{ "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": 9.102261763693875, "learning_rate": 1.0416666666666666e-08, "logits/chosen": -2.598843574523926, "logits/rejected": -2.5780863761901855, "logps/chosen": -269.870849609375, "logps/rejected": -361.686767578125, "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": 9.071970452275611, "learning_rate": 1.0416666666666667e-07, "logits/chosen": -2.4106454849243164, "logits/rejected": -2.3727071285247803, "logps/chosen": -269.8708801269531, "logps/rejected": -254.9169158935547, "loss": 0.6932, "rewards/accuracies": 0.4236111044883728, "rewards/chosen": 3.945813296013512e-05, "rewards/margins": -0.00021239757188595831, "rewards/rejected": 0.0002518556430004537, "step": 10 }, { "epoch": 0.04184100418410042, "grad_norm": 10.775263417671352, "learning_rate": 2.0833333333333333e-07, "logits/chosen": -2.4717860221862793, "logits/rejected": -2.407602310180664, "logps/chosen": -284.3338928222656, "logps/rejected": -297.76251220703125, "loss": 0.6925, "rewards/accuracies": 0.5249999761581421, "rewards/chosen": 0.004285941831767559, "rewards/margins": 0.0006065635825507343, "rewards/rejected": 0.00367937795817852, "step": 20 }, { "epoch": 0.06276150627615062, "grad_norm": 7.642238787668073, "learning_rate": 3.1249999999999997e-07, "logits/chosen": -2.5130527019500732, "logits/rejected": -2.4189674854278564, "logps/chosen": -302.9613952636719, "logps/rejected": -263.80633544921875, "loss": 0.6894, "rewards/accuracies": 0.6000000238418579, "rewards/chosen": 0.024250635877251625, "rewards/margins": 0.0054460749961435795, "rewards/rejected": 0.018804559484124184, "step": 30 }, { "epoch": 0.08368200836820083, "grad_norm": 7.769592061468885, "learning_rate": 4.1666666666666667e-07, "logits/chosen": -2.4025068283081055, "logits/rejected": -2.3595290184020996, "logps/chosen": -268.82928466796875, "logps/rejected": -243.6490936279297, "loss": 0.6827, "rewards/accuracies": 0.737500011920929, "rewards/chosen": 0.050604142248630524, "rewards/margins": 0.028540844097733498, "rewards/rejected": 0.022063296288251877, "step": 40 }, { "epoch": 0.10460251046025104, "grad_norm": 7.836365186344556, "learning_rate": 4.999733114418725e-07, "logits/chosen": -2.3352060317993164, "logits/rejected": -2.3020994663238525, "logps/chosen": -283.0472717285156, "logps/rejected": -297.33740234375, "loss": 0.6733, "rewards/accuracies": 0.637499988079071, "rewards/chosen": 0.014853554777801037, "rewards/margins": 0.029812458902597427, "rewards/rejected": -0.014958900399506092, "step": 50 }, { "epoch": 0.12552301255230125, "grad_norm": 15.966435105260237, "learning_rate": 4.990398100856366e-07, "logits/chosen": -2.396578311920166, "logits/rejected": -2.329003095626831, "logps/chosen": -274.9017333984375, "logps/rejected": -308.89105224609375, "loss": 0.6527, "rewards/accuracies": 0.699999988079071, "rewards/chosen": 0.003318754257634282, "rewards/margins": 0.09059403091669083, "rewards/rejected": -0.08727528154850006, "step": 60 }, { "epoch": 0.14644351464435146, "grad_norm": 9.243487328642475, "learning_rate": 4.967775735898179e-07, "logits/chosen": -2.244589328765869, "logits/rejected": -2.24755859375, "logps/chosen": -278.61004638671875, "logps/rejected": -280.5500183105469, "loss": 0.6348, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -0.08607344329357147, "rewards/margins": 0.11931805312633514, "rewards/rejected": -0.20539149641990662, "step": 70 }, { "epoch": 0.16736401673640167, "grad_norm": 14.953704290427632, "learning_rate": 4.931986719649298e-07, "logits/chosen": -2.3784019947052, "logits/rejected": -2.2930545806884766, "logps/chosen": -330.93719482421875, "logps/rejected": -288.47344970703125, "loss": 0.6192, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -0.03501243516802788, "rewards/margins": 0.18967404961585999, "rewards/rejected": -0.22468650341033936, "step": 80 }, { "epoch": 0.18828451882845187, "grad_norm": 15.867488807966232, "learning_rate": 4.883222001996351e-07, "logits/chosen": -2.0707411766052246, "logits/rejected": -1.9940516948699951, "logps/chosen": -288.53009033203125, "logps/rejected": -311.29840087890625, "loss": 0.6004, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -0.2625655233860016, "rewards/margins": 0.3136993646621704, "rewards/rejected": -0.5762647986412048, "step": 90 }, { "epoch": 0.20920502092050208, "grad_norm": 13.25483251042058, "learning_rate": 4.821741763807186e-07, "logits/chosen": -1.913400411605835, "logits/rejected": -1.7971922159194946, "logps/chosen": -292.3438415527344, "logps/rejected": -316.77410888671875, "loss": 0.5999, "rewards/accuracies": 0.6875, "rewards/chosen": -0.2575649917125702, "rewards/margins": 0.323893278837204, "rewards/rejected": -0.5814582705497742, "step": 100 }, { "epoch": 0.20920502092050208, "eval_logits/chosen": -1.794881820678711, "eval_logits/rejected": -1.6762588024139404, "eval_logps/chosen": -302.7283935546875, "eval_logps/rejected": -339.30548095703125, "eval_loss": 0.5966345071792603, "eval_rewards/accuracies": 0.7109375, "eval_rewards/chosen": -0.37388285994529724, "eval_rewards/margins": 0.371126264333725, "eval_rewards/rejected": -0.7450091242790222, "eval_runtime": 53.5348, "eval_samples_per_second": 37.359, "eval_steps_per_second": 0.598, "step": 100 }, { "epoch": 0.2301255230125523, "grad_norm": 15.2172693582291, "learning_rate": 4.747874028753375e-07, "logits/chosen": -1.628220796585083, "logits/rejected": -1.3428412675857544, "logps/chosen": -351.1053466796875, "logps/rejected": -339.97393798828125, "loss": 0.5893, "rewards/accuracies": 0.699999988079071, "rewards/chosen": -0.41141095757484436, "rewards/margins": 0.3850555717945099, "rewards/rejected": -0.7964665293693542, "step": 110 }, { "epoch": 0.2510460251046025, "grad_norm": 15.83081907349326, "learning_rate": 4.662012913161997e-07, "logits/chosen": -0.7602194547653198, "logits/rejected": -0.5894412994384766, "logps/chosen": -322.77508544921875, "logps/rejected": -345.4307861328125, "loss": 0.5779, "rewards/accuracies": 0.75, "rewards/chosen": -0.3538314998149872, "rewards/margins": 0.45046257972717285, "rewards/rejected": -0.8042941093444824, "step": 120 }, { "epoch": 0.2719665271966527, "grad_norm": 15.055402351260229, "learning_rate": 4.5646165232345103e-07, "logits/chosen": -0.357295423746109, "logits/rejected": -0.0887320414185524, "logps/chosen": -332.0826110839844, "logps/rejected": -368.31256103515625, "loss": 0.549, "rewards/accuracies": 0.75, "rewards/chosen": -0.441201388835907, "rewards/margins": 0.5332251787185669, "rewards/rejected": -0.9744265675544739, "step": 130 }, { "epoch": 0.2928870292887029, "grad_norm": 18.351529760698416, "learning_rate": 4.456204510851956e-07, "logits/chosen": -0.05173628777265549, "logits/rejected": 0.2410668432712555, "logps/chosen": -356.73333740234375, "logps/rejected": -390.5004577636719, "loss": 0.5555, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.37852978706359863, "rewards/margins": 0.49712079763412476, "rewards/rejected": -0.8756505846977234, "step": 140 }, { "epoch": 0.3138075313807531, "grad_norm": 22.63597752112208, "learning_rate": 4.337355301007335e-07, "logits/chosen": 0.6089886426925659, "logits/rejected": 0.9238691329956055, "logps/chosen": -365.7376403808594, "logps/rejected": -402.1335144042969, "loss": 0.5568, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -0.7808293104171753, "rewards/margins": 0.449969619512558, "rewards/rejected": -1.2307989597320557, "step": 150 }, { "epoch": 0.33472803347280333, "grad_norm": 20.03187629205258, "learning_rate": 4.2087030056579986e-07, "logits/chosen": 0.2681460976600647, "logits/rejected": 0.93768310546875, "logps/chosen": -325.05755615234375, "logps/rejected": -369.85015869140625, "loss": 0.5599, "rewards/accuracies": 0.75, "rewards/chosen": -0.4498773515224457, "rewards/margins": 0.700762927532196, "rewards/rejected": -1.1506402492523193, "step": 160 }, { "epoch": 0.35564853556485354, "grad_norm": 25.495361660969547, "learning_rate": 4.070934040463998e-07, "logits/chosen": 0.556589663028717, "logits/rejected": 0.9764513969421387, "logps/chosen": -326.57281494140625, "logps/rejected": -349.85894775390625, "loss": 0.5672, "rewards/accuracies": 0.65625, "rewards/chosen": -0.7536547780036926, "rewards/margins": 0.4030492901802063, "rewards/rejected": -1.1567041873931885, "step": 170 }, { "epoch": 0.37656903765690375, "grad_norm": 24.47540687241319, "learning_rate": 3.9247834624635404e-07, "logits/chosen": 1.273604393005371, "logits/rejected": 1.7295089960098267, "logps/chosen": -344.38751220703125, "logps/rejected": -365.9759521484375, "loss": 0.5254, "rewards/accuracies": 0.71875, "rewards/chosen": -0.8270662426948547, "rewards/margins": 0.5501136779785156, "rewards/rejected": -1.3771798610687256, "step": 180 }, { "epoch": 0.39748953974895396, "grad_norm": 22.91459265715999, "learning_rate": 3.7710310482256523e-07, "logits/chosen": 1.0068578720092773, "logits/rejected": 1.2909595966339111, "logps/chosen": -328.5067138671875, "logps/rejected": -372.14324951171875, "loss": 0.5444, "rewards/accuracies": 0.706250011920929, "rewards/chosen": -0.5530312061309814, "rewards/margins": 0.5563492178916931, "rewards/rejected": -1.1093804836273193, "step": 190 }, { "epoch": 0.41841004184100417, "grad_norm": 36.86200995874469, "learning_rate": 3.610497133404795e-07, "logits/chosen": 1.4111734628677368, "logits/rejected": 1.7813835144042969, "logps/chosen": -336.32794189453125, "logps/rejected": -386.16827392578125, "loss": 0.5606, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -0.7106108069419861, "rewards/margins": 0.5927971601486206, "rewards/rejected": -1.303407907485962, "step": 200 }, { "epoch": 0.41841004184100417, "eval_logits/chosen": 1.1101882457733154, "eval_logits/rejected": 1.5408776998519897, "eval_logps/chosen": -327.4931640625, "eval_logps/rejected": -387.65057373046875, "eval_loss": 0.5555490851402283, "eval_rewards/accuracies": 0.74609375, "eval_rewards/chosen": -0.6215302348136902, "eval_rewards/margins": 0.606930136680603, "eval_rewards/rejected": -1.228460431098938, "eval_runtime": 53.4692, "eval_samples_per_second": 37.405, "eval_steps_per_second": 0.598, "step": 200 }, { "epoch": 0.4393305439330544, "grad_norm": 19.538681466466336, "learning_rate": 3.4440382358952115e-07, "logits/chosen": 1.2745749950408936, "logits/rejected": 1.9181947708129883, "logps/chosen": -358.096435546875, "logps/rejected": -373.4177551269531, "loss": 0.5641, "rewards/accuracies": 0.75, "rewards/chosen": -0.5784404873847961, "rewards/margins": 0.6223729252815247, "rewards/rejected": -1.2008132934570312, "step": 210 }, { "epoch": 0.4602510460251046, "grad_norm": 20.86939524596463, "learning_rate": 3.272542485937368e-07, "logits/chosen": 0.8960593342781067, "logits/rejected": 1.4320571422576904, "logps/chosen": -332.91864013671875, "logps/rejected": -364.5188293457031, "loss": 0.5474, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.5551973581314087, "rewards/margins": 0.6443191170692444, "rewards/rejected": -1.1995164155960083, "step": 220 }, { "epoch": 0.4811715481171548, "grad_norm": 19.099059937166764, "learning_rate": 3.096924887558854e-07, "logits/chosen": 0.6132632493972778, "logits/rejected": 0.9618092775344849, "logps/chosen": -323.7577209472656, "logps/rejected": -385.98565673828125, "loss": 0.552, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -0.7150660753250122, "rewards/margins": 0.6639611721038818, "rewards/rejected": -1.3790271282196045, "step": 230 }, { "epoch": 0.502092050209205, "grad_norm": 17.891063709919703, "learning_rate": 2.9181224366319943e-07, "logits/chosen": 0.4524998664855957, "logits/rejected": 0.8361547589302063, "logps/chosen": -321.6592102050781, "logps/rejected": -369.956298828125, "loss": 0.5196, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -0.551439642906189, "rewards/margins": 0.6203877925872803, "rewards/rejected": -1.1718274354934692, "step": 240 }, { "epoch": 0.5230125523012552, "grad_norm": 19.34076101834313, "learning_rate": 2.7370891215954565e-07, "logits/chosen": 1.31077241897583, "logits/rejected": 1.824054479598999, "logps/chosen": -356.10235595703125, "logps/rejected": -388.8703918457031, "loss": 0.5268, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -0.48947277665138245, "rewards/margins": 0.7497965097427368, "rewards/rejected": -1.2392691373825073, "step": 250 }, { "epoch": 0.5439330543933054, "grad_norm": 26.294415278038993, "learning_rate": 2.55479083351317e-07, "logits/chosen": 1.545461893081665, "logits/rejected": 2.077069044113159, "logps/chosen": -352.08587646484375, "logps/rejected": -382.9310607910156, "loss": 0.5298, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -0.5431629419326782, "rewards/margins": 0.7132429480552673, "rewards/rejected": -1.2564058303833008, "step": 260 }, { "epoch": 0.5648535564853556, "grad_norm": 16.790572679540336, "learning_rate": 2.3722002126275822e-07, "logits/chosen": 1.6584622859954834, "logits/rejected": 2.1149163246154785, "logps/chosen": -342.0724182128906, "logps/rejected": -371.71588134765625, "loss": 0.5367, "rewards/accuracies": 0.668749988079071, "rewards/chosen": -0.615090548992157, "rewards/margins": 0.5203320980072021, "rewards/rejected": -1.1354224681854248, "step": 270 }, { "epoch": 0.5857740585774058, "grad_norm": 24.772008946191377, "learning_rate": 2.19029145890313e-07, "logits/chosen": 1.7721284627914429, "logits/rejected": 2.3066000938415527, "logps/chosen": -326.43603515625, "logps/rejected": -370.3632507324219, "loss": 0.5442, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.5788434147834778, "rewards/margins": 0.684018075466156, "rewards/rejected": -1.2628614902496338, "step": 280 }, { "epoch": 0.606694560669456, "grad_norm": 19.443360004007754, "learning_rate": 2.0100351342479216e-07, "logits/chosen": 1.2106627225875854, "logits/rejected": 1.4834706783294678, "logps/chosen": -333.8368225097656, "logps/rejected": -384.6605224609375, "loss": 0.5396, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -0.8036863207817078, "rewards/margins": 0.5939573645591736, "rewards/rejected": -1.397643804550171, "step": 290 }, { "epoch": 0.6276150627615062, "grad_norm": 24.510951465909894, "learning_rate": 1.8323929841460178e-07, "logits/chosen": 0.9158282279968262, "logits/rejected": 1.5702682733535767, "logps/chosen": -382.8146057128906, "logps/rejected": -400.11474609375, "loss": 0.5211, "rewards/accuracies": 0.6875, "rewards/chosen": -0.7668856382369995, "rewards/margins": 0.6154419183731079, "rewards/rejected": -1.382327675819397, "step": 300 }, { "epoch": 0.6276150627615062, "eval_logits/chosen": 0.7847145199775696, "eval_logits/rejected": 1.2079328298568726, "eval_logps/chosen": -329.06488037109375, "eval_logps/rejected": -404.1211242675781, "eval_loss": 0.5284228324890137, "eval_rewards/accuracies": 0.76171875, "eval_rewards/chosen": -0.6372475028038025, "eval_rewards/margins": 0.7559184432029724, "eval_rewards/rejected": -1.393165946006775, "eval_runtime": 53.4499, "eval_samples_per_second": 37.418, "eval_steps_per_second": 0.599, "step": 300 }, { "epoch": 0.6485355648535565, "grad_norm": 21.649958560460085, "learning_rate": 1.6583128063291573e-07, "logits/chosen": 0.9612062573432922, "logits/rejected": 1.3238331079483032, "logps/chosen": -379.32904052734375, "logps/rejected": -410.33685302734375, "loss": 0.5135, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -0.7942469120025635, "rewards/margins": 0.6349788904190063, "rewards/rejected": -1.4292259216308594, "step": 310 }, { "epoch": 0.6694560669456067, "grad_norm": 21.732265089014117, "learning_rate": 1.488723393865766e-07, "logits/chosen": 0.9882587194442749, "logits/rejected": 1.442718744277954, "logps/chosen": -375.4773254394531, "logps/rejected": -389.82733154296875, "loss": 0.506, "rewards/accuracies": 0.75, "rewards/chosen": -0.7661019563674927, "rewards/margins": 0.6393279433250427, "rewards/rejected": -1.4054299592971802, "step": 320 }, { "epoch": 0.6903765690376569, "grad_norm": 25.995557260279178, "learning_rate": 1.3245295796480788e-07, "logits/chosen": 0.8205612897872925, "logits/rejected": 1.306123971939087, "logps/chosen": -333.33074951171875, "logps/rejected": -386.80072021484375, "loss": 0.5192, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.5908290147781372, "rewards/margins": 0.5889186859130859, "rewards/rejected": -1.1797475814819336, "step": 330 }, { "epoch": 0.7112970711297071, "grad_norm": 34.06994351015907, "learning_rate": 1.1666074087171627e-07, "logits/chosen": 0.8795354962348938, "logits/rejected": 1.3036845922470093, "logps/chosen": -349.6972961425781, "logps/rejected": -425.7892150878906, "loss": 0.5208, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -0.5422636270523071, "rewards/margins": 0.8381665945053101, "rewards/rejected": -1.3804302215576172, "step": 340 }, { "epoch": 0.7322175732217573, "grad_norm": 21.28540346436605, "learning_rate": 1.0157994641835734e-07, "logits/chosen": 1.0639841556549072, "logits/rejected": 1.6034936904907227, "logps/chosen": -329.88690185546875, "logps/rejected": -385.4267578125, "loss": 0.4889, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.6864292621612549, "rewards/margins": 0.7708432674407959, "rewards/rejected": -1.4572727680206299, "step": 350 }, { "epoch": 0.7531380753138075, "grad_norm": 23.66473954551423, "learning_rate": 8.729103716819111e-08, "logits/chosen": 1.199178695678711, "logits/rejected": 1.9472402334213257, "logps/chosen": -380.9803161621094, "logps/rejected": -419.5332946777344, "loss": 0.5316, "rewards/accuracies": 0.768750011920929, "rewards/chosen": -0.7623031735420227, "rewards/margins": 0.7938345670700073, "rewards/rejected": -1.5561376810073853, "step": 360 }, { "epoch": 0.7740585774058577, "grad_norm": 22.814657698551834, "learning_rate": 7.387025063449081e-08, "logits/chosen": 1.5637246370315552, "logits/rejected": 2.090482234954834, "logps/chosen": -364.23126220703125, "logps/rejected": -389.804931640625, "loss": 0.5165, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.8549477458000183, "rewards/margins": 0.6235242486000061, "rewards/rejected": -1.4784719944000244, "step": 370 }, { "epoch": 0.7949790794979079, "grad_norm": 24.019268622369378, "learning_rate": 6.138919252022435e-08, "logits/chosen": 1.690513253211975, "logits/rejected": 1.896419882774353, "logps/chosen": -330.74395751953125, "logps/rejected": -429.8720703125, "loss": 0.5175, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -0.8654612302780151, "rewards/margins": 0.7997757196426392, "rewards/rejected": -1.6652368307113647, "step": 380 }, { "epoch": 0.8158995815899581, "grad_norm": 23.872152876539662, "learning_rate": 4.991445467064689e-08, "logits/chosen": 1.24801766872406, "logits/rejected": 1.5409824848175049, "logps/chosen": -374.94500732421875, "logps/rejected": -423.5159606933594, "loss": 0.5064, "rewards/accuracies": 0.731249988079071, "rewards/chosen": -0.623989462852478, "rewards/margins": 0.7279406785964966, "rewards/rejected": -1.351930022239685, "step": 390 }, { "epoch": 0.8368200836820083, "grad_norm": 22.17989276400099, "learning_rate": 3.9507259776993954e-08, "logits/chosen": 1.4913052320480347, "logits/rejected": 1.9780304431915283, "logps/chosen": -344.193359375, "logps/rejected": -412.5020446777344, "loss": 0.5189, "rewards/accuracies": 0.762499988079071, "rewards/chosen": -0.6954695582389832, "rewards/margins": 0.7355377078056335, "rewards/rejected": -1.4310071468353271, "step": 400 }, { "epoch": 0.8368200836820083, "eval_logits/chosen": 1.2049928903579712, "eval_logits/rejected": 1.687930941581726, "eval_logps/chosen": -323.4798889160156, "eval_logps/rejected": -403.6497802734375, "eval_loss": 0.5224906206130981, "eval_rewards/accuracies": 0.77734375, "eval_rewards/chosen": -0.5813976526260376, "eval_rewards/margins": 0.8070548176765442, "eval_rewards/rejected": -1.3884525299072266, "eval_runtime": 53.5131, "eval_samples_per_second": 37.374, "eval_steps_per_second": 0.598, "step": 400 }, { "epoch": 0.8577405857740585, "grad_norm": 24.78416106509579, "learning_rate": 3.022313472693447e-08, "logits/chosen": 1.401047945022583, "logits/rejected": 2.122811794281006, "logps/chosen": -370.56951904296875, "logps/rejected": -416.39080810546875, "loss": 0.5278, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -0.6760027408599854, "rewards/margins": 0.7805537581443787, "rewards/rejected": -1.4565564393997192, "step": 410 }, { "epoch": 0.8786610878661087, "grad_norm": 24.036813785272063, "learning_rate": 2.2111614344599684e-08, "logits/chosen": 1.3202136754989624, "logits/rejected": 1.759892225265503, "logps/chosen": -373.802734375, "logps/rejected": -416.2435607910156, "loss": 0.5136, "rewards/accuracies": 0.7124999761581421, "rewards/chosen": -0.7081748843193054, "rewards/margins": 0.6878335475921631, "rewards/rejected": -1.3960084915161133, "step": 420 }, { "epoch": 0.899581589958159, "grad_norm": 23.876888782252916, "learning_rate": 1.521597710086439e-08, "logits/chosen": 1.59340238571167, "logits/rejected": 2.011382579803467, "logps/chosen": -357.6289978027344, "logps/rejected": -414.678466796875, "loss": 0.4929, "rewards/accuracies": 0.7875000238418579, "rewards/chosen": -0.7411682605743408, "rewards/margins": 0.7640897631645203, "rewards/rejected": -1.5052578449249268, "step": 430 }, { "epoch": 0.9205020920502092, "grad_norm": 23.73816215215032, "learning_rate": 9.57301420397924e-09, "logits/chosen": 1.240863561630249, "logits/rejected": 1.8725379705429077, "logps/chosen": -354.1866149902344, "logps/rejected": -410.48187255859375, "loss": 0.5066, "rewards/accuracies": 0.7250000238418579, "rewards/chosen": -0.6514572501182556, "rewards/margins": 0.7045157551765442, "rewards/rejected": -1.3559730052947998, "step": 440 }, { "epoch": 0.9414225941422594, "grad_norm": 23.69382783496658, "learning_rate": 5.212833302556258e-09, "logits/chosen": 1.4327576160430908, "logits/rejected": 1.8339331150054932, "logps/chosen": -374.9490661621094, "logps/rejected": -459.17169189453125, "loss": 0.51, "rewards/accuracies": 0.6937500238418579, "rewards/chosen": -0.7898253202438354, "rewards/margins": 0.6647788286209106, "rewards/rejected": -1.454604148864746, "step": 450 }, { "epoch": 0.9623430962343096, "grad_norm": 28.599243740532902, "learning_rate": 2.158697848236607e-09, "logits/chosen": 1.528529405593872, "logits/rejected": 2.007927179336548, "logps/chosen": -352.3992004394531, "logps/rejected": -387.2112121582031, "loss": 0.5217, "rewards/accuracies": 0.7437499761581421, "rewards/chosen": -0.7364510893821716, "rewards/margins": 0.685036838054657, "rewards/rejected": -1.4214880466461182, "step": 460 }, { "epoch": 0.9832635983263598, "grad_norm": 20.454762644290014, "learning_rate": 4.269029751107489e-10, "logits/chosen": 1.380399227142334, "logits/rejected": 1.8892757892608643, "logps/chosen": -350.90447998046875, "logps/rejected": -421.0154724121094, "loss": 0.5088, "rewards/accuracies": 0.737500011920929, "rewards/chosen": -0.7164818644523621, "rewards/margins": 0.6821088194847107, "rewards/rejected": -1.3985908031463623, "step": 470 }, { "epoch": 1.0, "step": 478, "total_flos": 0.0, "train_loss": 0.5574632548886862, "train_runtime": 4290.8569, "train_samples_per_second": 14.248, "train_steps_per_second": 0.111 } ], "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 }