| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9994666666666666, | |
| "eval_steps": 500, | |
| "global_step": 937, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.319148936170213e-08, | |
| "logits/chosen": 0.4053989052772522, | |
| "logits/rejected": 0.1312936246395111, | |
| "logps/chosen": -434.00537109375, | |
| "logps/rejected": -516.5983276367188, | |
| "loss": 0.1853, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.319148936170213e-07, | |
| "logits/chosen": 0.15346182882785797, | |
| "logits/rejected": 0.2922114133834839, | |
| "logps/chosen": -365.6877746582031, | |
| "logps/rejected": -353.01446533203125, | |
| "loss": 0.2092, | |
| "rewards/accuracies": 0.2847222089767456, | |
| "rewards/chosen": -0.005939984694123268, | |
| "rewards/margins": -0.0001524988911114633, | |
| "rewards/rejected": -0.005787486210465431, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.1189204677939415, | |
| "logits/rejected": 0.20350512862205505, | |
| "logps/chosen": -340.10516357421875, | |
| "logps/rejected": -348.26995849609375, | |
| "loss": 0.2104, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": -0.006609903182834387, | |
| "rewards/margins": -0.0014991617063060403, | |
| "rewards/rejected": -0.005110741592943668, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.2546407878398895, | |
| "logits/rejected": 0.2511681616306305, | |
| "logps/chosen": -383.5899353027344, | |
| "logps/rejected": -364.458251953125, | |
| "loss": 0.2, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": -0.010028360411524773, | |
| "rewards/margins": 0.000567590759601444, | |
| "rewards/rejected": -0.010595952160656452, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.15378129482269287, | |
| "logits/rejected": 0.27011898159980774, | |
| "logps/chosen": -402.9029846191406, | |
| "logps/rejected": -387.1457214355469, | |
| "loss": 0.2097, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": -0.018913280218839645, | |
| "rewards/margins": -0.0008677333826199174, | |
| "rewards/rejected": -0.018045546486973763, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.23259440064430237, | |
| "logits/rejected": 0.27333545684814453, | |
| "logps/chosen": -444.83447265625, | |
| "logps/rejected": -435.3008728027344, | |
| "loss": 0.2034, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -0.039250001311302185, | |
| "rewards/margins": 0.002604855690151453, | |
| "rewards/rejected": -0.0418548583984375, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.1691911518573761, | |
| "logits/rejected": 0.3174379765987396, | |
| "logps/chosen": -413.50665283203125, | |
| "logps/rejected": -417.56591796875, | |
| "loss": 0.2144, | |
| "rewards/accuracies": 0.40625, | |
| "rewards/chosen": -0.04588966816663742, | |
| "rewards/margins": 0.002726958366110921, | |
| "rewards/rejected": -0.04861662536859512, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.12422730773687363, | |
| "logits/rejected": 0.2234603464603424, | |
| "logps/chosen": -361.4434814453125, | |
| "logps/rejected": -363.685302734375, | |
| "loss": 0.2037, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.06797133386135101, | |
| "rewards/margins": 0.009325895458459854, | |
| "rewards/rejected": -0.07729722559452057, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.14516139030456543, | |
| "logits/rejected": 0.25278085470199585, | |
| "logps/chosen": -403.75347900390625, | |
| "logps/rejected": -391.7194519042969, | |
| "loss": 0.2017, | |
| "rewards/accuracies": 0.4437499940395355, | |
| "rewards/chosen": -0.10898496210575104, | |
| "rewards/margins": 0.009633904322981834, | |
| "rewards/rejected": -0.11861886829137802, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.23768356442451477, | |
| "logits/rejected": 0.17344684898853302, | |
| "logps/chosen": -408.779052734375, | |
| "logps/rejected": -431.6903381347656, | |
| "loss": 0.1899, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": -0.15845593810081482, | |
| "rewards/margins": 0.01751708984375, | |
| "rewards/rejected": -0.17597301304340363, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999375059004058e-06, | |
| "logits/chosen": 0.15429627895355225, | |
| "logits/rejected": 0.1605452001094818, | |
| "logps/chosen": -438.6952209472656, | |
| "logps/rejected": -433.0787048339844, | |
| "loss": 0.1962, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.2150062620639801, | |
| "rewards/margins": 0.019437283277511597, | |
| "rewards/rejected": -0.2344435453414917, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9955571065548795e-06, | |
| "logits/chosen": 0.23242899775505066, | |
| "logits/rejected": 0.15503418445587158, | |
| "logps/chosen": -424.6685485839844, | |
| "logps/rejected": -408.0357666015625, | |
| "loss": 0.1863, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.055656641721725464, | |
| "rewards/margins": 0.023944241926074028, | |
| "rewards/rejected": -0.07960088551044464, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882736864879e-06, | |
| "logits/chosen": 0.09133955091238022, | |
| "logits/rejected": 0.2611747682094574, | |
| "logps/chosen": -403.81085205078125, | |
| "logps/rejected": -435.2984924316406, | |
| "loss": 0.1903, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.10940422862768173, | |
| "rewards/margins": 0.041554711759090424, | |
| "rewards/rejected": 0.06784951686859131, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977534912960124e-06, | |
| "logits/chosen": 0.16850045323371887, | |
| "logits/rejected": 0.2994873821735382, | |
| "logps/chosen": -397.17254638671875, | |
| "logps/rejected": -387.06561279296875, | |
| "loss": 0.1881, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.15394046902656555, | |
| "rewards/margins": 0.0355982705950737, | |
| "rewards/rejected": 0.11834220588207245, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.963355698422092e-06, | |
| "logits/chosen": 0.1726590096950531, | |
| "logits/rejected": 0.2367965281009674, | |
| "logps/chosen": -373.75689697265625, | |
| "logps/rejected": -357.42108154296875, | |
| "loss": 0.1843, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.10395370423793793, | |
| "rewards/margins": 0.031282939016819, | |
| "rewards/rejected": 0.07267077267169952, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945755732909625e-06, | |
| "logits/chosen": 0.04663983732461929, | |
| "logits/rejected": 0.09915662556886673, | |
| "logps/chosen": -379.2816162109375, | |
| "logps/rejected": -385.98834228515625, | |
| "loss": 0.1811, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.03478062525391579, | |
| "rewards/margins": 0.03926830738782883, | |
| "rewards/rejected": -0.004487680736929178, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924759456701167e-06, | |
| "logits/chosen": 0.1232941746711731, | |
| "logits/rejected": 0.20809748768806458, | |
| "logps/chosen": -426.62841796875, | |
| "logps/rejected": -422.14910888671875, | |
| "loss": 0.1715, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.03437420353293419, | |
| "rewards/margins": 0.0738130509853363, | |
| "rewards/rejected": -0.1081872433423996, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900396026378671e-06, | |
| "logits/chosen": 0.07926517724990845, | |
| "logits/rejected": 0.25820231437683105, | |
| "logps/chosen": -460.07012939453125, | |
| "logps/rejected": -439.01495361328125, | |
| "loss": 0.2022, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.19370847940444946, | |
| "rewards/margins": 0.05930668115615845, | |
| "rewards/rejected": -0.2530151605606079, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.872699274339169e-06, | |
| "logits/chosen": 0.19027003645896912, | |
| "logits/rejected": 0.20081575214862823, | |
| "logps/chosen": -411.0907287597656, | |
| "logps/rejected": -418.31689453125, | |
| "loss": 0.168, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.30076223611831665, | |
| "rewards/margins": 0.07709387689828873, | |
| "rewards/rejected": -0.3778560757637024, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8417076618132434e-06, | |
| "logits/chosen": 0.1264984905719757, | |
| "logits/rejected": 0.21260645985603333, | |
| "logps/chosen": -527.2296752929688, | |
| "logps/rejected": -502.3099670410156, | |
| "loss": 0.1784, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -0.43060359358787537, | |
| "rewards/margins": 0.06492508202791214, | |
| "rewards/rejected": -0.4955286979675293, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807464225455655e-06, | |
| "logits/chosen": 0.061636537313461304, | |
| "logits/rejected": 0.19780707359313965, | |
| "logps/chosen": -453.3473205566406, | |
| "logps/rejected": -477.33819580078125, | |
| "loss": 0.1818, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.4919506013393402, | |
| "rewards/margins": 0.06804502010345459, | |
| "rewards/rejected": -0.559995710849762, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770016517582283e-06, | |
| "logits/chosen": 0.12850946187973022, | |
| "logits/rejected": 0.12981989979743958, | |
| "logps/chosen": -447.4049377441406, | |
| "logps/rejected": -458.405029296875, | |
| "loss": 0.1796, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.4579945504665375, | |
| "rewards/margins": 0.07319770753383636, | |
| "rewards/rejected": -0.5311923027038574, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7294165401363616e-06, | |
| "logits/chosen": 0.12648232281208038, | |
| "logits/rejected": 0.10291828960180283, | |
| "logps/chosen": -447.4896545410156, | |
| "logps/rejected": -447.23248291015625, | |
| "loss": 0.1804, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.4558301568031311, | |
| "rewards/margins": 0.051274996250867844, | |
| "rewards/rejected": -0.5071052312850952, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68572067247573e-06, | |
| "logits/chosen": 0.11354929208755493, | |
| "logits/rejected": 0.13758793473243713, | |
| "logps/chosen": -435.9432678222656, | |
| "logps/rejected": -443.25250244140625, | |
| "loss": 0.1821, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.36004215478897095, | |
| "rewards/margins": 0.06442644447088242, | |
| "rewards/rejected": -0.42446860671043396, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.638989593081364e-06, | |
| "logits/chosen": -0.01750967465341091, | |
| "logits/rejected": 0.15084341168403625, | |
| "logps/chosen": -406.10284423828125, | |
| "logps/rejected": -415.41131591796875, | |
| "loss": 0.173, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.2852724492549896, | |
| "rewards/margins": 0.060403019189834595, | |
| "rewards/rejected": -0.3456754982471466, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5892881952959015e-06, | |
| "logits/chosen": 0.08206398785114288, | |
| "logits/rejected": 0.1491158902645111, | |
| "logps/chosen": -441.6056213378906, | |
| "logps/rejected": -435.2107849121094, | |
| "loss": 0.1863, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.2744644582271576, | |
| "rewards/margins": 0.07761115580797195, | |
| "rewards/rejected": -0.35207563638687134, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536685497209182e-06, | |
| "logits/chosen": 0.07012276351451874, | |
| "logits/rejected": 0.11525042355060577, | |
| "logps/chosen": -459.88323974609375, | |
| "logps/rejected": -479.2359313964844, | |
| "loss": 0.17, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.2863777279853821, | |
| "rewards/margins": 0.06567540019750595, | |
| "rewards/rejected": -0.35205310583114624, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481254545815943e-06, | |
| "logits/chosen": 0.03542683273553848, | |
| "logits/rejected": 0.08359168469905853, | |
| "logps/chosen": -459.03814697265625, | |
| "logps/rejected": -487.94781494140625, | |
| "loss": 0.1683, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.25155481696128845, | |
| "rewards/margins": 0.09583600610494614, | |
| "rewards/rejected": -0.3473908305168152, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.42307231557875e-06, | |
| "logits/chosen": 0.03823712840676308, | |
| "logits/rejected": 0.12090680748224258, | |
| "logps/chosen": -434.3282165527344, | |
| "logps/rejected": -428.10675048828125, | |
| "loss": 0.1758, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.2617419362068176, | |
| "rewards/margins": 0.07895456254482269, | |
| "rewards/rejected": -0.3406964838504791, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3622196015370305e-06, | |
| "logits/chosen": 0.04959043115377426, | |
| "logits/rejected": 0.1204204112291336, | |
| "logps/chosen": -440.5345764160156, | |
| "logps/rejected": -477.7173767089844, | |
| "loss": 0.184, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.37586796283721924, | |
| "rewards/margins": 0.07159671187400818, | |
| "rewards/rejected": -0.44746464490890503, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.298780907110648e-06, | |
| "logits/chosen": 0.09089604020118713, | |
| "logits/rejected": 0.0841570645570755, | |
| "logps/chosen": -469.25775146484375, | |
| "logps/rejected": -441.3363342285156, | |
| "loss": 0.1693, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.4600369930267334, | |
| "rewards/margins": 0.07286399602890015, | |
| "rewards/rejected": -0.5329009294509888, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.23284432675381e-06, | |
| "logits/chosen": -0.010215234942734241, | |
| "logits/rejected": 0.14281333982944489, | |
| "logps/chosen": -418.6227111816406, | |
| "logps/rejected": -425.4488830566406, | |
| "loss": 0.1808, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.39024704694747925, | |
| "rewards/margins": 0.07213394343852997, | |
| "rewards/rejected": -0.462380975484848, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.164501423622277e-06, | |
| "logits/chosen": 0.05900254100561142, | |
| "logits/rejected": 0.11304068565368652, | |
| "logps/chosen": -415.36883544921875, | |
| "logps/rejected": -410.0850524902344, | |
| "loss": 0.1688, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.3306947648525238, | |
| "rewards/margins": 0.06759382784366608, | |
| "rewards/rejected": -0.3982886075973511, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938471024237355e-06, | |
| "logits/chosen": 0.09281224012374878, | |
| "logits/rejected": 0.10557293891906738, | |
| "logps/chosen": -458.1168518066406, | |
| "logps/rejected": -433.6602478027344, | |
| "loss": 0.1818, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.2099827080965042, | |
| "rewards/margins": 0.07166289538145065, | |
| "rewards/rejected": -0.28164559602737427, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.020979477627907e-06, | |
| "logits/chosen": 0.1274840533733368, | |
| "logits/rejected": 0.1528143584728241, | |
| "logps/chosen": -436.2671813964844, | |
| "logps/rejected": -449.423095703125, | |
| "loss": 0.1728, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.16712762415409088, | |
| "rewards/margins": 0.0913640707731247, | |
| "rewards/rejected": -0.2584916949272156, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9459997372194105e-06, | |
| "logits/chosen": 0.03427785634994507, | |
| "logits/rejected": 0.1746084988117218, | |
| "logps/chosen": -448.252197265625, | |
| "logps/rejected": -450.52459716796875, | |
| "loss": 0.1746, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.1110633835196495, | |
| "rewards/margins": 0.07623620331287384, | |
| "rewards/rejected": -0.18729960918426514, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.869012002182573e-06, | |
| "logits/chosen": -0.044000182300806046, | |
| "logits/rejected": 0.06577601283788681, | |
| "logps/chosen": -455.25042724609375, | |
| "logps/rejected": -427.7881774902344, | |
| "loss": 0.1644, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.048846274614334106, | |
| "rewards/margins": 0.08700191974639893, | |
| "rewards/rejected": -0.13584819436073303, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7901231819133104e-06, | |
| "logits/chosen": 0.07364512979984283, | |
| "logits/rejected": 0.08710888028144836, | |
| "logps/chosen": -406.98309326171875, | |
| "logps/rejected": -417.3194274902344, | |
| "loss": 0.1731, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.07288642972707748, | |
| "rewards/margins": 0.06538988649845123, | |
| "rewards/rejected": -0.13827630877494812, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.709442825758875e-06, | |
| "logits/chosen": -0.05436893552541733, | |
| "logits/rejected": 0.11040735244750977, | |
| "logps/chosen": -409.92462158203125, | |
| "logps/rejected": -393.9139404296875, | |
| "loss": 0.1577, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": -0.05880141258239746, | |
| "rewards/margins": 0.08271567523479462, | |
| "rewards/rejected": -0.14151708781719208, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6270829708916113e-06, | |
| "logits/chosen": -0.007267399691045284, | |
| "logits/rejected": 0.06522424519062042, | |
| "logps/chosen": -431.5768127441406, | |
| "logps/rejected": -446.33001708984375, | |
| "loss": 0.1801, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.10045109689235687, | |
| "rewards/margins": 0.06214982271194458, | |
| "rewards/rejected": -0.16260093450546265, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543157986727991e-06, | |
| "logits/chosen": 0.07450960576534271, | |
| "logits/rejected": 0.09937725961208344, | |
| "logps/chosen": -435.92291259765625, | |
| "logps/rejected": -440.31182861328125, | |
| "loss": 0.1738, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -0.1277085244655609, | |
| "rewards/margins": 0.07721880823373795, | |
| "rewards/rejected": -0.20492735505104065, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4577844161089614e-06, | |
| "logits/chosen": 0.03635771945118904, | |
| "logits/rejected": 0.04270925745368004, | |
| "logps/chosen": -437.2557067871094, | |
| "logps/rejected": -465.94915771484375, | |
| "loss": 0.1651, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.15446950495243073, | |
| "rewards/margins": 0.09225080162286758, | |
| "rewards/rejected": -0.24672028422355652, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3710808134621577e-06, | |
| "logits/chosen": 0.0480308011174202, | |
| "logits/rejected": 0.09423025697469711, | |
| "logps/chosen": -461.2041931152344, | |
| "logps/rejected": -467.91644287109375, | |
| "loss": 0.1701, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.14544199407100677, | |
| "rewards/margins": 0.09961853921413422, | |
| "rewards/rejected": -0.245060533285141, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2831675801707126e-06, | |
| "logits/chosen": 0.0305793434381485, | |
| "logits/rejected": 0.033555738627910614, | |
| "logps/chosen": -377.18719482421875, | |
| "logps/rejected": -382.844970703125, | |
| "loss": 0.1698, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.09203015267848969, | |
| "rewards/margins": 0.07604523003101349, | |
| "rewards/rejected": -0.16807539761066437, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.194166797377289e-06, | |
| "logits/chosen": 0.046750448644161224, | |
| "logits/rejected": 0.07448974251747131, | |
| "logps/chosen": -459.547607421875, | |
| "logps/rejected": -455.531005859375, | |
| "loss": 0.175, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.08039159327745438, | |
| "rewards/margins": 0.0626746192574501, | |
| "rewards/rejected": -0.14306621253490448, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.104202056455501e-06, | |
| "logits/chosen": 0.03396008908748627, | |
| "logits/rejected": 0.09014374762773514, | |
| "logps/chosen": -429.5635681152344, | |
| "logps/rejected": -421.2688903808594, | |
| "loss": 0.1802, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.039100803434848785, | |
| "rewards/margins": 0.08359581232070923, | |
| "rewards/rejected": -0.12269661575555801, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.013398287384144e-06, | |
| "logits/chosen": 0.04828561842441559, | |
| "logits/rejected": 0.11091510206460953, | |
| "logps/chosen": -445.12713623046875, | |
| "logps/rejected": -433.92047119140625, | |
| "loss": 0.1619, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.06680645048618317, | |
| "rewards/margins": 0.0745815858244896, | |
| "rewards/rejected": -0.14138804376125336, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9218815852625717e-06, | |
| "logits/chosen": 0.04593321308493614, | |
| "logits/rejected": 0.07564611732959747, | |
| "logps/chosen": -422.21954345703125, | |
| "logps/rejected": -430.91802978515625, | |
| "loss": 0.1732, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.07942485064268112, | |
| "rewards/margins": 0.09848260879516602, | |
| "rewards/rejected": -0.17790746688842773, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.829779035208113e-06, | |
| "logits/chosen": -0.00015009715571068227, | |
| "logits/rejected": 0.11992751061916351, | |
| "logps/chosen": -397.8226318359375, | |
| "logps/rejected": -412.5755920410156, | |
| "loss": 0.1662, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.09762930124998093, | |
| "rewards/margins": 0.10028882324695587, | |
| "rewards/rejected": -0.1979181170463562, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.737218535878705e-06, | |
| "logits/chosen": 0.08348871767520905, | |
| "logits/rejected": 0.09916739165782928, | |
| "logps/chosen": -390.0538330078125, | |
| "logps/rejected": -407.1832275390625, | |
| "loss": 0.1732, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.12354443222284317, | |
| "rewards/margins": 0.10041115432977676, | |
| "rewards/rejected": -0.22395558655261993, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.64432862186579e-06, | |
| "logits/chosen": 0.0067026736214756966, | |
| "logits/rejected": -0.01416114903986454, | |
| "logps/chosen": -386.2361755371094, | |
| "logps/rejected": -397.430908203125, | |
| "loss": 0.1684, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.16371376812458038, | |
| "rewards/margins": 0.06288306415081024, | |
| "rewards/rejected": -0.22659683227539062, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.551238285204126e-06, | |
| "logits/chosen": 0.03226105123758316, | |
| "logits/rejected": 0.08519762009382248, | |
| "logps/chosen": -463.94879150390625, | |
| "logps/rejected": -457.156982421875, | |
| "loss": 0.1738, | |
| "rewards/accuracies": 0.5687500238418579, | |
| "rewards/chosen": -0.1887294501066208, | |
| "rewards/margins": 0.07099872827529907, | |
| "rewards/rejected": -0.25972819328308105, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4580767962463688e-06, | |
| "logits/chosen": -0.030474770814180374, | |
| "logits/rejected": 0.06743954867124557, | |
| "logps/chosen": -430.6739196777344, | |
| "logps/rejected": -406.80535888671875, | |
| "loss": 0.1712, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.1928265392780304, | |
| "rewards/margins": 0.08770117908716202, | |
| "rewards/rejected": -0.2805277109146118, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3649735241511546e-06, | |
| "logits/chosen": 0.1015111431479454, | |
| "logits/rejected": 0.06599382311105728, | |
| "logps/chosen": -437.9898986816406, | |
| "logps/rejected": -436.36004638671875, | |
| "loss": 0.179, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.2587732672691345, | |
| "rewards/margins": 0.0716363936662674, | |
| "rewards/rejected": -0.3304096758365631, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2720577572339914e-06, | |
| "logits/chosen": -0.04144694283604622, | |
| "logits/rejected": 0.055634863674640656, | |
| "logps/chosen": -430.6463317871094, | |
| "logps/rejected": -424.1119079589844, | |
| "loss": 0.1712, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.26249006390571594, | |
| "rewards/margins": 0.08809027820825577, | |
| "rewards/rejected": -0.3505803048610687, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1794585234303995e-06, | |
| "logits/chosen": -0.03823351487517357, | |
| "logits/rejected": 0.08283025026321411, | |
| "logps/chosen": -450.2481384277344, | |
| "logps/rejected": -451.69097900390625, | |
| "loss": 0.1741, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.31970635056495667, | |
| "rewards/margins": 0.07840566337108612, | |
| "rewards/rejected": -0.3981119990348816, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0873044111206407e-06, | |
| "logits/chosen": 0.04782441258430481, | |
| "logits/rejected": 0.0727071687579155, | |
| "logps/chosen": -412.89306640625, | |
| "logps/rejected": -450.1448669433594, | |
| "loss": 0.1816, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.3479355573654175, | |
| "rewards/margins": 0.07213512808084488, | |
| "rewards/rejected": -0.42007070779800415, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9957233905648293e-06, | |
| "logits/chosen": -0.041085027158260345, | |
| "logits/rejected": -0.010372793301939964, | |
| "logps/chosen": -402.28997802734375, | |
| "logps/rejected": -407.7055969238281, | |
| "loss": 0.1809, | |
| "rewards/accuracies": 0.48124998807907104, | |
| "rewards/chosen": -0.3540458381175995, | |
| "rewards/margins": 0.060313619673252106, | |
| "rewards/rejected": -0.4143594801425934, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.904842636196402e-06, | |
| "logits/chosen": 0.023486686870455742, | |
| "logits/rejected": 0.07347628474235535, | |
| "logps/chosen": -444.597412109375, | |
| "logps/rejected": -440.24407958984375, | |
| "loss": 0.1686, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.34095701575279236, | |
| "rewards/margins": 0.08776627480983734, | |
| "rewards/rejected": -0.4287233352661133, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.814788350020726e-06, | |
| "logits/chosen": 0.0032510061282664537, | |
| "logits/rejected": 0.12992344796657562, | |
| "logps/chosen": -465.0745544433594, | |
| "logps/rejected": -462.6073303222656, | |
| "loss": 0.1583, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.3434141278266907, | |
| "rewards/margins": 0.08694170415401459, | |
| "rewards/rejected": -0.43035584688186646, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.725685586364051e-06, | |
| "logits/chosen": -0.010292215272784233, | |
| "logits/rejected": 0.03650585561990738, | |
| "logps/chosen": -390.76690673828125, | |
| "logps/rejected": -422.77105712890625, | |
| "loss": 0.1712, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.3311789929866791, | |
| "rewards/margins": 0.07950376719236374, | |
| "rewards/rejected": -0.410682737827301, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6376580782162172e-06, | |
| "logits/chosen": -0.03925107792019844, | |
| "logits/rejected": -0.04231791943311691, | |
| "logps/chosen": -445.1822204589844, | |
| "logps/rejected": -434.329833984375, | |
| "loss": 0.1719, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.3707120418548584, | |
| "rewards/margins": 0.09252772480249405, | |
| "rewards/rejected": -0.46323972940444946, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.550828065408227e-06, | |
| "logits/chosen": 0.06463605910539627, | |
| "logits/rejected": -0.027300477027893066, | |
| "logps/chosen": -431.5687561035156, | |
| "logps/rejected": -470.56072998046875, | |
| "loss": 0.1793, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.37444019317626953, | |
| "rewards/margins": 0.07550361007452011, | |
| "rewards/rejected": -0.44994378089904785, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4653161248633053e-06, | |
| "logits/chosen": -0.04705502465367317, | |
| "logits/rejected": -0.02023206278681755, | |
| "logps/chosen": -402.87274169921875, | |
| "logps/rejected": -393.464599609375, | |
| "loss": 0.1722, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.3044665455818176, | |
| "rewards/margins": 0.08127270638942719, | |
| "rewards/rejected": -0.385739266872406, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.381241003157162e-06, | |
| "logits/chosen": -0.03925855830311775, | |
| "logits/rejected": 0.015137474052608013, | |
| "logps/chosen": -417.5736389160156, | |
| "logps/rejected": -411.40386962890625, | |
| "loss": 0.1753, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.2832847535610199, | |
| "rewards/margins": 0.10071472823619843, | |
| "rewards/rejected": -0.38399946689605713, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.298719451619979e-06, | |
| "logits/chosen": -0.03505668044090271, | |
| "logits/rejected": 0.16136914491653442, | |
| "logps/chosen": -457.20556640625, | |
| "logps/rejected": -451.82073974609375, | |
| "loss": 0.166, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.3379427492618561, | |
| "rewards/margins": 0.10498092323541641, | |
| "rewards/rejected": -0.4429236948490143, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2178660642091036e-06, | |
| "logits/chosen": -0.060994476079940796, | |
| "logits/rejected": 0.0816785916686058, | |
| "logps/chosen": -460.37445068359375, | |
| "logps/rejected": -450.03155517578125, | |
| "loss": 0.174, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.3193809986114502, | |
| "rewards/margins": 0.07850383222103119, | |
| "rewards/rejected": -0.3978848159313202, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1387931183775821e-06, | |
| "logits/chosen": 0.09993238747119904, | |
| "logits/rejected": 0.04898153617978096, | |
| "logps/chosen": -418.4270935058594, | |
| "logps/rejected": -422.74822998046875, | |
| "loss": 0.1796, | |
| "rewards/accuracies": 0.5687500238418579, | |
| "rewards/chosen": -0.29177647829055786, | |
| "rewards/margins": 0.07586276531219482, | |
| "rewards/rejected": -0.3676392436027527, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.061610419159532e-06, | |
| "logits/chosen": 0.04208558052778244, | |
| "logits/rejected": 0.044306449592113495, | |
| "logps/chosen": -393.4234924316406, | |
| "logps/rejected": -386.6611633300781, | |
| "loss": 0.1715, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.23296888172626495, | |
| "rewards/margins": 0.06498309224843979, | |
| "rewards/rejected": -0.29795199632644653, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.864251466888364e-07, | |
| "logits/chosen": -0.03271818161010742, | |
| "logits/rejected": 0.07850229740142822, | |
| "logps/chosen": -422.83489990234375, | |
| "logps/rejected": -424.16876220703125, | |
| "loss": 0.1734, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": -0.26918911933898926, | |
| "rewards/margins": 0.06875598430633545, | |
| "rewards/rejected": -0.3379451334476471, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.133417073629288e-07, | |
| "logits/chosen": -0.053741514682769775, | |
| "logits/rejected": 0.01618141494691372, | |
| "logps/chosen": -423.0733947753906, | |
| "logps/rejected": -434.5904235839844, | |
| "loss": 0.1454, | |
| "rewards/accuracies": 0.5687500238418579, | |
| "rewards/chosen": -0.2766970098018646, | |
| "rewards/margins": 0.07641127705574036, | |
| "rewards/rejected": -0.353108286857605, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.424615888583332e-07, | |
| "logits/chosen": -0.0247728880494833, | |
| "logits/rejected": 0.08638174831867218, | |
| "logps/chosen": -435.94366455078125, | |
| "logps/rejected": -447.54522705078125, | |
| "loss": 0.1705, | |
| "rewards/accuracies": 0.5687500238418579, | |
| "rewards/chosen": -0.26050230860710144, | |
| "rewards/margins": 0.10260170698165894, | |
| "rewards/rejected": -0.3631040155887604, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.738832191993092e-07, | |
| "logits/chosen": 0.014270253479480743, | |
| "logits/rejected": 0.035499878227710724, | |
| "logps/chosen": -450.578369140625, | |
| "logps/rejected": -448.53533935546875, | |
| "loss": 0.1607, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": -0.3028029799461365, | |
| "rewards/margins": 0.088174968957901, | |
| "rewards/rejected": -0.3909779191017151, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.077018300752917e-07, | |
| "logits/chosen": 0.037296146154403687, | |
| "logits/rejected": 0.01444760151207447, | |
| "logps/chosen": -452.60845947265625, | |
| "logps/rejected": -439.6041564941406, | |
| "loss": 0.1698, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.2787321209907532, | |
| "rewards/margins": 0.08983801305294037, | |
| "rewards/rejected": -0.36857014894485474, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.440093245969342e-07, | |
| "logits/chosen": -0.01685388572514057, | |
| "logits/rejected": 0.08555487543344498, | |
| "logps/chosen": -405.7493591308594, | |
| "logps/rejected": -383.0268859863281, | |
| "loss": 0.1631, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -0.2447400987148285, | |
| "rewards/margins": 0.08146357536315918, | |
| "rewards/rejected": -0.32620367407798767, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.828941496744075e-07, | |
| "logits/chosen": 4.3596326577244326e-05, | |
| "logits/rejected": 0.07585635036230087, | |
| "logps/chosen": -454.00299072265625, | |
| "logps/rejected": -468.2713317871094, | |
| "loss": 0.1772, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -0.3086394667625427, | |
| "rewards/margins": 0.08122680336236954, | |
| "rewards/rejected": -0.38986626267433167, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.244411731951671e-07, | |
| "logits/chosen": -0.05567328259348869, | |
| "logits/rejected": 0.16029450297355652, | |
| "logps/chosen": -415.29608154296875, | |
| "logps/rejected": -398.7294921875, | |
| "loss": 0.153, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.2835378050804138, | |
| "rewards/margins": 0.09213887155056, | |
| "rewards/rejected": -0.375676691532135, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6873156617173594e-07, | |
| "logits/chosen": -0.08789820969104767, | |
| "logits/rejected": -0.000363718718290329, | |
| "logps/chosen": -463.2998962402344, | |
| "logps/rejected": -440.36328125, | |
| "loss": 0.1584, | |
| "rewards/accuracies": 0.5687500238418579, | |
| "rewards/chosen": -0.29543575644493103, | |
| "rewards/margins": 0.09579849988222122, | |
| "rewards/rejected": -0.39123424887657166, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1584269002318653e-07, | |
| "logits/chosen": -0.030699348077178, | |
| "logits/rejected": 0.08616991341114044, | |
| "logps/chosen": -469.35833740234375, | |
| "logps/rejected": -482.96337890625, | |
| "loss": 0.1671, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.3277556598186493, | |
| "rewards/margins": 0.0916672796010971, | |
| "rewards/rejected": -0.4194229543209076, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.658479891468258e-07, | |
| "logits/chosen": 0.04981686919927597, | |
| "logits/rejected": 0.08207514882087708, | |
| "logps/chosen": -433.3348693847656, | |
| "logps/rejected": -440.70953369140625, | |
| "loss": 0.1727, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.31288522481918335, | |
| "rewards/margins": 0.08250272274017334, | |
| "rewards/rejected": -0.3953879475593567, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.18816888929272e-07, | |
| "logits/chosen": -0.040587715804576874, | |
| "logits/rejected": 0.07621972262859344, | |
| "logps/chosen": -451.85345458984375, | |
| "logps/rejected": -429.25634765625, | |
| "loss": 0.1697, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.3361504375934601, | |
| "rewards/margins": 0.09061960875988007, | |
| "rewards/rejected": -0.42677006125450134, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.748146993385484e-07, | |
| "logits/chosen": 0.033268146216869354, | |
| "logits/rejected": -0.0012865017633885145, | |
| "logps/chosen": -478.40673828125, | |
| "logps/rejected": -487.28680419921875, | |
| "loss": 0.1613, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.3244819939136505, | |
| "rewards/margins": 0.10256977379322052, | |
| "rewards/rejected": -0.42705172300338745, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3390252423108077e-07, | |
| "logits/chosen": -0.06332585215568542, | |
| "logits/rejected": 0.13626988232135773, | |
| "logps/chosen": -468.19378662109375, | |
| "logps/rejected": -437.52227783203125, | |
| "loss": 0.1594, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": -0.2964176535606384, | |
| "rewards/margins": 0.11590894311666489, | |
| "rewards/rejected": -0.4123266339302063, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.961371764995243e-07, | |
| "logits/chosen": 0.07440580427646637, | |
| "logits/rejected": 0.08452493697404861, | |
| "logps/chosen": -448.39862060546875, | |
| "logps/rejected": -437.47235107421875, | |
| "loss": 0.1661, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.29594478011131287, | |
| "rewards/margins": 0.09329672157764435, | |
| "rewards/rejected": -0.3892415463924408, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.61571099179261e-07, | |
| "logits/chosen": -0.06233375146985054, | |
| "logits/rejected": 0.027787405997514725, | |
| "logps/chosen": -398.1230773925781, | |
| "logps/rejected": -390.20489501953125, | |
| "loss": 0.1742, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.2930799126625061, | |
| "rewards/margins": 0.09176953136920929, | |
| "rewards/rejected": -0.3848494589328766, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3025229262312367e-07, | |
| "logits/chosen": -0.025651434436440468, | |
| "logits/rejected": 0.04196717590093613, | |
| "logps/chosen": -450.2112731933594, | |
| "logps/rejected": -449.22393798828125, | |
| "loss": 0.1564, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.31000474095344543, | |
| "rewards/margins": 0.09261561930179596, | |
| "rewards/rejected": -0.4026203751564026, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0222424784546853e-07, | |
| "logits/chosen": 0.1431967318058014, | |
| "logits/rejected": 0.003605568315833807, | |
| "logps/chosen": -432.6329650878906, | |
| "logps/rejected": -449.58282470703125, | |
| "loss": 0.1597, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.3245397210121155, | |
| "rewards/margins": 0.0643848329782486, | |
| "rewards/rejected": -0.3889245390892029, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.752588612816553e-08, | |
| "logits/chosen": 0.059076208621263504, | |
| "logits/rejected": -0.0008928939932957292, | |
| "logps/chosen": -480.4388732910156, | |
| "logps/rejected": -463.54522705078125, | |
| "loss": 0.1676, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": -0.31118619441986084, | |
| "rewards/margins": 0.08092699944972992, | |
| "rewards/rejected": -0.3921131491661072, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.619150497236991e-08, | |
| "logits/chosen": -0.019091134890913963, | |
| "logits/rejected": 0.034938834607601166, | |
| "logps/chosen": -435.51953125, | |
| "logps/rejected": -441.9061584472656, | |
| "loss": 0.1645, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.3230697214603424, | |
| "rewards/margins": 0.09634076803922653, | |
| "rewards/rejected": -0.41941046714782715, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.825073047112743e-08, | |
| "logits/chosen": -0.056211818009614944, | |
| "logits/rejected": 0.04693780839443207, | |
| "logps/chosen": -446.30224609375, | |
| "logps/rejected": -438.859375, | |
| "loss": 0.1766, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.30564260482788086, | |
| "rewards/margins": 0.0854407250881195, | |
| "rewards/rejected": -0.39108332991600037, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.372847616895685e-08, | |
| "logits/chosen": 0.026849299669265747, | |
| "logits/rejected": -0.015656864270567894, | |
| "logps/chosen": -413.23876953125, | |
| "logps/rejected": -446.19219970703125, | |
| "loss": 0.1638, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": -0.30960044264793396, | |
| "rewards/margins": 0.08425579965114594, | |
| "rewards/rejected": -0.3938562273979187, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.264490846553279e-08, | |
| "logits/chosen": 0.02639024890959263, | |
| "logits/rejected": 0.017060106620192528, | |
| "logps/chosen": -459.57318115234375, | |
| "logps/rejected": -468.657958984375, | |
| "loss": 0.1567, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.2922886312007904, | |
| "rewards/margins": 0.1054367646574974, | |
| "rewards/rejected": -0.3977254033088684, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.015418611516165e-09, | |
| "logits/chosen": -0.023000415414571762, | |
| "logits/rejected": -0.024313045665621758, | |
| "logps/chosen": -414.48370361328125, | |
| "logps/rejected": -406.40069580078125, | |
| "loss": 0.1696, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": -0.308437705039978, | |
| "rewards/margins": 0.0932714194059372, | |
| "rewards/rejected": -0.40170907974243164, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.506013354186993e-10, | |
| "logits/chosen": 0.016056960448622704, | |
| "logits/rejected": 0.02258588932454586, | |
| "logps/chosen": -415.54803466796875, | |
| "logps/rejected": -428.69775390625, | |
| "loss": 0.1713, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": -0.31125369668006897, | |
| "rewards/margins": 0.08610902726650238, | |
| "rewards/rejected": -0.39736273884773254, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 937, | |
| "total_flos": 0.0, | |
| "train_loss": 0.1752831537352046, | |
| "train_runtime": 7846.1192, | |
| "train_samples_per_second": 3.824, | |
| "train_steps_per_second": 0.119 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 937, | |
| "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 | |
| } | |