{ "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 100, "global_step": 46, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "debug/policy_chosen_logits": -2.5406789779663086, "debug/policy_chosen_logps": -174.41754150390625, "debug/policy_rejected_logits": -2.521660566329956, "debug/policy_rejected_logps": -159.5323486328125, "debug/reference_chosen_logps": -174.41754150390625, "debug/reference_rejected_logps": -159.5323486328125, "epoch": 0.021739130434782608, "grad_norm": 2.9138556694482802, "learning_rate": 1e-06, "logits/chosen": -2.5406789779663086, "logits/rejected": -2.521660566329956, "logps/chosen": -174.41754150390625, "logps/rejected": -159.5323486328125, "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": -2.3866376876831055, "debug/policy_chosen_logps": -191.08587646484375, "debug/policy_rejected_logits": -2.4478790760040283, "debug/policy_rejected_logps": -191.51112365722656, "debug/reference_chosen_logps": -191.0208740234375, "debug/reference_rejected_logps": -191.41552734375, "epoch": 0.043478260869565216, "grad_norm": 2.73946049555914, "learning_rate": 1e-06, "logits/chosen": -2.3866376876831055, "logits/rejected": -2.4478790760040283, "logps/chosen": -191.08587646484375, "logps/rejected": -191.51112365722656, "loss": 0.5002, "rewards/accuracies": 0.5, "rewards/chosen": -0.0006500434246845543, "rewards/margins": 0.0003061485185753554, "rewards/rejected": -0.0009561920305714011, "step": 2 }, { "debug/policy_chosen_logits": -2.465501308441162, "debug/policy_chosen_logps": -206.74551391601562, "debug/policy_rejected_logits": -2.5732452869415283, "debug/policy_rejected_logps": -183.27645874023438, "debug/reference_chosen_logps": -206.72006225585938, "debug/reference_rejected_logps": -183.27920532226562, "epoch": 0.06521739130434782, "grad_norm": 2.8074234755722296, "learning_rate": 1e-06, "logits/chosen": -2.465501308441162, "logits/rejected": -2.5732452869415283, "logps/chosen": -206.74551391601562, "logps/rejected": -183.27645874023438, "loss": 0.4997, "rewards/accuracies": 0.375, "rewards/chosen": -0.00025447848020121455, "rewards/margins": -0.00028175354236736894, "rewards/rejected": 2.7275120373815298e-05, "step": 3 }, { "debug/policy_chosen_logits": -2.4564504623413086, "debug/policy_chosen_logps": -196.91964721679688, "debug/policy_rejected_logits": -2.4699106216430664, "debug/policy_rejected_logps": -166.32009887695312, "debug/reference_chosen_logps": -197.21090698242188, "debug/reference_rejected_logps": -166.31948852539062, "epoch": 0.08695652173913043, "grad_norm": 3.206369897884046, "learning_rate": 1e-06, "logits/chosen": -2.4564504623413086, "logits/rejected": -2.4699106216430664, "logps/chosen": -196.91964721679688, "logps/rejected": -166.32009887695312, "loss": 0.4981, "rewards/accuracies": 0.75, "rewards/chosen": 0.002912597730755806, "rewards/margins": 0.0029187584295868874, "rewards/rejected": -6.160815246403217e-06, "step": 4 }, { "debug/policy_chosen_logits": -2.444270610809326, "debug/policy_chosen_logps": -189.8358154296875, "debug/policy_rejected_logits": -2.319845676422119, "debug/policy_rejected_logps": -174.91769409179688, "debug/reference_chosen_logps": -189.87257385253906, "debug/reference_rejected_logps": -174.7454071044922, "epoch": 0.10869565217391304, "grad_norm": 2.9622178910820045, "learning_rate": 1e-06, "logits/chosen": -2.444270610809326, "logits/rejected": -2.319845676422119, "logps/chosen": -189.8358154296875, "logps/rejected": -174.91769409179688, "loss": 0.4967, "rewards/accuracies": 0.625, "rewards/chosen": 0.0003675461048260331, "rewards/margins": 0.002090301364660263, "rewards/rejected": -0.0017227554926648736, "step": 5 }, { "debug/policy_chosen_logits": -2.393871307373047, "debug/policy_chosen_logps": -202.0091094970703, "debug/policy_rejected_logits": -2.4609367847442627, "debug/policy_rejected_logps": -192.79449462890625, "debug/reference_chosen_logps": -202.7803955078125, "debug/reference_rejected_logps": -192.9990997314453, "epoch": 0.13043478260869565, "grad_norm": 2.890923079465906, "learning_rate": 1e-06, "logits/chosen": -2.393871307373047, "logits/rejected": -2.4609367847442627, "logps/chosen": -202.0091094970703, "logps/rejected": -192.79449462890625, "loss": 0.4933, "rewards/accuracies": 0.625, "rewards/chosen": 0.007712879218161106, "rewards/margins": 0.005666866432875395, "rewards/rejected": 0.0020460127852857113, "step": 6 }, { "debug/policy_chosen_logits": -2.584399700164795, "debug/policy_chosen_logps": -201.59007263183594, "debug/policy_rejected_logits": -2.624255895614624, "debug/policy_rejected_logps": -177.23953247070312, "debug/reference_chosen_logps": -202.87220764160156, "debug/reference_rejected_logps": -176.957763671875, "epoch": 0.15217391304347827, "grad_norm": 2.957447005375319, "learning_rate": 1e-06, "logits/chosen": -2.584399700164795, "logits/rejected": -2.624255895614624, "logps/chosen": -201.59007263183594, "logps/rejected": -177.23953247070312, "loss": 0.495, "rewards/accuracies": 0.75, "rewards/chosen": 0.012821445241570473, "rewards/margins": 0.01563918963074684, "rewards/rejected": -0.0028177450876682997, "step": 7 }, { "debug/policy_chosen_logits": -2.5293426513671875, "debug/policy_chosen_logps": -191.92535400390625, "debug/policy_rejected_logits": -2.4522714614868164, "debug/policy_rejected_logps": -190.64605712890625, "debug/reference_chosen_logps": -192.4668426513672, "debug/reference_rejected_logps": -191.4180145263672, "epoch": 0.17391304347826086, "grad_norm": 2.857224539567903, "learning_rate": 1e-06, "logits/chosen": -2.5293426513671875, "logits/rejected": -2.4522714614868164, "logps/chosen": -191.92535400390625, "logps/rejected": -190.64605712890625, "loss": 0.4894, "rewards/accuracies": 0.375, "rewards/chosen": 0.005414981860667467, "rewards/margins": -0.002304553519934416, "rewards/rejected": 0.00771953584626317, "step": 8 }, { "debug/policy_chosen_logits": -2.524057149887085, "debug/policy_chosen_logps": -181.09461975097656, "debug/policy_rejected_logits": -2.53678035736084, "debug/policy_rejected_logps": -184.49664306640625, "debug/reference_chosen_logps": -181.84410095214844, "debug/reference_rejected_logps": -183.90103149414062, "epoch": 0.1956521739130435, "grad_norm": 2.659967589766961, "learning_rate": 1e-06, "logits/chosen": -2.524057149887085, "logits/rejected": -2.53678035736084, "logps/chosen": -181.09461975097656, "logps/rejected": -184.49664306640625, "loss": 0.4895, "rewards/accuracies": 0.625, "rewards/chosen": 0.00749471690505743, "rewards/margins": 0.013450756669044495, "rewards/rejected": -0.005956039298325777, "step": 9 }, { "debug/policy_chosen_logits": -2.3921713829040527, "debug/policy_chosen_logps": -179.6142578125, "debug/policy_rejected_logits": -2.3711440563201904, "debug/policy_rejected_logps": -162.54669189453125, "debug/reference_chosen_logps": -180.62747192382812, "debug/reference_rejected_logps": -160.78297424316406, "epoch": 0.21739130434782608, "grad_norm": 2.907825352825365, "learning_rate": 1e-06, "logits/chosen": -2.3921713829040527, "logits/rejected": -2.3711440563201904, "logps/chosen": -179.6142578125, "logps/rejected": -162.54669189453125, "loss": 0.4802, "rewards/accuracies": 0.875, "rewards/chosen": 0.010132160037755966, "rewards/margins": 0.027769355103373528, "rewards/rejected": -0.01763719692826271, "step": 10 }, { "debug/policy_chosen_logits": -2.521888494491577, "debug/policy_chosen_logps": -184.33038330078125, "debug/policy_rejected_logits": -2.344513416290283, "debug/policy_rejected_logps": -178.75074768066406, "debug/reference_chosen_logps": -184.63455200195312, "debug/reference_rejected_logps": -175.36129760742188, "epoch": 0.2391304347826087, "grad_norm": 2.4616167539116005, "learning_rate": 1e-06, "logits/chosen": -2.521888494491577, "logits/rejected": -2.344513416290283, "logps/chosen": -184.33038330078125, "logps/rejected": -178.75074768066406, "loss": 0.4843, "rewards/accuracies": 1.0, "rewards/chosen": 0.003041705582290888, "rewards/margins": 0.03693626448512077, "rewards/rejected": -0.033894557505846024, "step": 11 }, { "debug/policy_chosen_logits": -2.405729055404663, "debug/policy_chosen_logps": -175.63497924804688, "debug/policy_rejected_logits": -2.4264020919799805, "debug/policy_rejected_logps": -179.73696899414062, "debug/reference_chosen_logps": -174.1973876953125, "debug/reference_rejected_logps": -177.64727783203125, "epoch": 0.2608695652173913, "grad_norm": 2.736323742946957, "learning_rate": 1e-06, "logits/chosen": -2.405729055404663, "logits/rejected": -2.4264020919799805, "logps/chosen": -175.63497924804688, "logps/rejected": -179.73696899414062, "loss": 0.4859, "rewards/accuracies": 0.5, "rewards/chosen": -0.014375820755958557, "rewards/margins": 0.006521034054458141, "rewards/rejected": -0.020896855741739273, "step": 12 }, { "debug/policy_chosen_logits": -2.3890933990478516, "debug/policy_chosen_logps": -189.14987182617188, "debug/policy_rejected_logits": -2.305311918258667, "debug/policy_rejected_logps": -180.1926727294922, "debug/reference_chosen_logps": -191.95681762695312, "debug/reference_rejected_logps": -178.28973388671875, "epoch": 0.2826086956521739, "grad_norm": 2.7383468500636767, "learning_rate": 1e-06, "logits/chosen": -2.3890933990478516, "logits/rejected": -2.305311918258667, "logps/chosen": -189.14987182617188, "logps/rejected": -180.1926727294922, "loss": 0.4688, "rewards/accuracies": 1.0, "rewards/chosen": 0.028069380670785904, "rewards/margins": 0.047098807990550995, "rewards/rejected": -0.01902942731976509, "step": 13 }, { "debug/policy_chosen_logits": -2.4275922775268555, "debug/policy_chosen_logps": -181.49740600585938, "debug/policy_rejected_logits": -2.43969464302063, "debug/policy_rejected_logps": -199.9078369140625, "debug/reference_chosen_logps": -180.03472900390625, "debug/reference_rejected_logps": -198.68141174316406, "epoch": 0.30434782608695654, "grad_norm": 2.4272251174632857, "learning_rate": 1e-06, "logits/chosen": -2.4275922775268555, "logits/rejected": -2.43969464302063, "logps/chosen": -181.49740600585938, "logps/rejected": -199.9078369140625, "loss": 0.4772, "rewards/accuracies": 0.5, "rewards/chosen": -0.014626597985625267, "rewards/margins": -0.002362403552979231, "rewards/rejected": -0.012264193966984749, "step": 14 }, { "debug/policy_chosen_logits": -2.47592830657959, "debug/policy_chosen_logps": -164.09986877441406, "debug/policy_rejected_logits": -2.435157537460327, "debug/policy_rejected_logps": -176.61294555664062, "debug/reference_chosen_logps": -163.7836151123047, "debug/reference_rejected_logps": -173.32427978515625, "epoch": 0.32608695652173914, "grad_norm": 2.6909888775385027, "learning_rate": 1e-06, "logits/chosen": -2.47592830657959, "logits/rejected": -2.435157537460327, "logps/chosen": -164.09986877441406, "logps/rejected": -176.61294555664062, "loss": 0.4752, "rewards/accuracies": 0.75, "rewards/chosen": -0.0031626131385564804, "rewards/margins": 0.029723986983299255, "rewards/rejected": -0.032886601984500885, "step": 15 }, { "debug/policy_chosen_logits": -2.377653121948242, "debug/policy_chosen_logps": -173.7503662109375, "debug/policy_rejected_logits": -2.386281967163086, "debug/policy_rejected_logps": -194.0179443359375, "debug/reference_chosen_logps": -173.58297729492188, "debug/reference_rejected_logps": -192.9736328125, "epoch": 0.34782608695652173, "grad_norm": 2.5551392813174902, "learning_rate": 1e-06, "logits/chosen": -2.377653121948242, "logits/rejected": -2.386281967163086, "logps/chosen": -173.7503662109375, "logps/rejected": -194.0179443359375, "loss": 0.4761, "rewards/accuracies": 0.5, "rewards/chosen": -0.0016740784049034119, "rewards/margins": 0.0087689608335495, "rewards/rejected": -0.010443039238452911, "step": 16 }, { "debug/policy_chosen_logits": -2.529132843017578, "debug/policy_chosen_logps": -172.70501708984375, "debug/policy_rejected_logits": -2.4587604999542236, "debug/policy_rejected_logps": -186.12283325195312, "debug/reference_chosen_logps": -172.2464599609375, "debug/reference_rejected_logps": -186.38314819335938, "epoch": 0.3695652173913043, "grad_norm": 2.4118823503893054, "learning_rate": 1e-06, "logits/chosen": -2.529132843017578, "logits/rejected": -2.4587604999542236, "logps/chosen": -172.70501708984375, "logps/rejected": -186.12283325195312, "loss": 0.487, "rewards/accuracies": 0.75, "rewards/chosen": -0.0045854183845222, "rewards/margins": -0.007188738323748112, "rewards/rejected": 0.002603319473564625, "step": 17 }, { "debug/policy_chosen_logits": -2.4017813205718994, "debug/policy_chosen_logps": -166.11911010742188, "debug/policy_rejected_logits": -2.3311731815338135, "debug/policy_rejected_logps": -182.46627807617188, "debug/reference_chosen_logps": -168.52517700195312, "debug/reference_rejected_logps": -179.03460693359375, "epoch": 0.391304347826087, "grad_norm": 2.615012586102309, "learning_rate": 1e-06, "logits/chosen": -2.4017813205718994, "logits/rejected": -2.3311731815338135, "logps/chosen": -166.11911010742188, "logps/rejected": -182.46627807617188, "loss": 0.4631, "rewards/accuracies": 0.875, "rewards/chosen": 0.02406078390777111, "rewards/margins": 0.05837753415107727, "rewards/rejected": -0.03431674838066101, "step": 18 }, { "debug/policy_chosen_logits": -2.4313623905181885, "debug/policy_chosen_logps": -174.13351440429688, "debug/policy_rejected_logits": -2.3800158500671387, "debug/policy_rejected_logps": -218.58282470703125, "debug/reference_chosen_logps": -172.78656005859375, "debug/reference_rejected_logps": -214.70162963867188, "epoch": 0.41304347826086957, "grad_norm": 2.801324694410347, "learning_rate": 1e-06, "logits/chosen": -2.4313623905181885, "logits/rejected": -2.3800158500671387, "logps/chosen": -174.13351440429688, "logps/rejected": -218.58282470703125, "loss": 0.4571, "rewards/accuracies": 0.75, "rewards/chosen": -0.013469582423567772, "rewards/margins": 0.025342464447021484, "rewards/rejected": -0.038812048733234406, "step": 19 }, { "debug/policy_chosen_logits": -2.4221510887145996, "debug/policy_chosen_logps": -193.74337768554688, "debug/policy_rejected_logits": -2.429517984390259, "debug/policy_rejected_logps": -211.8312530517578, "debug/reference_chosen_logps": -197.71206665039062, "debug/reference_rejected_logps": -212.77467346191406, "epoch": 0.43478260869565216, "grad_norm": 2.452728910615928, "learning_rate": 1e-06, "logits/chosen": -2.4221510887145996, "logits/rejected": -2.429517984390259, "logps/chosen": -193.74337768554688, "logps/rejected": -211.8312530517578, "loss": 0.468, "rewards/accuracies": 0.5, "rewards/chosen": 0.039686813950538635, "rewards/margins": 0.030252492055296898, "rewards/rejected": 0.009434318169951439, "step": 20 }, { "debug/policy_chosen_logits": -2.3384790420532227, "debug/policy_chosen_logps": -168.90838623046875, "debug/policy_rejected_logits": -2.249765396118164, "debug/policy_rejected_logps": -170.010009765625, "debug/reference_chosen_logps": -168.66502380371094, "debug/reference_rejected_logps": -162.436279296875, "epoch": 0.45652173913043476, "grad_norm": 2.2296634238737805, "learning_rate": 1e-06, "logits/chosen": -2.3384790420532227, "logits/rejected": -2.249765396118164, "logps/chosen": -168.90838623046875, "logps/rejected": -170.010009765625, "loss": 0.4555, "rewards/accuracies": 0.875, "rewards/chosen": -0.002433512359857559, "rewards/margins": 0.07330392301082611, "rewards/rejected": -0.07573743164539337, "step": 21 }, { "debug/policy_chosen_logits": -2.49155855178833, "debug/policy_chosen_logps": -202.97955322265625, "debug/policy_rejected_logits": -2.5280747413635254, "debug/policy_rejected_logps": -187.58743286132812, "debug/reference_chosen_logps": -203.92587280273438, "debug/reference_rejected_logps": -185.87338256835938, "epoch": 0.4782608695652174, "grad_norm": 2.481538979707535, "learning_rate": 1e-06, "logits/chosen": -2.49155855178833, "logits/rejected": -2.5280747413635254, "logps/chosen": -202.97955322265625, "logps/rejected": -187.58743286132812, "loss": 0.4759, "rewards/accuracies": 0.5, "rewards/chosen": 0.009463195689022541, "rewards/margins": 0.02660369873046875, "rewards/rejected": -0.017140503972768784, "step": 22 }, { "debug/policy_chosen_logits": -2.5137152671813965, "debug/policy_chosen_logps": -181.94056701660156, "debug/policy_rejected_logits": -2.5435562133789062, "debug/policy_rejected_logps": -169.93206787109375, "debug/reference_chosen_logps": -188.2291717529297, "debug/reference_rejected_logps": -167.34231567382812, "epoch": 0.5, "grad_norm": 2.221961851151498, "learning_rate": 1e-06, "logits/chosen": -2.5137152671813965, "logits/rejected": -2.5435562133789062, "logps/chosen": -181.94056701660156, "logps/rejected": -169.93206787109375, "loss": 0.4645, "rewards/accuracies": 0.75, "rewards/chosen": 0.06288604438304901, "rewards/margins": 0.08878357708454132, "rewards/rejected": -0.025897536426782608, "step": 23 }, { "debug/policy_chosen_logits": -2.343871593475342, "debug/policy_chosen_logps": -172.3570556640625, "debug/policy_rejected_logits": -2.3897156715393066, "debug/policy_rejected_logps": -182.73715209960938, "debug/reference_chosen_logps": -171.8414306640625, "debug/reference_rejected_logps": -182.32119750976562, "epoch": 0.5217391304347826, "grad_norm": 2.3144852334581345, "learning_rate": 1e-06, "logits/chosen": -2.343871593475342, "logits/rejected": -2.3897156715393066, "logps/chosen": -172.3570556640625, "logps/rejected": -182.73715209960938, "loss": 0.4726, "rewards/accuracies": 0.25, "rewards/chosen": -0.005156230181455612, "rewards/margins": -0.0009968150407075882, "rewards/rejected": -0.0041594114154577255, "step": 24 }, { "debug/policy_chosen_logits": -2.4783878326416016, "debug/policy_chosen_logps": -181.16600036621094, "debug/policy_rejected_logits": -2.4538094997406006, "debug/policy_rejected_logps": -192.53765869140625, "debug/reference_chosen_logps": -185.05032348632812, "debug/reference_rejected_logps": -186.58950805664062, "epoch": 0.5434782608695652, "grad_norm": 2.4281399717516448, "learning_rate": 1e-06, "logits/chosen": -2.4783878326416016, "logits/rejected": -2.4538094997406006, "logps/chosen": -181.16600036621094, "logps/rejected": -192.53765869140625, "loss": 0.4538, "rewards/accuracies": 0.75, "rewards/chosen": 0.03884328901767731, "rewards/margins": 0.09832490980625153, "rewards/rejected": -0.05948162078857422, "step": 25 }, { "debug/policy_chosen_logits": -2.4092724323272705, "debug/policy_chosen_logps": -160.72732543945312, "debug/policy_rejected_logits": -2.477595329284668, "debug/policy_rejected_logps": -175.16604614257812, "debug/reference_chosen_logps": -166.90414428710938, "debug/reference_rejected_logps": -173.4667510986328, "epoch": 0.5652173913043478, "grad_norm": 2.4404949726132137, "learning_rate": 1e-06, "logits/chosen": -2.4092724323272705, "logits/rejected": -2.477595329284668, "logps/chosen": -160.72732543945312, "logps/rejected": -175.16604614257812, "loss": 0.4506, "rewards/accuracies": 0.75, "rewards/chosen": 0.06176815181970596, "rewards/margins": 0.07876109331846237, "rewards/rejected": -0.016992948949337006, "step": 26 }, { "debug/policy_chosen_logits": -2.4233715534210205, "debug/policy_chosen_logps": -190.73126220703125, "debug/policy_rejected_logits": -2.382371187210083, "debug/policy_rejected_logps": -197.88619995117188, "debug/reference_chosen_logps": -196.33099365234375, "debug/reference_rejected_logps": -193.63540649414062, "epoch": 0.5869565217391305, "grad_norm": 2.33043065534125, "learning_rate": 1e-06, "logits/chosen": -2.4233715534210205, "logits/rejected": -2.382371187210083, "logps/chosen": -190.73126220703125, "logps/rejected": -197.88619995117188, "loss": 0.4187, "rewards/accuracies": 0.875, "rewards/chosen": 0.05599748343229294, "rewards/margins": 0.0985054224729538, "rewards/rejected": -0.04250793531537056, "step": 27 }, { "debug/policy_chosen_logits": -2.5104820728302, "debug/policy_chosen_logps": -170.4730682373047, "debug/policy_rejected_logits": -2.4717085361480713, "debug/policy_rejected_logps": -189.20010375976562, "debug/reference_chosen_logps": -168.71560668945312, "debug/reference_rejected_logps": -185.61746215820312, "epoch": 0.6086956521739131, "grad_norm": 2.8275079552922446, "learning_rate": 1e-06, "logits/chosen": -2.5104820728302, "logits/rejected": -2.4717085361480713, "logps/chosen": -170.4730682373047, "logps/rejected": -189.20010375976562, "loss": 0.4666, "rewards/accuracies": 0.375, "rewards/chosen": -0.01757451891899109, "rewards/margins": 0.018252111971378326, "rewards/rejected": -0.035826630890369415, "step": 28 }, { "debug/policy_chosen_logits": -2.484487771987915, "debug/policy_chosen_logps": -186.02122497558594, "debug/policy_rejected_logits": -2.4840569496154785, "debug/policy_rejected_logps": -174.24012756347656, "debug/reference_chosen_logps": -190.0591278076172, "debug/reference_rejected_logps": -166.46307373046875, "epoch": 0.6304347826086957, "grad_norm": 2.5260328428308236, "learning_rate": 1e-06, "logits/chosen": -2.484487771987915, "logits/rejected": -2.4840569496154785, "logps/chosen": -186.02122497558594, "logps/rejected": -174.24012756347656, "loss": 0.4554, "rewards/accuracies": 0.75, "rewards/chosen": 0.04037889465689659, "rewards/margins": 0.11814939230680466, "rewards/rejected": -0.07777049392461777, "step": 29 }, { "debug/policy_chosen_logits": -2.4572229385375977, "debug/policy_chosen_logps": -177.25079345703125, "debug/policy_rejected_logits": -2.335451364517212, "debug/policy_rejected_logps": -211.37869262695312, "debug/reference_chosen_logps": -184.99151611328125, "debug/reference_rejected_logps": -207.14004516601562, "epoch": 0.6521739130434783, "grad_norm": 2.7172712356585476, "learning_rate": 1e-06, "logits/chosen": -2.4572229385375977, "logits/rejected": -2.335451364517212, "logps/chosen": -177.25079345703125, "logps/rejected": -211.37869262695312, "loss": 0.463, "rewards/accuracies": 0.625, "rewards/chosen": 0.07740737497806549, "rewards/margins": 0.11979396641254425, "rewards/rejected": -0.04238658770918846, "step": 30 }, { "debug/policy_chosen_logits": -2.4997236728668213, "debug/policy_chosen_logps": -176.38267517089844, "debug/policy_rejected_logits": -2.3669979572296143, "debug/policy_rejected_logps": -196.77554321289062, "debug/reference_chosen_logps": -178.01614379882812, "debug/reference_rejected_logps": -191.33193969726562, "epoch": 0.6739130434782609, "grad_norm": 2.353661137242158, "learning_rate": 1e-06, "logits/chosen": -2.4997236728668213, "logits/rejected": -2.3669979572296143, "logps/chosen": -176.38267517089844, "logps/rejected": -196.77554321289062, "loss": 0.4335, "rewards/accuracies": 0.625, "rewards/chosen": 0.01633470691740513, "rewards/margins": 0.07077069580554962, "rewards/rejected": -0.05443599820137024, "step": 31 }, { "debug/policy_chosen_logits": -2.4377670288085938, "debug/policy_chosen_logps": -173.71165466308594, "debug/policy_rejected_logits": -2.3115673065185547, "debug/policy_rejected_logps": -178.8059539794922, "debug/reference_chosen_logps": -179.61834716796875, "debug/reference_rejected_logps": -181.16517639160156, "epoch": 0.6956521739130435, "grad_norm": 2.3528189249640397, "learning_rate": 1e-06, "logits/chosen": -2.4377670288085938, "logits/rejected": -2.3115673065185547, "logps/chosen": -173.71165466308594, "logps/rejected": -178.8059539794922, "loss": 0.4606, "rewards/accuracies": 0.5, "rewards/chosen": 0.05906686931848526, "rewards/margins": 0.0354747548699379, "rewards/rejected": 0.023592106997966766, "step": 32 }, { "debug/policy_chosen_logits": -2.467334747314453, "debug/policy_chosen_logps": -188.59512329101562, "debug/policy_rejected_logits": -2.487856388092041, "debug/policy_rejected_logps": -200.94195556640625, "debug/reference_chosen_logps": -190.67750549316406, "debug/reference_rejected_logps": -195.129638671875, "epoch": 0.717391304347826, "grad_norm": 2.5411257448652553, "learning_rate": 1e-06, "logits/chosen": -2.467334747314453, "logits/rejected": -2.487856388092041, "logps/chosen": -188.59512329101562, "logps/rejected": -200.94195556640625, "loss": 0.4339, "rewards/accuracies": 0.75, "rewards/chosen": 0.020823916420340538, "rewards/margins": 0.07894693315029144, "rewards/rejected": -0.058123014867305756, "step": 33 }, { "debug/policy_chosen_logits": -2.3515822887420654, "debug/policy_chosen_logps": -172.7677001953125, "debug/policy_rejected_logits": -2.395439386367798, "debug/policy_rejected_logps": -188.29266357421875, "debug/reference_chosen_logps": -181.23960876464844, "debug/reference_rejected_logps": -179.37924194335938, "epoch": 0.7391304347826086, "grad_norm": 2.3779889701354375, "learning_rate": 1e-06, "logits/chosen": -2.3515822887420654, "logits/rejected": -2.395439386367798, "logps/chosen": -172.7677001953125, "logps/rejected": -188.29266357421875, "loss": 0.4141, "rewards/accuracies": 0.875, "rewards/chosen": 0.08471919596195221, "rewards/margins": 0.17385347187519073, "rewards/rejected": -0.08913427591323853, "step": 34 }, { "debug/policy_chosen_logits": -2.4782862663269043, "debug/policy_chosen_logps": -178.69082641601562, "debug/policy_rejected_logits": -2.456272840499878, "debug/policy_rejected_logps": -187.01419067382812, "debug/reference_chosen_logps": -177.35031127929688, "debug/reference_rejected_logps": -177.96685791015625, "epoch": 0.7608695652173914, "grad_norm": 2.3565954950933694, "learning_rate": 1e-06, "logits/chosen": -2.4782862663269043, "logits/rejected": -2.456272840499878, "logps/chosen": -178.69082641601562, "logps/rejected": -187.01419067382812, "loss": 0.4367, "rewards/accuracies": 0.625, "rewards/chosen": -0.013405149802565575, "rewards/margins": 0.07706813514232635, "rewards/rejected": -0.09047328680753708, "step": 35 }, { "debug/policy_chosen_logits": -2.457199811935425, "debug/policy_chosen_logps": -173.6107177734375, "debug/policy_rejected_logits": -2.484218120574951, "debug/policy_rejected_logps": -184.22190856933594, "debug/reference_chosen_logps": -185.9441375732422, "debug/reference_rejected_logps": -179.55136108398438, "epoch": 0.782608695652174, "grad_norm": 2.429486721658497, "learning_rate": 1e-06, "logits/chosen": -2.457199811935425, "logits/rejected": -2.484218120574951, "logps/chosen": -173.6107177734375, "logps/rejected": -184.22190856933594, "loss": 0.4355, "rewards/accuracies": 0.875, "rewards/chosen": 0.12333421409130096, "rewards/margins": 0.17003962397575378, "rewards/rejected": -0.04670540988445282, "step": 36 }, { "debug/policy_chosen_logits": -2.4880294799804688, "debug/policy_chosen_logps": -174.43316650390625, "debug/policy_rejected_logits": -2.4997189044952393, "debug/policy_rejected_logps": -177.84881591796875, "debug/reference_chosen_logps": -177.31240844726562, "debug/reference_rejected_logps": -165.77145385742188, "epoch": 0.8043478260869565, "grad_norm": 2.441269723279697, "learning_rate": 1e-06, "logits/chosen": -2.4880294799804688, "logits/rejected": -2.4997189044952393, "logps/chosen": -174.43316650390625, "logps/rejected": -177.84881591796875, "loss": 0.4394, "rewards/accuracies": 1.0, "rewards/chosen": 0.02879244089126587, "rewards/margins": 0.14956600964069366, "rewards/rejected": -0.12077357620000839, "step": 37 }, { "debug/policy_chosen_logits": -2.5809593200683594, "debug/policy_chosen_logps": -181.38897705078125, "debug/policy_rejected_logits": -2.483285665512085, "debug/policy_rejected_logps": -190.49742126464844, "debug/reference_chosen_logps": -187.25927734375, "debug/reference_rejected_logps": -184.10122680664062, "epoch": 0.8260869565217391, "grad_norm": 2.36617525939136, "learning_rate": 1e-06, "logits/chosen": -2.5809593200683594, "logits/rejected": -2.483285665512085, "logps/chosen": -181.38897705078125, "logps/rejected": -190.49742126464844, "loss": 0.4358, "rewards/accuracies": 0.75, "rewards/chosen": 0.05870307236909866, "rewards/margins": 0.12266507744789124, "rewards/rejected": -0.06396199762821198, "step": 38 }, { "debug/policy_chosen_logits": -2.449005126953125, "debug/policy_chosen_logps": -163.90740966796875, "debug/policy_rejected_logits": -2.4301159381866455, "debug/policy_rejected_logps": -194.1475067138672, "debug/reference_chosen_logps": -169.6490478515625, "debug/reference_rejected_logps": -184.88339233398438, "epoch": 0.8478260869565217, "grad_norm": 2.4604115357660445, "learning_rate": 1e-06, "logits/chosen": -2.449005126953125, "logits/rejected": -2.4301159381866455, "logps/chosen": -163.90740966796875, "logps/rejected": -194.1475067138672, "loss": 0.4329, "rewards/accuracies": 0.75, "rewards/chosen": 0.05741637200117111, "rewards/margins": 0.15005755424499512, "rewards/rejected": -0.0926411971449852, "step": 39 }, { "debug/policy_chosen_logits": -2.479243040084839, "debug/policy_chosen_logps": -177.88658142089844, "debug/policy_rejected_logits": -2.521466016769409, "debug/policy_rejected_logps": -174.86874389648438, "debug/reference_chosen_logps": -189.6613311767578, "debug/reference_rejected_logps": -172.8925323486328, "epoch": 0.8695652173913043, "grad_norm": 2.6005589820274, "learning_rate": 1e-06, "logits/chosen": -2.479243040084839, "logits/rejected": -2.521466016769409, "logps/chosen": -177.88658142089844, "logps/rejected": -174.86874389648438, "loss": 0.3689, "rewards/accuracies": 0.75, "rewards/chosen": 0.11774755269289017, "rewards/margins": 0.13750958442687988, "rewards/rejected": -0.01976201869547367, "step": 40 }, { "debug/policy_chosen_logits": -2.5032005310058594, "debug/policy_chosen_logps": -167.452392578125, "debug/policy_rejected_logits": -2.390470027923584, "debug/policy_rejected_logps": -187.92933654785156, "debug/reference_chosen_logps": -170.6841583251953, "debug/reference_rejected_logps": -191.27334594726562, "epoch": 0.8913043478260869, "grad_norm": 2.784651354754412, "learning_rate": 1e-06, "logits/chosen": -2.5032005310058594, "logits/rejected": -2.390470027923584, "logps/chosen": -167.452392578125, "logps/rejected": -187.92933654785156, "loss": 0.4514, "rewards/accuracies": 0.25, "rewards/chosen": 0.03231758251786232, "rewards/margins": -0.0011225640773773193, "rewards/rejected": 0.03344014659523964, "step": 41 }, { "debug/policy_chosen_logits": -2.5493698120117188, "debug/policy_chosen_logps": -172.9954071044922, "debug/policy_rejected_logits": -2.476661205291748, "debug/policy_rejected_logps": -179.5677947998047, "debug/reference_chosen_logps": -183.045654296875, "debug/reference_rejected_logps": -181.4873046875, "epoch": 0.9130434782608695, "grad_norm": 2.677440788734919, "learning_rate": 1e-06, "logits/chosen": -2.5493698120117188, "logits/rejected": -2.476661205291748, "logps/chosen": -172.9954071044922, "logps/rejected": -179.5677947998047, "loss": 0.4056, "rewards/accuracies": 0.625, "rewards/chosen": 0.10050234198570251, "rewards/margins": 0.08130710572004318, "rewards/rejected": 0.019195232540369034, "step": 42 }, { "debug/policy_chosen_logits": -2.5432963371276855, "debug/policy_chosen_logps": -175.61265563964844, "debug/policy_rejected_logits": -2.4852747917175293, "debug/policy_rejected_logps": -201.0880584716797, "debug/reference_chosen_logps": -176.60537719726562, "debug/reference_rejected_logps": -192.08706665039062, "epoch": 0.9347826086956522, "grad_norm": 2.7145887961284068, "learning_rate": 1e-06, "logits/chosen": -2.5432963371276855, "logits/rejected": -2.4852747917175293, "logps/chosen": -175.61265563964844, "logps/rejected": -201.0880584716797, "loss": 0.4241, "rewards/accuracies": 0.625, "rewards/chosen": 0.009927215054631233, "rewards/margins": 0.09993711113929749, "rewards/rejected": -0.0900098979473114, "step": 43 }, { "debug/policy_chosen_logits": -2.497959852218628, "debug/policy_chosen_logps": -174.22557067871094, "debug/policy_rejected_logits": -2.427154541015625, "debug/policy_rejected_logps": -183.83010864257812, "debug/reference_chosen_logps": -186.29788208007812, "debug/reference_rejected_logps": -183.1434326171875, "epoch": 0.9565217391304348, "grad_norm": 2.581573767785096, "learning_rate": 1e-06, "logits/chosen": -2.497959852218628, "logits/rejected": -2.427154541015625, "logps/chosen": -174.22557067871094, "logps/rejected": -183.83010864257812, "loss": 0.4177, "rewards/accuracies": 0.75, "rewards/chosen": 0.12072312831878662, "rewards/margins": 0.1275898963212967, "rewards/rejected": -0.006866758689284325, "step": 44 }, { "debug/policy_chosen_logits": -2.5587399005889893, "debug/policy_chosen_logps": -194.60675048828125, "debug/policy_rejected_logits": -2.4343531131744385, "debug/policy_rejected_logps": -195.62185668945312, "debug/reference_chosen_logps": -198.23883056640625, "debug/reference_rejected_logps": -180.94686889648438, "epoch": 0.9782608695652174, "grad_norm": 2.815838392583519, "learning_rate": 1e-06, "logits/chosen": -2.5587399005889893, "logits/rejected": -2.4343531131744385, "logps/chosen": -194.60675048828125, "logps/rejected": -195.62185668945312, "loss": 0.4548, "rewards/accuracies": 0.75, "rewards/chosen": 0.036320820450782776, "rewards/margins": 0.18307071924209595, "rewards/rejected": -0.14674988389015198, "step": 45 }, { "debug/policy_chosen_logits": -2.663532257080078, "debug/policy_chosen_logps": -181.77276611328125, "debug/policy_rejected_logits": -2.5467002391815186, "debug/policy_rejected_logps": -176.77719116210938, "debug/reference_chosen_logps": -182.98736572265625, "debug/reference_rejected_logps": -163.69749450683594, "epoch": 1.0, "grad_norm": 2.1986019501463825, "learning_rate": 1e-06, "logits/chosen": -2.663532257080078, "logits/rejected": -2.5467002391815186, "logps/chosen": -181.77276611328125, "logps/rejected": -176.77719116210938, "loss": 0.3328, "rewards/accuracies": 0.875, "rewards/chosen": 0.012146133929491043, "rewards/margins": 0.14294317364692688, "rewards/rejected": -0.13079702854156494, "step": 46 }, { "epoch": 1.0, "step": 46, "total_flos": 0.0, "train_loss": 0.4573696670324906, "train_runtime": 475.0336, "train_samples_per_second": 6.067, "train_steps_per_second": 0.097 } ], "logging_steps": 1, "max_steps": 46, "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 }