| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9993060374739764, | |
| "eval_steps": 1000, | |
| "global_step": 720, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0013879250520471894, | |
| "grad_norm": 7.102522482374833, | |
| "learning_rate": 6.9444444444444435e-09, | |
| "logits/chosen": -2.6390600204467773, | |
| "logits/rejected": -2.659458637237549, | |
| "logps/chosen": -349.352783203125, | |
| "logps/rejected": -372.3662109375, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.013879250520471894, | |
| "grad_norm": 7.0553818387235, | |
| "learning_rate": 6.944444444444444e-08, | |
| "logits/chosen": -2.701545238494873, | |
| "logits/rejected": -2.6816134452819824, | |
| "logps/chosen": -346.0178527832031, | |
| "logps/rejected": -351.35076904296875, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.4618055522441864, | |
| "rewards/chosen": 0.0002754588203970343, | |
| "rewards/margins": 0.0005549925845116377, | |
| "rewards/rejected": -0.0002795337641146034, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.027758501040943788, | |
| "grad_norm": 7.149876435606926, | |
| "learning_rate": 1.3888888888888888e-07, | |
| "logits/chosen": -2.7141003608703613, | |
| "logits/rejected": -2.635436773300171, | |
| "logps/chosen": -331.3048400878906, | |
| "logps/rejected": -326.703369140625, | |
| "loss": 0.6922, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.0008915923535823822, | |
| "rewards/margins": 0.00229089823551476, | |
| "rewards/rejected": -0.0013993059983476996, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.041637751561415685, | |
| "grad_norm": 6.6581229172191145, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "logits/chosen": -2.712292194366455, | |
| "logits/rejected": -2.695763349533081, | |
| "logps/chosen": -306.89031982421875, | |
| "logps/rejected": -346.85418701171875, | |
| "loss": 0.6868, | |
| "rewards/accuracies": 0.734375, | |
| "rewards/chosen": 0.008129820227622986, | |
| "rewards/margins": 0.011882511898875237, | |
| "rewards/rejected": -0.003752692136913538, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.055517002081887576, | |
| "grad_norm": 7.5041216352968245, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "logits/chosen": -2.7055609226226807, | |
| "logits/rejected": -2.694276809692383, | |
| "logps/chosen": -325.8503723144531, | |
| "logps/rejected": -350.899658203125, | |
| "loss": 0.6716, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": 0.03933032229542732, | |
| "rewards/margins": 0.04708551615476608, | |
| "rewards/rejected": -0.007755190134048462, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.06939625260235947, | |
| "grad_norm": 8.186890991027267, | |
| "learning_rate": 3.472222222222222e-07, | |
| "logits/chosen": -2.708888292312622, | |
| "logits/rejected": -2.6669304370880127, | |
| "logps/chosen": -343.9357604980469, | |
| "logps/rejected": -363.4876403808594, | |
| "loss": 0.6338, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": 0.05323966592550278, | |
| "rewards/margins": 0.13971659541130066, | |
| "rewards/rejected": -0.08647692948579788, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.08327550312283137, | |
| "grad_norm": 10.573196863586034, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "logits/chosen": -2.7201571464538574, | |
| "logits/rejected": -2.666919469833374, | |
| "logps/chosen": -341.60662841796875, | |
| "logps/rejected": -394.83935546875, | |
| "loss": 0.5595, | |
| "rewards/accuracies": 0.8031250238418579, | |
| "rewards/chosen": -0.09441100060939789, | |
| "rewards/margins": 0.3541037440299988, | |
| "rewards/rejected": -0.4485146999359131, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.09715475364330327, | |
| "grad_norm": 12.834200963453847, | |
| "learning_rate": 4.861111111111111e-07, | |
| "logits/chosen": -2.795583724975586, | |
| "logits/rejected": -2.743293285369873, | |
| "logps/chosen": -421.5907287597656, | |
| "logps/rejected": -483.6188049316406, | |
| "loss": 0.4814, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.7086803913116455, | |
| "rewards/margins": 0.6790022253990173, | |
| "rewards/rejected": -1.3876826763153076, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.11103400416377515, | |
| "grad_norm": 19.870733060862168, | |
| "learning_rate": 4.998119881260575e-07, | |
| "logits/chosen": -2.7935426235198975, | |
| "logits/rejected": -2.7506136894226074, | |
| "logps/chosen": -449.7024841308594, | |
| "logps/rejected": -549.8660888671875, | |
| "loss": 0.4425, | |
| "rewards/accuracies": 0.753125011920929, | |
| "rewards/chosen": -1.2756296396255493, | |
| "rewards/margins": 0.9220088124275208, | |
| "rewards/rejected": -2.1976382732391357, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.12491325468424705, | |
| "grad_norm": 13.159418295939288, | |
| "learning_rate": 4.990486745229364e-07, | |
| "logits/chosen": -2.780831813812256, | |
| "logits/rejected": -2.759742259979248, | |
| "logps/chosen": -451.6048278808594, | |
| "logps/rejected": -602.498046875, | |
| "loss": 0.4078, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -1.2828444242477417, | |
| "rewards/margins": 1.3387337923049927, | |
| "rewards/rejected": -2.6215782165527344, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.13879250520471895, | |
| "grad_norm": 19.37754158826291, | |
| "learning_rate": 4.977001008412112e-07, | |
| "logits/chosen": -2.5731492042541504, | |
| "logits/rejected": -2.49765682220459, | |
| "logps/chosen": -508.98223876953125, | |
| "logps/rejected": -693.7402954101562, | |
| "loss": 0.3795, | |
| "rewards/accuracies": 0.8218749761581421, | |
| "rewards/chosen": -1.520322322845459, | |
| "rewards/margins": 1.8594118356704712, | |
| "rewards/rejected": -3.3797340393066406, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.15267175572519084, | |
| "grad_norm": 15.404464079078549, | |
| "learning_rate": 4.957694362057149e-07, | |
| "logits/chosen": -2.0857653617858887, | |
| "logits/rejected": -1.8319637775421143, | |
| "logps/chosen": -497.1195373535156, | |
| "logps/rejected": -708.096923828125, | |
| "loss": 0.3493, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -1.465522289276123, | |
| "rewards/margins": 2.147498369216919, | |
| "rewards/rejected": -3.613020658493042, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.16655100624566274, | |
| "grad_norm": 19.947763315181913, | |
| "learning_rate": 4.932612176449559e-07, | |
| "logits/chosen": -1.201900839805603, | |
| "logits/rejected": -0.7779704928398132, | |
| "logps/chosen": -505.0912170410156, | |
| "logps/rejected": -777.6538696289062, | |
| "loss": 0.3487, | |
| "rewards/accuracies": 0.8218749761581421, | |
| "rewards/chosen": -1.800119400024414, | |
| "rewards/margins": 2.440018892288208, | |
| "rewards/rejected": -4.240138053894043, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.18043025676613464, | |
| "grad_norm": 14.14456268527709, | |
| "learning_rate": 4.901813394291801e-07, | |
| "logits/chosen": -0.6655147075653076, | |
| "logits/rejected": -0.07147368043661118, | |
| "logps/chosen": -448.736572265625, | |
| "logps/rejected": -710.1648559570312, | |
| "loss": 0.332, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -1.2649247646331787, | |
| "rewards/margins": 2.5081429481506348, | |
| "rewards/rejected": -3.7730679512023926, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.19430950728660654, | |
| "grad_norm": 14.401041246804821, | |
| "learning_rate": 4.865370392189376e-07, | |
| "logits/chosen": 0.013273833319544792, | |
| "logits/rejected": 0.503928542137146, | |
| "logps/chosen": -503.90087890625, | |
| "logps/rejected": -738.6141357421875, | |
| "loss": 0.3286, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -1.7297884225845337, | |
| "rewards/margins": 2.2531497478485107, | |
| "rewards/rejected": -3.982938051223755, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.2081887578070784, | |
| "grad_norm": 17.814943595433935, | |
| "learning_rate": 4.823368810567056e-07, | |
| "logits/chosen": 0.13247263431549072, | |
| "logits/rejected": 0.7856736183166504, | |
| "logps/chosen": -530.7867431640625, | |
| "logps/rejected": -812.2960205078125, | |
| "loss": 0.3307, | |
| "rewards/accuracies": 0.8531249761581421, | |
| "rewards/chosen": -1.9312444925308228, | |
| "rewards/margins": 2.786510705947876, | |
| "rewards/rejected": -4.71775484085083, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2220680083275503, | |
| "grad_norm": 26.180081364951256, | |
| "learning_rate": 4.775907352415367e-07, | |
| "logits/chosen": -0.29394057393074036, | |
| "logits/rejected": 0.6625006794929504, | |
| "logps/chosen": -522.6262817382812, | |
| "logps/rejected": -792.3927612304688, | |
| "loss": 0.3259, | |
| "rewards/accuracies": 0.8343750238418579, | |
| "rewards/chosen": -1.8135877847671509, | |
| "rewards/margins": 2.6192572116851807, | |
| "rewards/rejected": -4.432845115661621, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.2359472588480222, | |
| "grad_norm": 14.51543214840684, | |
| "learning_rate": 4.723097551340265e-07, | |
| "logits/chosen": -0.6154422760009766, | |
| "logits/rejected": 0.7039809226989746, | |
| "logps/chosen": -582.3150024414062, | |
| "logps/rejected": -870.4035034179688, | |
| "loss": 0.327, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -2.0321500301361084, | |
| "rewards/margins": 2.800929546356201, | |
| "rewards/rejected": -4.833079814910889, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.2498265093684941, | |
| "grad_norm": 16.28764148948749, | |
| "learning_rate": 4.6650635094610966e-07, | |
| "logits/chosen": -0.5453392267227173, | |
| "logits/rejected": 1.0312373638153076, | |
| "logps/chosen": -532.6235961914062, | |
| "logps/rejected": -847.57666015625, | |
| "loss": 0.3092, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -1.9564176797866821, | |
| "rewards/margins": 3.06388783454895, | |
| "rewards/rejected": -5.020305633544922, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.263705759888966, | |
| "grad_norm": 16.022768213210764, | |
| "learning_rate": 4.6019416057727577e-07, | |
| "logits/chosen": -0.6275483965873718, | |
| "logits/rejected": 0.9134941101074219, | |
| "logps/chosen": -546.0833129882812, | |
| "logps/rejected": -847.2735595703125, | |
| "loss": 0.3035, | |
| "rewards/accuracies": 0.871874988079071, | |
| "rewards/chosen": -1.9999825954437256, | |
| "rewards/margins": 2.9674413204193115, | |
| "rewards/rejected": -4.967423915863037, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.2775850104094379, | |
| "grad_norm": 17.512075327743087, | |
| "learning_rate": 4.5338801756574185e-07, | |
| "logits/chosen": -0.4514436721801758, | |
| "logits/rejected": 1.0739384889602661, | |
| "logps/chosen": -560.406494140625, | |
| "logps/rejected": -897.5866088867188, | |
| "loss": 0.306, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -2.1923580169677734, | |
| "rewards/margins": 3.19802188873291, | |
| "rewards/rejected": -5.390379905700684, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.2914642609299098, | |
| "grad_norm": 20.682964186544258, | |
| "learning_rate": 4.461039162298939e-07, | |
| "logits/chosen": 0.5358384847640991, | |
| "logits/rejected": 2.0921194553375244, | |
| "logps/chosen": -565.9454956054688, | |
| "logps/rejected": -872.009765625, | |
| "loss": 0.3124, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -2.333399534225464, | |
| "rewards/margins": 3.0657191276550293, | |
| "rewards/rejected": -5.399118900299072, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.3053435114503817, | |
| "grad_norm": 20.51613275970851, | |
| "learning_rate": 4.3835897408191513e-07, | |
| "logits/chosen": 0.03184742480516434, | |
| "logits/rejected": 1.515594244003296, | |
| "logps/chosen": -496.56744384765625, | |
| "logps/rejected": -793.4273071289062, | |
| "loss": 0.2886, | |
| "rewards/accuracies": 0.8343750238418579, | |
| "rewards/chosen": -1.7265708446502686, | |
| "rewards/margins": 2.774601459503174, | |
| "rewards/rejected": -4.501172065734863, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.3192227619708536, | |
| "grad_norm": 15.952868528864848, | |
| "learning_rate": 4.301713916019286e-07, | |
| "logits/chosen": -0.11686134338378906, | |
| "logits/rejected": 1.3415229320526123, | |
| "logps/chosen": -533.4548950195312, | |
| "logps/rejected": -846.3487548828125, | |
| "loss": 0.2971, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -1.9538986682891846, | |
| "rewards/margins": 3.150516986846924, | |
| "rewards/rejected": -5.104415416717529, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.3331020124913255, | |
| "grad_norm": 22.620955439230894, | |
| "learning_rate": 4.2156040946718343e-07, | |
| "logits/chosen": 0.23749442398548126, | |
| "logits/rejected": 1.675614595413208, | |
| "logps/chosen": -543.6336059570312, | |
| "logps/rejected": -936.5950317382812, | |
| "loss": 0.2868, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -2.0718159675598145, | |
| "rewards/margins": 3.8275322914123535, | |
| "rewards/rejected": -5.899348258972168, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.3469812630117974, | |
| "grad_norm": 20.46893017144138, | |
| "learning_rate": 4.125462633367959e-07, | |
| "logits/chosen": 0.0811614841222763, | |
| "logits/rejected": 1.8465839624404907, | |
| "logps/chosen": -561.4341430664062, | |
| "logps/rejected": -961.2340087890625, | |
| "loss": 0.2916, | |
| "rewards/accuracies": 0.8531249761581421, | |
| "rewards/chosen": -2.1782994270324707, | |
| "rewards/margins": 3.972548246383667, | |
| "rewards/rejected": -6.150847911834717, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.3608605135322693, | |
| "grad_norm": 14.598720710083828, | |
| "learning_rate": 4.031501362983007e-07, | |
| "logits/chosen": 0.12005454301834106, | |
| "logits/rejected": 1.7045762538909912, | |
| "logps/chosen": -555.3475952148438, | |
| "logps/rejected": -890.2991333007812, | |
| "loss": 0.3008, | |
| "rewards/accuracies": 0.846875011920929, | |
| "rewards/chosen": -2.2171506881713867, | |
| "rewards/margins": 3.3155696392059326, | |
| "rewards/rejected": -5.53272008895874, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.3747397640527412, | |
| "grad_norm": 16.026833129762366, | |
| "learning_rate": 3.933941090877615e-07, | |
| "logits/chosen": 0.3684059679508209, | |
| "logits/rejected": 1.7908920049667358, | |
| "logps/chosen": -571.7980346679688, | |
| "logps/rejected": -938.0518798828125, | |
| "loss": 0.2825, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -2.356633424758911, | |
| "rewards/margins": 3.5530152320861816, | |
| "rewards/rejected": -5.9096479415893555, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3886190145732131, | |
| "grad_norm": 17.23246443508795, | |
| "learning_rate": 3.833011082004228e-07, | |
| "logits/chosen": 0.3943202495574951, | |
| "logits/rejected": 2.144474506378174, | |
| "logps/chosen": -606.2532958984375, | |
| "logps/rejected": -1000.9134521484375, | |
| "loss": 0.2855, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -2.7293572425842285, | |
| "rewards/margins": 3.9710164070129395, | |
| "rewards/rejected": -6.70037317276001, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.4024982650936849, | |
| "grad_norm": 14.280408221009226, | |
| "learning_rate": 3.728948520138426e-07, | |
| "logits/chosen": 0.1209539920091629, | |
| "logits/rejected": 1.5210872888565063, | |
| "logps/chosen": -600.26611328125, | |
| "logps/rejected": -961.9885864257812, | |
| "loss": 0.3001, | |
| "rewards/accuracies": 0.8656250238418579, | |
| "rewards/chosen": -2.5707249641418457, | |
| "rewards/margins": 3.2284767627716064, | |
| "rewards/rejected": -5.799201965332031, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.4163775156141568, | |
| "grad_norm": 14.385360493039851, | |
| "learning_rate": 3.6219979505011555e-07, | |
| "logits/chosen": -0.4804894030094147, | |
| "logits/rejected": 1.5072767734527588, | |
| "logps/chosen": -552.1934814453125, | |
| "logps/rejected": -922.3473510742188, | |
| "loss": 0.2839, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -2.130495071411133, | |
| "rewards/margins": 3.7109711170196533, | |
| "rewards/rejected": -5.841465950012207, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.4302567661346287, | |
| "grad_norm": 17.33261427588973, | |
| "learning_rate": 3.512410705081684e-07, | |
| "logits/chosen": 0.34214338660240173, | |
| "logits/rejected": 2.0437850952148438, | |
| "logps/chosen": -627.4119262695312, | |
| "logps/rejected": -1092.1920166015625, | |
| "loss": 0.2926, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -2.9209723472595215, | |
| "rewards/margins": 4.4133782386779785, | |
| "rewards/rejected": -7.3343505859375, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.4441360166551006, | |
| "grad_norm": 14.401848583946974, | |
| "learning_rate": 3.400444312011776e-07, | |
| "logits/chosen": 0.3555607795715332, | |
| "logits/rejected": 2.2223145961761475, | |
| "logps/chosen": -568.3477783203125, | |
| "logps/rejected": -927.1591796875, | |
| "loss": 0.2822, | |
| "rewards/accuracies": 0.8531249761581421, | |
| "rewards/chosen": -2.276163101196289, | |
| "rewards/margins": 3.52215313911438, | |
| "rewards/rejected": -5.79831600189209, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.4580152671755725, | |
| "grad_norm": 18.609028672434732, | |
| "learning_rate": 3.286361890379034e-07, | |
| "logits/chosen": 0.6069589853286743, | |
| "logits/rejected": 2.4949374198913574, | |
| "logps/chosen": -575.1094970703125, | |
| "logps/rejected": -997.7161865234375, | |
| "loss": 0.2857, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -2.511615753173828, | |
| "rewards/margins": 4.006805896759033, | |
| "rewards/rejected": -6.518421173095703, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.4718945176960444, | |
| "grad_norm": 18.17335174994865, | |
| "learning_rate": 3.1704315319015936e-07, | |
| "logits/chosen": 0.4636520445346832, | |
| "logits/rejected": 2.732633113861084, | |
| "logps/chosen": -559.9303588867188, | |
| "logps/rejected": -1051.6278076171875, | |
| "loss": 0.2647, | |
| "rewards/accuracies": 0.8968750238418579, | |
| "rewards/chosen": -2.1130199432373047, | |
| "rewards/margins": 4.71722936630249, | |
| "rewards/rejected": -6.830248832702637, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.4857737682165163, | |
| "grad_norm": 21.753884723720933, | |
| "learning_rate": 3.052925670917219e-07, | |
| "logits/chosen": 1.2968404293060303, | |
| "logits/rejected": 3.403611421585083, | |
| "logps/chosen": -615.7195434570312, | |
| "logps/rejected": -1081.2171630859375, | |
| "loss": 0.2829, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": -2.809459686279297, | |
| "rewards/margins": 4.458012104034424, | |
| "rewards/rejected": -7.267472267150879, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.4996530187369882, | |
| "grad_norm": 19.099053383072434, | |
| "learning_rate": 2.934120444167326e-07, | |
| "logits/chosen": 1.4352385997772217, | |
| "logits/rejected": 3.273483991622925, | |
| "logps/chosen": -581.668701171875, | |
| "logps/rejected": -1042.809814453125, | |
| "loss": 0.2704, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -2.5318121910095215, | |
| "rewards/margins": 4.3836283683776855, | |
| "rewards/rejected": -6.915441036224365, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.5135322692574601, | |
| "grad_norm": 24.65293117421233, | |
| "learning_rate": 2.814295041880407e-07, | |
| "logits/chosen": 1.0912240743637085, | |
| "logits/rejected": 3.1343836784362793, | |
| "logps/chosen": -557.176513671875, | |
| "logps/rejected": -982.60107421875, | |
| "loss": 0.2701, | |
| "rewards/accuracies": 0.890625, | |
| "rewards/chosen": -2.306490898132324, | |
| "rewards/margins": 4.240821361541748, | |
| "rewards/rejected": -6.5473127365112305, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.527411519777932, | |
| "grad_norm": 16.00175200334935, | |
| "learning_rate": 2.6937310516798275e-07, | |
| "logits/chosen": 0.5683895349502563, | |
| "logits/rejected": 2.84633469581604, | |
| "logps/chosen": -578.7667236328125, | |
| "logps/rejected": -935.6932373046875, | |
| "loss": 0.2745, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -2.3604445457458496, | |
| "rewards/margins": 3.500983476638794, | |
| "rewards/rejected": -5.861428260803223, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.5412907702984039, | |
| "grad_norm": 20.233593507922254, | |
| "learning_rate": 2.5727117968577785e-07, | |
| "logits/chosen": 0.8911641240119934, | |
| "logits/rejected": 3.5523343086242676, | |
| "logps/chosen": -587.8340454101562, | |
| "logps/rejected": -973.2986450195312, | |
| "loss": 0.2781, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -2.5025112628936768, | |
| "rewards/margins": 3.980858325958252, | |
| "rewards/rejected": -6.483368873596191, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.5551700208188758, | |
| "grad_norm": 20.17598650271236, | |
| "learning_rate": 2.4515216705704393e-07, | |
| "logits/chosen": 1.0148788690567017, | |
| "logits/rejected": 3.686418056488037, | |
| "logps/chosen": -623.1129760742188, | |
| "logps/rejected": -1018.2747802734375, | |
| "loss": 0.2823, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -2.7866435050964355, | |
| "rewards/margins": 4.11606502532959, | |
| "rewards/rejected": -6.902708530426025, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.5690492713393477, | |
| "grad_norm": 15.785292476024692, | |
| "learning_rate": 2.330445467518977e-07, | |
| "logits/chosen": 0.4552311897277832, | |
| "logits/rejected": 2.6864824295043945, | |
| "logps/chosen": -626.0252685546875, | |
| "logps/rejected": -1022.4189453125, | |
| "loss": 0.2698, | |
| "rewards/accuracies": 0.8531249761581421, | |
| "rewards/chosen": -2.769418239593506, | |
| "rewards/margins": 3.919069290161133, | |
| "rewards/rejected": -6.6884870529174805, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.5829285218598196, | |
| "grad_norm": 24.14459841167646, | |
| "learning_rate": 2.209767714686924e-07, | |
| "logits/chosen": 1.0538936853408813, | |
| "logits/rejected": 3.6474509239196777, | |
| "logps/chosen": -675.6682739257812, | |
| "logps/rejected": -1114.697021484375, | |
| "loss": 0.2865, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -3.275700092315674, | |
| "rewards/margins": 4.374134063720703, | |
| "rewards/rejected": -7.649834632873535, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.5968077723802915, | |
| "grad_norm": 15.264745793489116, | |
| "learning_rate": 2.0897720027066897e-07, | |
| "logits/chosen": 1.596185326576233, | |
| "logits/rejected": 4.331744194030762, | |
| "logps/chosen": -679.60888671875, | |
| "logps/rejected": -1198.2357177734375, | |
| "loss": 0.277, | |
| "rewards/accuracies": 0.890625, | |
| "rewards/chosen": -3.4800593852996826, | |
| "rewards/margins": 5.010531902313232, | |
| "rewards/rejected": -8.490591049194336, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.6106870229007634, | |
| "grad_norm": 19.482670778404138, | |
| "learning_rate": 1.970740319426474e-07, | |
| "logits/chosen": 2.1650781631469727, | |
| "logits/rejected": 4.562286853790283, | |
| "logps/chosen": -713.8223266601562, | |
| "logps/rejected": -1141.03369140625, | |
| "loss": 0.2843, | |
| "rewards/accuracies": 0.8656250238418579, | |
| "rewards/chosen": -3.699204683303833, | |
| "rewards/margins": 4.300296783447266, | |
| "rewards/rejected": -7.9995012283325195, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.6245662734212353, | |
| "grad_norm": 16.787688600188858, | |
| "learning_rate": 1.8529523872436977e-07, | |
| "logits/chosen": 3.045531749725342, | |
| "logits/rejected": 4.9512128829956055, | |
| "logps/chosen": -796.2913818359375, | |
| "logps/rejected": -1225.3551025390625, | |
| "loss": 0.2792, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": -4.586967468261719, | |
| "rewards/margins": 4.1703386306762695, | |
| "rewards/rejected": -8.757307052612305, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.6384455239417072, | |
| "grad_norm": 16.212273972183578, | |
| "learning_rate": 1.7366850057622172e-07, | |
| "logits/chosen": 2.1972670555114746, | |
| "logits/rejected": 3.9832711219787598, | |
| "logps/chosen": -759.0075073242188, | |
| "logps/rejected": -1123.007080078125, | |
| "loss": 0.2771, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": -4.106420040130615, | |
| "rewards/margins": 3.6798558235168457, | |
| "rewards/rejected": -7.786275386810303, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.6523247744621791, | |
| "grad_norm": 18.354213726936635, | |
| "learning_rate": 1.622211401318028e-07, | |
| "logits/chosen": 1.4993724822998047, | |
| "logits/rejected": 4.176512241363525, | |
| "logps/chosen": -691.9547119140625, | |
| "logps/rejected": -1159.3623046875, | |
| "loss": 0.2635, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -3.5886549949645996, | |
| "rewards/margins": 4.674525260925293, | |
| "rewards/rejected": -8.26318073272705, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.666204024982651, | |
| "grad_norm": 17.687786590391063, | |
| "learning_rate": 1.5098005849021078e-07, | |
| "logits/chosen": 1.456290602684021, | |
| "logits/rejected": 4.025675296783447, | |
| "logps/chosen": -683.2501220703125, | |
| "logps/rejected": -1227.657470703125, | |
| "loss": 0.2727, | |
| "rewards/accuracies": 0.903124988079071, | |
| "rewards/chosen": -3.4943652153015137, | |
| "rewards/margins": 5.14384651184082, | |
| "rewards/rejected": -8.638211250305176, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.6800832755031229, | |
| "grad_norm": 15.302939524453814, | |
| "learning_rate": 1.3997167199892385e-07, | |
| "logits/chosen": 0.9439892768859863, | |
| "logits/rejected": 3.5333430767059326, | |
| "logps/chosen": -721.109375, | |
| "logps/rejected": -1133.033203125, | |
| "loss": 0.2586, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -3.779620409011841, | |
| "rewards/margins": 4.077832221984863, | |
| "rewards/rejected": -7.857451438903809, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.6939625260235948, | |
| "grad_norm": 20.366996379564934, | |
| "learning_rate": 1.2922185017584036e-07, | |
| "logits/chosen": 1.6919466257095337, | |
| "logits/rejected": 4.416836261749268, | |
| "logps/chosen": -755.7452392578125, | |
| "logps/rejected": -1292.0657958984375, | |
| "loss": 0.2637, | |
| "rewards/accuracies": 0.890625, | |
| "rewards/chosen": -4.0984649658203125, | |
| "rewards/margins": 5.336091041564941, | |
| "rewards/rejected": -9.434555053710938, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.7078417765440667, | |
| "grad_norm": 19.07449137750698, | |
| "learning_rate": 1.1875585491635998e-07, | |
| "logits/chosen": 1.5969302654266357, | |
| "logits/rejected": 3.998643398284912, | |
| "logps/chosen": -720.2608032226562, | |
| "logps/rejected": -1152.5308837890625, | |
| "loss": 0.2657, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -3.8675122261047363, | |
| "rewards/margins": 4.360492706298828, | |
| "rewards/rejected": -8.228004455566406, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.7217210270645386, | |
| "grad_norm": 20.85771539055888, | |
| "learning_rate": 1.0859828112836539e-07, | |
| "logits/chosen": 1.2433149814605713, | |
| "logits/rejected": 4.097987174987793, | |
| "logps/chosen": -671.2435913085938, | |
| "logps/rejected": -1159.519287109375, | |
| "loss": 0.2748, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -3.430670976638794, | |
| "rewards/margins": 4.710363388061523, | |
| "rewards/rejected": -8.141035079956055, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.7356002775850105, | |
| "grad_norm": 16.162845971874738, | |
| "learning_rate": 9.877299893461455e-08, | |
| "logits/chosen": 1.8318910598754883, | |
| "logits/rejected": 4.480563163757324, | |
| "logps/chosen": -694.7840576171875, | |
| "logps/rejected": -1228.274658203125, | |
| "loss": 0.2565, | |
| "rewards/accuracies": 0.878125011920929, | |
| "rewards/chosen": -3.6290040016174316, | |
| "rewards/margins": 5.196530818939209, | |
| "rewards/rejected": -8.82553482055664, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.7494795281054824, | |
| "grad_norm": 22.751190521542522, | |
| "learning_rate": 8.930309757836516e-08, | |
| "logits/chosen": 1.7889764308929443, | |
| "logits/rejected": 4.209009647369385, | |
| "logps/chosen": -711.76123046875, | |
| "logps/rejected": -1204.97265625, | |
| "loss": 0.2757, | |
| "rewards/accuracies": 0.871874988079071, | |
| "rewards/chosen": -3.8144354820251465, | |
| "rewards/margins": 4.804244041442871, | |
| "rewards/rejected": -8.618680000305176, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.7633587786259542, | |
| "grad_norm": 18.413955181298185, | |
| "learning_rate": 8.021083116405173e-08, | |
| "logits/chosen": 1.8144102096557617, | |
| "logits/rejected": 4.302700519561768, | |
| "logps/chosen": -712.6237182617188, | |
| "logps/rejected": -1180.5711669921875, | |
| "loss": 0.2783, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -3.794156312942505, | |
| "rewards/margins": 4.554933071136475, | |
| "rewards/rejected": -8.349088668823242, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.7772380291464261, | |
| "grad_norm": 15.840889959483494, | |
| "learning_rate": 7.151756636052527e-08, | |
| "logits/chosen": 1.8780730962753296, | |
| "logits/rejected": 3.9335429668426514, | |
| "logps/chosen": -710.3377075195312, | |
| "logps/rejected": -1109.950439453125, | |
| "loss": 0.2613, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -3.773430585861206, | |
| "rewards/margins": 3.8713042736053467, | |
| "rewards/rejected": -7.6447343826293945, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.7911172796668979, | |
| "grad_norm": 17.548413688482757, | |
| "learning_rate": 6.324373218975104e-08, | |
| "logits/chosen": 1.9608227014541626, | |
| "logits/rejected": 4.276215076446533, | |
| "logps/chosen": -716.494140625, | |
| "logps/rejected": -1132.06298828125, | |
| "loss": 0.2786, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -3.817418336868286, | |
| "rewards/margins": 4.1049346923828125, | |
| "rewards/rejected": -7.9223527908325195, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.8049965301873698, | |
| "grad_norm": 15.393822405492056, | |
| "learning_rate": 5.5408772018959996e-08, | |
| "logits/chosen": 1.5480540990829468, | |
| "logits/rejected": 4.223781585693359, | |
| "logps/chosen": -728.512451171875, | |
| "logps/rejected": -1185.4339599609375, | |
| "loss": 0.2625, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -3.6450915336608887, | |
| "rewards/margins": 4.617922306060791, | |
| "rewards/rejected": -8.26301383972168, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.8188757807078417, | |
| "grad_norm": 15.931503102859638, | |
| "learning_rate": 4.8031097869072225e-08, | |
| "logits/chosen": 1.2332836389541626, | |
| "logits/rejected": 3.650308609008789, | |
| "logps/chosen": -693.335205078125, | |
| "logps/rejected": -1153.408447265625, | |
| "loss": 0.2768, | |
| "rewards/accuracies": 0.859375, | |
| "rewards/chosen": -3.520477294921875, | |
| "rewards/margins": 4.526196479797363, | |
| "rewards/rejected": -8.046672821044922, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.8327550312283136, | |
| "grad_norm": 17.151234880694616, | |
| "learning_rate": 4.112804714676593e-08, | |
| "logits/chosen": 0.998208224773407, | |
| "logits/rejected": 3.692138195037842, | |
| "logps/chosen": -708.9652099609375, | |
| "logps/rejected": -1158.7760009765625, | |
| "loss": 0.26, | |
| "rewards/accuracies": 0.8531249761581421, | |
| "rewards/chosen": -3.4704201221466064, | |
| "rewards/margins": 4.559679985046387, | |
| "rewards/rejected": -8.030099868774414, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.8466342817487855, | |
| "grad_norm": 15.410888817871795, | |
| "learning_rate": 3.4715841901871545e-08, | |
| "logits/chosen": 1.2765249013900757, | |
| "logits/rejected": 3.9393792152404785, | |
| "logps/chosen": -706.2520141601562, | |
| "logps/rejected": -1225.5408935546875, | |
| "loss": 0.2575, | |
| "rewards/accuracies": 0.8656250238418579, | |
| "rewards/chosen": -3.693631410598755, | |
| "rewards/margins": 5.080514430999756, | |
| "rewards/rejected": -8.774145126342773, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.8605135322692574, | |
| "grad_norm": 20.44808883705845, | |
| "learning_rate": 2.8809550705835546e-08, | |
| "logits/chosen": 1.7312864065170288, | |
| "logits/rejected": 4.368432521820068, | |
| "logps/chosen": -755.4667358398438, | |
| "logps/rejected": -1226.697021484375, | |
| "loss": 0.2755, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -4.141783714294434, | |
| "rewards/margins": 4.687798500061035, | |
| "rewards/rejected": -8.829582214355469, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.8743927827897293, | |
| "grad_norm": 16.644054207216232, | |
| "learning_rate": 2.3423053240837514e-08, | |
| "logits/chosen": 1.7737137079238892, | |
| "logits/rejected": 4.126453399658203, | |
| "logps/chosen": -743.0369873046875, | |
| "logps/rejected": -1216.58203125, | |
| "loss": 0.2624, | |
| "rewards/accuracies": 0.8843749761581421, | |
| "rewards/chosen": -3.9477715492248535, | |
| "rewards/margins": 4.61209774017334, | |
| "rewards/rejected": -8.559869766235352, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.8882720333102012, | |
| "grad_norm": 16.30128360187393, | |
| "learning_rate": 1.8569007682777415e-08, | |
| "logits/chosen": 1.7834386825561523, | |
| "logits/rejected": 4.101445198059082, | |
| "logps/chosen": -724.764892578125, | |
| "logps/rejected": -1201.727783203125, | |
| "loss": 0.2596, | |
| "rewards/accuracies": 0.840624988079071, | |
| "rewards/chosen": -4.080270290374756, | |
| "rewards/margins": 4.567948341369629, | |
| "rewards/rejected": -8.648219108581543, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.9021512838306731, | |
| "grad_norm": 17.506230515738004, | |
| "learning_rate": 1.4258820954781037e-08, | |
| "logits/chosen": 2.0160067081451416, | |
| "logits/rejected": 4.396459102630615, | |
| "logps/chosen": -750.851318359375, | |
| "logps/rejected": -1249.35546875, | |
| "loss": 0.2658, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -4.158956050872803, | |
| "rewards/margins": 4.926741600036621, | |
| "rewards/rejected": -9.085699081420898, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.916030534351145, | |
| "grad_norm": 23.02776428486556, | |
| "learning_rate": 1.0502621921127774e-08, | |
| "logits/chosen": 1.575297474861145, | |
| "logits/rejected": 4.593780994415283, | |
| "logps/chosen": -728.802978515625, | |
| "logps/rejected": -1211.679931640625, | |
| "loss": 0.2587, | |
| "rewards/accuracies": 0.878125011920929, | |
| "rewards/chosen": -4.0324225425720215, | |
| "rewards/margins": 4.782987117767334, | |
| "rewards/rejected": -8.815410614013672, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.9299097848716169, | |
| "grad_norm": 16.504585016350322, | |
| "learning_rate": 7.309237584595007e-09, | |
| "logits/chosen": 1.5119308233261108, | |
| "logits/rejected": 4.147110939025879, | |
| "logps/chosen": -738.7492065429688, | |
| "logps/rejected": -1241.984619140625, | |
| "loss": 0.2632, | |
| "rewards/accuracies": 0.890625, | |
| "rewards/chosen": -3.941183090209961, | |
| "rewards/margins": 4.9969072341918945, | |
| "rewards/rejected": -8.938089370727539, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.9437890353920888, | |
| "grad_norm": 29.62214997576546, | |
| "learning_rate": 4.6861723431538265e-09, | |
| "logits/chosen": 1.4143844842910767, | |
| "logits/rejected": 4.09596586227417, | |
| "logps/chosen": -763.4151611328125, | |
| "logps/rejected": -1327.39599609375, | |
| "loss": 0.2657, | |
| "rewards/accuracies": 0.878125011920929, | |
| "rewards/chosen": -4.114924430847168, | |
| "rewards/margins": 5.448840141296387, | |
| "rewards/rejected": -9.563764572143555, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.9576682859125607, | |
| "grad_norm": 18.191320437540885, | |
| "learning_rate": 2.639590354763882e-09, | |
| "logits/chosen": 1.5057471990585327, | |
| "logits/rejected": 3.9795989990234375, | |
| "logps/chosen": -759.0143432617188, | |
| "logps/rejected": -1212.9630126953125, | |
| "loss": 0.2696, | |
| "rewards/accuracies": 0.809374988079071, | |
| "rewards/chosen": -4.070717811584473, | |
| "rewards/margins": 4.46467399597168, | |
| "rewards/rejected": -8.535390853881836, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.9715475364330326, | |
| "grad_norm": 19.757723334110935, | |
| "learning_rate": 1.1743010517085427e-09, | |
| "logits/chosen": 1.534629225730896, | |
| "logits/rejected": 4.418221473693848, | |
| "logps/chosen": -719.3196411132812, | |
| "logps/rejected": -1227.99609375, | |
| "loss": 0.2589, | |
| "rewards/accuracies": 0.8656250238418579, | |
| "rewards/chosen": -3.9217464923858643, | |
| "rewards/margins": 5.034605979919434, | |
| "rewards/rejected": -8.956352233886719, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.9854267869535045, | |
| "grad_norm": 18.633072832178254, | |
| "learning_rate": 2.9374783851240923e-10, | |
| "logits/chosen": 1.5875966548919678, | |
| "logits/rejected": 3.8790760040283203, | |
| "logps/chosen": -764.1870727539062, | |
| "logps/rejected": -1278.9044189453125, | |
| "loss": 0.2748, | |
| "rewards/accuracies": 0.878125011920929, | |
| "rewards/chosen": -4.169381618499756, | |
| "rewards/margins": 4.928309440612793, | |
| "rewards/rejected": -9.097692489624023, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.9993060374739764, | |
| "grad_norm": 17.747866961702037, | |
| "learning_rate": 0.0, | |
| "logits/chosen": 1.5639232397079468, | |
| "logits/rejected": 3.9590256214141846, | |
| "logps/chosen": -738.0526123046875, | |
| "logps/rejected": -1246.5972900390625, | |
| "loss": 0.2629, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -3.9611778259277344, | |
| "rewards/margins": 4.964480400085449, | |
| "rewards/rejected": -8.925657272338867, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.9993060374739764, | |
| "step": 720, | |
| "total_flos": 0.0, | |
| "train_loss": 0.32248947852187687, | |
| "train_runtime": 20768.7862, | |
| "train_samples_per_second": 8.881, | |
| "train_steps_per_second": 0.035 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 720, | |
| "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 | |
| } | |