| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.9992254066615027, | |
| "eval_steps": 100, | |
| "global_step": 726, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 6.84931506849315e-09, | |
| "logits/chosen": -2.3569769859313965, | |
| "logits/rejected": -2.397932529449463, | |
| "logps/chosen": -275.3341369628906, | |
| "logps/rejected": -209.60626220703125, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.84931506849315e-08, | |
| "logits/chosen": -2.4118523597717285, | |
| "logits/rejected": -2.3359429836273193, | |
| "logps/chosen": -296.8043212890625, | |
| "logps/rejected": -226.9486541748047, | |
| "loss": 0.6936, | |
| "rewards/accuracies": 0.4618055522441864, | |
| "rewards/chosen": -0.0005276877782307565, | |
| "rewards/margins": -0.0006244900869205594, | |
| "rewards/rejected": 9.680193034000695e-05, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.36986301369863e-07, | |
| "logits/chosen": -2.4295315742492676, | |
| "logits/rejected": -2.379143714904785, | |
| "logps/chosen": -281.6752014160156, | |
| "logps/rejected": -217.3535919189453, | |
| "loss": 0.6938, | |
| "rewards/accuracies": 0.4906249940395355, | |
| "rewards/chosen": 0.0024631484411656857, | |
| "rewards/margins": 5.021132437832421e-06, | |
| "rewards/rejected": 0.0024581279139965773, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.054794520547945e-07, | |
| "logits/chosen": -2.385784149169922, | |
| "logits/rejected": -2.3432505130767822, | |
| "logps/chosen": -255.5972442626953, | |
| "logps/rejected": -208.0276336669922, | |
| "loss": 0.6924, | |
| "rewards/accuracies": 0.5406249761581421, | |
| "rewards/chosen": 0.0008739754557609558, | |
| "rewards/margins": 0.002002383815124631, | |
| "rewards/rejected": -0.0011284081265330315, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 2.73972602739726e-07, | |
| "logits/chosen": -2.449314832687378, | |
| "logits/rejected": -2.3985402584075928, | |
| "logps/chosen": -286.9411926269531, | |
| "logps/rejected": -217.51119995117188, | |
| "loss": 0.6923, | |
| "rewards/accuracies": 0.5218750238418579, | |
| "rewards/chosen": 0.0011719572357833385, | |
| "rewards/margins": 0.002339401515200734, | |
| "rewards/rejected": -0.0011674443958327174, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 3.424657534246575e-07, | |
| "logits/chosen": -2.4232876300811768, | |
| "logits/rejected": -2.3819985389709473, | |
| "logps/chosen": -270.04620361328125, | |
| "logps/rejected": -224.44857788085938, | |
| "loss": 0.6904, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.0045755826868116856, | |
| "rewards/margins": 0.007746423594653606, | |
| "rewards/rejected": -0.003170841606333852, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.10958904109589e-07, | |
| "logits/chosen": -2.406489133834839, | |
| "logits/rejected": -2.376723289489746, | |
| "logps/chosen": -269.7476501464844, | |
| "logps/rejected": -216.1673583984375, | |
| "loss": 0.6893, | |
| "rewards/accuracies": 0.5609375238418579, | |
| "rewards/chosen": 0.00798078440129757, | |
| "rewards/margins": 0.009231673553586006, | |
| "rewards/rejected": -0.0012508893851190805, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.794520547945205e-07, | |
| "logits/chosen": -2.3951592445373535, | |
| "logits/rejected": -2.3819260597229004, | |
| "logps/chosen": -257.29620361328125, | |
| "logps/rejected": -215.2325897216797, | |
| "loss": 0.6874, | |
| "rewards/accuracies": 0.5609375238418579, | |
| "rewards/chosen": 0.012629570439457893, | |
| "rewards/margins": 0.013305542059242725, | |
| "rewards/rejected": -0.0006759737734682858, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.946401225114854e-07, | |
| "logits/chosen": -2.4296867847442627, | |
| "logits/rejected": -2.368792772293091, | |
| "logps/chosen": -267.8879699707031, | |
| "logps/rejected": -219.56350708007812, | |
| "loss": 0.6842, | |
| "rewards/accuracies": 0.621874988079071, | |
| "rewards/chosen": 0.019007008522748947, | |
| "rewards/margins": 0.0214321780949831, | |
| "rewards/rejected": -0.002425167942419648, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.869831546707504e-07, | |
| "logits/chosen": -2.463676929473877, | |
| "logits/rejected": -2.3972015380859375, | |
| "logps/chosen": -274.6397705078125, | |
| "logps/rejected": -225.1620635986328, | |
| "loss": 0.683, | |
| "rewards/accuracies": 0.604687511920929, | |
| "rewards/chosen": 0.024952612817287445, | |
| "rewards/margins": 0.02081022970378399, | |
| "rewards/rejected": 0.004142382647842169, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.793261868300153e-07, | |
| "logits/chosen": -2.4068779945373535, | |
| "logits/rejected": -2.375128746032715, | |
| "logps/chosen": -276.1687927246094, | |
| "logps/rejected": -228.1005096435547, | |
| "loss": 0.6783, | |
| "rewards/accuracies": 0.660937488079071, | |
| "rewards/chosen": 0.03607472777366638, | |
| "rewards/margins": 0.034400396049022675, | |
| "rewards/rejected": 0.001674329163506627, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.7166921898928023e-07, | |
| "logits/chosen": -2.445211887359619, | |
| "logits/rejected": -2.3825695514678955, | |
| "logps/chosen": -254.85311889648438, | |
| "logps/rejected": -222.86117553710938, | |
| "loss": 0.6761, | |
| "rewards/accuracies": 0.6390625238418579, | |
| "rewards/chosen": 0.034680236130952835, | |
| "rewards/margins": 0.035091597586870193, | |
| "rewards/rejected": -0.00041136034997180104, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.640122511485451e-07, | |
| "logits/chosen": -2.412264585494995, | |
| "logits/rejected": -2.3916800022125244, | |
| "logps/chosen": -259.5788879394531, | |
| "logps/rejected": -220.30831909179688, | |
| "loss": 0.6715, | |
| "rewards/accuracies": 0.6421874761581421, | |
| "rewards/chosen": 0.04985843971371651, | |
| "rewards/margins": 0.050119031220674515, | |
| "rewards/rejected": -0.00026059610536322, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.563552833078101e-07, | |
| "logits/chosen": -2.4411113262176514, | |
| "logits/rejected": -2.376892566680908, | |
| "logps/chosen": -268.37335205078125, | |
| "logps/rejected": -226.0688934326172, | |
| "loss": 0.666, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.05422825738787651, | |
| "rewards/margins": 0.05518989637494087, | |
| "rewards/rejected": -0.0009616309544071555, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.4869831546707505e-07, | |
| "logits/chosen": -2.448073387145996, | |
| "logits/rejected": -2.3861663341522217, | |
| "logps/chosen": -273.23272705078125, | |
| "logps/rejected": -230.17776489257812, | |
| "loss": 0.6681, | |
| "rewards/accuracies": 0.692187488079071, | |
| "rewards/chosen": 0.06362518668174744, | |
| "rewards/margins": 0.06625331938266754, | |
| "rewards/rejected": -0.002628129906952381, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.4104134762633994e-07, | |
| "logits/chosen": -2.480207920074463, | |
| "logits/rejected": -2.4179327487945557, | |
| "logps/chosen": -268.4297790527344, | |
| "logps/rejected": -226.82815551757812, | |
| "loss": 0.6596, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.07131167501211166, | |
| "rewards/margins": 0.0752154216170311, | |
| "rewards/rejected": -0.0039037547539919615, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.333843797856049e-07, | |
| "logits/chosen": -2.438366651535034, | |
| "logits/rejected": -2.3843159675598145, | |
| "logps/chosen": -281.81756591796875, | |
| "logps/rejected": -239.0724334716797, | |
| "loss": 0.6555, | |
| "rewards/accuracies": 0.7015625238418579, | |
| "rewards/chosen": 0.08467759937047958, | |
| "rewards/margins": 0.08220230042934418, | |
| "rewards/rejected": 0.002475299406796694, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.257274119448698e-07, | |
| "logits/chosen": -2.4684793949127197, | |
| "logits/rejected": -2.433481216430664, | |
| "logps/chosen": -272.9942932128906, | |
| "logps/rejected": -239.0500946044922, | |
| "loss": 0.6542, | |
| "rewards/accuracies": 0.6859375238418579, | |
| "rewards/chosen": 0.08272445946931839, | |
| "rewards/margins": 0.08306626230478287, | |
| "rewards/rejected": -0.00034180469810962677, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 4.180704441041347e-07, | |
| "logits/chosen": -2.429438829421997, | |
| "logits/rejected": -2.378554105758667, | |
| "logps/chosen": -259.08294677734375, | |
| "logps/rejected": -225.7974853515625, | |
| "loss": 0.6565, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.08147875219583511, | |
| "rewards/margins": 0.07715155184268951, | |
| "rewards/rejected": 0.004327205941081047, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 4.1041347626339966e-07, | |
| "logits/chosen": -2.4421346187591553, | |
| "logits/rejected": -2.3960843086242676, | |
| "logps/chosen": -263.78216552734375, | |
| "logps/rejected": -212.2428436279297, | |
| "loss": 0.6485, | |
| "rewards/accuracies": 0.684374988079071, | |
| "rewards/chosen": 0.09069393575191498, | |
| "rewards/margins": 0.1026170402765274, | |
| "rewards/rejected": -0.011923106387257576, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.027565084226646e-07, | |
| "logits/chosen": -2.463709831237793, | |
| "logits/rejected": -2.404499053955078, | |
| "logps/chosen": -264.03118896484375, | |
| "logps/rejected": -217.3473358154297, | |
| "loss": 0.6415, | |
| "rewards/accuracies": 0.7093750238418579, | |
| "rewards/chosen": 0.10673630237579346, | |
| "rewards/margins": 0.1264755129814148, | |
| "rewards/rejected": -0.019739216193556786, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.9509954058192954e-07, | |
| "logits/chosen": -2.458402633666992, | |
| "logits/rejected": -2.4066162109375, | |
| "logps/chosen": -286.2770690917969, | |
| "logps/rejected": -230.6191864013672, | |
| "loss": 0.6401, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.10997898876667023, | |
| "rewards/margins": 0.12333294004201889, | |
| "rewards/rejected": -0.013353955931961536, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.874425727411945e-07, | |
| "logits/chosen": -2.419983386993408, | |
| "logits/rejected": -2.386007308959961, | |
| "logps/chosen": -278.9669494628906, | |
| "logps/rejected": -227.8882293701172, | |
| "loss": 0.6399, | |
| "rewards/accuracies": 0.692187488079071, | |
| "rewards/chosen": 0.11453185975551605, | |
| "rewards/margins": 0.13373538851737976, | |
| "rewards/rejected": -0.01920352131128311, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.797856049004594e-07, | |
| "logits/chosen": -2.4336562156677246, | |
| "logits/rejected": -2.36842679977417, | |
| "logps/chosen": -266.24053955078125, | |
| "logps/rejected": -227.7322235107422, | |
| "loss": 0.6365, | |
| "rewards/accuracies": 0.6656249761581421, | |
| "rewards/chosen": 0.12494595348834991, | |
| "rewards/margins": 0.14494453370571136, | |
| "rewards/rejected": -0.019998596981167793, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.7212863705972436e-07, | |
| "logits/chosen": -2.45171856880188, | |
| "logits/rejected": -2.4028737545013428, | |
| "logps/chosen": -273.0599365234375, | |
| "logps/rejected": -221.7194366455078, | |
| "loss": 0.6313, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.13546457886695862, | |
| "rewards/margins": 0.1584644615650177, | |
| "rewards/rejected": -0.02299986407160759, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_logits/chosen": -2.1062474250793457, | |
| "eval_logits/rejected": -1.9863277673721313, | |
| "eval_logps/chosen": -266.3520812988281, | |
| "eval_logps/rejected": -220.30364990234375, | |
| "eval_loss": 0.6318375468254089, | |
| "eval_rewards/accuracies": 0.6600000262260437, | |
| "eval_rewards/chosen": 0.12281632423400879, | |
| "eval_rewards/margins": 0.1532445251941681, | |
| "eval_rewards/rejected": -0.030428189784288406, | |
| "eval_runtime": 170.6685, | |
| "eval_samples_per_second": 11.719, | |
| "eval_steps_per_second": 0.732, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.6447166921898925e-07, | |
| "logits/chosen": -2.3959295749664307, | |
| "logits/rejected": -2.3386054039001465, | |
| "logps/chosen": -259.109619140625, | |
| "logps/rejected": -209.0337677001953, | |
| "loss": 0.6337, | |
| "rewards/accuracies": 0.684374988079071, | |
| "rewards/chosen": 0.12544557452201843, | |
| "rewards/margins": 0.1500503271818161, | |
| "rewards/rejected": -0.024604763835668564, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.568147013782542e-07, | |
| "logits/chosen": -2.4076590538024902, | |
| "logits/rejected": -2.3606529235839844, | |
| "logps/chosen": -263.9520568847656, | |
| "logps/rejected": -225.4473876953125, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.721875011920929, | |
| "rewards/chosen": 0.1394219994544983, | |
| "rewards/margins": 0.1773197054862976, | |
| "rewards/rejected": -0.03789771348237991, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.4915773353751913e-07, | |
| "logits/chosen": -2.417232036590576, | |
| "logits/rejected": -2.346893072128296, | |
| "logps/chosen": -280.55291748046875, | |
| "logps/rejected": -234.11996459960938, | |
| "loss": 0.6203, | |
| "rewards/accuracies": 0.729687511920929, | |
| "rewards/chosen": 0.1420687586069107, | |
| "rewards/margins": 0.20756664872169495, | |
| "rewards/rejected": -0.06549788266420364, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.41500765696784e-07, | |
| "logits/chosen": -2.4278924465179443, | |
| "logits/rejected": -2.3850674629211426, | |
| "logps/chosen": -253.51803588867188, | |
| "logps/rejected": -224.95968627929688, | |
| "loss": 0.6242, | |
| "rewards/accuracies": 0.6968749761581421, | |
| "rewards/chosen": 0.13101010024547577, | |
| "rewards/margins": 0.16972467303276062, | |
| "rewards/rejected": -0.03871458023786545, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 3.33843797856049e-07, | |
| "logits/chosen": -2.3946681022644043, | |
| "logits/rejected": -2.380558967590332, | |
| "logps/chosen": -284.79608154296875, | |
| "logps/rejected": -226.66262817382812, | |
| "loss": 0.6194, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": 0.13935603201389313, | |
| "rewards/margins": 0.20914654433727264, | |
| "rewards/rejected": -0.06979051232337952, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 3.2618683001531396e-07, | |
| "logits/chosen": -2.3978686332702637, | |
| "logits/rejected": -2.3249762058258057, | |
| "logps/chosen": -263.2580871582031, | |
| "logps/rejected": -226.63101196289062, | |
| "loss": 0.6169, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": 0.14139556884765625, | |
| "rewards/margins": 0.1872793436050415, | |
| "rewards/rejected": -0.04588378220796585, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 3.1852986217457885e-07, | |
| "logits/chosen": -2.4475905895233154, | |
| "logits/rejected": -2.363696575164795, | |
| "logps/chosen": -262.8795471191406, | |
| "logps/rejected": -227.06930541992188, | |
| "loss": 0.619, | |
| "rewards/accuracies": 0.7203124761581421, | |
| "rewards/chosen": 0.15051104128360748, | |
| "rewards/margins": 0.1968255192041397, | |
| "rewards/rejected": -0.04631447046995163, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 3.108728943338438e-07, | |
| "logits/chosen": -2.438253164291382, | |
| "logits/rejected": -2.421774387359619, | |
| "logps/chosen": -253.95639038085938, | |
| "logps/rejected": -236.4005584716797, | |
| "loss": 0.6208, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": 0.13959024846553802, | |
| "rewards/margins": 0.1706792563199997, | |
| "rewards/rejected": -0.031088998541235924, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 3.0321592649310873e-07, | |
| "logits/chosen": -2.4183051586151123, | |
| "logits/rejected": -2.339179039001465, | |
| "logps/chosen": -254.692626953125, | |
| "logps/rejected": -222.8722381591797, | |
| "loss": 0.6141, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.1566968411207199, | |
| "rewards/margins": 0.19593168795108795, | |
| "rewards/rejected": -0.03923482820391655, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.955589586523736e-07, | |
| "logits/chosen": -2.4377613067626953, | |
| "logits/rejected": -2.366072177886963, | |
| "logps/chosen": -267.422607421875, | |
| "logps/rejected": -224.8284149169922, | |
| "loss": 0.6188, | |
| "rewards/accuracies": 0.6890624761581421, | |
| "rewards/chosen": 0.14487192034721375, | |
| "rewards/margins": 0.2071552723646164, | |
| "rewards/rejected": -0.06228336691856384, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.8790199081163856e-07, | |
| "logits/chosen": -2.435045003890991, | |
| "logits/rejected": -2.3581719398498535, | |
| "logps/chosen": -267.0567626953125, | |
| "logps/rejected": -223.5650177001953, | |
| "loss": 0.6136, | |
| "rewards/accuracies": 0.6890624761581421, | |
| "rewards/chosen": 0.1428806185722351, | |
| "rewards/margins": 0.20400968194007874, | |
| "rewards/rejected": -0.06112906336784363, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.802450229709035e-07, | |
| "logits/chosen": -2.3934175968170166, | |
| "logits/rejected": -2.3315463066101074, | |
| "logps/chosen": -256.1235656738281, | |
| "logps/rejected": -211.8139190673828, | |
| "loss": 0.6093, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": 0.14465923607349396, | |
| "rewards/margins": 0.20695683360099792, | |
| "rewards/rejected": -0.06229761987924576, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.725880551301684e-07, | |
| "logits/chosen": -2.4487271308898926, | |
| "logits/rejected": -2.3933699131011963, | |
| "logps/chosen": -275.8642578125, | |
| "logps/rejected": -231.2057342529297, | |
| "loss": 0.6129, | |
| "rewards/accuracies": 0.692187488079071, | |
| "rewards/chosen": 0.14190760254859924, | |
| "rewards/margins": 0.22471928596496582, | |
| "rewards/rejected": -0.08281168341636658, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.649310872894334e-07, | |
| "logits/chosen": -2.4377894401550293, | |
| "logits/rejected": -2.406348466873169, | |
| "logps/chosen": -276.6672058105469, | |
| "logps/rejected": -221.1224365234375, | |
| "loss": 0.603, | |
| "rewards/accuracies": 0.7093750238418579, | |
| "rewards/chosen": 0.15706932544708252, | |
| "rewards/margins": 0.24053001403808594, | |
| "rewards/rejected": -0.08346069604158401, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.572741194486983e-07, | |
| "logits/chosen": -2.4386019706726074, | |
| "logits/rejected": -2.3754172325134277, | |
| "logps/chosen": -274.6893005371094, | |
| "logps/rejected": -206.1323699951172, | |
| "loss": 0.6004, | |
| "rewards/accuracies": 0.715624988079071, | |
| "rewards/chosen": 0.167318657040596, | |
| "rewards/margins": 0.25797349214553833, | |
| "rewards/rejected": -0.09065485745668411, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.496171516079632e-07, | |
| "logits/chosen": -2.4488420486450195, | |
| "logits/rejected": -2.3803772926330566, | |
| "logps/chosen": -294.0111389160156, | |
| "logps/rejected": -237.24282836914062, | |
| "loss": 0.6033, | |
| "rewards/accuracies": 0.714062511920929, | |
| "rewards/chosen": 0.17194847762584686, | |
| "rewards/margins": 0.2693827748298645, | |
| "rewards/rejected": -0.09743430465459824, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.4196018376722816e-07, | |
| "logits/chosen": -2.413655996322632, | |
| "logits/rejected": -2.38279390335083, | |
| "logps/chosen": -279.74090576171875, | |
| "logps/rejected": -226.3095245361328, | |
| "loss": 0.6081, | |
| "rewards/accuracies": 0.6953125, | |
| "rewards/chosen": 0.16821043193340302, | |
| "rewards/margins": 0.26299187541007996, | |
| "rewards/rejected": -0.09478144347667694, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.343032159264931e-07, | |
| "logits/chosen": -2.390028476715088, | |
| "logits/rejected": -2.3162546157836914, | |
| "logps/chosen": -270.34320068359375, | |
| "logps/rejected": -221.6541290283203, | |
| "loss": 0.6019, | |
| "rewards/accuracies": 0.721875011920929, | |
| "rewards/chosen": 0.17296920716762543, | |
| "rewards/margins": 0.2894430458545685, | |
| "rewards/rejected": -0.11647380888462067, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.26646248085758e-07, | |
| "logits/chosen": -2.3680365085601807, | |
| "logits/rejected": -2.3160338401794434, | |
| "logps/chosen": -260.14593505859375, | |
| "logps/rejected": -214.6702423095703, | |
| "loss": 0.5951, | |
| "rewards/accuracies": 0.721875011920929, | |
| "rewards/chosen": 0.16743981838226318, | |
| "rewards/margins": 0.2915259897708893, | |
| "rewards/rejected": -0.1240861564874649, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 2.1898928024502298e-07, | |
| "logits/chosen": -2.443164825439453, | |
| "logits/rejected": -2.3816096782684326, | |
| "logps/chosen": -264.55914306640625, | |
| "logps/rejected": -219.47341918945312, | |
| "loss": 0.6012, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.15291035175323486, | |
| "rewards/margins": 0.2545214593410492, | |
| "rewards/rejected": -0.10161112248897552, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 2.113323124042879e-07, | |
| "logits/chosen": -2.416532039642334, | |
| "logits/rejected": -2.3757832050323486, | |
| "logps/chosen": -272.0744934082031, | |
| "logps/rejected": -228.9375, | |
| "loss": 0.5995, | |
| "rewards/accuracies": 0.698437511920929, | |
| "rewards/chosen": 0.17259187996387482, | |
| "rewards/margins": 0.2831020951271057, | |
| "rewards/rejected": -0.1105102151632309, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 2.036753445635528e-07, | |
| "logits/chosen": -2.4263906478881836, | |
| "logits/rejected": -2.4025325775146484, | |
| "logps/chosen": -284.97576904296875, | |
| "logps/rejected": -235.3527069091797, | |
| "loss": 0.5904, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": 0.16448193788528442, | |
| "rewards/margins": 0.32010418176651, | |
| "rewards/rejected": -0.1556222140789032, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.9601837672281775e-07, | |
| "logits/chosen": -2.402890920639038, | |
| "logits/rejected": -2.3698153495788574, | |
| "logps/chosen": -271.0093688964844, | |
| "logps/rejected": -234.02481079101562, | |
| "loss": 0.5976, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.14684638381004333, | |
| "rewards/margins": 0.2707447409629822, | |
| "rewards/rejected": -0.12389836460351944, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.883614088820827e-07, | |
| "logits/chosen": -2.425776958465576, | |
| "logits/rejected": -2.371450901031494, | |
| "logps/chosen": -269.1894226074219, | |
| "logps/rejected": -222.6921844482422, | |
| "loss": 0.6013, | |
| "rewards/accuracies": 0.6968749761581421, | |
| "rewards/chosen": 0.1555326133966446, | |
| "rewards/margins": 0.28726813197135925, | |
| "rewards/rejected": -0.13173556327819824, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_logits/chosen": -2.09140682220459, | |
| "eval_logits/rejected": -1.9723255634307861, | |
| "eval_logps/chosen": -266.0958557128906, | |
| "eval_logps/rejected": -221.3338165283203, | |
| "eval_loss": 0.598257839679718, | |
| "eval_rewards/accuracies": 0.6759999990463257, | |
| "eval_rewards/chosen": 0.14844101667404175, | |
| "eval_rewards/margins": 0.28188496828079224, | |
| "eval_rewards/rejected": -0.1334439367055893, | |
| "eval_runtime": 171.7034, | |
| "eval_samples_per_second": 11.648, | |
| "eval_steps_per_second": 0.728, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.807044410413476e-07, | |
| "logits/chosen": -2.3953592777252197, | |
| "logits/rejected": -2.363999843597412, | |
| "logps/chosen": -258.8542175292969, | |
| "logps/rejected": -233.6983642578125, | |
| "loss": 0.5992, | |
| "rewards/accuracies": 0.7171875238418579, | |
| "rewards/chosen": 0.1408676654100418, | |
| "rewards/margins": 0.27908438444137573, | |
| "rewards/rejected": -0.13821670413017273, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.7304747320061255e-07, | |
| "logits/chosen": -2.3902556896209717, | |
| "logits/rejected": -2.3884072303771973, | |
| "logps/chosen": -265.0683288574219, | |
| "logps/rejected": -228.423828125, | |
| "loss": 0.5955, | |
| "rewards/accuracies": 0.684374988079071, | |
| "rewards/chosen": 0.17755301296710968, | |
| "rewards/margins": 0.3005513548851013, | |
| "rewards/rejected": -0.12299831956624985, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.6539050535987747e-07, | |
| "logits/chosen": -2.473872661590576, | |
| "logits/rejected": -2.3617444038391113, | |
| "logps/chosen": -270.9250793457031, | |
| "logps/rejected": -222.18264770507812, | |
| "loss": 0.5912, | |
| "rewards/accuracies": 0.714062511920929, | |
| "rewards/chosen": 0.16197429597377777, | |
| "rewards/margins": 0.3233007788658142, | |
| "rewards/rejected": -0.16132643818855286, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.5773353751914243e-07, | |
| "logits/chosen": -2.4702870845794678, | |
| "logits/rejected": -2.4053397178649902, | |
| "logps/chosen": -285.6563720703125, | |
| "logps/rejected": -227.668212890625, | |
| "loss": 0.5944, | |
| "rewards/accuracies": 0.7046874761581421, | |
| "rewards/chosen": 0.1673307716846466, | |
| "rewards/margins": 0.3238461911678314, | |
| "rewards/rejected": -0.15651538968086243, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.5007656967840735e-07, | |
| "logits/chosen": -2.381838798522949, | |
| "logits/rejected": -2.3299927711486816, | |
| "logps/chosen": -254.9144287109375, | |
| "logps/rejected": -228.2997589111328, | |
| "loss": 0.5935, | |
| "rewards/accuracies": 0.721875011920929, | |
| "rewards/chosen": 0.16740316152572632, | |
| "rewards/margins": 0.29852622747421265, | |
| "rewards/rejected": -0.13112305104732513, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.4241960183767226e-07, | |
| "logits/chosen": -2.4169716835021973, | |
| "logits/rejected": -2.3592681884765625, | |
| "logps/chosen": -283.3446350097656, | |
| "logps/rejected": -233.24679565429688, | |
| "loss": 0.5918, | |
| "rewards/accuracies": 0.684374988079071, | |
| "rewards/chosen": 0.1610698103904724, | |
| "rewards/margins": 0.3080851435661316, | |
| "rewards/rejected": -0.1470153033733368, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.347626339969372e-07, | |
| "logits/chosen": -2.408930540084839, | |
| "logits/rejected": -2.37349009513855, | |
| "logps/chosen": -271.91973876953125, | |
| "logps/rejected": -240.18679809570312, | |
| "loss": 0.5949, | |
| "rewards/accuracies": 0.692187488079071, | |
| "rewards/chosen": 0.17847296595573425, | |
| "rewards/margins": 0.29651308059692383, | |
| "rewards/rejected": -0.118040069937706, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.2710566615620215e-07, | |
| "logits/chosen": -2.367671489715576, | |
| "logits/rejected": -2.320443868637085, | |
| "logps/chosen": -275.58294677734375, | |
| "logps/rejected": -222.78842163085938, | |
| "loss": 0.5862, | |
| "rewards/accuracies": 0.7109375, | |
| "rewards/chosen": 0.17732994258403778, | |
| "rewards/margins": 0.3325561583042145, | |
| "rewards/rejected": -0.1552262306213379, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.1944869831546706e-07, | |
| "logits/chosen": -2.388644218444824, | |
| "logits/rejected": -2.3782949447631836, | |
| "logps/chosen": -268.08172607421875, | |
| "logps/rejected": -236.77163696289062, | |
| "loss": 0.5938, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": 0.1551700383424759, | |
| "rewards/margins": 0.3083476424217224, | |
| "rewards/rejected": -0.15317757427692413, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.11791730474732e-07, | |
| "logits/chosen": -2.398094654083252, | |
| "logits/rejected": -2.3552727699279785, | |
| "logps/chosen": -263.4698181152344, | |
| "logps/rejected": -223.05538940429688, | |
| "loss": 0.5874, | |
| "rewards/accuracies": 0.715624988079071, | |
| "rewards/chosen": 0.16049396991729736, | |
| "rewards/margins": 0.35638219118118286, | |
| "rewards/rejected": -0.1958882361650467, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 1.0413476263399694e-07, | |
| "logits/chosen": -2.4469425678253174, | |
| "logits/rejected": -2.392503261566162, | |
| "logps/chosen": -270.57635498046875, | |
| "logps/rejected": -216.8152618408203, | |
| "loss": 0.5878, | |
| "rewards/accuracies": 0.7109375, | |
| "rewards/chosen": 0.15162338316440582, | |
| "rewards/margins": 0.3167043924331665, | |
| "rewards/rejected": -0.1650809943675995, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 9.647779479326186e-08, | |
| "logits/chosen": -2.3848066329956055, | |
| "logits/rejected": -2.3738114833831787, | |
| "logps/chosen": -251.4600372314453, | |
| "logps/rejected": -222.6741180419922, | |
| "loss": 0.5932, | |
| "rewards/accuracies": 0.690625011920929, | |
| "rewards/chosen": 0.16963472962379456, | |
| "rewards/margins": 0.2706124782562256, | |
| "rewards/rejected": -0.10097774118185043, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 8.88208269525268e-08, | |
| "logits/chosen": -2.377875804901123, | |
| "logits/rejected": -2.374070644378662, | |
| "logps/chosen": -261.51788330078125, | |
| "logps/rejected": -221.13034057617188, | |
| "loss": 0.5909, | |
| "rewards/accuracies": 0.690625011920929, | |
| "rewards/chosen": 0.1429089605808258, | |
| "rewards/margins": 0.29002851247787476, | |
| "rewards/rejected": -0.14711955189704895, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 8.116385911179173e-08, | |
| "logits/chosen": -2.3656888008117676, | |
| "logits/rejected": -2.3523547649383545, | |
| "logps/chosen": -281.01617431640625, | |
| "logps/rejected": -217.3480682373047, | |
| "loss": 0.5924, | |
| "rewards/accuracies": 0.7109375, | |
| "rewards/chosen": 0.17781397700309753, | |
| "rewards/margins": 0.3453444540500641, | |
| "rewards/rejected": -0.16753047704696655, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 7.350689127105667e-08, | |
| "logits/chosen": -2.4372496604919434, | |
| "logits/rejected": -2.3591084480285645, | |
| "logps/chosen": -273.20965576171875, | |
| "logps/rejected": -229.5889434814453, | |
| "loss": 0.5833, | |
| "rewards/accuracies": 0.6890624761581421, | |
| "rewards/chosen": 0.17219647765159607, | |
| "rewards/margins": 0.35740959644317627, | |
| "rewards/rejected": -0.185213103890419, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 6.584992343032159e-08, | |
| "logits/chosen": -2.351811647415161, | |
| "logits/rejected": -2.3350961208343506, | |
| "logps/chosen": -263.6435241699219, | |
| "logps/rejected": -226.4728240966797, | |
| "loss": 0.5973, | |
| "rewards/accuracies": 0.682812511920929, | |
| "rewards/chosen": 0.15416522324085236, | |
| "rewards/margins": 0.29506421089172363, | |
| "rewards/rejected": -0.14089898765087128, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.819295558958652e-08, | |
| "logits/chosen": -2.390491008758545, | |
| "logits/rejected": -2.3410990238189697, | |
| "logps/chosen": -289.5431823730469, | |
| "logps/rejected": -219.86788940429688, | |
| "loss": 0.5824, | |
| "rewards/accuracies": 0.7328125238418579, | |
| "rewards/chosen": 0.18201851844787598, | |
| "rewards/margins": 0.38538652658462524, | |
| "rewards/rejected": -0.20336803793907166, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 5.0535987748851455e-08, | |
| "logits/chosen": -2.449439525604248, | |
| "logits/rejected": -2.3970067501068115, | |
| "logps/chosen": -276.04620361328125, | |
| "logps/rejected": -229.76620483398438, | |
| "loss": 0.5751, | |
| "rewards/accuracies": 0.7093750238418579, | |
| "rewards/chosen": 0.15049387514591217, | |
| "rewards/margins": 0.32289570569992065, | |
| "rewards/rejected": -0.17240183055400848, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 4.287901990811638e-08, | |
| "logits/chosen": -2.436511993408203, | |
| "logits/rejected": -2.341296672821045, | |
| "logps/chosen": -269.1939392089844, | |
| "logps/rejected": -242.32968139648438, | |
| "loss": 0.583, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.16247320175170898, | |
| "rewards/margins": 0.3472925126552582, | |
| "rewards/rejected": -0.1848193258047104, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.522205206738132e-08, | |
| "logits/chosen": -2.3986496925354004, | |
| "logits/rejected": -2.3604884147644043, | |
| "logps/chosen": -276.03436279296875, | |
| "logps/rejected": -227.72573852539062, | |
| "loss": 0.5863, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.16009405255317688, | |
| "rewards/margins": 0.34014397859573364, | |
| "rewards/rejected": -0.18004995584487915, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.7565084226646246e-08, | |
| "logits/chosen": -2.375347137451172, | |
| "logits/rejected": -2.3558340072631836, | |
| "logps/chosen": -265.17486572265625, | |
| "logps/rejected": -234.3466796875, | |
| "loss": 0.5896, | |
| "rewards/accuracies": 0.721875011920929, | |
| "rewards/chosen": 0.16287991404533386, | |
| "rewards/margins": 0.35990676283836365, | |
| "rewards/rejected": -0.19702686369419098, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.9908116385911178e-08, | |
| "logits/chosen": -2.4065492153167725, | |
| "logits/rejected": -2.36779522895813, | |
| "logps/chosen": -261.1010437011719, | |
| "logps/rejected": -224.9178009033203, | |
| "loss": 0.5892, | |
| "rewards/accuracies": 0.6546875238418579, | |
| "rewards/chosen": 0.15390679240226746, | |
| "rewards/margins": 0.2883544862270355, | |
| "rewards/rejected": -0.13444769382476807, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.225114854517611e-08, | |
| "logits/chosen": -2.433964252471924, | |
| "logits/rejected": -2.3655142784118652, | |
| "logps/chosen": -281.37139892578125, | |
| "logps/rejected": -219.88296508789062, | |
| "loss": 0.5838, | |
| "rewards/accuracies": 0.7093750238418579, | |
| "rewards/chosen": 0.1487404853105545, | |
| "rewards/margins": 0.35009315609931946, | |
| "rewards/rejected": -0.20135268568992615, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.594180704441042e-09, | |
| "logits/chosen": -2.3966562747955322, | |
| "logits/rejected": -2.368652820587158, | |
| "logps/chosen": -258.764892578125, | |
| "logps/rejected": -223.00405883789062, | |
| "loss": 0.5889, | |
| "rewards/accuracies": 0.707812488079071, | |
| "rewards/chosen": 0.14300301671028137, | |
| "rewards/margins": 0.3152904212474823, | |
| "rewards/rejected": -0.17228738963603973, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_logits/chosen": -2.085902214050293, | |
| "eval_logits/rejected": -1.9675065279006958, | |
| "eval_logps/chosen": -266.14141845703125, | |
| "eval_logps/rejected": -221.83285522460938, | |
| "eval_loss": 0.5895045399665833, | |
| "eval_rewards/accuracies": 0.6880000233650208, | |
| "eval_rewards/chosen": 0.14388784766197205, | |
| "eval_rewards/margins": 0.3272360563278198, | |
| "eval_rewards/rejected": -0.1833481788635254, | |
| "eval_runtime": 170.0, | |
| "eval_samples_per_second": 11.765, | |
| "eval_steps_per_second": 0.735, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 726, | |
| "total_flos": 0.0, | |
| "train_loss": 0.622471270236102, | |
| "train_runtime": 20371.8366, | |
| "train_samples_per_second": 9.125, | |
| "train_steps_per_second": 0.036 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 726, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |