| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 100, | |
| "global_step": 478, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0020920502092050207, | |
| "grad_norm": 9.170599089942366, | |
| "learning_rate": 1.0416666666666666e-08, | |
| "logits/chosen": -2.7655434608459473, | |
| "logits/rejected": -2.7158141136169434, | |
| "logps/chosen": -269.8756103515625, | |
| "logps/rejected": -360.62164306640625, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02092050209205021, | |
| "grad_norm": 9.150565723604885, | |
| "learning_rate": 1.0416666666666667e-07, | |
| "logits/chosen": -2.5923984050750732, | |
| "logits/rejected": -2.562947988510132, | |
| "logps/chosen": -264.8230895996094, | |
| "logps/rejected": -251.5508575439453, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.3888888955116272, | |
| "rewards/chosen": -0.0003005376784130931, | |
| "rewards/margins": -0.00048651432734914124, | |
| "rewards/rejected": 0.0001859766198322177, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04184100418410042, | |
| "grad_norm": 8.401794731870115, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "logits/chosen": -2.6533799171447754, | |
| "logits/rejected": -2.6054489612579346, | |
| "logps/chosen": -281.46368408203125, | |
| "logps/rejected": -296.8436584472656, | |
| "loss": 0.6924, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.0004009083495475352, | |
| "rewards/margins": 0.000830893695820123, | |
| "rewards/rejected": -0.00042998549179174006, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06276150627615062, | |
| "grad_norm": 8.275952755978503, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": -2.6644034385681152, | |
| "logits/rejected": -2.591947078704834, | |
| "logps/chosen": -299.4920959472656, | |
| "logps/rejected": -263.7144470214844, | |
| "loss": 0.6885, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.011641914024949074, | |
| "rewards/margins": 0.009539209306240082, | |
| "rewards/rejected": 0.0021027035545557737, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08368200836820083, | |
| "grad_norm": 8.165628930324349, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "logits/chosen": -2.557978391647339, | |
| "logits/rejected": -2.5226192474365234, | |
| "logps/chosen": -263.15728759765625, | |
| "logps/rejected": -245.1326141357422, | |
| "loss": 0.6776, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": 0.03938329964876175, | |
| "rewards/margins": 0.04691743105649948, | |
| "rewards/rejected": -0.00753413001075387, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.10460251046025104, | |
| "grad_norm": 8.408127463371278, | |
| "learning_rate": 4.999733114418725e-07, | |
| "logits/chosen": -2.499746084213257, | |
| "logits/rejected": -2.4598052501678467, | |
| "logps/chosen": -279.6321716308594, | |
| "logps/rejected": -297.48968505859375, | |
| "loss": 0.6615, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.0010753454407677054, | |
| "rewards/margins": 0.04832205921411514, | |
| "rewards/rejected": -0.04724670574069023, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12552301255230125, | |
| "grad_norm": 11.367364527138742, | |
| "learning_rate": 4.990398100856366e-07, | |
| "logits/chosen": -2.5166015625, | |
| "logits/rejected": -2.4452641010284424, | |
| "logps/chosen": -273.73309326171875, | |
| "logps/rejected": -312.8514709472656, | |
| "loss": 0.639, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.02069356106221676, | |
| "rewards/margins": 0.1263645738363266, | |
| "rewards/rejected": -0.1470581293106079, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.14644351464435146, | |
| "grad_norm": 10.797907272646912, | |
| "learning_rate": 4.967775735898179e-07, | |
| "logits/chosen": -2.369143009185791, | |
| "logits/rejected": -2.366969585418701, | |
| "logps/chosen": -270.88665771484375, | |
| "logps/rejected": -283.3267517089844, | |
| "loss": 0.6162, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.05217716097831726, | |
| "rewards/margins": 0.21364183723926544, | |
| "rewards/rejected": -0.2658190131187439, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.16736401673640167, | |
| "grad_norm": 20.445807364960142, | |
| "learning_rate": 4.931986719649298e-07, | |
| "logits/chosen": -2.560523509979248, | |
| "logits/rejected": -2.4790120124816895, | |
| "logps/chosen": -336.60650634765625, | |
| "logps/rejected": -304.67822265625, | |
| "loss": 0.6025, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.13829118013381958, | |
| "rewards/margins": 0.26742225885391235, | |
| "rewards/rejected": -0.4057134687900543, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.18828451882845187, | |
| "grad_norm": 16.762310140078398, | |
| "learning_rate": 4.883222001996351e-07, | |
| "logits/chosen": -2.073870897293091, | |
| "logits/rejected": -2.0239508152008057, | |
| "logps/chosen": -294.5444641113281, | |
| "logps/rejected": -334.00042724609375, | |
| "loss": 0.5699, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.3539494276046753, | |
| "rewards/margins": 0.46201545000076294, | |
| "rewards/rejected": -0.8159648180007935, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20920502092050208, | |
| "grad_norm": 18.455144867656706, | |
| "learning_rate": 4.821741763807186e-07, | |
| "logits/chosen": -1.0050065517425537, | |
| "logits/rejected": -0.8190995454788208, | |
| "logps/chosen": -317.0809631347656, | |
| "logps/rejected": -373.81805419921875, | |
| "loss": 0.5655, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.5671249628067017, | |
| "rewards/margins": 0.6001720428466797, | |
| "rewards/rejected": -1.1672970056533813, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.20920502092050208, | |
| "eval_logits/chosen": -0.6500550508499146, | |
| "eval_logits/rejected": -0.4365238547325134, | |
| "eval_logps/chosen": -334.2371520996094, | |
| "eval_logps/rejected": -392.31927490234375, | |
| "eval_loss": 0.5745880603790283, | |
| "eval_rewards/accuracies": 0.70703125, | |
| "eval_rewards/chosen": -0.7160719633102417, | |
| "eval_rewards/margins": 0.5804991126060486, | |
| "eval_rewards/rejected": -1.2965710163116455, | |
| "eval_runtime": 110.283, | |
| "eval_samples_per_second": 18.135, | |
| "eval_steps_per_second": 0.29, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2301255230125523, | |
| "grad_norm": 17.43281841430265, | |
| "learning_rate": 4.747874028753375e-07, | |
| "logits/chosen": -0.7553631067276001, | |
| "logits/rejected": -0.2396622896194458, | |
| "logps/chosen": -366.02532958984375, | |
| "logps/rejected": -376.58050537109375, | |
| "loss": 0.568, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.61496901512146, | |
| "rewards/margins": 0.5723368525505066, | |
| "rewards/rejected": -1.1873060464859009, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2510460251046025, | |
| "grad_norm": 19.437266977582095, | |
| "learning_rate": 4.662012913161997e-07, | |
| "logits/chosen": -0.5649857521057129, | |
| "logits/rejected": -0.25259464979171753, | |
| "logps/chosen": -341.97308349609375, | |
| "logps/rejected": -376.1507568359375, | |
| "loss": 0.5592, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.6079371571540833, | |
| "rewards/margins": 0.5546227693557739, | |
| "rewards/rejected": -1.162559986114502, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2719665271966527, | |
| "grad_norm": 25.375107282483974, | |
| "learning_rate": 4.5646165232345103e-07, | |
| "logits/chosen": -0.06077578663825989, | |
| "logits/rejected": 0.2923804521560669, | |
| "logps/chosen": -373.219482421875, | |
| "logps/rejected": -418.9559631347656, | |
| "loss": 0.5347, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.9123528599739075, | |
| "rewards/margins": 0.592673659324646, | |
| "rewards/rejected": -1.5050265789031982, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2928870292887029, | |
| "grad_norm": 21.746466725148508, | |
| "learning_rate": 4.456204510851956e-07, | |
| "logits/chosen": -0.2530488073825836, | |
| "logits/rejected": 0.0992981344461441, | |
| "logps/chosen": -384.7760009765625, | |
| "logps/rejected": -437.4297790527344, | |
| "loss": 0.5317, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.6880978345870972, | |
| "rewards/margins": 0.6797319650650024, | |
| "rewards/rejected": -1.3678295612335205, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3138075313807531, | |
| "grad_norm": 26.091710204113372, | |
| "learning_rate": 4.337355301007335e-07, | |
| "logits/chosen": 0.01746506430208683, | |
| "logits/rejected": 0.5026353597640991, | |
| "logps/chosen": -387.0420227050781, | |
| "logps/rejected": -434.38580322265625, | |
| "loss": 0.5459, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.0380537509918213, | |
| "rewards/margins": 0.5438278913497925, | |
| "rewards/rejected": -1.5818815231323242, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33472803347280333, | |
| "grad_norm": 23.15637544568347, | |
| "learning_rate": 4.2087030056579986e-07, | |
| "logits/chosen": -0.19735758006572723, | |
| "logits/rejected": 0.6574625968933105, | |
| "logps/chosen": -339.94073486328125, | |
| "logps/rejected": -404.662109375, | |
| "loss": 0.5375, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.668976902961731, | |
| "rewards/margins": 0.8034380078315735, | |
| "rewards/rejected": -1.4724147319793701, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.35564853556485354, | |
| "grad_norm": 26.27445620588168, | |
| "learning_rate": 4.070934040463998e-07, | |
| "logits/chosen": 0.8700563311576843, | |
| "logits/rejected": 1.3576207160949707, | |
| "logps/chosen": -358.09051513671875, | |
| "logps/rejected": -400.4471740722656, | |
| "loss": 0.5423, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -1.0877528190612793, | |
| "rewards/margins": 0.6053714156150818, | |
| "rewards/rejected": -1.6931244134902954, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.37656903765690375, | |
| "grad_norm": 36.79313621128292, | |
| "learning_rate": 3.9247834624635404e-07, | |
| "logits/chosen": 1.2770946025848389, | |
| "logits/rejected": 1.9397356510162354, | |
| "logps/chosen": -342.9526672363281, | |
| "logps/rejected": -384.0509338378906, | |
| "loss": 0.5076, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.8662456274032593, | |
| "rewards/margins": 0.6985085010528564, | |
| "rewards/rejected": -1.5647542476654053, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.39748953974895396, | |
| "grad_norm": 23.134859194828028, | |
| "learning_rate": 3.7710310482256523e-07, | |
| "logits/chosen": 0.7933472990989685, | |
| "logits/rejected": 1.352903127670288, | |
| "logps/chosen": -354.90484619140625, | |
| "logps/rejected": -407.89801025390625, | |
| "loss": 0.5173, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.8360937237739563, | |
| "rewards/margins": 0.6451088190078735, | |
| "rewards/rejected": -1.481202483177185, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.41841004184100417, | |
| "grad_norm": 51.000005370435, | |
| "learning_rate": 3.610497133404795e-07, | |
| "logits/chosen": 0.7528682947158813, | |
| "logits/rejected": 1.2566933631896973, | |
| "logps/chosen": -360.9665832519531, | |
| "logps/rejected": -426.8341369628906, | |
| "loss": 0.5457, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.0107109546661377, | |
| "rewards/margins": 0.7371687293052673, | |
| "rewards/rejected": -1.7478797435760498, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.41841004184100417, | |
| "eval_logits/chosen": 0.3159633278846741, | |
| "eval_logits/rejected": 0.9213072061538696, | |
| "eval_logps/chosen": -361.0535888671875, | |
| "eval_logps/rejected": -441.9471740722656, | |
| "eval_loss": 0.5249255299568176, | |
| "eval_rewards/accuracies": 0.77734375, | |
| "eval_rewards/chosen": -0.9842361211776733, | |
| "eval_rewards/margins": 0.808614194393158, | |
| "eval_rewards/rejected": -1.7928502559661865, | |
| "eval_runtime": 111.6649, | |
| "eval_samples_per_second": 17.911, | |
| "eval_steps_per_second": 0.287, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.4393305439330544, | |
| "grad_norm": 25.554968491047823, | |
| "learning_rate": 3.4440382358952115e-07, | |
| "logits/chosen": 0.3864581882953644, | |
| "logits/rejected": 1.292891263961792, | |
| "logps/chosen": -393.22100830078125, | |
| "logps/rejected": -420.07012939453125, | |
| "loss": 0.5389, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -1.0355665683746338, | |
| "rewards/margins": 0.652549684047699, | |
| "rewards/rejected": -1.6881163120269775, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.4602510460251046, | |
| "grad_norm": 29.830177463941375, | |
| "learning_rate": 3.272542485937368e-07, | |
| "logits/chosen": 0.6986919641494751, | |
| "logits/rejected": 1.5640161037445068, | |
| "logps/chosen": -370.61199951171875, | |
| "logps/rejected": -428.12554931640625, | |
| "loss": 0.5263, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -1.0523474216461182, | |
| "rewards/margins": 0.7719280123710632, | |
| "rewards/rejected": -1.8242753744125366, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4811715481171548, | |
| "grad_norm": 21.180298956536415, | |
| "learning_rate": 3.096924887558854e-07, | |
| "logits/chosen": 0.7276783585548401, | |
| "logits/rejected": 1.484283447265625, | |
| "logps/chosen": -353.75164794921875, | |
| "logps/rejected": -443.76275634765625, | |
| "loss": 0.5424, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -1.0479848384857178, | |
| "rewards/margins": 0.9119135737419128, | |
| "rewards/rejected": -1.959898591041565, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.502092050209205, | |
| "grad_norm": 23.974967302833218, | |
| "learning_rate": 2.9181224366319943e-07, | |
| "logits/chosen": 0.2961747646331787, | |
| "logits/rejected": 1.008892297744751, | |
| "logps/chosen": -349.42010498046875, | |
| "logps/rejected": -405.09637451171875, | |
| "loss": 0.5026, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.8568391799926758, | |
| "rewards/margins": 0.7035945057868958, | |
| "rewards/rejected": -1.5604336261749268, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5230125523012552, | |
| "grad_norm": 28.98490083610951, | |
| "learning_rate": 2.7370891215954565e-07, | |
| "logits/chosen": 1.0544288158416748, | |
| "logits/rejected": 1.983277678489685, | |
| "logps/chosen": -397.53851318359375, | |
| "logps/rejected": -447.08734130859375, | |
| "loss": 0.5082, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.9933551549911499, | |
| "rewards/margins": 0.8820589780807495, | |
| "rewards/rejected": -1.8754141330718994, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5439330543933054, | |
| "grad_norm": 34.570387827045344, | |
| "learning_rate": 2.55479083351317e-07, | |
| "logits/chosen": 1.1669882535934448, | |
| "logits/rejected": 2.0455715656280518, | |
| "logps/chosen": -404.6676940917969, | |
| "logps/rejected": -453.14019775390625, | |
| "loss": 0.5115, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -1.1036981344223022, | |
| "rewards/margins": 0.8884363174438477, | |
| "rewards/rejected": -1.992134690284729, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5648535564853556, | |
| "grad_norm": 23.541591723414815, | |
| "learning_rate": 2.3722002126275822e-07, | |
| "logits/chosen": 0.38921916484832764, | |
| "logits/rejected": 1.070547342300415, | |
| "logps/chosen": -363.3959655761719, | |
| "logps/rejected": -411.51025390625, | |
| "loss": 0.5271, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.8641554117202759, | |
| "rewards/margins": 0.6744648814201355, | |
| "rewards/rejected": -1.5386203527450562, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5857740585774058, | |
| "grad_norm": 25.076270854562903, | |
| "learning_rate": 2.19029145890313e-07, | |
| "logits/chosen": 0.5906327962875366, | |
| "logits/rejected": 1.5835926532745361, | |
| "logps/chosen": -364.9591369628906, | |
| "logps/rejected": -425.6160583496094, | |
| "loss": 0.53, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.0323399305343628, | |
| "rewards/margins": 0.8016790151596069, | |
| "rewards/rejected": -1.8340189456939697, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.606694560669456, | |
| "grad_norm": 26.744088332600196, | |
| "learning_rate": 2.0100351342479216e-07, | |
| "logits/chosen": 0.8191035985946655, | |
| "logits/rejected": 1.6216096878051758, | |
| "logps/chosen": -373.52532958984375, | |
| "logps/rejected": -437.02728271484375, | |
| "loss": 0.5156, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.2243343591690063, | |
| "rewards/margins": 0.7263683080673218, | |
| "rewards/rejected": -1.9507026672363281, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6276150627615062, | |
| "grad_norm": 26.465017968091658, | |
| "learning_rate": 1.8323929841460178e-07, | |
| "logits/chosen": 0.5103687047958374, | |
| "logits/rejected": 1.6966192722320557, | |
| "logps/chosen": -411.9766540527344, | |
| "logps/rejected": -446.80499267578125, | |
| "loss": 0.4955, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.1094756126403809, | |
| "rewards/margins": 0.7578223943710327, | |
| "rewards/rejected": -1.867297887802124, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6276150627615062, | |
| "eval_logits/chosen": 0.4705168604850769, | |
| "eval_logits/rejected": 1.2340543270111084, | |
| "eval_logps/chosen": -359.1224060058594, | |
| "eval_logps/rejected": -449.2220153808594, | |
| "eval_loss": 0.5091525912284851, | |
| "eval_rewards/accuracies": 0.765625, | |
| "eval_rewards/chosen": -0.9649244546890259, | |
| "eval_rewards/margins": 0.9006747603416443, | |
| "eval_rewards/rejected": -1.865599274635315, | |
| "eval_runtime": 113.4568, | |
| "eval_samples_per_second": 17.628, | |
| "eval_steps_per_second": 0.282, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6485355648535565, | |
| "grad_norm": 29.87262228525191, | |
| "learning_rate": 1.6583128063291573e-07, | |
| "logits/chosen": 0.7746948003768921, | |
| "logits/rejected": 1.4845497608184814, | |
| "logps/chosen": -408.56512451171875, | |
| "logps/rejected": -445.75543212890625, | |
| "loss": 0.4961, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.1326206922531128, | |
| "rewards/margins": 0.7399401664733887, | |
| "rewards/rejected": -1.8725608587265015, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.6694560669456067, | |
| "grad_norm": 31.16422293780737, | |
| "learning_rate": 1.488723393865766e-07, | |
| "logits/chosen": 1.0418922901153564, | |
| "logits/rejected": 2.0347073078155518, | |
| "logps/chosen": -416.1197204589844, | |
| "logps/rejected": -444.3402404785156, | |
| "loss": 0.49, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.2240684032440186, | |
| "rewards/margins": 0.7707505226135254, | |
| "rewards/rejected": -1.9948190450668335, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6903765690376569, | |
| "grad_norm": 32.83677496957721, | |
| "learning_rate": 1.3245295796480788e-07, | |
| "logits/chosen": 0.1572292298078537, | |
| "logits/rejected": 1.1684186458587646, | |
| "logps/chosen": -365.3023986816406, | |
| "logps/rejected": -434.7137145996094, | |
| "loss": 0.4983, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.0306357145309448, | |
| "rewards/margins": 0.7212126851081848, | |
| "rewards/rejected": -1.7518484592437744, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7112970711297071, | |
| "grad_norm": 34.84216926965088, | |
| "learning_rate": 1.1666074087171627e-07, | |
| "logits/chosen": 0.42324066162109375, | |
| "logits/rejected": 1.437130093574524, | |
| "logps/chosen": -386.4037170410156, | |
| "logps/rejected": -480.5757751464844, | |
| "loss": 0.5021, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.963259220123291, | |
| "rewards/margins": 1.04145348072052, | |
| "rewards/rejected": -2.0047125816345215, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7322175732217573, | |
| "grad_norm": 26.071760002983055, | |
| "learning_rate": 1.0157994641835734e-07, | |
| "logits/chosen": 0.6516727209091187, | |
| "logits/rejected": 1.5883102416992188, | |
| "logps/chosen": -368.0363464355469, | |
| "logps/rejected": -438.5006408691406, | |
| "loss": 0.4783, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.098336935043335, | |
| "rewards/margins": 0.8796666264533997, | |
| "rewards/rejected": -1.9780035018920898, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7531380753138075, | |
| "grad_norm": 26.502609026351305, | |
| "learning_rate": 8.729103716819111e-08, | |
| "logits/chosen": 0.5831906199455261, | |
| "logits/rejected": 1.8786232471466064, | |
| "logps/chosen": -415.86541748046875, | |
| "logps/rejected": -466.9419860839844, | |
| "loss": 0.5141, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.1316957473754883, | |
| "rewards/margins": 0.9251440763473511, | |
| "rewards/rejected": -2.056839942932129, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.7740585774058577, | |
| "grad_norm": 28.539635727706617, | |
| "learning_rate": 7.387025063449081e-08, | |
| "logits/chosen": 1.116403341293335, | |
| "logits/rejected": 1.9285640716552734, | |
| "logps/chosen": -403.40655517578125, | |
| "logps/rejected": -438.885498046875, | |
| "loss": 0.5091, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -1.280707597732544, | |
| "rewards/margins": 0.7252805829048157, | |
| "rewards/rejected": -2.005988121032715, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7949790794979079, | |
| "grad_norm": 26.46363151511675, | |
| "learning_rate": 6.138919252022435e-08, | |
| "logits/chosen": 1.3486121892929077, | |
| "logits/rejected": 2.0518882274627686, | |
| "logps/chosen": -381.86395263671875, | |
| "logps/rejected": -493.0929260253906, | |
| "loss": 0.5042, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.423814058303833, | |
| "rewards/margins": 0.9205785989761353, | |
| "rewards/rejected": -2.344392776489258, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8158995815899581, | |
| "grad_norm": 23.582088705423008, | |
| "learning_rate": 4.991445467064689e-08, | |
| "logits/chosen": 0.8774948120117188, | |
| "logits/rejected": 1.5052356719970703, | |
| "logps/chosen": -426.742919921875, | |
| "logps/rejected": -485.6712341308594, | |
| "loss": 0.4942, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -1.211219310760498, | |
| "rewards/margins": 0.8056495785713196, | |
| "rewards/rejected": -2.016869068145752, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8368200836820083, | |
| "grad_norm": 25.52262025467119, | |
| "learning_rate": 3.9507259776993954e-08, | |
| "logits/chosen": 0.8794261813163757, | |
| "logits/rejected": 1.9908300638198853, | |
| "logps/chosen": -388.5022277832031, | |
| "logps/rejected": -473.42755126953125, | |
| "loss": 0.5033, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.17954683303833, | |
| "rewards/margins": 0.8852158784866333, | |
| "rewards/rejected": -2.064762592315674, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8368200836820083, | |
| "eval_logits/chosen": 0.5144522190093994, | |
| "eval_logits/rejected": 1.390769600868225, | |
| "eval_logps/chosen": -362.76513671875, | |
| "eval_logps/rejected": -458.7183837890625, | |
| "eval_loss": 0.5032494068145752, | |
| "eval_rewards/accuracies": 0.7734375, | |
| "eval_rewards/chosen": -1.0013517141342163, | |
| "eval_rewards/margins": 0.9592112302780151, | |
| "eval_rewards/rejected": -1.9605629444122314, | |
| "eval_runtime": 104.6004, | |
| "eval_samples_per_second": 19.12, | |
| "eval_steps_per_second": 0.306, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8577405857740585, | |
| "grad_norm": 24.81847546133683, | |
| "learning_rate": 3.022313472693447e-08, | |
| "logits/chosen": 0.6560511589050293, | |
| "logits/rejected": 1.8641235828399658, | |
| "logps/chosen": -406.5023193359375, | |
| "logps/rejected": -463.3063049316406, | |
| "loss": 0.5077, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -1.0756852626800537, | |
| "rewards/margins": 0.8691245913505554, | |
| "rewards/rejected": -1.9448096752166748, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.8786610878661087, | |
| "grad_norm": 25.68538032885617, | |
| "learning_rate": 2.2111614344599684e-08, | |
| "logits/chosen": 0.39021772146224976, | |
| "logits/rejected": 1.3714463710784912, | |
| "logps/chosen": -403.7474670410156, | |
| "logps/rejected": -460.71478271484375, | |
| "loss": 0.4954, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -1.066702961921692, | |
| "rewards/margins": 0.7960401773452759, | |
| "rewards/rejected": -1.8627431392669678, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.899581589958159, | |
| "grad_norm": 28.26563566454169, | |
| "learning_rate": 1.521597710086439e-08, | |
| "logits/chosen": 0.8775579333305359, | |
| "logits/rejected": 1.8475208282470703, | |
| "logps/chosen": -390.8741149902344, | |
| "logps/rejected": -458.987060546875, | |
| "loss": 0.4745, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -1.1183277368545532, | |
| "rewards/margins": 0.8938654661178589, | |
| "rewards/rejected": -2.012192964553833, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9205020920502092, | |
| "grad_norm": 28.547233961480817, | |
| "learning_rate": 9.57301420397924e-09, | |
| "logits/chosen": 0.518905520439148, | |
| "logits/rejected": 1.5218467712402344, | |
| "logps/chosen": -385.42559814453125, | |
| "logps/rejected": -450.37603759765625, | |
| "loss": 0.5019, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.0109506845474243, | |
| "rewards/margins": 0.8017943501472473, | |
| "rewards/rejected": -1.8127450942993164, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.9414225941422594, | |
| "grad_norm": 23.78987065435005, | |
| "learning_rate": 5.212833302556258e-09, | |
| "logits/chosen": 0.5571370124816895, | |
| "logits/rejected": 1.4383113384246826, | |
| "logps/chosen": -402.0091552734375, | |
| "logps/rejected": -497.11376953125, | |
| "loss": 0.4972, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -1.0828384160995483, | |
| "rewards/margins": 0.8171746134757996, | |
| "rewards/rejected": -1.9000129699707031, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9623430962343096, | |
| "grad_norm": 25.601990523329984, | |
| "learning_rate": 2.158697848236607e-09, | |
| "logits/chosen": 0.730278491973877, | |
| "logits/rejected": 1.7400062084197998, | |
| "logps/chosen": -374.495361328125, | |
| "logps/rejected": -426.7510681152344, | |
| "loss": 0.5018, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -1.025061011314392, | |
| "rewards/margins": 0.8159972429275513, | |
| "rewards/rejected": -1.8410581350326538, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.9832635983263598, | |
| "grad_norm": 25.12747739082461, | |
| "learning_rate": 4.269029751107489e-10, | |
| "logits/chosen": 0.5695275068283081, | |
| "logits/rejected": 1.7241246700286865, | |
| "logps/chosen": -380.4364013671875, | |
| "logps/rejected": -468.3048400878906, | |
| "loss": 0.4924, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.0173852443695068, | |
| "rewards/margins": 0.8419061899185181, | |
| "rewards/rejected": -1.859291434288025, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 478, | |
| "total_flos": 0.0, | |
| "train_loss": 0.5406385645208, | |
| "train_runtime": 10904.9554, | |
| "train_samples_per_second": 5.606, | |
| "train_steps_per_second": 0.044 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 478, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "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 | |
| } | |