| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 2000, | |
| "global_step": 1911, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0005232862375719519, | |
| "grad_norm": 5.2912027840794105, | |
| "learning_rate": 2.6041666666666664e-09, | |
| "logits/chosen": -0.613072395324707, | |
| "logits/rejected": -0.9113049507141113, | |
| "logps/chosen": -675.448974609375, | |
| "logps/rejected": -580.3436279296875, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0052328623757195184, | |
| "grad_norm": 5.207863571024578, | |
| "learning_rate": 2.6041666666666667e-08, | |
| "logits/chosen": -0.8929427266120911, | |
| "logits/rejected": -0.9015587568283081, | |
| "logps/chosen": -316.1885070800781, | |
| "logps/rejected": -265.829833984375, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.3888888955116272, | |
| "rewards/chosen": -0.0001502325030742213, | |
| "rewards/margins": -0.00036717823240906, | |
| "rewards/rejected": 0.00021694571478292346, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.010465724751439037, | |
| "grad_norm": 4.518662580055778, | |
| "learning_rate": 5.208333333333333e-08, | |
| "logits/chosen": -1.0493892431259155, | |
| "logits/rejected": -1.0917696952819824, | |
| "logps/chosen": -225.9092254638672, | |
| "logps/rejected": -229.95114135742188, | |
| "loss": 0.6934, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": -0.0009278160287067294, | |
| "rewards/margins": -0.0010514187160879374, | |
| "rewards/rejected": 0.00012360289110802114, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.015698587127158554, | |
| "grad_norm": 4.590969349531553, | |
| "learning_rate": 7.812499999999999e-08, | |
| "logits/chosen": -0.993874192237854, | |
| "logits/rejected": -1.0379687547683716, | |
| "logps/chosen": -285.5773010253906, | |
| "logps/rejected": -273.5916748046875, | |
| "loss": 0.6932, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.0005485258880071342, | |
| "rewards/margins": 0.0011458387598395348, | |
| "rewards/rejected": -0.0005973129882477224, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.020931449502878074, | |
| "grad_norm": 4.856489671520613, | |
| "learning_rate": 1.0416666666666667e-07, | |
| "logits/chosen": -0.8481732606887817, | |
| "logits/rejected": -0.9292299151420593, | |
| "logps/chosen": -298.5322570800781, | |
| "logps/rejected": -292.2106018066406, | |
| "loss": 0.6932, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 3.153112993459217e-05, | |
| "rewards/margins": 0.0002942727878689766, | |
| "rewards/rejected": -0.000262741552432999, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.026164311878597593, | |
| "grad_norm": 5.145809878741566, | |
| "learning_rate": 1.3020833333333334e-07, | |
| "logits/chosen": -0.7799001932144165, | |
| "logits/rejected": -0.9963434338569641, | |
| "logps/chosen": -285.77923583984375, | |
| "logps/rejected": -258.2279357910156, | |
| "loss": 0.6928, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.0015076332492753863, | |
| "rewards/margins": 0.001345943077467382, | |
| "rewards/rejected": 0.00016169001173693687, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03139717425431711, | |
| "grad_norm": 4.7530882145258, | |
| "learning_rate": 1.5624999999999999e-07, | |
| "logits/chosen": -1.0012855529785156, | |
| "logits/rejected": -1.029201626777649, | |
| "logps/chosen": -265.9154968261719, | |
| "logps/rejected": -263.63165283203125, | |
| "loss": 0.6929, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.0019283884903416038, | |
| "rewards/margins": -0.0014839128125458956, | |
| "rewards/rejected": -0.0004444758524186909, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03663003663003663, | |
| "grad_norm": 5.411287853510262, | |
| "learning_rate": 1.8229166666666666e-07, | |
| "logits/chosen": -0.7499306797981262, | |
| "logits/rejected": -0.9206749200820923, | |
| "logps/chosen": -337.06011962890625, | |
| "logps/rejected": -261.22991943359375, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.0009921113960444927, | |
| "rewards/margins": 0.00047098720096983016, | |
| "rewards/rejected": -0.0014630984514951706, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04186289900575615, | |
| "grad_norm": 4.971639223521402, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "logits/chosen": -0.8354660868644714, | |
| "logits/rejected": -0.8838351368904114, | |
| "logps/chosen": -306.07379150390625, | |
| "logps/rejected": -292.2297058105469, | |
| "loss": 0.6927, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -8.171911031240597e-05, | |
| "rewards/margins": -0.0005383099196478724, | |
| "rewards/rejected": 0.00045659096213057637, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04709576138147567, | |
| "grad_norm": 4.558913262944195, | |
| "learning_rate": 2.3437499999999998e-07, | |
| "logits/chosen": -0.8262732625007629, | |
| "logits/rejected": -0.9426995515823364, | |
| "logps/chosen": -250.512451171875, | |
| "logps/rejected": -203.21292114257812, | |
| "loss": 0.6926, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.000877672981005162, | |
| "rewards/margins": 0.0007694482337683439, | |
| "rewards/rejected": 0.00010822457261383533, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.052328623757195186, | |
| "grad_norm": 4.502101598833216, | |
| "learning_rate": 2.604166666666667e-07, | |
| "logits/chosen": -0.9312704205513, | |
| "logits/rejected": -1.0538091659545898, | |
| "logps/chosen": -209.8478546142578, | |
| "logps/rejected": -190.95339965820312, | |
| "loss": 0.692, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.0008268986130133271, | |
| "rewards/margins": 0.0016605403507128358, | |
| "rewards/rejected": -0.0024874391965568066, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0575614861329147, | |
| "grad_norm": 4.621365428807003, | |
| "learning_rate": 2.864583333333333e-07, | |
| "logits/chosen": -0.8299196362495422, | |
| "logits/rejected": -0.9797964096069336, | |
| "logps/chosen": -249.6680908203125, | |
| "logps/rejected": -200.65777587890625, | |
| "loss": 0.6913, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.0012093835975974798, | |
| "rewards/margins": 0.004302759654819965, | |
| "rewards/rejected": -0.0030933755915611982, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06279434850863422, | |
| "grad_norm": 4.60719554844544, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": -0.8148070573806763, | |
| "logits/rejected": -0.8203374147415161, | |
| "logps/chosen": -379.8021240234375, | |
| "logps/rejected": -344.6637878417969, | |
| "loss": 0.6915, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.0022750557400286198, | |
| "rewards/margins": 0.0023332713171839714, | |
| "rewards/rejected": -5.8215529861627147e-05, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06802721088435375, | |
| "grad_norm": 4.4712927326454155, | |
| "learning_rate": 3.3854166666666667e-07, | |
| "logits/chosen": -0.8715853691101074, | |
| "logits/rejected": -0.9833810925483704, | |
| "logps/chosen": -205.82373046875, | |
| "logps/rejected": -231.65127563476562, | |
| "loss": 0.6903, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.00287313642911613, | |
| "rewards/margins": 0.006604389753192663, | |
| "rewards/rejected": -0.0037312530912458897, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07326007326007326, | |
| "grad_norm": 4.733449427919051, | |
| "learning_rate": 3.645833333333333e-07, | |
| "logits/chosen": -1.0960773229599, | |
| "logits/rejected": -1.0255062580108643, | |
| "logps/chosen": -282.6625671386719, | |
| "logps/rejected": -292.23785400390625, | |
| "loss": 0.689, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.002779006026685238, | |
| "rewards/margins": 0.012505276128649712, | |
| "rewards/rejected": -0.009726271964609623, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07849293563579278, | |
| "grad_norm": 4.813942709544535, | |
| "learning_rate": 3.9062499999999997e-07, | |
| "logits/chosen": -0.8290227055549622, | |
| "logits/rejected": -0.9419819116592407, | |
| "logps/chosen": -260.06561279296875, | |
| "logps/rejected": -244.0886688232422, | |
| "loss": 0.6879, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.004813476465642452, | |
| "rewards/margins": -0.0007598908850923181, | |
| "rewards/rejected": -0.004053585696965456, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0837257980115123, | |
| "grad_norm": 5.027647064510372, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "logits/chosen": -0.8543462753295898, | |
| "logits/rejected": -0.9786416292190552, | |
| "logps/chosen": -244.1794891357422, | |
| "logps/rejected": -233.8755645751953, | |
| "loss": 0.6866, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.00423551257699728, | |
| "rewards/margins": 0.01706615462899208, | |
| "rewards/rejected": -0.0128306420519948, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.08895866038723181, | |
| "grad_norm": 5.173910287170609, | |
| "learning_rate": 4.427083333333333e-07, | |
| "logits/chosen": -0.827562153339386, | |
| "logits/rejected": -0.8305305242538452, | |
| "logps/chosen": -227.57394409179688, | |
| "logps/rejected": -246.3840789794922, | |
| "loss": 0.6836, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.004262433387339115, | |
| "rewards/margins": 0.008676105178892612, | |
| "rewards/rejected": -0.00441367132589221, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.09419152276295134, | |
| "grad_norm": 3.942933028309808, | |
| "learning_rate": 4.6874999999999996e-07, | |
| "logits/chosen": -0.9691295623779297, | |
| "logits/rejected": -1.081464171409607, | |
| "logps/chosen": -338.87176513671875, | |
| "logps/rejected": -273.46832275390625, | |
| "loss": 0.6863, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.009511679410934448, | |
| "rewards/margins": 0.0174461267888546, | |
| "rewards/rejected": -0.026957804337143898, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.09942438513867086, | |
| "grad_norm": 5.933795378160298, | |
| "learning_rate": 4.947916666666667e-07, | |
| "logits/chosen": -0.9045713543891907, | |
| "logits/rejected": -1.0874099731445312, | |
| "logps/chosen": -289.9124450683594, | |
| "logps/rejected": -246.2819366455078, | |
| "loss": 0.6778, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.007765337824821472, | |
| "rewards/margins": 0.03508571907877922, | |
| "rewards/rejected": -0.0273203793913126, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.10465724751439037, | |
| "grad_norm": 4.772143956400877, | |
| "learning_rate": 4.999732803821339e-07, | |
| "logits/chosen": -0.8650393486022949, | |
| "logits/rejected": -0.926235556602478, | |
| "logps/chosen": -262.9915771484375, | |
| "logps/rejected": -298.93353271484375, | |
| "loss": 0.6771, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.011056855320930481, | |
| "rewards/margins": 0.02525464817881584, | |
| "rewards/rejected": -0.036311499774456024, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.10989010989010989, | |
| "grad_norm": 4.996286850099119, | |
| "learning_rate": 4.998647417232375e-07, | |
| "logits/chosen": -0.9325754046440125, | |
| "logits/rejected": -1.0693069696426392, | |
| "logps/chosen": -203.27798461914062, | |
| "logps/rejected": -191.7810821533203, | |
| "loss": 0.6841, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.0018788937013596296, | |
| "rewards/margins": 0.009139192290604115, | |
| "rewards/rejected": -0.011018086224794388, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.1151229722658294, | |
| "grad_norm": 4.64438234097018, | |
| "learning_rate": 4.996727502703357e-07, | |
| "logits/chosen": -0.8251157999038696, | |
| "logits/rejected": -0.8886043429374695, | |
| "logps/chosen": -297.6938781738281, | |
| "logps/rejected": -257.5809631347656, | |
| "loss": 0.6773, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.028462525457143784, | |
| "rewards/margins": 0.07731024920940399, | |
| "rewards/rejected": -0.04884772375226021, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.12035583464154893, | |
| "grad_norm": 4.84006289249012, | |
| "learning_rate": 4.993973701470142e-07, | |
| "logits/chosen": -0.7838829755783081, | |
| "logits/rejected": -0.837600827217102, | |
| "logps/chosen": -234.60653686523438, | |
| "logps/rejected": -311.21441650390625, | |
| "loss": 0.6664, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.0013715565437451005, | |
| "rewards/margins": 0.05342661589384079, | |
| "rewards/rejected": -0.054798174649477005, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.12558869701726844, | |
| "grad_norm": 5.056335509183621, | |
| "learning_rate": 4.990386933279972e-07, | |
| "logits/chosen": -0.926543116569519, | |
| "logits/rejected": -1.05828857421875, | |
| "logps/chosen": -227.8069305419922, | |
| "logps/rejected": -217.5619659423828, | |
| "loss": 0.6696, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.03816365823149681, | |
| "rewards/margins": 0.036327291280031204, | |
| "rewards/rejected": -0.07449094951152802, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.13082155939298795, | |
| "grad_norm": 4.903194310499359, | |
| "learning_rate": 4.985968396084284e-07, | |
| "logits/chosen": -0.9419911503791809, | |
| "logits/rejected": -0.9368138313293457, | |
| "logps/chosen": -273.85748291015625, | |
| "logps/rejected": -251.20681762695312, | |
| "loss": 0.6636, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.03060590662062168, | |
| "rewards/margins": 0.0605587437748909, | |
| "rewards/rejected": -0.09116465598344803, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1360544217687075, | |
| "grad_norm": 5.658505335877987, | |
| "learning_rate": 4.98071956563861e-07, | |
| "logits/chosen": -0.845447838306427, | |
| "logits/rejected": -0.9603404998779297, | |
| "logps/chosen": -291.99737548828125, | |
| "logps/rejected": -279.862548828125, | |
| "loss": 0.6656, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.013078232295811176, | |
| "rewards/margins": 0.10106952488422394, | |
| "rewards/rejected": -0.11414775997400284, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.141287284144427, | |
| "grad_norm": 4.523399348408552, | |
| "learning_rate": 4.97464219500968e-07, | |
| "logits/chosen": -0.9943469166755676, | |
| "logits/rejected": -1.018686294555664, | |
| "logps/chosen": -284.5821228027344, | |
| "logps/rejected": -299.7913513183594, | |
| "loss": 0.6482, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.09529539942741394, | |
| "rewards/margins": 0.04227287694811821, | |
| "rewards/rejected": -0.13756826519966125, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.14652014652014653, | |
| "grad_norm": 5.272141127441927, | |
| "learning_rate": 4.967738313989918e-07, | |
| "logits/chosen": -0.9240992665290833, | |
| "logits/rejected": -0.9420005679130554, | |
| "logps/chosen": -306.8502197265625, | |
| "logps/rejected": -290.08953857421875, | |
| "loss": 0.6627, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.052370209246873856, | |
| "rewards/margins": 0.10898049920797348, | |
| "rewards/rejected": -0.16135069727897644, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.15175300889586604, | |
| "grad_norm": 4.892220404106949, | |
| "learning_rate": 4.960010228419499e-07, | |
| "logits/chosen": -0.83436119556427, | |
| "logits/rejected": -1.0053937435150146, | |
| "logps/chosen": -327.9340515136719, | |
| "logps/rejected": -256.15679931640625, | |
| "loss": 0.6634, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.11021896451711655, | |
| "rewards/margins": 0.08143133670091629, | |
| "rewards/rejected": -0.19165030121803284, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.15698587127158556, | |
| "grad_norm": 5.0307188285371245, | |
| "learning_rate": 4.951460519416227e-07, | |
| "logits/chosen": -0.7597275972366333, | |
| "logits/rejected": -0.8329798579216003, | |
| "logps/chosen": -309.31396484375, | |
| "logps/rejected": -261.9037170410156, | |
| "loss": 0.6581, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.1599477231502533, | |
| "rewards/margins": 0.05440790206193924, | |
| "rewards/rejected": -0.21435561776161194, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.16221873364730507, | |
| "grad_norm": 5.10301703350847, | |
| "learning_rate": 4.942092042513458e-07, | |
| "logits/chosen": -0.8226385116577148, | |
| "logits/rejected": -0.975459098815918, | |
| "logps/chosen": -312.9759826660156, | |
| "logps/rejected": -300.53985595703125, | |
| "loss": 0.6553, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.0716322511434555, | |
| "rewards/margins": 0.10881064087152481, | |
| "rewards/rejected": -0.1804428994655609, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.1674515960230246, | |
| "grad_norm": 5.700591112038071, | |
| "learning_rate": 4.931907926706373e-07, | |
| "logits/chosen": -0.8691483736038208, | |
| "logits/rejected": -1.0902544260025024, | |
| "logps/chosen": -347.06158447265625, | |
| "logps/rejected": -232.77938842773438, | |
| "loss": 0.6474, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.17011448740959167, | |
| "rewards/margins": 0.08706648647785187, | |
| "rewards/rejected": -0.25718098878860474, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.1726844583987441, | |
| "grad_norm": 5.816432334802341, | |
| "learning_rate": 4.920911573406924e-07, | |
| "logits/chosen": -0.9927451014518738, | |
| "logits/rejected": -1.230741262435913, | |
| "logps/chosen": -251.46133422851562, | |
| "logps/rejected": -209.3514404296875, | |
| "loss": 0.6502, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.14585557579994202, | |
| "rewards/margins": 0.09222600609064102, | |
| "rewards/rejected": -0.23808160424232483, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.17791732077446362, | |
| "grad_norm": 6.704270656287176, | |
| "learning_rate": 4.909106655307787e-07, | |
| "logits/chosen": -0.9669449925422668, | |
| "logits/rejected": -0.8335038423538208, | |
| "logps/chosen": -284.52716064453125, | |
| "logps/rejected": -351.88427734375, | |
| "loss": 0.63, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.14778541028499603, | |
| "rewards/margins": 0.1832939088344574, | |
| "rewards/rejected": -0.3310793340206146, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.18315018315018314, | |
| "grad_norm": 6.276672163574993, | |
| "learning_rate": 4.896497115155709e-07, | |
| "logits/chosen": -1.0247383117675781, | |
| "logits/rejected": -0.9595943689346313, | |
| "logps/chosen": -198.9410858154297, | |
| "logps/rejected": -248.18814086914062, | |
| "loss": 0.6228, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.1372702717781067, | |
| "rewards/margins": 0.25034016370773315, | |
| "rewards/rejected": -0.38761046528816223, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.18838304552590268, | |
| "grad_norm": 5.746693380854425, | |
| "learning_rate": 4.883087164434672e-07, | |
| "logits/chosen": -0.8813273310661316, | |
| "logits/rejected": -1.021053433418274, | |
| "logps/chosen": -287.7932434082031, | |
| "logps/rejected": -295.81683349609375, | |
| "loss": 0.6309, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.26605018973350525, | |
| "rewards/margins": 0.08224769681692123, | |
| "rewards/rejected": -0.3482978940010071, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.1936159079016222, | |
| "grad_norm": 6.815865277948172, | |
| "learning_rate": 4.868881281959282e-07, | |
| "logits/chosen": -0.8798068165779114, | |
| "logits/rejected": -1.0722447633743286, | |
| "logps/chosen": -305.884765625, | |
| "logps/rejected": -305.1341247558594, | |
| "loss": 0.6222, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.30294957756996155, | |
| "rewards/margins": 0.2897435128688812, | |
| "rewards/rejected": -0.5926931500434875, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.1988487702773417, | |
| "grad_norm": 6.505239560678682, | |
| "learning_rate": 4.853884212378889e-07, | |
| "logits/chosen": -1.0362972021102905, | |
| "logits/rejected": -1.0566703081130981, | |
| "logps/chosen": -260.21527099609375, | |
| "logps/rejected": -412.50079345703125, | |
| "loss": 0.6119, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2835327982902527, | |
| "rewards/margins": 0.32830673456192017, | |
| "rewards/rejected": -0.6118394732475281, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.20408163265306123, | |
| "grad_norm": 6.743940065948427, | |
| "learning_rate": 4.838100964592904e-07, | |
| "logits/chosen": -0.8700660467147827, | |
| "logits/rejected": -1.0450454950332642, | |
| "logps/chosen": -416.87078857421875, | |
| "logps/rejected": -305.87506103515625, | |
| "loss": 0.6197, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.31748878955841064, | |
| "rewards/margins": 0.21450409293174744, | |
| "rewards/rejected": -0.5319928526878357, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.20931449502878074, | |
| "grad_norm": 6.828641090762346, | |
| "learning_rate": 4.821536810077878e-07, | |
| "logits/chosen": -0.8353579640388489, | |
| "logits/rejected": -1.031461477279663, | |
| "logps/chosen": -313.9386901855469, | |
| "logps/rejected": -301.3431396484375, | |
| "loss": 0.6001, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.4328695237636566, | |
| "rewards/margins": 0.27747058868408203, | |
| "rewards/rejected": -0.7103400230407715, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.21454735740450026, | |
| "grad_norm": 6.905075751508215, | |
| "learning_rate": 4.804197281126862e-07, | |
| "logits/chosen": -1.030688762664795, | |
| "logits/rejected": -1.0457452535629272, | |
| "logps/chosen": -334.46826171875, | |
| "logps/rejected": -377.8026428222656, | |
| "loss": 0.6288, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.6228912472724915, | |
| "rewards/margins": 0.17153984308242798, | |
| "rewards/rejected": -0.7944310903549194, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.21978021978021978, | |
| "grad_norm": 8.196590997844144, | |
| "learning_rate": 4.786088169001671e-07, | |
| "logits/chosen": -0.8937980532646179, | |
| "logits/rejected": -0.9507759809494019, | |
| "logps/chosen": -329.95458984375, | |
| "logps/rejected": -369.8622741699219, | |
| "loss": 0.5992, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.5947745442390442, | |
| "rewards/margins": 0.21435976028442383, | |
| "rewards/rejected": -0.8091343641281128, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.2250130821559393, | |
| "grad_norm": 7.961032416351325, | |
| "learning_rate": 4.767215521998648e-07, | |
| "logits/chosen": -0.8109152913093567, | |
| "logits/rejected": -0.9932589530944824, | |
| "logps/chosen": -330.60394287109375, | |
| "logps/rejected": -318.09564208984375, | |
| "loss": 0.5913, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.3654792308807373, | |
| "rewards/margins": 0.4697452485561371, | |
| "rewards/rejected": -0.8352245092391968, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.2302459445316588, | |
| "grad_norm": 8.367649286131634, | |
| "learning_rate": 4.7475856434285853e-07, | |
| "logits/chosen": -0.9328480958938599, | |
| "logits/rejected": -1.0322339534759521, | |
| "logps/chosen": -305.74041748046875, | |
| "logps/rejected": -315.39947509765625, | |
| "loss": 0.5974, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.4867183566093445, | |
| "rewards/margins": 0.32400593161582947, | |
| "rewards/rejected": -0.8107242584228516, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.23547880690737832, | |
| "grad_norm": 8.474835469399126, | |
| "learning_rate": 4.727205089511466e-07, | |
| "logits/chosen": -0.9545785784721375, | |
| "logits/rejected": -0.9672814607620239, | |
| "logps/chosen": -307.53125, | |
| "logps/rejected": -335.9497985839844, | |
| "loss": 0.5514, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.6702186465263367, | |
| "rewards/margins": 0.2993699610233307, | |
| "rewards/rejected": -0.9695886373519897, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.24071166928309787, | |
| "grad_norm": 8.59698500773277, | |
| "learning_rate": 4.706080667186738e-07, | |
| "logits/chosen": -0.8426378965377808, | |
| "logits/rejected": -0.922264575958252, | |
| "logps/chosen": -324.0062561035156, | |
| "logps/rejected": -368.1070251464844, | |
| "loss": 0.5884, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.4376835823059082, | |
| "rewards/margins": 0.5105610489845276, | |
| "rewards/rejected": -0.9482446908950806, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.24594453165881738, | |
| "grad_norm": 9.852887746579286, | |
| "learning_rate": 4.68421943183986e-07, | |
| "logits/chosen": -0.8820059895515442, | |
| "logits/rejected": -0.9242187738418579, | |
| "logps/chosen": -324.6005859375, | |
| "logps/rejected": -369.9400939941406, | |
| "loss": 0.5528, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.538652241230011, | |
| "rewards/margins": 0.3738628029823303, | |
| "rewards/rejected": -0.9125150442123413, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.25117739403453687, | |
| "grad_norm": 8.603529097346893, | |
| "learning_rate": 4.661628684945851e-07, | |
| "logits/chosen": -1.0860228538513184, | |
| "logits/rejected": -1.1342618465423584, | |
| "logps/chosen": -266.6770935058594, | |
| "logps/rejected": -296.2251281738281, | |
| "loss": 0.567, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.3188680410385132, | |
| "rewards/margins": 0.47449684143066406, | |
| "rewards/rejected": -0.7933648824691772, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.2564102564102564, | |
| "grad_norm": 8.5211812968141, | |
| "learning_rate": 4.638315971630662e-07, | |
| "logits/chosen": -1.0311534404754639, | |
| "logits/rejected": -1.0557152032852173, | |
| "logps/chosen": -331.73358154296875, | |
| "logps/rejected": -337.86724853515625, | |
| "loss": 0.5574, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.5146857500076294, | |
| "rewards/margins": 0.3701913356781006, | |
| "rewards/rejected": -0.88487708568573, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.2616431187859759, | |
| "grad_norm": 8.377843401032223, | |
| "learning_rate": 4.6142890781511635e-07, | |
| "logits/chosen": -1.0599453449249268, | |
| "logits/rejected": -1.076552391052246, | |
| "logps/chosen": -302.8187561035156, | |
| "logps/rejected": -389.29205322265625, | |
| "loss": 0.5551, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.5295077562332153, | |
| "rewards/margins": 0.6217503547668457, | |
| "rewards/rejected": -1.151257872581482, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.2668759811616955, | |
| "grad_norm": 8.512186027439014, | |
| "learning_rate": 4.5895560292945996e-07, | |
| "logits/chosen": -0.9369387626647949, | |
| "logits/rejected": -1.1153618097305298, | |
| "logps/chosen": -341.99407958984375, | |
| "logps/rejected": -331.5032653808594, | |
| "loss": 0.5868, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.6802676916122437, | |
| "rewards/margins": 0.2563009560108185, | |
| "rewards/rejected": -0.9365686178207397, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.272108843537415, | |
| "grad_norm": 10.104036998058119, | |
| "learning_rate": 4.5641250856983743e-07, | |
| "logits/chosen": -1.0164803266525269, | |
| "logits/rejected": -1.0712977647781372, | |
| "logps/chosen": -339.76318359375, | |
| "logps/rejected": -381.4029541015625, | |
| "loss": 0.5763, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.7330660820007324, | |
| "rewards/margins": 0.2603517472743988, | |
| "rewards/rejected": -0.9934177398681641, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.2773417059131345, | |
| "grad_norm": 10.06969780430147, | |
| "learning_rate": 4.5380047410910655e-07, | |
| "logits/chosen": -1.067757248878479, | |
| "logits/rejected": -1.210282564163208, | |
| "logps/chosen": -369.5272521972656, | |
| "logps/rejected": -336.33172607421875, | |
| "loss": 0.5628, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.40420642495155334, | |
| "rewards/margins": 0.6128471493721008, | |
| "rewards/rejected": -1.017053484916687, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.282574568288854, | |
| "grad_norm": 8.93627123396453, | |
| "learning_rate": 4.5112037194555876e-07, | |
| "logits/chosen": -1.0878922939300537, | |
| "logits/rejected": -1.141874074935913, | |
| "logps/chosen": -336.7235412597656, | |
| "logps/rejected": -364.3137512207031, | |
| "loss": 0.5363, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.6179962158203125, | |
| "rewards/margins": 0.45184534788131714, | |
| "rewards/rejected": -1.0698416233062744, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.28780743066457354, | |
| "grad_norm": 8.720601331273834, | |
| "learning_rate": 4.4837309721154536e-07, | |
| "logits/chosen": -0.9832679629325867, | |
| "logits/rejected": -1.089292287826538, | |
| "logps/chosen": -380.1823425292969, | |
| "logps/rejected": -369.4367370605469, | |
| "loss": 0.6046, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.49308666586875916, | |
| "rewards/margins": 0.5638622641563416, | |
| "rewards/rejected": -1.0569490194320679, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.29304029304029305, | |
| "grad_norm": 8.122021775438958, | |
| "learning_rate": 4.4555956747451065e-07, | |
| "logits/chosen": -0.9626067876815796, | |
| "logits/rejected": -1.0849710702896118, | |
| "logps/chosen": -331.22589111328125, | |
| "logps/rejected": -364.7495422363281, | |
| "loss": 0.5482, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.4361996054649353, | |
| "rewards/margins": 0.5262717008590698, | |
| "rewards/rejected": -0.9624713659286499, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.29827315541601257, | |
| "grad_norm": 9.058735023964035, | |
| "learning_rate": 4.426807224305315e-07, | |
| "logits/chosen": -0.9820379018783569, | |
| "logits/rejected": -1.1078466176986694, | |
| "logps/chosen": -328.267822265625, | |
| "logps/rejected": -369.1883239746094, | |
| "loss": 0.5588, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.597852885723114, | |
| "rewards/margins": 0.5885254740715027, | |
| "rewards/rejected": -1.1863784790039062, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.3035060177917321, | |
| "grad_norm": 10.393753670315041, | |
| "learning_rate": 4.397375235904669e-07, | |
| "logits/chosen": -0.9954863786697388, | |
| "logits/rejected": -1.1284685134887695, | |
| "logps/chosen": -390.5664978027344, | |
| "logps/rejected": -338.8497009277344, | |
| "loss": 0.5495, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.7412960529327393, | |
| "rewards/margins": 0.5047825574874878, | |
| "rewards/rejected": -1.2460787296295166, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.3087388801674516, | |
| "grad_norm": 10.230885742921803, | |
| "learning_rate": 4.3673095395882074e-07, | |
| "logits/chosen": -0.9860847592353821, | |
| "logits/rejected": -1.067049503326416, | |
| "logps/chosen": -270.33978271484375, | |
| "logps/rejected": -309.6357727050781, | |
| "loss": 0.5701, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.6597366333007812, | |
| "rewards/margins": 0.40602293610572815, | |
| "rewards/rejected": -1.065759539604187, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.3139717425431711, | |
| "grad_norm": 11.671000583145103, | |
| "learning_rate": 4.3366201770542687e-07, | |
| "logits/chosen": -1.0431239604949951, | |
| "logits/rejected": -1.1043084859848022, | |
| "logps/chosen": -349.1619873046875, | |
| "logps/rejected": -369.27593994140625, | |
| "loss": 0.5881, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.7552496194839478, | |
| "rewards/margins": 0.28604671359062195, | |
| "rewards/rejected": -1.041296362876892, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.31920460491889063, | |
| "grad_norm": 12.220956059151984, | |
| "learning_rate": 4.3053173983006395e-07, | |
| "logits/chosen": -1.039014458656311, | |
| "logits/rejected": -1.2115468978881836, | |
| "logps/chosen": -268.86944580078125, | |
| "logps/rejected": -332.51348876953125, | |
| "loss": 0.5508, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.7133229970932007, | |
| "rewards/margins": 0.5132843255996704, | |
| "rewards/rejected": -1.2266072034835815, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.32443746729461015, | |
| "grad_norm": 11.379646125310213, | |
| "learning_rate": 4.2734116582011403e-07, | |
| "logits/chosen": -0.86652010679245, | |
| "logits/rejected": -1.2231082916259766, | |
| "logps/chosen": -414.2841796875, | |
| "logps/rejected": -326.4421691894531, | |
| "loss": 0.5612, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.6138820648193359, | |
| "rewards/margins": 0.6353493332862854, | |
| "rewards/rejected": -1.2492313385009766, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.32967032967032966, | |
| "grad_norm": 7.7814508252983385, | |
| "learning_rate": 4.2409136130137845e-07, | |
| "logits/chosen": -1.1352870464324951, | |
| "logits/rejected": -1.1191558837890625, | |
| "logps/chosen": -278.9501647949219, | |
| "logps/rejected": -384.22528076171875, | |
| "loss": 0.605, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.6691921949386597, | |
| "rewards/margins": 0.47407063841819763, | |
| "rewards/rejected": -1.1432626247406006, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.3349031920460492, | |
| "grad_norm": 11.028751044651262, | |
| "learning_rate": 4.207834116821672e-07, | |
| "logits/chosen": -0.9769953489303589, | |
| "logits/rejected": -1.1401973962783813, | |
| "logps/chosen": -330.58074951171875, | |
| "logps/rejected": -386.44622802734375, | |
| "loss": 0.5514, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.4714907705783844, | |
| "rewards/margins": 0.5420767068862915, | |
| "rewards/rejected": -1.0135674476623535, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.3401360544217687, | |
| "grad_norm": 9.056266468898434, | |
| "learning_rate": 4.174184217907818e-07, | |
| "logits/chosen": -0.9672349095344543, | |
| "logits/rejected": -1.060320496559143, | |
| "logps/chosen": -332.55218505859375, | |
| "logps/rejected": -368.73651123046875, | |
| "loss": 0.5554, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.6068300008773804, | |
| "rewards/margins": 0.6446004509925842, | |
| "rewards/rejected": -1.2514302730560303, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.3453689167974882, | |
| "grad_norm": 8.814652599770767, | |
| "learning_rate": 4.1399751550651084e-07, | |
| "logits/chosen": -1.0880647897720337, | |
| "logits/rejected": -1.1898562908172607, | |
| "logps/chosen": -315.17413330078125, | |
| "logps/rejected": -347.0714111328125, | |
| "loss": 0.5482, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.8068042993545532, | |
| "rewards/margins": 0.4671156406402588, | |
| "rewards/rejected": -1.2739198207855225, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.35060177917320773, | |
| "grad_norm": 14.937683752075078, | |
| "learning_rate": 4.1052183538426426e-07, | |
| "logits/chosen": -1.2153905630111694, | |
| "logits/rejected": -1.29157555103302, | |
| "logps/chosen": -331.7286682128906, | |
| "logps/rejected": -331.05328369140625, | |
| "loss": 0.5794, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.7625390291213989, | |
| "rewards/margins": 0.35178858041763306, | |
| "rewards/rejected": -1.1143276691436768, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.35583464154892724, | |
| "grad_norm": 12.003219375511447, | |
| "learning_rate": 4.0699254227296884e-07, | |
| "logits/chosen": -1.116801381111145, | |
| "logits/rejected": -1.1564903259277344, | |
| "logps/chosen": -306.5473937988281, | |
| "logps/rejected": -331.49139404296875, | |
| "loss": 0.5429, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.757170557975769, | |
| "rewards/margins": 0.5436859130859375, | |
| "rewards/rejected": -1.300856351852417, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.36106750392464676, | |
| "grad_norm": 9.112786075319724, | |
| "learning_rate": 4.034108149278543e-07, | |
| "logits/chosen": -1.034462332725525, | |
| "logits/rejected": -1.1058332920074463, | |
| "logps/chosen": -296.75445556640625, | |
| "logps/rejected": -326.24468994140625, | |
| "loss": 0.5479, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.5596586465835571, | |
| "rewards/margins": 0.48433390259742737, | |
| "rewards/rejected": -1.0439926385879517, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.3663003663003663, | |
| "grad_norm": 9.548789994350182, | |
| "learning_rate": 3.9977784961675833e-07, | |
| "logits/chosen": -1.1233961582183838, | |
| "logits/rejected": -1.1769764423370361, | |
| "logps/chosen": -317.4129638671875, | |
| "logps/rejected": -339.80096435546875, | |
| "loss": 0.5418, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.735338568687439, | |
| "rewards/margins": 0.4009101390838623, | |
| "rewards/rejected": -1.1362487077713013, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.3715332286760858, | |
| "grad_norm": 9.744434333555398, | |
| "learning_rate": 3.96094859720583e-07, | |
| "logits/chosen": -1.0553786754608154, | |
| "logits/rejected": -1.1237189769744873, | |
| "logps/chosen": -386.6427001953125, | |
| "logps/rejected": -401.42901611328125, | |
| "loss": 0.5079, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.7077308297157288, | |
| "rewards/margins": 0.6166214346885681, | |
| "rewards/rejected": -1.3243521451950073, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.37676609105180536, | |
| "grad_norm": 8.917911314746286, | |
| "learning_rate": 3.923630753280357e-07, | |
| "logits/chosen": -1.1400585174560547, | |
| "logits/rejected": -1.1914727687835693, | |
| "logps/chosen": -292.65716552734375, | |
| "logps/rejected": -340.9034729003906, | |
| "loss": 0.523, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.7678954005241394, | |
| "rewards/margins": 0.3283305764198303, | |
| "rewards/rejected": -1.0962259769439697, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.3819989534275249, | |
| "grad_norm": 10.682611358880774, | |
| "learning_rate": 3.8858374282478893e-07, | |
| "logits/chosen": -1.1396077871322632, | |
| "logits/rejected": -1.0520281791687012, | |
| "logps/chosen": -351.7391052246094, | |
| "logps/rejected": -429.90301513671875, | |
| "loss": 0.5649, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.9305592775344849, | |
| "rewards/margins": 0.651319682598114, | |
| "rewards/rejected": -1.581878900527954, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.3872318158032444, | |
| "grad_norm": 10.046409983126262, | |
| "learning_rate": 3.8475812447719823e-07, | |
| "logits/chosen": -1.0544391870498657, | |
| "logits/rejected": -1.036026120185852, | |
| "logps/chosen": -306.620849609375, | |
| "logps/rejected": -323.1575927734375, | |
| "loss": 0.533, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.7031368017196655, | |
| "rewards/margins": 0.4120270609855652, | |
| "rewards/rejected": -1.115163803100586, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.3924646781789639, | |
| "grad_norm": 12.554983975343603, | |
| "learning_rate": 3.8088749801071496e-07, | |
| "logits/chosen": -1.1147099733352661, | |
| "logits/rejected": -1.1453278064727783, | |
| "logps/chosen": -351.16778564453125, | |
| "logps/rejected": -433.20355224609375, | |
| "loss": 0.5283, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.6787683367729187, | |
| "rewards/margins": 0.4242841303348541, | |
| "rewards/rejected": -1.1030524969100952, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.3976975405546834, | |
| "grad_norm": 13.756345420434204, | |
| "learning_rate": 3.7697315618313644e-07, | |
| "logits/chosen": -1.2937476634979248, | |
| "logits/rejected": -1.324729561805725, | |
| "logps/chosen": -272.1294250488281, | |
| "logps/rejected": -297.9911193847656, | |
| "loss": 0.5791, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.49157842993736267, | |
| "rewards/margins": 0.5376006960868835, | |
| "rewards/rejected": -1.0291790962219238, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.40293040293040294, | |
| "grad_norm": 8.445800627854018, | |
| "learning_rate": 3.7301640635283584e-07, | |
| "logits/chosen": -1.0468685626983643, | |
| "logits/rejected": -1.0729619264602661, | |
| "logps/chosen": -320.27886962890625, | |
| "logps/rejected": -397.4753723144531, | |
| "loss": 0.5853, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.8435429334640503, | |
| "rewards/margins": 0.4757511019706726, | |
| "rewards/rejected": -1.3192939758300781, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 10.23215084321274, | |
| "learning_rate": 3.6901857004211443e-07, | |
| "logits/chosen": -1.1420189142227173, | |
| "logits/rejected": -1.1897895336151123, | |
| "logps/chosen": -306.6218566894531, | |
| "logps/rejected": -333.86529541015625, | |
| "loss": 0.5477, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.8528299331665039, | |
| "rewards/margins": 0.570170521736145, | |
| "rewards/rejected": -1.4230002164840698, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.413396127681842, | |
| "grad_norm": 7.768713138499057, | |
| "learning_rate": 3.6498098249582444e-07, | |
| "logits/chosen": -1.1434110403060913, | |
| "logits/rejected": -1.1015931367874146, | |
| "logps/chosen": -289.0810241699219, | |
| "logps/rejected": -400.203857421875, | |
| "loss": 0.535, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.738580584526062, | |
| "rewards/margins": 0.36460793018341064, | |
| "rewards/rejected": -1.1031885147094727, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.4186289900575615, | |
| "grad_norm": 9.222279959215516, | |
| "learning_rate": 3.6090499223540757e-07, | |
| "logits/chosen": -1.0318394899368286, | |
| "logits/rejected": -1.096234917640686, | |
| "logps/chosen": -372.46697998046875, | |
| "logps/rejected": -402.5672607421875, | |
| "loss": 0.5629, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.6092363595962524, | |
| "rewards/margins": 0.5345269441604614, | |
| "rewards/rejected": -1.1437633037567139, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.423861852433281, | |
| "grad_norm": 11.495873829346216, | |
| "learning_rate": 3.5679196060850034e-07, | |
| "logits/chosen": -1.0856099128723145, | |
| "logits/rejected": -1.189156174659729, | |
| "logps/chosen": -354.1376037597656, | |
| "logps/rejected": -397.07867431640625, | |
| "loss": 0.5489, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.6487225294113159, | |
| "rewards/margins": 0.575424313545227, | |
| "rewards/rejected": -1.224146842956543, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.4290947148090005, | |
| "grad_norm": 9.191445360758081, | |
| "learning_rate": 3.5264326133425464e-07, | |
| "logits/chosen": -1.1175934076309204, | |
| "logits/rejected": -1.1940885782241821, | |
| "logps/chosen": -336.60986328125, | |
| "logps/rejected": -362.86102294921875, | |
| "loss": 0.5637, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.7536613941192627, | |
| "rewards/margins": 0.709743320941925, | |
| "rewards/rejected": -1.463404655456543, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.43432757718472004, | |
| "grad_norm": 8.657867428538292, | |
| "learning_rate": 3.4846028004452693e-07, | |
| "logits/chosen": -1.0080922842025757, | |
| "logits/rejected": -1.0725328922271729, | |
| "logps/chosen": -341.5125427246094, | |
| "logps/rejected": -365.8641662597656, | |
| "loss": 0.5806, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.8481569290161133, | |
| "rewards/margins": 0.4187043607234955, | |
| "rewards/rejected": -1.2668612003326416, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.43956043956043955, | |
| "grad_norm": 14.202414150454052, | |
| "learning_rate": 3.4424441382108826e-07, | |
| "logits/chosen": -0.9394620656967163, | |
| "logits/rejected": -1.249443769454956, | |
| "logps/chosen": -447.23114013671875, | |
| "logps/rejected": -404.6824951171875, | |
| "loss": 0.5523, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.6666282415390015, | |
| "rewards/margins": 0.6986402273178101, | |
| "rewards/rejected": -1.3652684688568115, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.44479330193615907, | |
| "grad_norm": 10.60914774682239, | |
| "learning_rate": 3.399970707290105e-07, | |
| "logits/chosen": -1.121051549911499, | |
| "logits/rejected": -1.2734081745147705, | |
| "logps/chosen": -319.57073974609375, | |
| "logps/rejected": -331.162353515625, | |
| "loss": 0.5707, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.7978970408439636, | |
| "rewards/margins": 0.3845141530036926, | |
| "rewards/rejected": -1.1824110746383667, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.4500261643118786, | |
| "grad_norm": 9.92330100833777, | |
| "learning_rate": 3.3571966934638376e-07, | |
| "logits/chosen": -1.1604969501495361, | |
| "logits/rejected": -1.2318819761276245, | |
| "logps/chosen": -235.0649871826172, | |
| "logps/rejected": -369.85101318359375, | |
| "loss": 0.5412, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.5089398622512817, | |
| "rewards/margins": 0.6952416300773621, | |
| "rewards/rejected": -1.204181432723999, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.4552590266875981, | |
| "grad_norm": 10.13666736824925, | |
| "learning_rate": 3.314136382905234e-07, | |
| "logits/chosen": -1.2827130556106567, | |
| "logits/rejected": -1.2579691410064697, | |
| "logps/chosen": -267.91741943359375, | |
| "logps/rejected": -337.76373291015625, | |
| "loss": 0.5464, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.665598452091217, | |
| "rewards/margins": 0.6469942331314087, | |
| "rewards/rejected": -1.3125927448272705, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.4604918890633176, | |
| "grad_norm": 11.300262953662553, | |
| "learning_rate": 3.270804157408225e-07, | |
| "logits/chosen": -1.1657836437225342, | |
| "logits/rejected": -1.204975962638855, | |
| "logps/chosen": -356.23004150390625, | |
| "logps/rejected": -367.14715576171875, | |
| "loss": 0.5503, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.9099416732788086, | |
| "rewards/margins": 0.3574753701686859, | |
| "rewards/rejected": -1.2674171924591064, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.46572475143903713, | |
| "grad_norm": 13.16396572942542, | |
| "learning_rate": 3.227214489584128e-07, | |
| "logits/chosen": -0.9544562101364136, | |
| "logits/rejected": -1.0509642362594604, | |
| "logps/chosen": -374.04803466796875, | |
| "logps/rejected": -359.88800048828125, | |
| "loss": 0.5354, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.7183581590652466, | |
| "rewards/margins": 0.5582870244979858, | |
| "rewards/rejected": -1.2766451835632324, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.47095761381475665, | |
| "grad_norm": 10.532502099774513, | |
| "learning_rate": 3.1833819380279023e-07, | |
| "logits/chosen": -1.101965308189392, | |
| "logits/rejected": -1.2270119190216064, | |
| "logps/chosen": -309.1413879394531, | |
| "logps/rejected": -422.55731201171875, | |
| "loss": 0.5354, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.671983003616333, | |
| "rewards/margins": 0.9125014543533325, | |
| "rewards/rejected": -1.5844844579696655, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.47619047619047616, | |
| "grad_norm": 12.350587135347572, | |
| "learning_rate": 3.139321142455703e-07, | |
| "logits/chosen": -1.092979907989502, | |
| "logits/rejected": -1.1662163734436035, | |
| "logps/chosen": -211.8175811767578, | |
| "logps/rejected": -299.8162536621094, | |
| "loss": 0.5752, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.49126997590065, | |
| "rewards/margins": 0.7602614164352417, | |
| "rewards/rejected": -1.2515313625335693, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.48142333856619574, | |
| "grad_norm": 9.43043477617756, | |
| "learning_rate": 3.095046818815331e-07, | |
| "logits/chosen": -0.962468147277832, | |
| "logits/rejected": -1.1546117067337036, | |
| "logps/chosen": -343.376953125, | |
| "logps/rejected": -345.5092468261719, | |
| "loss": 0.5556, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.615440845489502, | |
| "rewards/margins": 0.5722676515579224, | |
| "rewards/rejected": -1.1877084970474243, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.48665620094191525, | |
| "grad_norm": 8.135223143391158, | |
| "learning_rate": 3.0505737543712275e-07, | |
| "logits/chosen": -1.0646668672561646, | |
| "logits/rejected": -1.0549968481063843, | |
| "logps/chosen": -324.72381591796875, | |
| "logps/rejected": -370.3115539550781, | |
| "loss": 0.5388, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.643751859664917, | |
| "rewards/margins": 0.4123232364654541, | |
| "rewards/rejected": -1.0560752153396606, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.49188906331763477, | |
| "grad_norm": 10.054077403477091, | |
| "learning_rate": 3.0059168027656475e-07, | |
| "logits/chosen": -1.0609228610992432, | |
| "logits/rejected": -1.1830861568450928, | |
| "logps/chosen": -350.5185852050781, | |
| "logps/rejected": -352.5489196777344, | |
| "loss": 0.509, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.7355393171310425, | |
| "rewards/margins": 0.5512505769729614, | |
| "rewards/rejected": -1.286789894104004, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.4971219256933543, | |
| "grad_norm": 11.74797414461764, | |
| "learning_rate": 2.9610908790576663e-07, | |
| "logits/chosen": -0.9819517135620117, | |
| "logits/rejected": -1.0499684810638428, | |
| "logps/chosen": -335.88037109375, | |
| "logps/rejected": -419.1771545410156, | |
| "loss": 0.5209, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.6323331594467163, | |
| "rewards/margins": 0.839043915271759, | |
| "rewards/rejected": -1.4713770151138306, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.5023547880690737, | |
| "grad_norm": 9.96668666601236, | |
| "learning_rate": 2.9161109547416667e-07, | |
| "logits/chosen": -1.037734866142273, | |
| "logits/rejected": -1.1786705255508423, | |
| "logps/chosen": -347.0103759765625, | |
| "logps/rejected": -399.40478515625, | |
| "loss": 0.5471, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.743817925453186, | |
| "rewards/margins": 0.5720555782318115, | |
| "rewards/rejected": -1.3158735036849976, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.5075876504447933, | |
| "grad_norm": 10.601625897857508, | |
| "learning_rate": 2.8709920527469834e-07, | |
| "logits/chosen": -1.1739659309387207, | |
| "logits/rejected": -1.3225023746490479, | |
| "logps/chosen": -343.32000732421875, | |
| "logps/rejected": -377.5054016113281, | |
| "loss": 0.5309, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.6069706082344055, | |
| "rewards/margins": 0.7253375053405762, | |
| "rewards/rejected": -1.332308053970337, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.5128205128205128, | |
| "grad_norm": 7.289801762662034, | |
| "learning_rate": 2.8257492424203685e-07, | |
| "logits/chosen": -1.1249244213104248, | |
| "logits/rejected": -1.2511053085327148, | |
| "logps/chosen": -373.466552734375, | |
| "logps/rejected": -389.02410888671875, | |
| "loss": 0.5167, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.5530604124069214, | |
| "rewards/margins": 0.7902454137802124, | |
| "rewards/rejected": -1.3433058261871338, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.5180533751962323, | |
| "grad_norm": 10.09098324922051, | |
| "learning_rate": 2.780397634492949e-07, | |
| "logits/chosen": -1.054896354675293, | |
| "logits/rejected": -1.1181640625, | |
| "logps/chosen": -296.7220764160156, | |
| "logps/rejected": -363.0633239746094, | |
| "loss": 0.5387, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.5451050400733948, | |
| "rewards/margins": 0.8848390579223633, | |
| "rewards/rejected": -1.4299442768096924, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.5232862375719518, | |
| "grad_norm": 9.787172704225766, | |
| "learning_rate": 2.7349523760333674e-07, | |
| "logits/chosen": -1.0413906574249268, | |
| "logits/rejected": -1.117727279663086, | |
| "logps/chosen": -294.98175048828125, | |
| "logps/rejected": -349.99237060546875, | |
| "loss": 0.5253, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.7654287219047546, | |
| "rewards/margins": 0.6528303623199463, | |
| "rewards/rejected": -1.4182591438293457, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.5285190999476713, | |
| "grad_norm": 14.621424246568333, | |
| "learning_rate": 2.6894286453887827e-07, | |
| "logits/chosen": -1.1582221984863281, | |
| "logits/rejected": -1.2330009937286377, | |
| "logps/chosen": -280.18353271484375, | |
| "logps/rejected": -368.38482666015625, | |
| "loss": 0.5355, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.6598084568977356, | |
| "rewards/margins": 0.6717857718467712, | |
| "rewards/rejected": -1.3315941095352173, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.533751962323391, | |
| "grad_norm": 10.241549905247137, | |
| "learning_rate": 2.6438416471154273e-07, | |
| "logits/chosen": -1.1604335308074951, | |
| "logits/rejected": -1.2005010843276978, | |
| "logps/chosen": -332.10205078125, | |
| "logps/rejected": -378.48541259765625, | |
| "loss": 0.5089, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.6059886813163757, | |
| "rewards/margins": 0.849461555480957, | |
| "rewards/rejected": -1.455450415611267, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.5389848246991105, | |
| "grad_norm": 9.013727354575442, | |
| "learning_rate": 2.598206606900406e-07, | |
| "logits/chosen": -1.1394898891448975, | |
| "logits/rejected": -1.2945622205734253, | |
| "logps/chosen": -316.51446533203125, | |
| "logps/rejected": -328.1947937011719, | |
| "loss": 0.5333, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.7797945737838745, | |
| "rewards/margins": 0.45246514678001404, | |
| "rewards/rejected": -1.232259750366211, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.54421768707483, | |
| "grad_norm": 8.111336781776568, | |
| "learning_rate": 2.552538766476443e-07, | |
| "logits/chosen": -1.1438872814178467, | |
| "logits/rejected": -1.1238300800323486, | |
| "logps/chosen": -323.6461181640625, | |
| "logps/rejected": -386.71478271484375, | |
| "loss": 0.5564, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.7553135752677917, | |
| "rewards/margins": 0.5062183141708374, | |
| "rewards/rejected": -1.2615320682525635, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.5494505494505495, | |
| "grad_norm": 10.703676009985099, | |
| "learning_rate": 2.5068533785312666e-07, | |
| "logits/chosen": -1.0496041774749756, | |
| "logits/rejected": -1.1529366970062256, | |
| "logps/chosen": -375.7614440917969, | |
| "logps/rejected": -418.11248779296875, | |
| "loss": 0.5067, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.5294119119644165, | |
| "rewards/margins": 0.8402576446533203, | |
| "rewards/rejected": -1.3696695566177368, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.554683411826269, | |
| "grad_norm": 12.77058304597108, | |
| "learning_rate": 2.461165701613333e-07, | |
| "logits/chosen": -1.1512768268585205, | |
| "logits/rejected": -1.184362530708313, | |
| "logps/chosen": -310.32611083984375, | |
| "logps/rejected": -410.4712829589844, | |
| "loss": 0.5389, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.45870405435562134, | |
| "rewards/margins": 0.9488838315010071, | |
| "rewards/rejected": -1.407588005065918, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.5599162742019885, | |
| "grad_norm": 9.668288275009493, | |
| "learning_rate": 2.415490995035596e-07, | |
| "logits/chosen": -1.1707966327667236, | |
| "logits/rejected": -1.2220295667648315, | |
| "logps/chosen": -371.3827819824219, | |
| "logps/rejected": -394.5265197753906, | |
| "loss": 0.5431, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.6321583986282349, | |
| "rewards/margins": 0.6839585304260254, | |
| "rewards/rejected": -1.3161169290542603, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.565149136577708, | |
| "grad_norm": 12.531799821628884, | |
| "learning_rate": 2.3698445137790258e-07, | |
| "logits/chosen": -1.1429023742675781, | |
| "logits/rejected": -1.2653216123580933, | |
| "logps/chosen": -314.1465759277344, | |
| "logps/rejected": -360.99285888671875, | |
| "loss": 0.5355, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.7225407958030701, | |
| "rewards/margins": 0.6546424031257629, | |
| "rewards/rejected": -1.377183198928833, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.5703819989534276, | |
| "grad_norm": 17.81295562815824, | |
| "learning_rate": 2.3242415033975575e-07, | |
| "logits/chosen": -1.0764009952545166, | |
| "logits/rejected": -1.2749533653259277, | |
| "logps/chosen": -387.68292236328125, | |
| "logps/rejected": -314.9404602050781, | |
| "loss": 0.5505, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.9429619908332825, | |
| "rewards/margins": 0.37492847442626953, | |
| "rewards/rejected": -1.3178904056549072, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.5756148613291471, | |
| "grad_norm": 9.692932105761217, | |
| "learning_rate": 2.2786971949262134e-07, | |
| "logits/chosen": -1.1522599458694458, | |
| "logits/rejected": -1.2962963581085205, | |
| "logps/chosen": -344.22064208984375, | |
| "logps/rejected": -393.7493591308594, | |
| "loss": 0.5266, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.6750838160514832, | |
| "rewards/margins": 0.5769845843315125, | |
| "rewards/rejected": -1.252068281173706, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.5808477237048666, | |
| "grad_norm": 11.579642689145317, | |
| "learning_rate": 2.2332267997940513e-07, | |
| "logits/chosen": -1.2705551385879517, | |
| "logits/rejected": -1.2792996168136597, | |
| "logps/chosen": -203.71925354003906, | |
| "logps/rejected": -326.6751708984375, | |
| "loss": 0.5749, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.4526478350162506, | |
| "rewards/margins": 0.8654245138168335, | |
| "rewards/rejected": -1.3180721998214722, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.5860805860805861, | |
| "grad_norm": 12.621329221135278, | |
| "learning_rate": 2.1878455047436753e-07, | |
| "logits/chosen": -1.1624226570129395, | |
| "logits/rejected": -1.2655895948410034, | |
| "logps/chosen": -341.18841552734375, | |
| "logps/rejected": -382.3958435058594, | |
| "loss": 0.5163, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.7849606275558472, | |
| "rewards/margins": 0.6679478883743286, | |
| "rewards/rejected": -1.4529087543487549, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.5913134484563056, | |
| "grad_norm": 12.668403273893347, | |
| "learning_rate": 2.1425684667589852e-07, | |
| "logits/chosen": -1.3048256635665894, | |
| "logits/rejected": -1.2398827075958252, | |
| "logps/chosen": -347.1092224121094, | |
| "logps/rejected": -453.8326721191406, | |
| "loss": 0.5516, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -1.2174538373947144, | |
| "rewards/margins": 0.5305573344230652, | |
| "rewards/rejected": -1.7480109930038452, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.5965463108320251, | |
| "grad_norm": 12.023926242949333, | |
| "learning_rate": 2.0974108080028692e-07, | |
| "logits/chosen": -1.06456458568573, | |
| "logits/rejected": -1.0955144166946411, | |
| "logps/chosen": -325.31622314453125, | |
| "logps/rejected": -423.0694274902344, | |
| "loss": 0.5159, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.8166521787643433, | |
| "rewards/margins": 0.6923406720161438, | |
| "rewards/rejected": -1.5089927911758423, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.6017791732077447, | |
| "grad_norm": 10.799097847904763, | |
| "learning_rate": 2.0523876107665194e-07, | |
| "logits/chosen": -1.2194546461105347, | |
| "logits/rejected": -1.280065894126892, | |
| "logps/chosen": -336.81768798828125, | |
| "logps/rejected": -378.32562255859375, | |
| "loss": 0.5344, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.7641147971153259, | |
| "rewards/margins": 0.6722861528396606, | |
| "rewards/rejected": -1.4364010095596313, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.6070120355834642, | |
| "grad_norm": 12.522164728436978, | |
| "learning_rate": 2.0075139124320787e-07, | |
| "logits/chosen": -1.2554112672805786, | |
| "logits/rejected": -1.3160804510116577, | |
| "logps/chosen": -269.1253356933594, | |
| "logps/rejected": -307.1266784667969, | |
| "loss": 0.5443, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.6767506003379822, | |
| "rewards/margins": 0.6231390237808228, | |
| "rewards/rejected": -1.2998898029327393, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.6122448979591837, | |
| "grad_norm": 12.094410032709751, | |
| "learning_rate": 1.962804700450265e-07, | |
| "logits/chosen": -1.1103699207305908, | |
| "logits/rejected": -1.2241857051849365, | |
| "logps/chosen": -358.9649963378906, | |
| "logps/rejected": -439.03521728515625, | |
| "loss": 0.5465, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.7203751802444458, | |
| "rewards/margins": 0.9127132296562195, | |
| "rewards/rejected": -1.6330883502960205, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.6174777603349032, | |
| "grad_norm": 8.270545089093483, | |
| "learning_rate": 1.9182749073346943e-07, | |
| "logits/chosen": -1.0301766395568848, | |
| "logits/rejected": -1.1534416675567627, | |
| "logps/chosen": -394.6572265625, | |
| "logps/rejected": -391.7749938964844, | |
| "loss": 0.4896, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.9890543818473816, | |
| "rewards/margins": 0.38350680470466614, | |
| "rewards/rejected": -1.3725612163543701, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.6227106227106227, | |
| "grad_norm": 9.554247673666037, | |
| "learning_rate": 1.8739394056745372e-07, | |
| "logits/chosen": -1.1898452043533325, | |
| "logits/rejected": -1.28109610080719, | |
| "logps/chosen": -451.5150451660156, | |
| "logps/rejected": -415.083251953125, | |
| "loss": 0.4908, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.704225480556488, | |
| "rewards/margins": 0.5918093919754028, | |
| "rewards/rejected": -1.2960349321365356, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.6279434850863422, | |
| "grad_norm": 9.027062976356286, | |
| "learning_rate": 1.8298130031671972e-07, | |
| "logits/chosen": -1.2719876766204834, | |
| "logits/rejected": -1.3323533535003662, | |
| "logps/chosen": -381.16339111328125, | |
| "logps/rejected": -408.12457275390625, | |
| "loss": 0.5196, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.9807897806167603, | |
| "rewards/margins": 0.5190504193305969, | |
| "rewards/rejected": -1.4998401403427124, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.6331763474620618, | |
| "grad_norm": 8.68315282690499, | |
| "learning_rate": 1.785910437672658e-07, | |
| "logits/chosen": -1.1212913990020752, | |
| "logits/rejected": -1.1531600952148438, | |
| "logps/chosen": -383.25640869140625, | |
| "logps/rejected": -408.6227722167969, | |
| "loss": 0.5249, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.9310611486434937, | |
| "rewards/margins": 0.5751183032989502, | |
| "rewards/rejected": -1.5061795711517334, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.6384092098377813, | |
| "grad_norm": 11.716703884026428, | |
| "learning_rate": 1.7422463722911624e-07, | |
| "logits/chosen": -1.110461711883545, | |
| "logits/rejected": -1.1867046356201172, | |
| "logps/chosen": -368.7302551269531, | |
| "logps/rejected": -419.4322204589844, | |
| "loss": 0.5433, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.7190608978271484, | |
| "rewards/margins": 0.8615682721138, | |
| "rewards/rejected": -1.5806291103363037, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.6436420722135008, | |
| "grad_norm": 10.184537916022915, | |
| "learning_rate": 1.6988353904658492e-07, | |
| "logits/chosen": -1.0632117986679077, | |
| "logits/rejected": -1.2908556461334229, | |
| "logps/chosen": -423.96270751953125, | |
| "logps/rejected": -391.506591796875, | |
| "loss": 0.4879, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.7387353777885437, | |
| "rewards/margins": 0.8664577603340149, | |
| "rewards/rejected": -1.6051928997039795, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.6488749345892203, | |
| "grad_norm": 12.166712100792914, | |
| "learning_rate": 1.6556919911120081e-07, | |
| "logits/chosen": -1.178629994392395, | |
| "logits/rejected": -1.1636600494384766, | |
| "logps/chosen": -295.05731201171875, | |
| "logps/rejected": -349.7823486328125, | |
| "loss": 0.5021, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.6630123257637024, | |
| "rewards/margins": 0.7607491612434387, | |
| "rewards/rejected": -1.4237613677978516, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.6541077969649398, | |
| "grad_norm": 8.365177431092949, | |
| "learning_rate": 1.6128305837745546e-07, | |
| "logits/chosen": -1.134745717048645, | |
| "logits/rejected": -1.2268315553665161, | |
| "logps/chosen": -335.88470458984375, | |
| "logps/rejected": -419.7283630371094, | |
| "loss": 0.5064, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.7585079669952393, | |
| "rewards/margins": 0.6967343091964722, | |
| "rewards/rejected": -1.4552422761917114, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.6593406593406593, | |
| "grad_norm": 10.11243118504911, | |
| "learning_rate": 1.570265483815364e-07, | |
| "logits/chosen": -1.2292219400405884, | |
| "logits/rejected": -1.3019181489944458, | |
| "logps/chosen": -311.6837158203125, | |
| "logps/rejected": -287.0787048339844, | |
| "loss": 0.5112, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.6337876319885254, | |
| "rewards/margins": 0.5905357599258423, | |
| "rewards/rejected": -1.2243235111236572, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.6645735217163788, | |
| "grad_norm": 10.434415583794985, | |
| "learning_rate": 1.5280109076320506e-07, | |
| "logits/chosen": -1.3286902904510498, | |
| "logits/rejected": -1.3394997119903564, | |
| "logps/chosen": -293.60614013671875, | |
| "logps/rejected": -345.57568359375, | |
| "loss": 0.4924, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.7150704264640808, | |
| "rewards/margins": 0.7000972628593445, | |
| "rewards/rejected": -1.4151678085327148, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.6698063840920984, | |
| "grad_norm": 10.400220883666378, | |
| "learning_rate": 1.4860809679098158e-07, | |
| "logits/chosen": -1.2018976211547852, | |
| "logits/rejected": -1.325392246246338, | |
| "logps/chosen": -316.5107421875, | |
| "logps/rejected": -351.9391174316406, | |
| "loss": 0.5292, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.816146969795227, | |
| "rewards/margins": 0.6872614622116089, | |
| "rewards/rejected": -1.503408432006836, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.6750392464678179, | |
| "grad_norm": 8.95257167533715, | |
| "learning_rate": 1.444489668907914e-07, | |
| "logits/chosen": -1.2102806568145752, | |
| "logits/rejected": -1.2517168521881104, | |
| "logps/chosen": -324.88275146484375, | |
| "logps/rejected": -426.87713623046875, | |
| "loss": 0.5292, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.6853327751159668, | |
| "rewards/margins": 0.8505553007125854, | |
| "rewards/rejected": -1.5358881950378418, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.6802721088435374, | |
| "grad_norm": 10.218025125539159, | |
| "learning_rate": 1.403250901782354e-07, | |
| "logits/chosen": -1.2663801908493042, | |
| "logits/rejected": -1.2181060314178467, | |
| "logps/chosen": -361.97479248046875, | |
| "logps/rejected": -415.7010803222656, | |
| "loss": 0.5265, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.9003621935844421, | |
| "rewards/margins": 0.5406725406646729, | |
| "rewards/rejected": -1.4410346746444702, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.6855049712192569, | |
| "grad_norm": 12.330414497118486, | |
| "learning_rate": 1.3623784399463584e-07, | |
| "logits/chosen": -1.126432180404663, | |
| "logits/rejected": -1.1779577732086182, | |
| "logps/chosen": -310.89398193359375, | |
| "logps/rejected": -351.1187744140625, | |
| "loss": 0.5091, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.7655483484268188, | |
| "rewards/margins": 0.66068434715271, | |
| "rewards/rejected": -1.4262326955795288, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.6907378335949764, | |
| "grad_norm": 13.55513826496698, | |
| "learning_rate": 1.3218859344701632e-07, | |
| "logits/chosen": -1.0685323476791382, | |
| "logits/rejected": -1.0921872854232788, | |
| "logps/chosen": -357.3135986328125, | |
| "logps/rejected": -424.2374572753906, | |
| "loss": 0.496, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.8321298360824585, | |
| "rewards/margins": 0.6854997873306274, | |
| "rewards/rejected": -1.5176297426223755, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.6959706959706959, | |
| "grad_norm": 11.973601998113498, | |
| "learning_rate": 1.2817869095216624e-07, | |
| "logits/chosen": -1.0887490510940552, | |
| "logits/rejected": -1.0985252857208252, | |
| "logps/chosen": -329.9906311035156, | |
| "logps/rejected": -428.9029235839844, | |
| "loss": 0.5013, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.6890138387680054, | |
| "rewards/margins": 0.7360657453536987, | |
| "rewards/rejected": -1.4250797033309937, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.7012035583464155, | |
| "grad_norm": 9.263981118702059, | |
| "learning_rate": 1.2420947578494522e-07, | |
| "logits/chosen": -1.1637510061264038, | |
| "logits/rejected": -1.3139464855194092, | |
| "logps/chosen": -347.0451965332031, | |
| "logps/rejected": -364.57647705078125, | |
| "loss": 0.5143, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.7794694304466248, | |
| "rewards/margins": 0.8562557101249695, | |
| "rewards/rejected": -1.6357250213623047, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.706436420722135, | |
| "grad_norm": 13.571595353683206, | |
| "learning_rate": 1.202822736309758e-07, | |
| "logits/chosen": -1.0500030517578125, | |
| "logits/rejected": -1.0962427854537964, | |
| "logps/chosen": -360.6551208496094, | |
| "logps/rejected": -407.03448486328125, | |
| "loss": 0.5026, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.8235113024711609, | |
| "rewards/margins": 0.6789143681526184, | |
| "rewards/rejected": -1.5024257898330688, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.7116692830978545, | |
| "grad_norm": 9.965928066396195, | |
| "learning_rate": 1.1639839614387572e-07, | |
| "logits/chosen": -1.1780047416687012, | |
| "logits/rejected": -1.2829917669296265, | |
| "logps/chosen": -435.4798278808594, | |
| "logps/rejected": -439.09173583984375, | |
| "loss": 0.5544, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.0139061212539673, | |
| "rewards/margins": 0.5282729864120483, | |
| "rewards/rejected": -1.542178988456726, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.716902145473574, | |
| "grad_norm": 11.275357703138836, | |
| "learning_rate": 1.1255914050717552e-07, | |
| "logits/chosen": -1.1181185245513916, | |
| "logits/rejected": -1.3721811771392822, | |
| "logps/chosen": -379.4562683105469, | |
| "logps/rejected": -353.43157958984375, | |
| "loss": 0.4895, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.8644318580627441, | |
| "rewards/margins": 0.5917305946350098, | |
| "rewards/rejected": -1.4561625719070435, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.7221350078492935, | |
| "grad_norm": 12.737103206928401, | |
| "learning_rate": 1.0876578900107053e-07, | |
| "logits/chosen": -1.2187843322753906, | |
| "logits/rejected": -1.2873694896697998, | |
| "logps/chosen": -264.6595153808594, | |
| "logps/rejected": -384.25225830078125, | |
| "loss": 0.4935, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.5753992795944214, | |
| "rewards/margins": 0.8757054209709167, | |
| "rewards/rejected": -1.4511046409606934, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.727367870225013, | |
| "grad_norm": 12.97168258093361, | |
| "learning_rate": 1.050196085741491e-07, | |
| "logits/chosen": -1.2466976642608643, | |
| "logits/rejected": -1.3973594903945923, | |
| "logps/chosen": -346.0795593261719, | |
| "logps/rejected": -363.08380126953125, | |
| "loss": 0.489, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.8371197581291199, | |
| "rewards/margins": 0.7653623819351196, | |
| "rewards/rejected": -1.6024821996688843, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.7326007326007326, | |
| "grad_norm": 13.41322662896709, | |
| "learning_rate": 1.0132185042024246e-07, | |
| "logits/chosen": -1.3399550914764404, | |
| "logits/rejected": -1.3757188320159912, | |
| "logps/chosen": -323.0008544921875, | |
| "logps/rejected": -408.1412658691406, | |
| "loss": 0.5233, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.9458025097846985, | |
| "rewards/margins": 0.7458266019821167, | |
| "rewards/rejected": -1.6916290521621704, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.7378335949764521, | |
| "grad_norm": 12.965281325135477, | |
| "learning_rate": 9.767374956053584e-08, | |
| "logits/chosen": -1.347015619277954, | |
| "logits/rejected": -1.3667693138122559, | |
| "logps/chosen": -324.0329895019531, | |
| "logps/rejected": -401.6748352050781, | |
| "loss": 0.5521, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.8981969952583313, | |
| "rewards/margins": 0.8032199144363403, | |
| "rewards/rejected": -1.7014169692993164, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.7430664573521716, | |
| "grad_norm": 13.396320385256042, | |
| "learning_rate": 9.407652443108192e-08, | |
| "logits/chosen": -1.1957058906555176, | |
| "logits/rejected": -1.2673814296722412, | |
| "logps/chosen": -389.23223876953125, | |
| "logps/rejected": -418.9557189941406, | |
| "loss": 0.5632, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.886542797088623, | |
| "rewards/margins": 0.7842100262641907, | |
| "rewards/rejected": -1.670752763748169, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.7482993197278912, | |
| "grad_norm": 12.063700551819691, | |
| "learning_rate": 9.053137647585229e-08, | |
| "logits/chosen": -1.19754159450531, | |
| "logits/rejected": -1.3801628351211548, | |
| "logps/chosen": -395.425537109375, | |
| "logps/rejected": -390.66278076171875, | |
| "loss": 0.5038, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.8353336453437805, | |
| "rewards/margins": 0.6488715410232544, | |
| "rewards/rejected": -1.4842052459716797, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.7535321821036107, | |
| "grad_norm": 11.990445284897602, | |
| "learning_rate": 8.70394897454659e-08, | |
| "logits/chosen": -1.1735503673553467, | |
| "logits/rejected": -1.292515516281128, | |
| "logps/chosen": -401.2131042480469, | |
| "logps/rejected": -466.5420837402344, | |
| "loss": 0.522, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.558559775352478, | |
| "rewards/margins": 0.9399889707565308, | |
| "rewards/rejected": -1.4985487461090088, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.7587650444793302, | |
| "grad_norm": 11.935902040542206, | |
| "learning_rate": 8.360203050172488e-08, | |
| "logits/chosen": -1.2571841478347778, | |
| "logits/rejected": -1.2958558797836304, | |
| "logps/chosen": -362.5431213378906, | |
| "logps/rejected": -411.53997802734375, | |
| "loss": 0.5259, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.9163681268692017, | |
| "rewards/margins": 0.7593438029289246, | |
| "rewards/rejected": -1.675711989402771, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.7639979068550498, | |
| "grad_norm": 9.068298634968928, | |
| "learning_rate": 8.022014682809305e-08, | |
| "logits/chosen": -1.3620612621307373, | |
| "logits/rejected": -1.3937609195709229, | |
| "logps/chosen": -273.1943664550781, | |
| "logps/rejected": -345.25555419921875, | |
| "loss": 0.5177, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.8040696978569031, | |
| "rewards/margins": 0.6031185984611511, | |
| "rewards/rejected": -1.4071884155273438, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 11.382282536026993, | |
| "learning_rate": 7.689496824624525e-08, | |
| "logits/chosen": -1.3386561870574951, | |
| "logits/rejected": -1.527148962020874, | |
| "logps/chosen": -396.3229064941406, | |
| "logps/rejected": -402.32891845703125, | |
| "loss": 0.5072, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.8049358129501343, | |
| "rewards/margins": 0.9081543684005737, | |
| "rewards/rejected": -1.7130903005599976, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.7744636316064888, | |
| "grad_norm": 12.33392884893423, | |
| "learning_rate": 7.362760533881649e-08, | |
| "logits/chosen": -1.422590732574463, | |
| "logits/rejected": -1.4653226137161255, | |
| "logps/chosen": -330.05987548828125, | |
| "logps/rejected": -365.05535888671875, | |
| "loss": 0.5244, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.9965084195137024, | |
| "rewards/margins": 0.5925046801567078, | |
| "rewards/rejected": -1.5890133380889893, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.7796964939822083, | |
| "grad_norm": 12.928253984063549, | |
| "learning_rate": 7.041914937847584e-08, | |
| "logits/chosen": -1.3457154035568237, | |
| "logits/rejected": -1.3661472797393799, | |
| "logps/chosen": -366.29339599609375, | |
| "logps/rejected": -452.2115173339844, | |
| "loss": 0.51, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -1.0464023351669312, | |
| "rewards/margins": 0.8855581283569336, | |
| "rewards/rejected": -1.9319604635238647, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.7849293563579278, | |
| "grad_norm": 11.846016623066228, | |
| "learning_rate": 6.727067196345099e-08, | |
| "logits/chosen": -1.2849265336990356, | |
| "logits/rejected": -1.3893473148345947, | |
| "logps/chosen": -296.6019592285156, | |
| "logps/rejected": -444.0691833496094, | |
| "loss": 0.4968, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.9065812826156616, | |
| "rewards/margins": 1.098984718322754, | |
| "rewards/rejected": -2.005566120147705, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.7901622187336473, | |
| "grad_norm": 9.960268668010652, | |
| "learning_rate": 6.418322465962233e-08, | |
| "logits/chosen": -1.3249355554580688, | |
| "logits/rejected": -1.3122931718826294, | |
| "logps/chosen": -381.8565979003906, | |
| "logps/rejected": -499.72161865234375, | |
| "loss": 0.5356, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.3371002674102783, | |
| "rewards/margins": 0.8308144807815552, | |
| "rewards/rejected": -2.167914628982544, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.7953950811093669, | |
| "grad_norm": 12.546874816293252, | |
| "learning_rate": 6.115783864930905e-08, | |
| "logits/chosen": -1.3687621355056763, | |
| "logits/rejected": -1.45345139503479, | |
| "logps/chosen": -298.7080383300781, | |
| "logps/rejected": -421.26422119140625, | |
| "loss": 0.5174, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.8631349802017212, | |
| "rewards/margins": 1.0178935527801514, | |
| "rewards/rejected": -1.8810285329818726, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.8006279434850864, | |
| "grad_norm": 15.449951101743657, | |
| "learning_rate": 5.8195524386862374e-08, | |
| "logits/chosen": -1.1471742391586304, | |
| "logits/rejected": -1.1375999450683594, | |
| "logps/chosen": -405.28619384765625, | |
| "logps/rejected": -550.5733642578125, | |
| "loss": 0.4858, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.0993586778640747, | |
| "rewards/margins": 0.7847630381584167, | |
| "rewards/rejected": -1.8841215372085571, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.8058608058608059, | |
| "grad_norm": 7.57695758921922, | |
| "learning_rate": 5.529727126118228e-08, | |
| "logits/chosen": -1.2175819873809814, | |
| "logits/rejected": -1.2101088762283325, | |
| "logps/chosen": -447.8209533691406, | |
| "logps/rejected": -454.62249755859375, | |
| "loss": 0.5209, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -1.1148792505264282, | |
| "rewards/margins": 0.4590207040309906, | |
| "rewards/rejected": -1.5738999843597412, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.8110936682365254, | |
| "grad_norm": 16.121099871171214, | |
| "learning_rate": 5.246404726526918e-08, | |
| "logits/chosen": -1.2117811441421509, | |
| "logits/rejected": -1.4320929050445557, | |
| "logps/chosen": -382.09588623046875, | |
| "logps/rejected": -381.78546142578125, | |
| "loss": 0.4976, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.7104379534721375, | |
| "rewards/margins": 0.8749149441719055, | |
| "rewards/rejected": -1.585352897644043, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.8163265306122449, | |
| "grad_norm": 14.686715983716352, | |
| "learning_rate": 4.969679867292276e-08, | |
| "logits/chosen": -1.1892309188842773, | |
| "logits/rejected": -1.2661142349243164, | |
| "logps/chosen": -429.66680908203125, | |
| "logps/rejected": -465.50543212890625, | |
| "loss": 0.5241, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.9120133519172668, | |
| "rewards/margins": 0.9419037699699402, | |
| "rewards/rejected": -1.853917121887207, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.8215593929879644, | |
| "grad_norm": 8.406024980782155, | |
| "learning_rate": 4.6996449722693315e-08, | |
| "logits/chosen": -1.3920047283172607, | |
| "logits/rejected": -1.4546654224395752, | |
| "logps/chosen": -303.8490295410156, | |
| "logps/rejected": -380.0191650390625, | |
| "loss": 0.5046, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.7151414155960083, | |
| "rewards/margins": 0.7757319808006287, | |
| "rewards/rejected": -1.4908732175827026, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.826792255363684, | |
| "grad_norm": 10.726509305028333, | |
| "learning_rate": 4.436390230919465e-08, | |
| "logits/chosen": -1.309987187385559, | |
| "logits/rejected": -1.4876197576522827, | |
| "logps/chosen": -366.0071716308594, | |
| "logps/rejected": -395.597412109375, | |
| "loss": 0.5465, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.959511399269104, | |
| "rewards/margins": 0.7645119428634644, | |
| "rewards/rejected": -1.724023461341858, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.8320251177394035, | |
| "grad_norm": 12.88345857405619, | |
| "learning_rate": 4.180003568187776e-08, | |
| "logits/chosen": -1.1906282901763916, | |
| "logits/rejected": -1.4163811206817627, | |
| "logps/chosen": -328.5754699707031, | |
| "logps/rejected": -397.17022705078125, | |
| "loss": 0.504, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.827189564704895, | |
| "rewards/margins": 0.9950035214424133, | |
| "rewards/rejected": -1.822192907333374, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.837257980115123, | |
| "grad_norm": 11.786525332826088, | |
| "learning_rate": 3.930570615136919e-08, | |
| "logits/chosen": -1.3638622760772705, | |
| "logits/rejected": -1.3613183498382568, | |
| "logps/chosen": -340.79046630859375, | |
| "logps/rejected": -425.66632080078125, | |
| "loss": 0.5046, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.038352608680725, | |
| "rewards/margins": 0.6946929693222046, | |
| "rewards/rejected": -1.7330455780029297, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.8424908424908425, | |
| "grad_norm": 10.921746476297674, | |
| "learning_rate": 3.6881746803469756e-08, | |
| "logits/chosen": -1.2174428701400757, | |
| "logits/rejected": -1.4488804340362549, | |
| "logps/chosen": -471.79669189453125, | |
| "logps/rejected": -489.78411865234375, | |
| "loss": 0.5129, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.9443795084953308, | |
| "rewards/margins": 0.927892804145813, | |
| "rewards/rejected": -1.872272253036499, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.847723704866562, | |
| "grad_norm": 9.455246482209075, | |
| "learning_rate": 3.452896722091128e-08, | |
| "logits/chosen": -1.2315971851348877, | |
| "logits/rejected": -1.416680097579956, | |
| "logps/chosen": -377.7510681152344, | |
| "logps/rejected": -402.55242919921875, | |
| "loss": 0.5117, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.1194517612457275, | |
| "rewards/margins": 0.7395168542861938, | |
| "rewards/rejected": -1.8589684963226318, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.8529565672422815, | |
| "grad_norm": 11.403985101827214, | |
| "learning_rate": 3.2248153212961677e-08, | |
| "logits/chosen": -1.237486481666565, | |
| "logits/rejected": -1.262434959411621, | |
| "logps/chosen": -348.9461669921875, | |
| "logps/rejected": -417.15863037109375, | |
| "loss": 0.5023, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.8699714541435242, | |
| "rewards/margins": 0.9124903678894043, | |
| "rewards/rejected": -1.7824617624282837, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.858189429618001, | |
| "grad_norm": 10.508074438895047, | |
| "learning_rate": 3.004006655297209e-08, | |
| "logits/chosen": -1.2522094249725342, | |
| "logits/rejected": -1.2810122966766357, | |
| "logps/chosen": -351.80853271484375, | |
| "logps/rejected": -417.1268615722656, | |
| "loss": 0.5365, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.7258558869361877, | |
| "rewards/margins": 0.8368725776672363, | |
| "rewards/rejected": -1.5627284049987793, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.8634222919937206, | |
| "grad_norm": 11.237693491740915, | |
| "learning_rate": 2.7905444723949762e-08, | |
| "logits/chosen": -1.2839200496673584, | |
| "logits/rejected": -1.3117491006851196, | |
| "logps/chosen": -376.00311279296875, | |
| "logps/rejected": -447.16436767578125, | |
| "loss": 0.5161, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.0290104150772095, | |
| "rewards/margins": 0.6223047971725464, | |
| "rewards/rejected": -1.6513150930404663, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.8686551543694401, | |
| "grad_norm": 11.12452801324459, | |
| "learning_rate": 2.5845000672245572e-08, | |
| "logits/chosen": -1.328969955444336, | |
| "logits/rejected": -1.3776309490203857, | |
| "logps/chosen": -274.914794921875, | |
| "logps/rejected": -414.8409118652344, | |
| "loss": 0.5253, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.8333606719970703, | |
| "rewards/margins": 1.0471667051315308, | |
| "rewards/rejected": -1.8805274963378906, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.8738880167451596, | |
| "grad_norm": 12.769128426893182, | |
| "learning_rate": 2.385942256943499e-08, | |
| "logits/chosen": -1.0632429122924805, | |
| "logits/rejected": -1.2329734563827515, | |
| "logps/chosen": -372.83538818359375, | |
| "logps/rejected": -433.05462646484375, | |
| "loss": 0.4994, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -1.1593245267868042, | |
| "rewards/margins": 0.6629003286361694, | |
| "rewards/rejected": -1.8222248554229736, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.8791208791208791, | |
| "grad_norm": 11.097228517795964, | |
| "learning_rate": 2.194937358247506e-08, | |
| "logits/chosen": -1.355252981185913, | |
| "logits/rejected": -1.4283853769302368, | |
| "logps/chosen": -370.66558837890625, | |
| "logps/rejected": -464.18096923828125, | |
| "loss": 0.4679, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.9099235534667969, | |
| "rewards/margins": 0.9922647476196289, | |
| "rewards/rejected": -1.9021883010864258, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.8843537414965986, | |
| "grad_norm": 14.718338385740743, | |
| "learning_rate": 2.011549165221127e-08, | |
| "logits/chosen": -1.252229928970337, | |
| "logits/rejected": -1.31440007686615, | |
| "logps/chosen": -340.9755859375, | |
| "logps/rejected": -405.0323791503906, | |
| "loss": 0.4857, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.9458065032958984, | |
| "rewards/margins": 0.9342392086982727, | |
| "rewards/rejected": -1.8800458908081055, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.8895866038723181, | |
| "grad_norm": 12.150291217399005, | |
| "learning_rate": 1.8358389280311303e-08, | |
| "logits/chosen": -1.4331414699554443, | |
| "logits/rejected": -1.4398243427276611, | |
| "logps/chosen": -375.02655029296875, | |
| "logps/rejected": -449.79510498046875, | |
| "loss": 0.4881, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.0200358629226685, | |
| "rewards/margins": 0.7897581458091736, | |
| "rewards/rejected": -1.8097938299179077, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.8948194662480377, | |
| "grad_norm": 16.03783513635316, | |
| "learning_rate": 1.6678653324693787e-08, | |
| "logits/chosen": -1.295977234840393, | |
| "logits/rejected": -1.335905909538269, | |
| "logps/chosen": -410.98858642578125, | |
| "logps/rejected": -475.57720947265625, | |
| "loss": 0.5011, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.9554780125617981, | |
| "rewards/margins": 0.7775296568870544, | |
| "rewards/rejected": -1.733007788658142, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.9000523286237572, | |
| "grad_norm": 13.521037055172467, | |
| "learning_rate": 1.507684480352292e-08, | |
| "logits/chosen": -1.1388336420059204, | |
| "logits/rejected": -1.3264272212982178, | |
| "logps/chosen": -376.89617919921875, | |
| "logps/rejected": -386.2228088378906, | |
| "loss": 0.5166, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.0570180416107178, | |
| "rewards/margins": 0.4624674320220947, | |
| "rewards/rejected": -1.519485354423523, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.9052851909994767, | |
| "grad_norm": 10.015858817548864, | |
| "learning_rate": 1.3553498707832761e-08, | |
| "logits/chosen": -1.2846730947494507, | |
| "logits/rejected": -1.3853490352630615, | |
| "logps/chosen": -296.6762390136719, | |
| "logps/rejected": -344.22509765625, | |
| "loss": 0.4947, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.7420059442520142, | |
| "rewards/margins": 0.8823955655097961, | |
| "rewards/rejected": -1.6244014501571655, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.9105180533751962, | |
| "grad_norm": 10.663148094639538, | |
| "learning_rate": 1.2109123822844653e-08, | |
| "logits/chosen": -1.2846484184265137, | |
| "logits/rejected": -1.3924165964126587, | |
| "logps/chosen": -337.274658203125, | |
| "logps/rejected": -369.1552429199219, | |
| "loss": 0.5413, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.9676492810249329, | |
| "rewards/margins": 0.6571919322013855, | |
| "rewards/rejected": -1.624841332435608, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.9157509157509157, | |
| "grad_norm": 11.75016661537887, | |
| "learning_rate": 1.0744202558037014e-08, | |
| "logits/chosen": -1.2299830913543701, | |
| "logits/rejected": -1.314736247062683, | |
| "logps/chosen": -416.3876037597656, | |
| "logps/rejected": -447.00653076171875, | |
| "loss": 0.5348, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.9475942850112915, | |
| "rewards/margins": 0.7942700982093811, | |
| "rewards/rejected": -1.7418642044067383, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.9209837781266352, | |
| "grad_norm": 10.166512726785776, | |
| "learning_rate": 9.459190786024696e-09, | |
| "logits/chosen": -1.3395822048187256, | |
| "logits/rejected": -1.3754020929336548, | |
| "logps/chosen": -295.82745361328125, | |
| "logps/rejected": -336.22552490234375, | |
| "loss": 0.4754, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.8604118227958679, | |
| "rewards/margins": 0.5532194375991821, | |
| "rewards/rejected": -1.4136312007904053, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.9262166405023547, | |
| "grad_norm": 11.291165980710177, | |
| "learning_rate": 8.254517690300944e-09, | |
| "logits/chosen": -1.1100029945373535, | |
| "logits/rejected": -1.2225925922393799, | |
| "logps/chosen": -354.47979736328125, | |
| "logps/rejected": -421.1259765625, | |
| "loss": 0.5095, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.8763915300369263, | |
| "rewards/margins": 0.9879170656204224, | |
| "rewards/rejected": -1.8643085956573486, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.9314495028780743, | |
| "grad_norm": 11.362297064245054, | |
| "learning_rate": 7.130585621893809e-09, | |
| "logits/chosen": -1.3329540491104126, | |
| "logits/rejected": -1.432592511177063, | |
| "logps/chosen": -308.84466552734375, | |
| "logps/rejected": -347.50848388671875, | |
| "loss": 0.546, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.8689080476760864, | |
| "rewards/margins": 0.6261805295944214, | |
| "rewards/rejected": -1.4950885772705078, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.9366823652537938, | |
| "grad_norm": 11.948692977185269, | |
| "learning_rate": 6.0877699649840574e-09, | |
| "logits/chosen": -1.1590728759765625, | |
| "logits/rejected": -1.1631317138671875, | |
| "logps/chosen": -414.06805419921875, | |
| "logps/rejected": -474.5601501464844, | |
| "loss": 0.5078, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.9375759959220886, | |
| "rewards/margins": 0.5639073848724365, | |
| "rewards/rejected": -1.50148344039917, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.9419152276295133, | |
| "grad_norm": 13.428163080513379, | |
| "learning_rate": 5.126419011529992e-09, | |
| "logits/chosen": -1.2970398664474487, | |
| "logits/rejected": -1.4397534132003784, | |
| "logps/chosen": -339.0660705566406, | |
| "logps/rejected": -468.02801513671875, | |
| "loss": 0.4911, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -1.0843138694763184, | |
| "rewards/margins": 1.0834088325500488, | |
| "rewards/rejected": -2.167722463607788, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.9471480900052328, | |
| "grad_norm": 13.583498099799197, | |
| "learning_rate": 4.246853844940723e-09, | |
| "logits/chosen": -1.2331116199493408, | |
| "logits/rejected": -1.3253166675567627, | |
| "logps/chosen": -342.36871337890625, | |
| "logps/rejected": -370.36529541015625, | |
| "loss": 0.5195, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.9595893621444702, | |
| "rewards/margins": 0.6066954135894775, | |
| "rewards/rejected": -1.5662847757339478, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.9523809523809523, | |
| "grad_norm": 11.121286337310199, | |
| "learning_rate": 3.449368232836869e-09, | |
| "logits/chosen": -1.3873889446258545, | |
| "logits/rejected": -1.5287644863128662, | |
| "logps/chosen": -303.5126647949219, | |
| "logps/rejected": -344.46868896484375, | |
| "loss": 0.4997, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.8543957471847534, | |
| "rewards/margins": 0.759514570236206, | |
| "rewards/rejected": -1.613910436630249, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.957613814756672, | |
| "grad_norm": 16.825805293360574, | |
| "learning_rate": 2.734228528934679e-09, | |
| "logits/chosen": -1.2536653280258179, | |
| "logits/rejected": -1.3062855005264282, | |
| "logps/chosen": -343.5291442871094, | |
| "logps/rejected": -403.6029968261719, | |
| "loss": 0.5206, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.8069379925727844, | |
| "rewards/margins": 0.8626105189323425, | |
| "rewards/rejected": -1.6695486307144165, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.9628466771323915, | |
| "grad_norm": 11.493646679091231, | |
| "learning_rate": 2.1016735840859447e-09, | |
| "logits/chosen": -1.2242063283920288, | |
| "logits/rejected": -1.4777610301971436, | |
| "logps/chosen": -426.0459899902344, | |
| "logps/rejected": -438.92108154296875, | |
| "loss": 0.4916, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -1.0979913473129272, | |
| "rewards/margins": 0.821546733379364, | |
| "rewards/rejected": -1.919538140296936, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.968079539508111, | |
| "grad_norm": 9.306375726544099, | |
| "learning_rate": 1.551914666503812e-09, | |
| "logits/chosen": -1.2090791463851929, | |
| "logits/rejected": -1.3315274715423584, | |
| "logps/chosen": -424.6327209472656, | |
| "logps/rejected": -421.3246154785156, | |
| "loss": 0.4724, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.7541013360023499, | |
| "rewards/margins": 0.6384197473526001, | |
| "rewards/rejected": -1.3925210237503052, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.9733124018838305, | |
| "grad_norm": 11.04339214589406, | |
| "learning_rate": 1.0851353912008642e-09, | |
| "logits/chosen": -1.3252540826797485, | |
| "logits/rejected": -1.5138733386993408, | |
| "logps/chosen": -393.14459228515625, | |
| "logps/rejected": -373.80706787109375, | |
| "loss": 0.4887, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.9482460021972656, | |
| "rewards/margins": 0.6573212742805481, | |
| "rewards/rejected": -1.605567216873169, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.97854526425955, | |
| "grad_norm": 12.262645565063035, | |
| "learning_rate": 7.014916586632336e-10, | |
| "logits/chosen": -1.2882921695709229, | |
| "logits/rejected": -1.4113324880599976, | |
| "logps/chosen": -311.13079833984375, | |
| "logps/rejected": -411.09759521484375, | |
| "loss": 0.5073, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.7763317227363586, | |
| "rewards/margins": 0.7843965291976929, | |
| "rewards/rejected": -1.5607283115386963, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.9837781266352695, | |
| "grad_norm": 9.069425686799821, | |
| "learning_rate": 4.011116027811956e-10, | |
| "logits/chosen": -1.3659498691558838, | |
| "logits/rejected": -1.2314040660858154, | |
| "logps/chosen": -339.2212219238281, | |
| "logps/rejected": -494.2679138183594, | |
| "loss": 0.4938, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.8353347778320312, | |
| "rewards/margins": 0.8836865425109863, | |
| "rewards/rejected": -1.719021201133728, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.989010989010989, | |
| "grad_norm": 13.468123125126455, | |
| "learning_rate": 1.840955480532924e-10, | |
| "logits/chosen": -1.1797103881835938, | |
| "logits/rejected": -1.274377703666687, | |
| "logps/chosen": -481.68157958984375, | |
| "logps/rejected": -480.9568786621094, | |
| "loss": 0.5258, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.9562680125236511, | |
| "rewards/margins": 0.5136700868606567, | |
| "rewards/rejected": -1.4699381589889526, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.9942438513867086, | |
| "grad_norm": 16.36790843814101, | |
| "learning_rate": 5.051597607894087e-11, | |
| "logits/chosen": -1.2505313158035278, | |
| "logits/rejected": -1.3999385833740234, | |
| "logps/chosen": -310.5320739746094, | |
| "logps/rejected": -401.0025939941406, | |
| "loss": 0.5107, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.9702110290527344, | |
| "rewards/margins": 0.921745777130127, | |
| "rewards/rejected": -1.8919566869735718, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.9994767137624281, | |
| "grad_norm": 12.286951734943862, | |
| "learning_rate": 4.1750135001961117e-13, | |
| "logits/chosen": -1.396078109741211, | |
| "logits/rejected": -1.3952105045318604, | |
| "logps/chosen": -433.47412109375, | |
| "logps/rejected": -525.8956298828125, | |
| "loss": 0.5001, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.7201281189918518, | |
| "rewards/margins": 1.2082884311676025, | |
| "rewards/rejected": -1.9284168481826782, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1911, | |
| "total_flos": 0.0, | |
| "train_loss": 0.5605571830478531, | |
| "train_runtime": 8972.3869, | |
| "train_samples_per_second": 6.814, | |
| "train_steps_per_second": 0.213 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1911, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |