| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9996190476190476, | |
| "eval_steps": 500, | |
| "global_step": 1312, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 3.787878787878788e-08, | |
| "logits/chosen": 0.3414086103439331, | |
| "logits/rejected": 0.3618736267089844, | |
| "logps/chosen": -457.8209228515625, | |
| "logps/rejected": -477.39923095703125, | |
| "loss": 0.2223, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.787878787878788e-07, | |
| "logits/chosen": 0.3330618441104889, | |
| "logits/rejected": 0.3205517530441284, | |
| "logps/chosen": -306.0744323730469, | |
| "logps/rejected": -326.5745849609375, | |
| "loss": 0.2099, | |
| "rewards/accuracies": 0.2777777910232544, | |
| "rewards/chosen": -0.00023350439732894301, | |
| "rewards/margins": 6.652800948359072e-05, | |
| "rewards/rejected": -0.0003000324359163642, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.575757575757576e-07, | |
| "logits/chosen": 0.3300573229789734, | |
| "logits/rejected": 0.3766365945339203, | |
| "logps/chosen": -294.01226806640625, | |
| "logps/rejected": -328.8618469238281, | |
| "loss": 0.2087, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 9.059391595656052e-05, | |
| "rewards/margins": 1.657838947721757e-05, | |
| "rewards/rejected": 7.401553739327937e-05, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.1363636363636364e-06, | |
| "logits/chosen": 0.33157598972320557, | |
| "logits/rejected": 0.3647070527076721, | |
| "logps/chosen": -308.50250244140625, | |
| "logps/rejected": -316.5055236816406, | |
| "loss": 0.21, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": -9.306786523666233e-05, | |
| "rewards/margins": -5.975523890811019e-06, | |
| "rewards/rejected": -8.709232497494668e-05, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.5151515151515152e-06, | |
| "logits/chosen": 0.30125534534454346, | |
| "logits/rejected": 0.3123665452003479, | |
| "logps/chosen": -278.3026123046875, | |
| "logps/rejected": -290.43365478515625, | |
| "loss": 0.221, | |
| "rewards/accuracies": 0.20624999701976776, | |
| "rewards/chosen": -1.4783197912038304e-05, | |
| "rewards/margins": 5.556107498705387e-05, | |
| "rewards/rejected": -7.034426380414516e-05, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.8939393939393941e-06, | |
| "logits/chosen": 0.33402004837989807, | |
| "logits/rejected": 0.3797036111354828, | |
| "logps/chosen": -250.8700408935547, | |
| "logps/rejected": -274.0110778808594, | |
| "loss": 0.2, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": 0.0001303141616517678, | |
| "rewards/margins": 2.952014074253384e-05, | |
| "rewards/rejected": 0.00010079403000418097, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.2727272727272728e-06, | |
| "logits/chosen": 0.3142412304878235, | |
| "logits/rejected": 0.3297632336616516, | |
| "logps/chosen": -261.3359069824219, | |
| "logps/rejected": -272.4247131347656, | |
| "loss": 0.216, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 7.56668159738183e-05, | |
| "rewards/margins": 0.0001469437702326104, | |
| "rewards/rejected": -7.12769542587921e-05, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6515151515151514e-06, | |
| "logits/chosen": 0.31074008345603943, | |
| "logits/rejected": 0.3008107542991638, | |
| "logps/chosen": -305.2731628417969, | |
| "logps/rejected": -301.9687805175781, | |
| "loss": 0.21, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.0002583070017863065, | |
| "rewards/margins": -3.105907671852037e-05, | |
| "rewards/rejected": -0.0002272479177918285, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.0303030303030305e-06, | |
| "logits/chosen": 0.310350626707077, | |
| "logits/rejected": 0.3315640389919281, | |
| "logps/chosen": -232.4719696044922, | |
| "logps/rejected": -277.22833251953125, | |
| "loss": 0.1962, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": 9.660933574195951e-05, | |
| "rewards/margins": 0.00016913158469833434, | |
| "rewards/rejected": -7.252227078424767e-05, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.409090909090909e-06, | |
| "logits/chosen": 0.28212469816207886, | |
| "logits/rejected": 0.35147595405578613, | |
| "logps/chosen": -291.93121337890625, | |
| "logps/rejected": -309.50341796875, | |
| "loss": 0.2222, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": -0.0006029252544976771, | |
| "rewards/margins": 0.00015305924171116203, | |
| "rewards/rejected": -0.0007559844525530934, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.7878787878787882e-06, | |
| "logits/chosen": 0.2799969017505646, | |
| "logits/rejected": 0.33390945196151733, | |
| "logps/chosen": -313.59515380859375, | |
| "logps/rejected": -333.5437316894531, | |
| "loss": 0.2158, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": -0.0013320285361260176, | |
| "rewards/margins": 0.00041345407953485847, | |
| "rewards/rejected": -0.001745482673868537, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.166666666666667e-06, | |
| "logits/chosen": 0.3202076852321625, | |
| "logits/rejected": 0.31327325105667114, | |
| "logps/chosen": -339.7900085449219, | |
| "logps/rejected": -391.5009765625, | |
| "loss": 0.2034, | |
| "rewards/accuracies": 0.39375001192092896, | |
| "rewards/chosen": -0.0016250547487288713, | |
| "rewards/margins": 0.000587599235586822, | |
| "rewards/rejected": -0.002212654100731015, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.5454545454545455e-06, | |
| "logits/chosen": 0.2636483311653137, | |
| "logits/rejected": 0.3315524756908417, | |
| "logps/chosen": -257.42376708984375, | |
| "logps/rejected": -279.6871643066406, | |
| "loss": 0.2195, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": -0.0009949387749657035, | |
| "rewards/margins": 0.0003876467817462981, | |
| "rewards/rejected": -0.0013825856149196625, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.924242424242425e-06, | |
| "logits/chosen": 0.25745201110839844, | |
| "logits/rejected": 0.3214574456214905, | |
| "logps/chosen": -302.13787841796875, | |
| "logps/rejected": -334.5384826660156, | |
| "loss": 0.1975, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": -0.0025944842491298914, | |
| "rewards/margins": 0.0014145069289952517, | |
| "rewards/rejected": -0.004008991178125143, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999432965739786e-06, | |
| "logits/chosen": 0.26108819246292114, | |
| "logits/rejected": 0.21493813395500183, | |
| "logps/chosen": -321.00067138671875, | |
| "logps/rejected": -349.4842834472656, | |
| "loss": 0.2053, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": -0.003485953900963068, | |
| "rewards/margins": 0.0010465523228049278, | |
| "rewards/rejected": -0.0045325057581067085, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.997129829895409e-06, | |
| "logits/chosen": 0.16764743626117706, | |
| "logits/rejected": 0.20418021082878113, | |
| "logps/chosen": -216.89199829101562, | |
| "logps/rejected": -234.42935180664062, | |
| "loss": 0.208, | |
| "rewards/accuracies": 0.2562499940395355, | |
| "rewards/chosen": -0.0010050686541944742, | |
| "rewards/margins": 0.0011736680753529072, | |
| "rewards/rejected": -0.0021787367295473814, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9930567839810125e-06, | |
| "logits/chosen": 0.13134139776229858, | |
| "logits/rejected": 0.14908884465694427, | |
| "logps/chosen": -268.61419677734375, | |
| "logps/rejected": -301.322265625, | |
| "loss": 0.2203, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": -0.0038432697765529156, | |
| "rewards/margins": 0.005182007793337107, | |
| "rewards/rejected": -0.009025278501212597, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.987216714880929e-06, | |
| "logits/chosen": 0.1437113881111145, | |
| "logits/rejected": 0.13215014338493347, | |
| "logps/chosen": -259.3898010253906, | |
| "logps/rejected": -267.3370056152344, | |
| "loss": 0.2012, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -0.01510892529040575, | |
| "rewards/margins": 0.0029346179217100143, | |
| "rewards/rejected": -0.01804354228079319, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.979613761906212e-06, | |
| "logits/chosen": 0.07960179448127747, | |
| "logits/rejected": 0.08153261244297028, | |
| "logps/chosen": -357.0888366699219, | |
| "logps/rejected": -368.5904846191406, | |
| "loss": 0.2207, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": -0.021911371499300003, | |
| "rewards/margins": 0.0033388130832463503, | |
| "rewards/rejected": -0.025250181555747986, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.970253313860788e-06, | |
| "logits/chosen": 0.12686966359615326, | |
| "logits/rejected": 0.092551589012146, | |
| "logps/chosen": -323.23162841796875, | |
| "logps/rejected": -354.8710632324219, | |
| "loss": 0.2065, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": -0.027633970603346825, | |
| "rewards/margins": 0.007219144143164158, | |
| "rewards/rejected": -0.03485311195254326, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.959142005221991e-06, | |
| "logits/chosen": 0.030479246750473976, | |
| "logits/rejected": 0.03255733102560043, | |
| "logps/chosen": -324.7137145996094, | |
| "logps/rejected": -360.15802001953125, | |
| "loss": 0.2118, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": -0.036979734897613525, | |
| "rewards/margins": 0.0006652610609307885, | |
| "rewards/rejected": -0.037644993513822556, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.94628771143819e-06, | |
| "logits/chosen": 0.011406493373215199, | |
| "logits/rejected": 0.0717320591211319, | |
| "logps/chosen": -335.43402099609375, | |
| "logps/rejected": -334.8843078613281, | |
| "loss": 0.2127, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": -0.02695571817457676, | |
| "rewards/margins": 0.0005523961153812706, | |
| "rewards/rejected": -0.027508113533258438, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.931699543346854e-06, | |
| "logits/chosen": 0.11805645376443863, | |
| "logits/rejected": 0.13258619606494904, | |
| "logps/chosen": -321.3063049316406, | |
| "logps/rejected": -337.80084228515625, | |
| "loss": 0.221, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": -0.009172406047582626, | |
| "rewards/margins": 0.0013926494866609573, | |
| "rewards/rejected": -0.010565054602921009, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9153878407169815e-06, | |
| "logits/chosen": 0.04168776422739029, | |
| "logits/rejected": 0.052837539464235306, | |
| "logps/chosen": -321.91546630859375, | |
| "logps/rejected": -337.6319274902344, | |
| "loss": 0.2142, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.00017253603436984122, | |
| "rewards/margins": 0.0029792343266308308, | |
| "rewards/rejected": -0.002806698437780142, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.897364164920515e-06, | |
| "logits/chosen": 0.0712689757347107, | |
| "logits/rejected": 0.07041509449481964, | |
| "logps/chosen": -275.91632080078125, | |
| "logps/rejected": -272.06927490234375, | |
| "loss": 0.2087, | |
| "rewards/accuracies": 0.2562499940395355, | |
| "rewards/chosen": 0.0034352331422269344, | |
| "rewards/margins": 0.0012143031926825643, | |
| "rewards/rejected": 0.002220930065959692, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.8776412907378845e-06, | |
| "logits/chosen": 0.0784018337726593, | |
| "logits/rejected": 0.07138744741678238, | |
| "logps/chosen": -270.2529296875, | |
| "logps/rejected": -276.2494812011719, | |
| "loss": 0.2158, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.004147468134760857, | |
| "rewards/margins": 0.0022368025965988636, | |
| "rewards/rejected": 0.0019106656545773149, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8562331973035396e-06, | |
| "logits/chosen": 0.04922521486878395, | |
| "logits/rejected": 0.03360098600387573, | |
| "logps/chosen": -256.4879455566406, | |
| "logps/rejected": -295.20220947265625, | |
| "loss": 0.2052, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.0018207021057605743, | |
| "rewards/margins": -4.667430039262399e-05, | |
| "rewards/rejected": 0.0018673762679100037, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.833155058197842e-06, | |
| "logits/chosen": 0.003297035349532962, | |
| "logits/rejected": 0.05275546759366989, | |
| "logps/chosen": -241.99462890625, | |
| "logps/rejected": -263.92120361328125, | |
| "loss": 0.2039, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.005943115334957838, | |
| "rewards/margins": 0.004178833216428757, | |
| "rewards/rejected": 0.0017642822349444032, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.808423230692374e-06, | |
| "logits/chosen": 0.05449346452951431, | |
| "logits/rejected": 0.06165250390768051, | |
| "logps/chosen": -253.69369506835938, | |
| "logps/rejected": -293.31585693359375, | |
| "loss": 0.2029, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.012908290140330791, | |
| "rewards/margins": 0.007234405726194382, | |
| "rewards/rejected": 0.005673886742442846, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.7820552441562625e-06, | |
| "logits/chosen": -0.005410021636635065, | |
| "logits/rejected": 0.010305285453796387, | |
| "logps/chosen": -270.04083251953125, | |
| "logps/rejected": -316.2156066894531, | |
| "loss": 0.1928, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": 0.010301874950528145, | |
| "rewards/margins": 0.007835518568754196, | |
| "rewards/rejected": 0.002466357545927167, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.754069787631761e-06, | |
| "logits/chosen": -0.05866356939077377, | |
| "logits/rejected": -0.09499181061983109, | |
| "logps/chosen": -276.0648193359375, | |
| "logps/rejected": -317.32037353515625, | |
| "loss": 0.203, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.008139651268720627, | |
| "rewards/margins": 0.008907916024327278, | |
| "rewards/rejected": -0.0007682637078687549, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.724486696587862e-06, | |
| "logits/chosen": -0.08903555572032928, | |
| "logits/rejected": -0.07178256660699844, | |
| "logps/chosen": -291.4513244628906, | |
| "logps/rejected": -304.63726806640625, | |
| "loss": 0.2106, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": 0.001623433781787753, | |
| "rewards/margins": 0.0021732584573328495, | |
| "rewards/rejected": -0.0005498243262991309, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.693326938861367e-06, | |
| "logits/chosen": -0.12407398223876953, | |
| "logits/rejected": -0.06067138910293579, | |
| "logps/chosen": -249.7843475341797, | |
| "logps/rejected": -290.00408935546875, | |
| "loss": 0.2082, | |
| "rewards/accuracies": 0.34375, | |
| "rewards/chosen": 0.010884495452046394, | |
| "rewards/margins": 0.008355050347745419, | |
| "rewards/rejected": 0.002529443707317114, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.660612599795343e-06, | |
| "logits/chosen": -0.11239596456289291, | |
| "logits/rejected": -0.12089481204748154, | |
| "logps/chosen": -290.61737060546875, | |
| "logps/rejected": -312.4447937011719, | |
| "loss": 0.2046, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.007386790122836828, | |
| "rewards/margins": 0.009295441210269928, | |
| "rewards/rejected": -0.0019086506217718124, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.626366866585528e-06, | |
| "logits/chosen": -0.12021325528621674, | |
| "logits/rejected": -0.11920841783285141, | |
| "logps/chosen": -334.55303955078125, | |
| "logps/rejected": -341.6528625488281, | |
| "loss": 0.2095, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.005449886433780193, | |
| "rewards/margins": -0.0001580852986080572, | |
| "rewards/rejected": 0.0056079719215631485, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.590614011845758e-06, | |
| "logits/chosen": -0.0888114720582962, | |
| "logits/rejected": -0.11601515114307404, | |
| "logps/chosen": -275.80316162109375, | |
| "logps/rejected": -300.4934997558594, | |
| "loss": 0.1974, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": 0.01670825481414795, | |
| "rewards/margins": 0.00685430783778429, | |
| "rewards/rejected": 0.009853946976363659, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.553379376404085e-06, | |
| "logits/chosen": -0.1058160662651062, | |
| "logits/rejected": -0.11791692674160004, | |
| "logps/chosen": -266.61322021484375, | |
| "logps/rejected": -279.0782470703125, | |
| "loss": 0.1983, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.021091196686029434, | |
| "rewards/margins": 0.004850545898079872, | |
| "rewards/rejected": 0.01624065265059471, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.514689351341751e-06, | |
| "logits/chosen": -0.16150256991386414, | |
| "logits/rejected": -0.1245572566986084, | |
| "logps/chosen": -277.8587341308594, | |
| "logps/rejected": -272.55792236328125, | |
| "loss": 0.208, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.017171379178762436, | |
| "rewards/margins": 8.804257959127426e-05, | |
| "rewards/rejected": 0.017083335667848587, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.474571359287791e-06, | |
| "logits/chosen": -0.1678142547607422, | |
| "logits/rejected": -0.1356990933418274, | |
| "logps/chosen": -280.3147277832031, | |
| "logps/rejected": -318.61224365234375, | |
| "loss": 0.2096, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 0.024694612249732018, | |
| "rewards/margins": 0.010597179643809795, | |
| "rewards/rejected": 0.014097435399889946, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.4330538349824684e-06, | |
| "logits/chosen": -0.12732991576194763, | |
| "logits/rejected": -0.12050570547580719, | |
| "logps/chosen": -287.99615478515625, | |
| "logps/rejected": -286.3104553222656, | |
| "loss": 0.2079, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.017507802695035934, | |
| "rewards/margins": 1.922808587551117e-05, | |
| "rewards/rejected": 0.017488572746515274, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.3901662051233755e-06, | |
| "logits/chosen": -0.18693214654922485, | |
| "logits/rejected": -0.11668721586465836, | |
| "logps/chosen": -293.45611572265625, | |
| "logps/rejected": -291.56011962890625, | |
| "loss": 0.2156, | |
| "rewards/accuracies": 0.2562499940395355, | |
| "rewards/chosen": 0.011068233288824558, | |
| "rewards/margins": -0.005082954186946154, | |
| "rewards/rejected": 0.01615118607878685, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.345938867508439e-06, | |
| "logits/chosen": -0.06379786878824234, | |
| "logits/rejected": -0.07422082126140594, | |
| "logps/chosen": -299.4942626953125, | |
| "logps/rejected": -322.50201416015625, | |
| "loss": 0.209, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.015014107339084148, | |
| "rewards/margins": 0.003927754703909159, | |
| "rewards/rejected": 0.011086350306868553, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.30040316949064e-06, | |
| "logits/chosen": -0.0699755996465683, | |
| "logits/rejected": -0.08232170343399048, | |
| "logps/chosen": -288.906494140625, | |
| "logps/rejected": -310.1832580566406, | |
| "loss": 0.2186, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.015213333070278168, | |
| "rewards/margins": 0.0060978480614721775, | |
| "rewards/rejected": 0.009115484543144703, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.253591385759705e-06, | |
| "logits/chosen": -0.10615179687738419, | |
| "logits/rejected": -0.06481219828128815, | |
| "logps/chosen": -278.0174255371094, | |
| "logps/rejected": -334.2236328125, | |
| "loss": 0.2108, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.012311488389968872, | |
| "rewards/margins": 0.00595985259860754, | |
| "rewards/rejected": 0.006351633928716183, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.205536695466524e-06, | |
| "logits/chosen": -0.13743802905082703, | |
| "logits/rejected": -0.133576899766922, | |
| "logps/chosen": -274.42333984375, | |
| "logps/rejected": -305.0398864746094, | |
| "loss": 0.1894, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.00840366818010807, | |
| "rewards/margins": 0.004010985605418682, | |
| "rewards/rejected": 0.004392682109028101, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.15627315870651e-06, | |
| "logits/chosen": -0.2103540152311325, | |
| "logits/rejected": -0.19342456758022308, | |
| "logps/chosen": -282.6641540527344, | |
| "logps/rejected": -310.85113525390625, | |
| "loss": 0.1974, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.00448573287576437, | |
| "rewards/margins": 0.0029663294553756714, | |
| "rewards/rejected": 0.0015194038860499859, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.105835692378557e-06, | |
| "logits/chosen": -0.1941739320755005, | |
| "logits/rejected": -0.22001402080059052, | |
| "logps/chosen": -267.8673095703125, | |
| "logps/rejected": -265.15618896484375, | |
| "loss": 0.2101, | |
| "rewards/accuracies": 0.2562499940395355, | |
| "rewards/chosen": 0.007565724663436413, | |
| "rewards/margins": 0.0029779626056551933, | |
| "rewards/rejected": 0.004587762989103794, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.05426004543672e-06, | |
| "logits/chosen": -0.2290777713060379, | |
| "logits/rejected": -0.22092656791210175, | |
| "logps/chosen": -267.7095947265625, | |
| "logps/rejected": -307.23468017578125, | |
| "loss": 0.2034, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.013477683067321777, | |
| "rewards/margins": 0.006231679115444422, | |
| "rewards/rejected": 0.007246003951877356, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.001582773552153e-06, | |
| "logits/chosen": -0.2154918909072876, | |
| "logits/rejected": -0.2263382226228714, | |
| "logps/chosen": -236.22079467773438, | |
| "logps/rejected": -281.47308349609375, | |
| "loss": 0.208, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.008721614256501198, | |
| "rewards/margins": 0.004514150787144899, | |
| "rewards/rejected": 0.004207463003695011, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.947841213203262e-06, | |
| "logits/chosen": -0.25622615218162537, | |
| "logits/rejected": -0.28361567854881287, | |
| "logps/chosen": -307.4781494140625, | |
| "logps/rejected": -311.0342712402344, | |
| "loss": 0.2118, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.00395470205694437, | |
| "rewards/margins": 0.005559145472943783, | |
| "rewards/rejected": -0.001604443765245378, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.893073455212438e-06, | |
| "logits/chosen": -0.2454405277967453, | |
| "logits/rejected": -0.23036327958106995, | |
| "logps/chosen": -305.7126159667969, | |
| "logps/rejected": -320.9441833496094, | |
| "loss": 0.1978, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.006379422731697559, | |
| "rewards/margins": 0.0032655359245836735, | |
| "rewards/rejected": 0.003113886807113886, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.837318317748134e-06, | |
| "logits/chosen": -0.23187056183815002, | |
| "logits/rejected": -0.25806108117103577, | |
| "logps/chosen": -295.49847412109375, | |
| "logps/rejected": -312.9036560058594, | |
| "loss": 0.2095, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": 0.010056470520794392, | |
| "rewards/margins": 0.005588938947767019, | |
| "rewards/rejected": 0.004467530641704798, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.7806153188114027e-06, | |
| "logits/chosen": -0.28555235266685486, | |
| "logits/rejected": -0.27182719111442566, | |
| "logps/chosen": -278.7008361816406, | |
| "logps/rejected": -286.72845458984375, | |
| "loss": 0.2112, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.0004449138359632343, | |
| "rewards/margins": 0.0027928180061280727, | |
| "rewards/rejected": -0.002347904024645686, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.7230046482264256e-06, | |
| "logits/chosen": -0.29695600271224976, | |
| "logits/rejected": -0.30071455240249634, | |
| "logps/chosen": -303.6511535644531, | |
| "logps/rejected": -311.7902526855469, | |
| "loss": 0.2241, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.0028660064563155174, | |
| "rewards/margins": -0.001624327152967453, | |
| "rewards/rejected": 0.00449033360928297, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.6645271391548542e-06, | |
| "logits/chosen": -0.2697906196117401, | |
| "logits/rejected": -0.2420235425233841, | |
| "logps/chosen": -287.1878662109375, | |
| "logps/rejected": -289.3710632324219, | |
| "loss": 0.1866, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.010242622345685959, | |
| "rewards/margins": 0.0021990591194480658, | |
| "rewards/rejected": 0.008043562062084675, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6052242391541746e-06, | |
| "logits/chosen": -0.2508285641670227, | |
| "logits/rejected": -0.22416014969348907, | |
| "logps/chosen": -282.50885009765625, | |
| "logps/rejected": -304.0210876464844, | |
| "loss": 0.1951, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.014059506356716156, | |
| "rewards/margins": 0.006549051962792873, | |
| "rewards/rejected": 0.007510454393923283, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.5451379808006014e-06, | |
| "logits/chosen": -0.2925412654876709, | |
| "logits/rejected": -0.27184224128723145, | |
| "logps/chosen": -226.45083618164062, | |
| "logps/rejected": -252.47207641601562, | |
| "loss": 0.2026, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": 0.006832784973084927, | |
| "rewards/margins": 0.006407728884369135, | |
| "rewards/rejected": 0.00042505591409280896, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.484310951897323e-06, | |
| "logits/chosen": -0.280683308839798, | |
| "logits/rejected": -0.2360977679491043, | |
| "logps/chosen": -272.83953857421875, | |
| "logps/rejected": -299.12847900390625, | |
| "loss": 0.1933, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.0044427914544939995, | |
| "rewards/margins": 0.007019062992185354, | |
| "rewards/rejected": -0.0025762729346752167, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4227862652892106e-06, | |
| "logits/chosen": -0.23029498755931854, | |
| "logits/rejected": -0.30575060844421387, | |
| "logps/chosen": -284.2434997558594, | |
| "logps/rejected": -325.15606689453125, | |
| "loss": 0.2113, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.0008473257767036557, | |
| "rewards/margins": 0.0063803717494010925, | |
| "rewards/rejected": -0.007227697875350714, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3606075283054005e-06, | |
| "logits/chosen": -0.21216976642608643, | |
| "logits/rejected": -0.24321305751800537, | |
| "logps/chosen": -289.8013610839844, | |
| "logps/rejected": -292.233642578125, | |
| "loss": 0.2076, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -0.006494040135294199, | |
| "rewards/margins": 0.0029291484970599413, | |
| "rewards/rejected": -0.009423188865184784, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2978188118513814e-06, | |
| "logits/chosen": -0.3345397114753723, | |
| "logits/rejected": -0.2698872983455658, | |
| "logps/chosen": -288.9801025390625, | |
| "logps/rejected": -325.5001525878906, | |
| "loss": 0.2204, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": -0.005164691712707281, | |
| "rewards/margins": 0.006596796214580536, | |
| "rewards/rejected": -0.011761486530303955, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.234464619172522e-06, | |
| "logits/chosen": -0.28216275572776794, | |
| "logits/rejected": -0.36517441272735596, | |
| "logps/chosen": -275.3839416503906, | |
| "logps/rejected": -297.3681640625, | |
| "loss": 0.2018, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.007699407637119293, | |
| "rewards/margins": 0.0056534623727202415, | |
| "rewards/rejected": -0.01335287094116211, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.1705898543111576e-06, | |
| "logits/chosen": -0.32491374015808105, | |
| "logits/rejected": -0.31058216094970703, | |
| "logps/chosen": -318.451904296875, | |
| "logps/rejected": -336.5116271972656, | |
| "loss": 0.2052, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": -0.007857877761125565, | |
| "rewards/margins": 0.005408051423728466, | |
| "rewards/rejected": -0.013265928253531456, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.106239790279606e-06, | |
| "logits/chosen": -0.3316212594509125, | |
| "logits/rejected": -0.27467650175094604, | |
| "logps/chosen": -306.818359375, | |
| "logps/rejected": -310.139892578125, | |
| "loss": 0.2198, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.008624967187643051, | |
| "rewards/margins": 0.0013632235350087285, | |
| "rewards/rejected": -0.009988190606236458, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.041460036971664e-06, | |
| "logits/chosen": -0.30510929226875305, | |
| "logits/rejected": -0.2651751637458801, | |
| "logps/chosen": -289.76177978515625, | |
| "logps/rejected": -306.5693664550781, | |
| "loss": 0.2091, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": -0.0034012228716164827, | |
| "rewards/margins": 0.005689640529453754, | |
| "rewards/rejected": -0.009090864099562168, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.976296508835326e-06, | |
| "logits/chosen": -0.27622857689857483, | |
| "logits/rejected": -0.3029255270957947, | |
| "logps/chosen": -254.8801727294922, | |
| "logps/rejected": -280.85552978515625, | |
| "loss": 0.2118, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": 0.003135355655103922, | |
| "rewards/margins": 0.00885056797415018, | |
| "rewards/rejected": -0.005715211853384972, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.910795392329649e-06, | |
| "logits/chosen": -0.29605865478515625, | |
| "logits/rejected": -0.27770885825157166, | |
| "logps/chosen": -277.80804443359375, | |
| "logps/rejected": -309.5155029296875, | |
| "loss": 0.2133, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": 0.0010512445587664843, | |
| "rewards/margins": 0.008048586547374725, | |
| "rewards/rejected": -0.006997342221438885, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.8450031131888147e-06, | |
| "logits/chosen": -0.35795870423316956, | |
| "logits/rejected": -0.37449702620506287, | |
| "logps/chosen": -289.4441833496094, | |
| "logps/rejected": -332.01055908203125, | |
| "loss": 0.1977, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.004084504209458828, | |
| "rewards/margins": 0.011728955432772636, | |
| "rewards/rejected": -0.0076444498263299465, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7789663035166035e-06, | |
| "logits/chosen": -0.35911503434181213, | |
| "logits/rejected": -0.3697941303253174, | |
| "logps/chosen": -251.0248565673828, | |
| "logps/rejected": -283.61279296875, | |
| "loss": 0.2036, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.0005092100473120809, | |
| "rewards/margins": 0.011741384863853455, | |
| "rewards/rejected": -0.011232174932956696, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.7127317687345973e-06, | |
| "logits/chosen": -0.4329107403755188, | |
| "logits/rejected": -0.38218554854393005, | |
| "logps/chosen": -275.99176025390625, | |
| "logps/rejected": -282.9251708984375, | |
| "loss": 0.2255, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": -0.0015629607951268554, | |
| "rewards/margins": 0.005656304769217968, | |
| "rewards/rejected": -0.007219265215098858, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.6463464544075344e-06, | |
| "logits/chosen": -0.38624441623687744, | |
| "logits/rejected": -0.4270317554473877, | |
| "logps/chosen": -313.55047607421875, | |
| "logps/rejected": -360.2120361328125, | |
| "loss": 0.1998, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": -0.0040723588317632675, | |
| "rewards/margins": 0.013303135521709919, | |
| "rewards/rejected": -0.01737549528479576, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.579857412969345e-06, | |
| "logits/chosen": -0.4271603226661682, | |
| "logits/rejected": -0.4234057366847992, | |
| "logps/chosen": -304.10211181640625, | |
| "logps/rejected": -354.1361389160156, | |
| "loss": 0.1979, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": -0.007728138472884893, | |
| "rewards/margins": 0.016483409330248833, | |
| "rewards/rejected": -0.024211544543504715, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.513311770373421e-06, | |
| "logits/chosen": -0.5024863481521606, | |
| "logits/rejected": -0.47420111298561096, | |
| "logps/chosen": -317.60174560546875, | |
| "logps/rejected": -362.4761657714844, | |
| "loss": 0.192, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": -0.011080259457230568, | |
| "rewards/margins": 0.010900650173425674, | |
| "rewards/rejected": -0.021980909630656242, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.446756692690804e-06, | |
| "logits/chosen": -0.4976615905761719, | |
| "logits/rejected": -0.4921053349971771, | |
| "logps/chosen": -306.67547607421875, | |
| "logps/rejected": -361.6275939941406, | |
| "loss": 0.2069, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": -0.006612093187868595, | |
| "rewards/margins": 0.014301796443760395, | |
| "rewards/rejected": -0.02091388963162899, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.380239352679908e-06, | |
| "logits/chosen": -0.45196834206581116, | |
| "logits/rejected": -0.48185840249061584, | |
| "logps/chosen": -313.2336730957031, | |
| "logps/rejected": -313.48260498046875, | |
| "loss": 0.2149, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": -0.010375128127634525, | |
| "rewards/margins": 0.004926495254039764, | |
| "rewards/rejected": -0.015301624312996864, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.313806896351529e-06, | |
| "logits/chosen": -0.5004686117172241, | |
| "logits/rejected": -0.47765883803367615, | |
| "logps/chosen": -274.7811279296875, | |
| "logps/rejected": -311.44207763671875, | |
| "loss": 0.1904, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.002466167788952589, | |
| "rewards/margins": 0.012883700430393219, | |
| "rewards/rejected": -0.010417533107101917, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.247506409552795e-06, | |
| "logits/chosen": -0.563784122467041, | |
| "logits/rejected": -0.5083848237991333, | |
| "logps/chosen": -286.28277587890625, | |
| "logps/rejected": -303.66925048828125, | |
| "loss": 0.2037, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.0052384668961167336, | |
| "rewards/margins": 0.004287950228899717, | |
| "rewards/rejected": -0.009526416659355164, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1813848845937695e-06, | |
| "logits/chosen": -0.5057908296585083, | |
| "logits/rejected": -0.48972076177597046, | |
| "logps/chosen": -297.547607421875, | |
| "logps/rejected": -294.28863525390625, | |
| "loss": 0.2101, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": -0.0030983686447143555, | |
| "rewards/margins": 0.0027166479267179966, | |
| "rewards/rejected": -0.00581501517444849, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1154891869403436e-06, | |
| "logits/chosen": -0.472622811794281, | |
| "logits/rejected": -0.4686339497566223, | |
| "logps/chosen": -241.92630004882812, | |
| "logps/rejected": -279.33062744140625, | |
| "loss": 0.2091, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.009422892704606056, | |
| "rewards/margins": 0.009348099119961262, | |
| "rewards/rejected": 7.479395571863279e-05, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0498660219970395e-06, | |
| "logits/chosen": -0.44067639112472534, | |
| "logits/rejected": -0.46692126989364624, | |
| "logps/chosen": -285.769287109375, | |
| "logps/rejected": -305.04296875, | |
| "loss": 0.2271, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.008178983815014362, | |
| "rewards/margins": 0.007535718381404877, | |
| "rewards/rejected": 0.0006432650843635201, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9845619020032552e-06, | |
| "logits/chosen": -0.4349413514137268, | |
| "logits/rejected": -0.4484075605869293, | |
| "logps/chosen": -305.90972900390625, | |
| "logps/rejected": -287.2285461425781, | |
| "loss": 0.2105, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.006356480531394482, | |
| "rewards/margins": 0.0014576372923329473, | |
| "rewards/rejected": 0.004898843355476856, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.9196231130664282e-06, | |
| "logits/chosen": -0.45880216360092163, | |
| "logits/rejected": -0.44850587844848633, | |
| "logps/chosen": -263.9127502441406, | |
| "logps/rejected": -292.25177001953125, | |
| "loss": 0.2033, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.0159307774156332, | |
| "rewards/margins": 0.009099806658923626, | |
| "rewards/rejected": 0.006830970756709576, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8550956823554708e-06, | |
| "logits/chosen": -0.4190438687801361, | |
| "logits/rejected": -0.39324015378952026, | |
| "logps/chosen": -297.383056640625, | |
| "logps/rejected": -307.9434814453125, | |
| "loss": 0.2115, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.008881737478077412, | |
| "rewards/margins": 0.0028643650002777576, | |
| "rewards/rejected": 0.006017371080815792, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.7910253454777346e-06, | |
| "logits/chosen": -0.4339013993740082, | |
| "logits/rejected": -0.4780551791191101, | |
| "logps/chosen": -264.0601501464844, | |
| "logps/rejected": -309.7989807128906, | |
| "loss": 0.2071, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": 0.01502480823546648, | |
| "rewards/margins": 0.013155427761375904, | |
| "rewards/rejected": 0.0018693817546591163, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.7274575140626318e-06, | |
| "logits/chosen": -0.413571834564209, | |
| "logits/rejected": -0.4285905361175537, | |
| "logps/chosen": -263.08453369140625, | |
| "logps/rejected": -286.2958068847656, | |
| "loss": 0.2184, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": 0.008341384120285511, | |
| "rewards/margins": 0.0012967393267899752, | |
| "rewards/rejected": 0.007044644560664892, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6644372435748823e-06, | |
| "logits/chosen": -0.408627986907959, | |
| "logits/rejected": -0.4011750817298889, | |
| "logps/chosen": -229.0798797607422, | |
| "logps/rejected": -262.96075439453125, | |
| "loss": 0.2043, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.0140065373852849, | |
| "rewards/margins": 0.007881510071456432, | |
| "rewards/rejected": 0.006125027313828468, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.6020092013802002e-06, | |
| "logits/chosen": -0.43513408303260803, | |
| "logits/rejected": -0.4182719588279724, | |
| "logps/chosen": -283.6298828125, | |
| "logps/rejected": -299.7771301269531, | |
| "loss": 0.2094, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.015776053071022034, | |
| "rewards/margins": 0.005827675573527813, | |
| "rewards/rejected": 0.00994837749749422, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5402176350860653e-06, | |
| "logits/chosen": -0.4543350338935852, | |
| "logits/rejected": -0.43857789039611816, | |
| "logps/chosen": -258.48822021484375, | |
| "logps/rejected": -309.0599365234375, | |
| "loss": 0.2144, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.01519864983856678, | |
| "rewards/margins": 0.0170084610581398, | |
| "rewards/rejected": -0.0018098097061738372, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4791063411799938e-06, | |
| "logits/chosen": -0.4367285370826721, | |
| "logits/rejected": -0.4124454855918884, | |
| "logps/chosen": -258.4836120605469, | |
| "logps/rejected": -256.78948974609375, | |
| "loss": 0.2125, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": 0.010697437450289726, | |
| "rewards/margins": 0.0008281940827146173, | |
| "rewards/rejected": 0.009869244880974293, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.4187186339875697e-06, | |
| "logits/chosen": -0.3717727065086365, | |
| "logits/rejected": -0.3890214264392853, | |
| "logps/chosen": -238.35391235351562, | |
| "logps/rejected": -270.5140686035156, | |
| "loss": 0.202, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": 0.015574706718325615, | |
| "rewards/margins": 0.009048102423548698, | |
| "rewards/rejected": 0.006526602897793055, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3590973149722103e-06, | |
| "logits/chosen": -0.4208298325538635, | |
| "logits/rejected": -0.4319419860839844, | |
| "logps/chosen": -272.0270690917969, | |
| "logps/rejected": -303.1417541503906, | |
| "loss": 0.2083, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.015811121091246605, | |
| "rewards/margins": 0.010845163837075233, | |
| "rewards/rejected": 0.004965959116816521, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.300284642398445e-06, | |
| "logits/chosen": -0.47379741072654724, | |
| "logits/rejected": -0.4520508348941803, | |
| "logps/chosen": -262.06292724609375, | |
| "logps/rejected": -311.0146789550781, | |
| "loss": 0.1942, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.014805925078690052, | |
| "rewards/margins": 0.01618744060397148, | |
| "rewards/rejected": -0.0013815152924507856, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2423223013801946e-06, | |
| "logits/chosen": -0.45992302894592285, | |
| "logits/rejected": -0.4493522644042969, | |
| "logps/chosen": -277.81231689453125, | |
| "logps/rejected": -303.0189514160156, | |
| "loss": 0.2059, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.007296368479728699, | |
| "rewards/margins": 0.007703060749918222, | |
| "rewards/rejected": -0.00040669291047379375, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1852513743352886e-06, | |
| "logits/chosen": -0.4434199929237366, | |
| "logits/rejected": -0.4254834055900574, | |
| "logps/chosen": -236.1838836669922, | |
| "logps/rejected": -280.4830627441406, | |
| "loss": 0.2005, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.00997086614370346, | |
| "rewards/margins": 0.012648127973079681, | |
| "rewards/rejected": -0.0026772604323923588, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.1291123118671665e-06, | |
| "logits/chosen": -0.4678889811038971, | |
| "logits/rejected": -0.4497453570365906, | |
| "logps/chosen": -260.39630126953125, | |
| "logps/rejected": -310.59783935546875, | |
| "loss": 0.2033, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.013568739406764507, | |
| "rewards/margins": 0.013490801677107811, | |
| "rewards/rejected": 7.793791883159429e-05, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.073944904094385e-06, | |
| "logits/chosen": -0.4590323865413666, | |
| "logits/rejected": -0.4857376217842102, | |
| "logps/chosen": -250.05941772460938, | |
| "logps/rejected": -269.81182861328125, | |
| "loss": 0.2081, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": 0.009169178083539009, | |
| "rewards/margins": 0.0009465640177950263, | |
| "rewards/rejected": 0.008222613483667374, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.019788252448267e-06, | |
| "logits/chosen": -0.42771902680397034, | |
| "logits/rejected": -0.4784491956233978, | |
| "logps/chosen": -286.91033935546875, | |
| "logps/rejected": -356.3207092285156, | |
| "loss": 0.2176, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.009989687241613865, | |
| "rewards/margins": 0.01874762959778309, | |
| "rewards/rejected": -0.008757943287491798, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.66680741958685e-07, | |
| "logits/chosen": -0.5021045804023743, | |
| "logits/rejected": -0.4944073557853699, | |
| "logps/chosen": -256.72845458984375, | |
| "logps/rejected": -282.7173767089844, | |
| "loss": 0.2023, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.008730728179216385, | |
| "rewards/margins": 0.010655781254172325, | |
| "rewards/rejected": -0.001925053307786584, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.146600140475945e-07, | |
| "logits/chosen": -0.49039751291275024, | |
| "logits/rejected": -0.4602015018463135, | |
| "logps/chosen": -319.5450134277344, | |
| "logps/rejected": -356.46478271484375, | |
| "loss": 0.2141, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.0025001950562000275, | |
| "rewards/margins": 0.0142419608309865, | |
| "rewards/rejected": -0.011741766706109047, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.637629398496378e-07, | |
| "logits/chosen": -0.4612942636013031, | |
| "logits/rejected": -0.4650956988334656, | |
| "logps/chosen": -314.468017578125, | |
| "logps/rejected": -336.6165466308594, | |
| "loss": 0.2201, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.004640791565179825, | |
| "rewards/margins": 0.011841602623462677, | |
| "rewards/rejected": -0.0072008101269602776, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.140255940787059e-07, | |
| "logits/chosen": -0.4051927924156189, | |
| "logits/rejected": -0.41572269797325134, | |
| "logps/chosen": -266.1190185546875, | |
| "logps/rejected": -307.9905090332031, | |
| "loss": 0.1983, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.005232260562479496, | |
| "rewards/margins": 0.011333976872265339, | |
| "rewards/rejected": -0.006101716309785843, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.654832294589776e-07, | |
| "logits/chosen": -0.433448851108551, | |
| "logits/rejected": -0.4530878961086273, | |
| "logps/chosen": -264.8626403808594, | |
| "logps/rejected": -305.2262878417969, | |
| "loss": 0.2114, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.004321249667555094, | |
| "rewards/margins": 0.009284446947276592, | |
| "rewards/rejected": -0.004963197745382786, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.181702517385789e-07, | |
| "logits/chosen": -0.470588743686676, | |
| "logits/rejected": -0.4782032370567322, | |
| "logps/chosen": -274.71527099609375, | |
| "logps/rejected": -307.9129943847656, | |
| "loss": 0.1997, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.007287648506462574, | |
| "rewards/margins": 0.00972357951104641, | |
| "rewards/rejected": -0.0024359312374144793, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.721201953035511e-07, | |
| "logits/chosen": -0.4322214126586914, | |
| "logits/rejected": -0.458743155002594, | |
| "logps/chosen": -262.1369323730469, | |
| "logps/rejected": -275.8923034667969, | |
| "loss": 0.2108, | |
| "rewards/accuracies": 0.23749999701976776, | |
| "rewards/chosen": 0.0005865416023880243, | |
| "rewards/margins": 0.000941948383115232, | |
| "rewards/rejected": -0.00035540637327358127, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.273656994094232e-07, | |
| "logits/chosen": -0.4811081886291504, | |
| "logits/rejected": -0.40350890159606934, | |
| "logps/chosen": -272.11419677734375, | |
| "logps/rejected": -307.12371826171875, | |
| "loss": 0.1991, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.005647097248584032, | |
| "rewards/margins": 0.007634321693331003, | |
| "rewards/rejected": -0.0019872249104082584, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.839384850472359e-07, | |
| "logits/chosen": -0.46519404649734497, | |
| "logits/rejected": -0.40799570083618164, | |
| "logps/chosen": -289.98883056640625, | |
| "logps/rejected": -292.3999938964844, | |
| "loss": 0.2163, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.008130472153425217, | |
| "rewards/margins": 0.002877952065318823, | |
| "rewards/rejected": 0.005252520553767681, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.418693324604082e-07, | |
| "logits/chosen": -0.5294805765151978, | |
| "logits/rejected": -0.5169610977172852, | |
| "logps/chosen": -272.34686279296875, | |
| "logps/rejected": -307.16412353515625, | |
| "loss": 0.1938, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": 0.012028077617287636, | |
| "rewards/margins": 0.013656511902809143, | |
| "rewards/rejected": -0.0016284309094771743, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 5.01188059328386e-07, | |
| "logits/chosen": -0.5478223562240601, | |
| "logits/rejected": -0.4814823269844055, | |
| "logps/chosen": -289.8566589355469, | |
| "logps/rejected": -289.57525634765625, | |
| "loss": 0.2065, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.008872238919138908, | |
| "rewards/margins": 0.006515635643154383, | |
| "rewards/rejected": 0.0023566028103232384, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.619234996325314e-07, | |
| "logits/chosen": -0.4742591977119446, | |
| "logits/rejected": -0.45757928490638733, | |
| "logps/chosen": -328.6953125, | |
| "logps/rejected": -340.1205139160156, | |
| "loss": 0.2115, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.006806342396885157, | |
| "rewards/margins": 0.0063201361335814, | |
| "rewards/rejected": 0.000486206088680774, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.241034832192434e-07, | |
| "logits/chosen": -0.5209555625915527, | |
| "logits/rejected": -0.5151633024215698, | |
| "logps/chosen": -276.89581298828125, | |
| "logps/rejected": -273.1256103515625, | |
| "loss": 0.2135, | |
| "rewards/accuracies": 0.24375000596046448, | |
| "rewards/chosen": 0.0027219385374337435, | |
| "rewards/margins": 0.0031283677089959383, | |
| "rewards/rejected": -0.00040642902604304254, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.877548160747768e-07, | |
| "logits/chosen": -0.4666425585746765, | |
| "logits/rejected": -0.48001235723495483, | |
| "logps/chosen": -280.6470031738281, | |
| "logps/rejected": -315.48443603515625, | |
| "loss": 0.2048, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.002273663878440857, | |
| "rewards/margins": 0.008108812384307384, | |
| "rewards/rejected": -0.0058351485058665276, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.529032613257574e-07, | |
| "logits/chosen": -0.4815247654914856, | |
| "logits/rejected": -0.45564061403274536, | |
| "logps/chosen": -300.01959228515625, | |
| "logps/rejected": -330.40899658203125, | |
| "loss": 0.2076, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.0031171857845038176, | |
| "rewards/margins": 0.009788629598915577, | |
| "rewards/rejected": -0.006671445909887552, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.195735209788528e-07, | |
| "logits/chosen": -0.4933408796787262, | |
| "logits/rejected": -0.5491820573806763, | |
| "logps/chosen": -258.11822509765625, | |
| "logps/rejected": -296.2974853515625, | |
| "loss": 0.213, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.008377974852919579, | |
| "rewards/margins": 0.012130925431847572, | |
| "rewards/rejected": -0.003752949880436063, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.8778921841253774e-07, | |
| "logits/chosen": -0.43251729011535645, | |
| "logits/rejected": -0.5054982900619507, | |
| "logps/chosen": -281.94403076171875, | |
| "logps/rejected": -280.0823669433594, | |
| "loss": 0.2204, | |
| "rewards/accuracies": 0.23125000298023224, | |
| "rewards/chosen": 0.0010522855445742607, | |
| "rewards/margins": 0.00011768531840061769, | |
| "rewards/rejected": 0.0009346003644168377, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.5757288163336806e-07, | |
| "logits/chosen": -0.4674537181854248, | |
| "logits/rejected": -0.43040981888771057, | |
| "logps/chosen": -267.8955078125, | |
| "logps/rejected": -268.64312744140625, | |
| "loss": 0.2134, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -0.00031215575290843844, | |
| "rewards/margins": -0.00015194570005405694, | |
| "rewards/rejected": -0.00016021067858673632, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.2894592730863336e-07, | |
| "logits/chosen": -0.4933416247367859, | |
| "logits/rejected": -0.4184231758117676, | |
| "logps/chosen": -274.43487548828125, | |
| "logps/rejected": -287.72259521484375, | |
| "loss": 0.2058, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.012380121275782585, | |
| "rewards/margins": 0.007005924824625254, | |
| "rewards/rejected": 0.005374195985496044, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.019286455866981e-07, | |
| "logits/chosen": -0.5218170881271362, | |
| "logits/rejected": -0.46864786744117737, | |
| "logps/chosen": -298.1045227050781, | |
| "logps/rejected": -322.55181884765625, | |
| "loss": 0.1925, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": -0.0008321186760440469, | |
| "rewards/margins": 0.004535870626568794, | |
| "rewards/rejected": -0.005367989186197519, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.7654018571579557e-07, | |
| "logits/chosen": -0.48008307814598083, | |
| "logits/rejected": -0.4698426127433777, | |
| "logps/chosen": -307.2906494140625, | |
| "logps/rejected": -356.3848571777344, | |
| "loss": 0.2072, | |
| "rewards/accuracies": 0.3812499940395355, | |
| "rewards/chosen": 0.013308033347129822, | |
| "rewards/margins": 0.019972598180174828, | |
| "rewards/rejected": -0.00666456576436758, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.5279854247146703e-07, | |
| "logits/chosen": -0.46893006563186646, | |
| "logits/rejected": -0.4585728645324707, | |
| "logps/chosen": -340.1523132324219, | |
| "logps/rejected": -351.4723205566406, | |
| "loss": 0.2135, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.0029886148404330015, | |
| "rewards/margins": 0.005356112495064735, | |
| "rewards/rejected": -0.002367498353123665, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.307205434022671e-07, | |
| "logits/chosen": -0.4998309016227722, | |
| "logits/rejected": -0.49961596727371216, | |
| "logps/chosen": -272.09832763671875, | |
| "logps/rejected": -272.3662109375, | |
| "loss": 0.1976, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": 0.0032546469010412693, | |
| "rewards/margins": 0.003977665212005377, | |
| "rewards/rejected": -0.0007230177288874984, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1032183690276754e-07, | |
| "logits/chosen": -0.518333911895752, | |
| "logits/rejected": -0.5002647638320923, | |
| "logps/chosen": -280.9651794433594, | |
| "logps/rejected": -300.6822204589844, | |
| "loss": 0.2098, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": 0.002680544275790453, | |
| "rewards/margins": 0.003743491368368268, | |
| "rewards/rejected": -0.001062947092577815, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.161688112232836e-08, | |
| "logits/chosen": -0.46787554025650024, | |
| "logits/rejected": -0.45076021552085876, | |
| "logps/chosen": -282.44903564453125, | |
| "logps/rejected": -328.4521789550781, | |
| "loss": 0.2076, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": 0.012471871450543404, | |
| "rewards/margins": 0.01295357383787632, | |
| "rewards/rejected": -0.00048170099034905434, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.46189337174788e-08, | |
| "logits/chosen": -0.4574583172798157, | |
| "logits/rejected": -0.48262113332748413, | |
| "logps/chosen": -287.9381408691406, | |
| "logps/rejected": -307.37017822265625, | |
| "loss": 0.2135, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.01407331507652998, | |
| "rewards/margins": 0.009314117953181267, | |
| "rewards/rejected": 0.004759198985993862, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.934004245518793e-08, | |
| "logits/chosen": -0.5343510508537292, | |
| "logits/rejected": -0.42381030321121216, | |
| "logps/chosen": -301.0657958984375, | |
| "logps/rejected": -314.10260009765625, | |
| "loss": 0.2144, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.011122044175863266, | |
| "rewards/margins": 0.00779320765286684, | |
| "rewards/rejected": 0.0033288367558270693, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.579103667367385e-08, | |
| "logits/chosen": -0.5401466488838196, | |
| "logits/rejected": -0.4905480444431305, | |
| "logps/chosen": -302.44818115234375, | |
| "logps/rejected": -316.00640869140625, | |
| "loss": 0.2009, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.009371964260935783, | |
| "rewards/margins": 0.005004683043807745, | |
| "rewards/rejected": 0.004367280751466751, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.398151960681162e-08, | |
| "logits/chosen": -0.4864376485347748, | |
| "logits/rejected": -0.474257230758667, | |
| "logps/chosen": -272.18450927734375, | |
| "logps/rejected": -293.8534851074219, | |
| "loss": 0.212, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.010555078275501728, | |
| "rewards/margins": 0.007184301503002644, | |
| "rewards/rejected": 0.003370775608345866, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.3919861577572924e-08, | |
| "logits/chosen": -0.48694509267807007, | |
| "logits/rejected": -0.4881502091884613, | |
| "logps/chosen": -285.6047668457031, | |
| "logps/rejected": -344.7032470703125, | |
| "loss": 0.207, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.010286719538271427, | |
| "rewards/margins": 0.011380873620510101, | |
| "rewards/rejected": -0.0010941538494080305, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.5613194065327854e-08, | |
| "logits/chosen": -0.4984816610813141, | |
| "logits/rejected": -0.4886694550514221, | |
| "logps/chosen": -341.03369140625, | |
| "logps/rejected": -376.7138977050781, | |
| "loss": 0.1955, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 0.008053203113377094, | |
| "rewards/margins": 0.012355836108326912, | |
| "rewards/rejected": -0.004302632994949818, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.067404651211808e-09, | |
| "logits/chosen": -0.47365856170654297, | |
| "logits/rejected": -0.463469922542572, | |
| "logps/chosen": -296.1658630371094, | |
| "logps/rejected": -324.7825012207031, | |
| "loss": 0.2016, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": 0.00670830812305212, | |
| "rewards/margins": 0.009644337929785252, | |
| "rewards/rejected": -0.0029360298067331314, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.287132845137709e-09, | |
| "logits/chosen": -0.4601454734802246, | |
| "logits/rejected": -0.44960451126098633, | |
| "logps/chosen": -305.7701110839844, | |
| "logps/rejected": -349.0350341796875, | |
| "loss": 0.1976, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.007751205004751682, | |
| "rewards/margins": 0.011196482926607132, | |
| "rewards/rejected": -0.0034452793188393116, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.2757667974155896e-09, | |
| "logits/chosen": -0.4950791001319885, | |
| "logits/rejected": -0.47172340750694275, | |
| "logps/chosen": -312.46453857421875, | |
| "logps/rejected": -315.9293212890625, | |
| "loss": 0.2221, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.00022761887521483004, | |
| "rewards/margins": -0.004159080795943737, | |
| "rewards/rejected": 0.0043866997584700584, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.544089730633804e-11, | |
| "logits/chosen": -0.48454493284225464, | |
| "logits/rejected": -0.46828117966651917, | |
| "logps/chosen": -290.0190734863281, | |
| "logps/rejected": -295.50665283203125, | |
| "loss": 0.1916, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.0072511411271989346, | |
| "rewards/margins": 0.00542284082621336, | |
| "rewards/rejected": 0.001828300068154931, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1312, | |
| "total_flos": 0.0, | |
| "train_loss": 0.20755663267677513, | |
| "train_runtime": 11765.5627, | |
| "train_samples_per_second": 1.785, | |
| "train_steps_per_second": 0.112 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1312, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |