rewards/rejected": -0.008769340813159943, "step": 9 }, { "debug/policy_chosen_logits": -2.4346506595611572, "debug/policy_chosen_logps": -184.35189819335938, "debug/policy_rejected_logits": -2.3823659420013428, "debug/policy_rejected_logps": -177.3807373046875, "debug/reference_chosen_logps": -183.94378662109375, "debug/reference_rejected_logps": -176.2843017578125, "epoch": 0.19230769230769232, "grad_norm": 2.5515027926271254, "learning_rate": 1e-06, "logits/chosen": -2.4346506595611572, "logits/rejected": -2.3823659420013428, "logps/chosen": -184.35189819335938, "logps/rejected": -177.3807373046875, "loss": 0.4849, "rewards/accuracies": 0.625, "rewards/chosen": -0.004081211052834988, "rewards/margins": 0.006883163936436176, "rewards/rejected": -0.010964374989271164, "step": 10 }, { "debug/policy_chosen_logits": -2.4018287658691406, "debug/policy_chosen_logps": -202.3935546875, "debug/policy_rejected_logits": -2.393723249435425, "debug/policy_rejected_logps": -176.10626220703125, "debug/reference_chosen_logps": -204.30197143554688, "debug/reference_rejected_logps": -172.5755615234375, "epoch": 0.21153846153846154, "grad_norm": 2.5842830353224504, "learning_rate": 1e-06, "logits/chosen": -2.4018287658691406, "logits/rejected": -2.393723249435425, "logps/chosen": -202.3935546875, "logps/rejected": -176.10626220703125, "loss": 0.4826, "rewards/accuracies": 0.75, "rewards/chosen": 0.01908406987786293, "rewards/margins": 0.05439120531082153, "rewards/rejected": -0.0353071391582489, "step": 11 }, { "debug/policy_chosen_logits": -2.5084776878356934, "debug/policy_chosen_logps": -171.409912109375, "debug/policy_rejected_logits": -2.3861539363861084, "debug/policy_rejected_logps": -177.23403930664062, "debug/reference_chosen_logps": -171.54713439941406, "debug/reference_rejected_logps": -176.790283203125, "epoch": 0.23076923076923078, "grad_norm": 13.924057089587274, "learning_rate": 1e-06, "logits/chosen": -2.5084776878356934, "logits/rejected": -2.3861539363861084, "logps/chosen": -171.409912109375, "logps/rejected": -177.23403930664062, "loss": 0.4839, "rewards/accuracies": 0.625, "rewards/chosen": 0.0013722041621804237, "rewards/margins": 0.005809726193547249, "rewards/rejected": -0.004437522031366825, "step": 12 }, { "debug/policy_chosen_logits": -2.4744768142700195, "debug/policy_chosen_logps": -193.55789184570312, "debug/policy_rejected_logits": -2.4974329471588135, "debug/policy_rejected_logps": -168.37875366210938, "debug/reference_chosen_logps": -200.3148193359375, "debug/reference_rejected_logps": -167.8997344970703, "epoch": 0.25, "grad_norm": 2.4746191677166034, "learning_rate": 1e-06, "logits/chosen": -2.4744768142700195, "logits/rejected": -2.4974329471588135, "logps/chosen": -193.55789184570312, "logps/rejected": -168.37875366210938, "loss": 0.4767, "rewards/accuracies": 0.875, "rewards/chosen": 0.06756922602653503, "rewards/margins": 0.07235933095216751, "rewards/rejected": -0.004790096543729305, "step": 13 }, { "debug/policy_chosen_logits": -2.476685047149658, "debug/policy_chosen_logps": -185.6330108642578, "debug/policy_rejected_logits": -2.4483656883239746, "debug/policy_rejected_logps": -168.87298583984375, "debug/reference_chosen_logps": -186.21258544921875, "debug/reference_rejected_logps": -167.95176696777344, "epoch": 0.2692307692307692, "grad_norm": 2.899101440660964, "learning_rate": 1e-06, "logits/chosen": -2.476685047149658, "logits/rejected": -2.4483656883239746, "logps/chosen": -185.6330108642578, "logps/rejected": -168.87298583984375, "loss": 0.4745, "rewards/accuracies": 0.5, "rewards/chosen": 0.005795860663056374, "rewards/margins": 0.015008030459284782, "rewards/rejected": -0.009212171658873558, "step": 14 }, { "debug/policy_chosen_logits": -2.5017175674438477, "debug/policy_chosen_logps": -190.4012451171875, "debug/policy_rejected_logits": -2.491224765777588, "debug/policy_rejected_logps": -184.8310089111328, "debug/reference_chosen_logps": -190.4630126953125, "debug/reference_rejected_logps": -184.58352661132812, "epoch": 0.28846153846153844, "grad_norm": 2.707026111376091, "learning_rate": 1e-06, "logits/chosen": -2.5017175674438477, "logits/rejected": -2.491224765777588, "logps/chosen": -190.4012451171875, "logps/rejected": -184.8310089111328, "loss": 0.4844, "rewards/accuracies": 0.375, "rewards/chosen": 0.0006176568567752838, "rewards/margins": 0.003092498518526554, "rewards/rejected": -0.0024748416617512703, "step": 15 }, { "debug/policy_chosen_logits": -2.504624128341675, "debug/policy_chosen_logps": -198.4542999267578, "debug/policy_rejected_logits": -2.5721426010131836, "debug/policy_rejected_logps": -184.6616668701172, "debug/reference_chosen_logps": -198.76397705078125, "debug/reference_rejected_logps": -184.24363708496094, "epoch": 0.3076923076923077, "grad_norm": 2.451867353363919, "learning_rate": 1e-06, "logits/chosen": -2.504624128341675, "logits/rejected": -2.5721426010131836, "logps/chosen": -198.4542999267578, "logps/rejected": -184.6616668701172, "loss": 0.4723, "rewards/accuracies": 0.375, "rewards/chosen": 0.0030968282371759415, "rewards/margins": 0.007277088705450296, "rewards/rejected": -0.004180260002613068, "step": 16 }, { "debug/policy_chosen_logits": -2.412865400314331, "debug/policy_chosen_logps": -188.2651824951172, "debug/policy_rejected_logits": -2.619596004486084, "debug/policy_rejected_logps": -181.06619262695312, "debug/reference_chosen_logps": -193.1811065673828, "debug/reference_rejected_logps": -177.83963012695312, "epoch": 0.3269230769230769, "grad_norm": 2.7151091944601835, "learning_rate": 1e-06, "logits/chosen": -2.412865400314331, "logits/rejected": -2.619596004486084, "logps/chosen": -188.2651824951172, "logps/rejected": -181.06619262695312, "loss": 0.4597, "rewards/accuracies": 0.875, "rewards/chosen": 0.04915916174650192, "rewards/margins": 0.08142490684986115, "rewards/rejected": -0.032265737652778625, "step": 17 }, { "debug/policy_chosen_logits": -2.5042190551757812, "debug/policy_chosen_logps": -173.1564483642578, "debug/policy_rejected_logits": -2.437241315841675, "debug/policy_rejected_logps": -169.3559112548828, "debug/reference_chosen_logps": -176.03759765625, "debug/reference_rejected_logps": -165.6055145263672, "epoch": 0.34615384615384615, "grad_norm": 2.4735514132532463, "learning_rate": 1e-06, "logits/chosen": -2.5042190551757812, "logits/rejected": -2.437241315841675, "logps/chosen": -173.1564483642578, "logps/rejected": -169.3559112548828, "loss": 0.4615, "rewards/accuracies": 0.75, "rewards/chosen": 0.02881147339940071, "rewards/margins": 0.06631540507078171, "rewards/rejected": -0.0375039279460907, "step": 18 }, { "debug/policy_chosen_logits": -2.575247049331665, "debug/policy_chosen_logps": -177.78562927246094, "debug/policy_rejected_logits": -2.478971242904663, "debug/policy_rejected_logps": -197.25088500976562, "debug/reference_chosen_logps": -185.2480926513672, "debug/reference_rejected_logps": -191.05230712890625, "epoch": 0.36538461538461536, "grad_norm": 2.5059572820368983, "learning_rate": 1e-06, "logits/chosen": -2.575247049331665, "logits/rejected": -2.478971242904663, "logps/chosen": -177.78562927246094, "logps/rejected": -197.25088500976562, "loss": 0.4556, "rewards/accuracies": 0.75, "rewards/chosen": 0.0746246874332428, "rewards/margins": 0.13661034405231476, "rewards/rejected": -0.06198566406965256, "step": 19 }, { "debug/policy_chosen_logits": -2.4564669132232666, "debug/policy_chosen_logps": -187.01968383789062, "debug/policy_rejected_logits": -2.4763052463531494, "debug/policy_rejected_logps": -167.06741333007812, "debug/reference_chosen_logps": -191.00048828125, "debug/reference_rejected_logps": -161.8093719482422, "epoch": 0.38461538461538464, "grad_norm": 2.425342166211571, "learning_rate": 1e-06, "logits/chosen": -2.4564669132232666, "logits/rejected": -2.4763052463531494, "logps/chosen": -187.01968383789062, "logps/rejected": -167.06741333007812, "loss": 0.4602, "rewards/accuracies": 1.0, "rewards/chosen": 0.03980813920497894, "rewards/margins": 0.0923885703086853, "rewards/rejected": -0.05258043110370636, "step": 20 }, { "debug/policy_chosen_logits": -2.608337879180908, "debug/policy_chosen_logps": -187.06781005859375, "debug/policy_rejected_logits": -2.5621848106384277, "debug/policy_rejected_logps": -198.1059112548828, "debug/reference_chosen_logps": -190.21240234375, "debug/reference_rejected_logps": -199.0448455810547, "epoch": 0.40384615384615385, "grad_norm": 2.7000635287256856, "learning_rate": 1e-06, "logits/chosen": -2.608337879180908, "logits/rejected": -2.5621848106384277, "logps/chosen": -187.06781005859375, "logps/rejected": -198.1059112548828, "loss": 0.4668, "rewards/accuracies": 0.5, "rewards/chosen": 0.03144590184092522, "rewards/margins": 0.022056521847844124, "rewards/rejected": 0.009389379993081093, "step": 21 }, { "debug/policy_chosen_logits": -2.4417309761047363, "debug/policy_chosen_logps": -182.86297607421875, "debug/policy_rejected_logits": -2.4356002807617188, "debug/policy_rejected_logps": -183.1719207763672, "debug/reference_chosen_logps": -186.6331787109375, "debug/reference_rejected_logps": -176.64874267578125, "epoch": 0.4230769230769231, "grad_norm": 2.4584380266236465, "learning_rate": 1e-06, "logits/chosen": -2.4417309761047363, "logits/rejected": -2.4356002807617188, "logps/chosen": -182.86297607421875, "logps/rejected": -183.1719207763672, "loss": 0.4312, "rewards/accuracies": 0.75, "rewards/chosen": 0.037701837718486786, "rewards/margins": 0.10293363034725189, "rewards/rejected": -0.0652318000793457, "step": 22 }, { "debug/policy_chosen_logits": -2.3962759971618652, "debug/policy_chosen_logps": -188.2168426513672, "debug/policy_rejected_logits": -2.3940060138702393, "debug/policy_rejected_logps": -194.57363891601562, "debug/reference_chosen_logps": -188.746337890625, "debug/reference_rejected_logps": -188.7298583984375, "epoch": 0.4423076923076923, "grad_norm": 2.480095762764339, "learning_rate": 1e-06, "logits/chosen": -2.3962759971618652, "logits/rejected": -2.3940060138702393, "logps/chosen": -188.2168426513672, "logps/rejected": -194.57363891601562, "loss": 0.4706, "rewards/accuracies": 0.875, "rewards/chosen": 0.005294991657137871, "rewards/margins": 0.06373281031847, "rewards/rejected": -0.05843781679868698, "step": 23 }, { "debug/policy_chosen_logits": -2.4898786544799805, "debug/policy_chosen_logps": -181.02059936523438, "debug/policy_rejected_logits": -2.4966273307800293, "debug/policy_rejected_logps": -178.0485382080078, "debug/reference_chosen_logps": -186.77435302734375, "debug/reference_rejected_logps": -165.88906860351562, "epoch": 0.46153846153846156, "grad_norm": 3.310010224714518, "learning_rate": 1e-06, "logits/chosen": -2.4898786544799805, "logits/rejected": -2.4966273307800293, "logps/chosen": -181.02059936523438, "logps/rejected": -178.0485382080078, "loss": 0.4563, "rewards/accuracies": 0.625, "rewards/chosen": 0.057537782937288284, "rewards/margins": 0.17913252115249634, "rewards/rejected": -0.12159475684165955, "step": 24 }, { "debug/policy_chosen_logits": -2.432927131652832, "debug/policy_chosen_logps": -193.996337890625, "debug/policy_rejected_logits": -2.3908307552337646, "debug/policy_rejected_logps": -179.30120849609375, "debug/reference_chosen_logps": -194.2586669921875, "debug/reference_rejected_logps": -175.4796600341797, "epoch": 0.4807692307692308, "grad_norm": 2.734643941898122, "learning_rate": 1e-06, "logits/chosen": -2.432927131652832, "logits/rejected": -2.3908307552337646, "logps/chosen": -193.996337890625, "logps/rejected": -179.30120849609375, "loss": 0.4619, "rewards/accuracies": 0.625, "rewards/chosen": 0.0026233294047415257, "rewards/margins": 0.040838733315467834, "rewards/rejected": -0.038215406239032745, "step": 25 }, { "debug/policy_chosen_logits": -2.633073568344116, "debug/policy_chosen_logps": -169.07635498046875, "debug/policy_rejected_logits": -2.499666452407837, "debug/policy_rejected_logps": -181.62399291992188, "debug/reference_chosen_logps": -170.44346618652344, "debug/reference_rejected_logps": -182.32366943359375, "epoch": 0.5, "grad_norm": 2.4564174502500187, "learning_rate": 1e-06, "logits/chosen": -2.633073568344116, "logits/rejected": -2.499666452407837, "logps/chosen": -169.07635498046875, "logps/rejected": -181.62399291992188, "loss": 0.4155, "rewards/accuracies": 0.375, "rewards/chosen": 0.013671131804585457, "rewards/margins": 0.006674385629594326, "rewards/rejected": 0.006996745243668556, "step": 26 }, { "debug/policy_chosen_logits": -2.4202969074249268, "debug/policy_chosen_logps": -169.46751403808594, "debug/policy_rejected_logits": -2.3937838077545166, "debug/policy_rejected_logps": -202.80528259277344, "debug/reference_chosen_logps": -174.74176025390625, "debug/reference_rejected_logps": -188.62229919433594, "epoch": 0.5192307692307693, "grad_norm": 2.5559014487276897, "learning_rate": 1e-06, "logits/chosen": -2.4202969074249268, "logits/rejected": -2.3937838077545166, "logps/chosen": -169.46751403808594, "logps/rejected": -202.80528259277344, "loss": 0.4164, "rewards/accuracies": 0.625, "rewards/chosen": 0.05274249613285065, "rewards/margins": 0.1945723593235016, "rewards/rejected": -0.14182986319065094, "step": 27 }, { "debug/policy_chosen_logits": -2.5047430992126465, "debug/policy_chosen_logps": -193.3773193359375, "debug/policy_rejected_logits": -2.5947458744049072, "debug/policy_rejected_logps": -190.57015991210938, "debug/reference_chosen_logps": -197.58242797851562, "debug/reference_rejected_logps": -192.86123657226562, "epoch": 0.5384615384615384, "grad_norm": 2.557372552942143, "learning_rate": 1e-06, "logits/chosen": -2.5047430992126465, "logits/rejected": -2.5947458744049072, "logps/chosen": -193.3773193359375, "logps/rejected": -190.57015991210938, "loss": 0.4418, "rewards/accuracies": 0.5, "rewards/chosen": 0.04205124080181122, "rewards/margins": 0.019140375778079033, "rewards/rejected": 0.022910863161087036, "step": 28 }, { "debug/policy_chosen_logits": -2.5177297592163086, "debug/policy_chosen_logps": -177.6782684326172, "debug/policy_rejected_logits": -2.553908586502075, "debug/policy_rejected_logps": -180.67721557617188, "debug/reference_chosen_logps": -188.78192138671875, "debug/reference_rejected_logps": -176.42282104492188, "epoch": 0.5576923076923077, "grad_norm": 2.8504632090769895, "learning_rate": 1e-06, "logits/chosen": -2.5177297592163086, "logits/rejected": -2.553908586502075, "logps/chosen": -177.6782684326172, "logps/rejected": -180.67721557617188, "loss": 0.4267, "rewards/accuracies": 0.875, "rewards/chosen": 0.11103656888008118, "rewards/margins": 0.15358057618141174, "rewards/rejected": -0.04254401847720146, "step": 29 }, { "debug/policy_chosen_logits": -2.581258535385132, "debug/policy_chosen_logps": -179.66339111328125, "debug/policy_rejected_logits": -2.608623504638672, "debug/policy_rejected_logps": -189.11404418945312, "debug/reference_chosen_logps": -195.3540496826172, "debug/reference_rejected_logps": -183.5130615234375, "epoch": 0.5769230769230769, "grad_norm": 2.46712437577004, "learning_rate": 1e-06, "logits/chosen": -2.581258535385132, "logits/rejected": -2.608623504638672, "logps/chosen": -179.66339111328125, "logps/rejected": -189.11404418945312, "loss": 0.4377, "rewards/accuracies": 0.875, "rewards/chosen": 0.1569066047668457, "rewards/margins": 0.21291649341583252, "rewards/rejected": -0.05600988492369652, "step": 30 }, { "debug/policy_chosen_logits": -2.6142561435699463, "debug/policy_chosen_logps": -171.92503356933594, "debug/policy_rejected_logits": -2.6900107860565186, "debug/policy_rejected_logps": -191.13990783691406, "debug/reference_chosen_logps": -182.1653594970703, "debug/reference_rejected_logps": -179.43771362304688, "epoch": 0.5961538461538461, "grad_norm": 2.721519686537735, "learning_rate": 1e-06, "logits/chosen": -2.6142561435699463, "logits/rejected": -2.6900107860565186, "logps/chosen": -171.92503356933594, "logps/rejected": -191.13990783691406, "loss": 0.4596, "rewards/accuracies": 0.75, "rewards/chosen": 0.10240327566862106, "rewards/margins": 0.2194252908229828, "rewards/rejected": -0.11702201515436172, "step": 31 }, { "debug/policy_chosen_logits": -2.6169021129608154, "debug/policy_chosen_logps": -188.87440490722656, "debug/policy_rejected_logits": -2.6965301036834717, "debug/policy_rejected_logps": -171.771728515625, "debug/reference_chosen_logps": -195.43609619140625, "debug/reference_rejected_logps": -163.79580688476562, "epoch": 0.6153846153846154, "grad_norm": 2.3832321574951383, "learning_rate": 1e-06, "logits/chosen": -2.6169021129608154, "logits/rejected": -2.6965301036834717, "logps/chosen": -188.87440490722656, "logps/rejected": -171.771728515625, "loss": 0.4268, "rewards/accuracies": 0.625, "rewards/chosen": 0.06561696529388428, "rewards/margins": 0.14537619054317474, "rewards/rejected": -0.07975921779870987, "step": 32 }, { "debug/policy_chosen_logits": -2.609198570251465, "debug/policy_chosen_logps": -177.60379028320312, "debug/policy_rejected_logits": -2.612945556640625, "debug/policy_rejected_logps": -180.07342529296875, "debug/reference_chosen_logps": -178.36428833007812, "debug/reference_rejected_logps": -184.2348175048828, "epoch": 0.6346153846153846, "grad_norm": 2.547530468856055, "learning_rate": 1e-06, "logits/chosen": -2.609198570251465, "logits/rejected": -2.612945556640625, "logps/chosen": -177.60379028320312, "logps/rejected": -180.07342529296875, "loss": 0.4551, "rewards/accuracies": 0.25, "rewards/chosen": 0.007604943588376045, "rewards/margins": -0.034009091556072235, "rewards/rejected": 0.04161403328180313, "step": 33 }, { "debug/policy_chosen_logits": -2.5101234912872314, "debug/policy_chosen_logps": -175.4557647705078, "debug/policy_rejected_logits": -2.437258720397949, "debug/policy_rejected_logps": -179.3190460205078, "debug/reference_chosen_logps": -187.11509704589844, "debug/reference_rejected_logps": -165.7357635498047, "epoch": 0.6538461538461539, "grad_norm": 2.5305119814539445, "learning_rate": 1e-06, "logits/chosen": -2.5101234912872314, "logits/rejected": -2.437258720397949, "logps/chosen": -175.4557647705078, "logps/rejected": -179.3190460205078, "loss": 0.4091, "rewards/accuracies": 0.875, "rewards/chosen": 0.11659333854913712, "rewards/margins": 0.2524262070655823, "rewards/rejected": -0.13583287596702576, "step": 34 }, { "debug/policy_chosen_logits": -2.468809127807617, "debug/policy_chosen_logps": -185.04544067382812, "debug/policy_rejected_logits": -2.469884157180786, "debug/policy_rejected_logps": -184.70066833496094, "debug/reference_chosen_logps": -185.5817108154297, "debug/reference_rejected_logps": -172.92080688476562, "epoch": 0.6730769230769231, "grad_norm": 3.1480814962694685, "learning_rate": 1e-06, "logits/chosen": -2.468809127807617, "logits/rejected": -2.469884157180786, "logps/chosen": -185.04544067382812, "logps/rejected": -184.70066833496094, "loss": 0.4113, "rewards/accuracies": 0.875, "rewards/chosen": 0.005362699739634991, "rewards/margins": 0.12316124886274338, "rewards/rejected": -0.11779855191707611, "step": 35 }, { "debug/policy_chosen_logits": -2.5778415203094482, "debug/policy_chosen_logps": -181.67620849609375, "debug/policy_rejected_logits": -2.547286033630371, "debug/policy_rejected_logps": -169.78890991210938, "debug/reference_chosen_logps": -186.6635284423828, "debug/reference_rejected_logps": -164.9740753173828, "epoch": 0.6923076923076923, "grad_norm": 2.7680443110481425, "learning_rate": 1e-06, "logits/chosen": -2.5778415203094482, "logits/rejected": -2.547286033630371, "logps/chosen": -181.67620849609375, "logps/rejected": -169.78890991210938, "loss": 0.3974, "rewards/accuracies": 0.75, "rewards/chosen": 0.049873240292072296, "rewards/margins": 0.09802162647247314, "rewards/rejected": -0.04814840480685234, "step": 36 }, { "debug/policy_chosen_logits": -2.3479855060577393, "debug/policy_chosen_logps": -193.39633178710938, "debug/policy_rejected_logits": -2.406175136566162, "debug/policy_rejected_logps": -210.29010009765625, "debug/reference_chosen_logps": -195.6107177734375, "debug/reference_rejected_logps": -210.36666870117188, "epoch": 0.7115384615384616, "grad_norm": 2.8459592127259077, "learning_rate": 1e-06, "logits/chosen": -2.3479855060577393, "logits/rejected": -2.406175136566162, "logps/chosen": -193.39633178710938, "logps/rejected": -210.29010009765625, "loss": 0.4494, "rewards/accuracies": 0.5, "rewards/chosen": 0.022143803536891937, "rewards/margins": 0.021378114819526672, "rewards/rejected": 0.0007656887173652649, "step": 37 }, { "debug/policy_chosen_logits": -2.5617218017578125, "debug/policy_chosen_logps": -169.8967742919922, "debug/policy_rejected_logits": -2.553598642349243, "debug/policy_rejected_logps": -184.04336547851562, "debug/reference_chosen_logps": -171.13760375976562, "debug/reference_rejected_logps": -173.55960083007812, "epoch": 0.7307692307692307, "grad_norm": 2.4998422725548872, "learning_rate": 1e-06, "logits/chosen": -2.5617218017578125, "logits/rejected": -2.553598642349243, "logps/chosen": -169.8967742919922, "logps/rejected": -184.04336547851562, "loss": 0.4182, "rewards/accuracies": 0.625, "rewards/chosen": 0.012408370152115822, "rewards/margins": 0.1172461062669754, "rewards/rejected": -0.10483773797750473, "step": 38 }, { "debug/policy_chosen_logits": -2.6159751415252686, "debug/policy_chosen_logps": -177.14169311523438, "debug/policy_rejected_logits": -2.470346212387085, "debug/policy_rejected_logps": -185.30313110351562, "debug/reference_chosen_logps": -176.63275146484375, "debug/reference_rejected_logps": -173.42752075195312, "epoch": 0.75, "grad_norm": 2.4903873108561334, "learning_rate": 1e-06, "logits/chosen": -2.6159751415252686, "logits/rejected": -2.470346212387085, "logps/chosen": -177.14169311523438, "logps/rejected": -185.30313110351562, "loss": 0.4231, "rewards/accuracies": 0.625, "rewards/chosen": -0.00508943572640419, "rewards/margins": 0.11366665363311768, "rewards/rejected": -0.11875607818365097, "step": 39 }, { "debug/policy_chosen_logits": -2.559603214263916, "debug/policy_chosen_logps": -163.67063903808594, "debug/policy_rejected_logits": -2.5933189392089844, "debug/policy_rejected_logps": -204.8035888671875, "debug/reference_chosen_logps": -181.83868408203125, "debug/reference_rejected_logps": -176.70587158203125, "epoch": 0.7692307692307693, "grad_norm": 2.6366677610786207, "learning_rate": 1e-06, "logits/chosen": -2.559603214263916, "logits/rejected": -2.5933189392089844, "logps/chosen": -163.67063903808594, "logps/rejected": -204.8035888671875, "loss": 0.4185, "rewards/accuracies": 1.0, "rewards/chosen": 0.18168054521083832, "rewards/margins": 0.46265774965286255, "rewards/rejected": -0.28097718954086304, "step": 40 }, { "debug/policy_chosen_logits": -2.5421648025512695, "debug/policy_chosen_logps": -177.27471923828125, "debug/policy_rejected_logits": -2.505100727081299, "debug/policy_rejected_logps": -160.61941528320312, "debug/reference_chosen_logps": -187.25704956054688, "debug/reference_rejected_logps": -167.02279663085938, "epoch": 0.7884615384615384, "grad_norm": 3.236312060323122, "learning_rate": 1e-06, "logits/chosen": -2.5421648025512695, "logits/rejected": -2.505100727081299, "logps/chosen": -177.27471923828125, "logps/rejected": -160.61941528320312, "loss": 0.4289, "rewards/accuracies": 0.5, "rewards/chosen": 0.09982329607009888, "rewards/margins": 0.03578958660364151, "rewards/rejected": 0.06403371691703796, "step": 41 }, { "debug/policy_chosen_logits": -2.4895424842834473, "debug/policy_chosen_logps": -175.51809692382812, "debug/policy_rejected_logits": -2.51467227935791, "debug/policy_rejected_logps": -183.83267211914062, "debug/reference_chosen_logps": -185.79779052734375, "debug/reference_rejected_logps": -182.77545166015625, "epoch": 0.8076923076923077, "grad_norm": 2.6383530556095534, "learning_rate": 1e-06, "logits/chosen": -2.4895424842834473, "logits/rejected": -2.51467227935791, "logps/chosen": -175.51809692382812, "logps/rejected": -183.83267211914062, "loss": 0.4038, "rewards/accuracies": 0.625, "rewards/chosen": 0.10279695689678192, "rewards/margins": 0.11336928606033325, "rewards/rejected": -0.010572336614131927, "step": 42 }, { "debug/policy_chosen_logits": -2.4962143898010254, "debug/policy_chosen_logps": -158.62835693359375, "debug/policy_rejected_logits": -2.4434423446655273, "debug/policy_rejected_logps": -185.53860473632812, "debug/reference_chosen_logps": -164.46080017089844, "debug/reference_rejected_logps": -176.23179626464844, "epoch": 0.8269230769230769, "grad_norm": 2.6112399691060086, "learning_rate": 1e-06, "logits/chosen": -2.4962143898010254, "logits/rejected": -2.4434423446655273, "logps/chosen": -158.62835693359375, "logps/rejected": -185.53860473632812, "loss": 0.4235, "rewards/accuracies": 0.75, "rewards/chosen": 0.05832444876432419, "rewards/margins": 0.15139251947402954, "rewards/rejected": -0.09306806325912476, "step": 43 }, { "debug/policy_chosen_logits": -2.5579187870025635, "debug/policy_chosen_logps": -170.46934509277344, "debug/policy_rejected_logits": -2.6019985675811768, "debug/policy_rejected_logps": -178.60887145996094, "debug/reference_chosen_logps": -172.26303100585938, "debug/reference_rejected_logps": -175.90789794921875, "epoch": 0.8461538461538461, "grad_norm": 2.3532832206319196, "learning_rate": 1e-06, "logits/chosen": -2.5579187870025635, "logits/rejected": -2.6019985675811768, "logps/chosen": -170.46934509277344, "logps/rejected": -178.60887145996094, "loss": 0.4057, "rewards/accuracies": 0.625, "rewards/chosen": 0.017936859279870987, "rewards/margins": 0.044946540147066116, "rewards/rejected": -0.02700968086719513, "step": 44 }, { "debug/policy_chosen_logits": -2.469608783721924, "debug/policy_chosen_logps": -172.97830200195312, "debug/policy_rejected_logits": -2.3801496028900146, "debug/policy_rejected_logps": -187.92433166503906, "debug/reference_chosen_logps": -172.52529907226562, "debug/reference_rejected_logps": -192.7061309814453, "epoch": 0.8653846153846154, "grad_norm": 2.812631576781413, "learning_rate": 1e-06, "logits/chosen": -2.469608783721924, "logits/rejected": -2.3801496028900146, "logps/chosen": -172.97830200195312, "logps/rejected": -187.92433166503906, "loss": 0.4785, "rewards/accuracies": 0.5, "rewards/chosen": -0.0045300861820578575, "rewards/margins": -0.05234817415475845, "rewards/rejected": 0.04781809076666832, "step": 45 }, { "debug/policy_chosen_logits": -2.5066044330596924, "debug/policy_chosen_logps": -179.63990783691406, "debug/policy_rejected_logits": -2.453065872192383, "debug/policy_rejected_logps": -186.52072143554688, "debug/reference_chosen_logps": -187.28431701660156, "debug/reference_rejected_logps": -177.356201171875, "epoch": 0.8846153846153846, "grad_norm": 2.504767872387961, "learning_rate": 1e-06, "logits/chosen": -2.5066044330596924, "logits/rejected": -2.453065872192383, "logps/chosen": -179.63990783691406, "logps/rejected": -186.52072143554688, "loss": 0.4148, "rewards/accuracies": 0.625, "rewards/chosen": 0.07644408941268921, "rewards/margins": 0.1680891513824463, "rewards/rejected": -0.09164506196975708, "step": 46 }, { "debug/policy_chosen_logits": -2.654778242111206, "debug/policy_chosen_logps": -179.33819580078125, "debug/policy_rejected_logits": -2.682997941970825, "debug/policy_rejected_logps": -192.85226440429688, "debug/reference_chosen_logps": -183.18214416503906, "debug/reference_rejected_logps": -187.54226684570312, "epoch": 0.9038461538461539, "grad_norm": 2.562280161360506, "learning_rate": 1e-06, "logits/chosen": -2.654778242111206, "logits/rejected": -2.682997941970825, "logps/chosen": -179.33819580078125, "logps/rejected": -192.85226440429688, "loss": 0.4253, "rewards/accuracies": 0.625, "rewards/chosen": 0.038439467549324036, "rewards/margins": 0.09153944253921509, "rewards/rejected": -0.05309997498989105, "step": 47 }, { "debug/policy_chosen_logits": -2.5873446464538574, "debug/policy_chosen_logps": -189.9923095703125, "debug/policy_rejected_logits": -2.5469915866851807, "debug/policy_rejected_logps": -217.79244995117188, "debug/reference_chosen_logps": -195.66209411621094, "debug/reference_rejected_logps": -211.9183807373047, "epoch": 0.9230769230769231, "grad_norm": 3.1238863952973825, "learning_rate": 1e-06, "logits/chosen": -2.5873446464538574, "logits/rejected": -2.5469915866851807, "logps/chosen": -189.9923095703125, "logps/rejected": -217.79244995117188, "loss": 0.4156, "rewards/accuracies": 0.75, "rewards/chosen": 0.056697867810726166, "rewards/margins": 0.11543838679790497, "rewards/rejected": -0.0587405264377594, "step": 48 }, { "debug/policy_chosen_logits": -2.523024320602417, "debug/policy_chosen_logps": -174.21859741210938, "debug/policy_rejected_logits": -2.6089887619018555, "debug/policy_rejected_logps": -189.6505584716797, "debug/reference_chosen_logps": -188.59774780273438, "debug/reference_rejected_logps": -175.30157470703125, "epoch": 0.9423076923076923, "grad_norm": 4.820015921348583, "learning_rate": 1e-06, "logits/chosen": -2.523024320602417, "logits/rejected": -2.6089887619018555, "logps/chosen": -174.21859741210938, "logps/rejected": -189.6505584716797, "loss": 0.4273, "rewards/accuracies": 1.0, "rewards/chosen": 0.14379161596298218, "rewards/margins": 0.28728151321411133, "rewards/rejected": -0.14348991215229034, "step": 49 }, { "debug/policy_chosen_logits": -2.497637987136841, "debug/policy_chosen_logps": -173.8629150390625, "debug/policy_rejected_logits": -2.571160316467285, "debug/policy_rejected_logps": -172.56173706054688, "debug/reference_chosen_logps": -177.63519287109375, "debug/reference_rejected_logps": -179.11053466796875, "epoch": 0.9615384615384616, "grad_norm": 2.514694194877925, "learning_rate": 1e-06, "logits/chosen": -2.497637987136841, "logits/rejected": -2.571160316467285, "logps/chosen": -173.8629150390625, "logps/rejected": -172.56173706054688, "loss": 0.4278, "rewards/accuracies": 0.25, "rewards/chosen": 0.037722814828157425, "rewards/margins": -0.027765290811657906, "rewards/rejected": 0.06548811495304108, "step": 50 }, { "debug/policy_chosen_logits": -2.562901496887207, "debug/policy_chosen_logps": -171.64630126953125, "debug/policy_rejected_logits": -2.454517364501953, "debug/policy_rejected_logps": -202.7394256591797, "debug/reference_chosen_logps": -184.54364013671875, "debug/reference_rejected_logps": -186.64706420898438, "epoch": 0.9807692307692307, "grad_norm": 2.312459713777107, "learning_rate": 1e-06, "logits/chosen": -2.562901496887207, "logits/rejected": -2.454517364501953, "logps/chosen": -171.64630126953125, "logps/rejected": -202.7394256591797, "loss": 0.3931, "rewards/accuracies": 0.75, "rewards/chosen": 0.12897342443466187, "rewards/margins": 0.2898971438407898, "rewards/rejected": -0.16092371940612793, "step": 51 }, { "debug/policy_chosen_logits": -2.5228018760681152, "debug/policy_chosen_logps": -183.95437622070312, "debug/policy_rejected_logits": -2.5072357654571533, "debug/policy_rejected_logps": -203.783935546875, "debug/reference_chosen_logps": -192.70321655273438, "debug/reference_rejected_logps": -193.47909545898438, "epoch": 1.0, "grad_norm": 2.616108850938675, "learning_rate": 1e-06, "logits/chosen": -2.5228018760681152, "logits/rejected": -2.5072357654571533, "logps/chosen": -183.95437622070312, "logps/rejected": -203.783935546875, "loss": 0.4113, "rewards/accuracies": 0.75, "rewards/chosen": 0.08748845756053925, "rewards/margins": 0.19053685665130615, "rewards/rejected": -0.10304839164018631, "step": 52 }, { "epoch": 1.0, "step": 52, "total_flos": 0.0, "train_loss": 0.4499474299641756, "train_runtime": 514.8505, "train_samples_per_second": 6.447, "train_steps_per_second": 0.101 } ], "logging_steps": 1, "max_steps": 52, "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 }