| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1875, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 2.6595744680851065e-08, | |
| "logits/chosen": 0.41994422674179077, | |
| "logits/rejected": 0.44097208976745605, | |
| "logps/chosen": -255.3050537109375, | |
| "logps/rejected": -278.9874572753906, | |
| "loss": 0.247, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.6595744680851066e-07, | |
| "logits/chosen": 0.3012881875038147, | |
| "logits/rejected": 0.3096545338630676, | |
| "logps/chosen": -280.16510009765625, | |
| "logps/rejected": -291.20611572265625, | |
| "loss": 0.2072, | |
| "rewards/accuracies": 0.2083333283662796, | |
| "rewards/chosen": -0.00013855236466042697, | |
| "rewards/margins": -6.587021289305994e-06, | |
| "rewards/rejected": -0.0001319653238169849, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.319148936170213e-07, | |
| "logits/chosen": 0.31470876932144165, | |
| "logits/rejected": 0.31193265318870544, | |
| "logps/chosen": -327.54095458984375, | |
| "logps/rejected": -344.1315612792969, | |
| "loss": 0.2101, | |
| "rewards/accuracies": 0.23125000298023224, | |
| "rewards/chosen": -0.00020987121388316154, | |
| "rewards/margins": -7.895669841673225e-05, | |
| "rewards/rejected": -0.0001309145154664293, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.97872340425532e-07, | |
| "logits/chosen": 0.3390754163265228, | |
| "logits/rejected": 0.35239648818969727, | |
| "logps/chosen": -248.12741088867188, | |
| "logps/rejected": -250.649658203125, | |
| "loss": 0.2173, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": -0.00013244706497061998, | |
| "rewards/margins": 4.99083471368067e-05, | |
| "rewards/rejected": -0.00018235540483146906, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.3102382719516754, | |
| "logits/rejected": 0.32991108298301697, | |
| "logps/chosen": -301.28411865234375, | |
| "logps/rejected": -329.452880859375, | |
| "loss": 0.2161, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.00040395656833425164, | |
| "rewards/margins": 8.279903340735473e-06, | |
| "rewards/rejected": -0.00041223643347620964, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.3297872340425533e-06, | |
| "logits/chosen": 0.34786778688430786, | |
| "logits/rejected": 0.35041943192481995, | |
| "logps/chosen": -254.66000366210938, | |
| "logps/rejected": -254.5922393798828, | |
| "loss": 0.2075, | |
| "rewards/accuracies": 0.24375000596046448, | |
| "rewards/chosen": -0.00043535622535273433, | |
| "rewards/margins": 2.4446379029541276e-05, | |
| "rewards/rejected": -0.0004598026571329683, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.28234267234802246, | |
| "logits/rejected": 0.3171209692955017, | |
| "logps/chosen": -284.116943359375, | |
| "logps/rejected": -319.9707946777344, | |
| "loss": 0.2174, | |
| "rewards/accuracies": 0.24375000596046448, | |
| "rewards/chosen": -0.00010556467896094546, | |
| "rewards/margins": -1.569558480696287e-05, | |
| "rewards/rejected": -8.986910688690841e-05, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.8617021276595745e-06, | |
| "logits/chosen": 0.32085567712783813, | |
| "logits/rejected": 0.3497418761253357, | |
| "logps/chosen": -317.13531494140625, | |
| "logps/rejected": -336.9044494628906, | |
| "loss": 0.1988, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.0005014509661123157, | |
| "rewards/margins": 4.7156230721157044e-05, | |
| "rewards/rejected": -0.0005486071459017694, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.28880470991134644, | |
| "logits/rejected": 0.3247111141681671, | |
| "logps/chosen": -284.1805114746094, | |
| "logps/rejected": -306.4732666015625, | |
| "loss": 0.1987, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.00016313868400175124, | |
| "rewards/margins": 0.00013786499039269984, | |
| "rewards/rejected": 2.5273693609051406e-05, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.393617021276596e-06, | |
| "logits/chosen": 0.33655500411987305, | |
| "logits/rejected": 0.3271638751029968, | |
| "logps/chosen": -257.6504821777344, | |
| "logps/rejected": -280.1950988769531, | |
| "loss": 0.205, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": -0.00013443852367345244, | |
| "rewards/margins": 3.99285345338285e-05, | |
| "rewards/rejected": -0.00017436707275919616, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.30501195788383484, | |
| "logits/rejected": 0.30879053473472595, | |
| "logps/chosen": -291.9537048339844, | |
| "logps/rejected": -314.95623779296875, | |
| "loss": 0.2111, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -1.4989857845648658e-05, | |
| "rewards/margins": 0.0001315394474659115, | |
| "rewards/rejected": -0.0001465293171349913, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.9255319148936174e-06, | |
| "logits/chosen": 0.3248497545719147, | |
| "logits/rejected": 0.32843533158302307, | |
| "logps/chosen": -305.5889587402344, | |
| "logps/rejected": -315.61578369140625, | |
| "loss": 0.2001, | |
| "rewards/accuracies": 0.23125000298023224, | |
| "rewards/chosen": -0.0013872724957764149, | |
| "rewards/margins": -4.216738307150081e-05, | |
| "rewards/rejected": -0.0013451050035655499, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": 0.2866339087486267, | |
| "logits/rejected": 0.315407931804657, | |
| "logps/chosen": -271.4537048339844, | |
| "logps/rejected": -288.8186950683594, | |
| "loss": 0.2086, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": -0.0016517008189111948, | |
| "rewards/margins": 0.00013306428445503116, | |
| "rewards/rejected": -0.0017847649287432432, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.457446808510639e-06, | |
| "logits/chosen": 0.231134295463562, | |
| "logits/rejected": 0.3182562291622162, | |
| "logps/chosen": -296.9430847167969, | |
| "logps/rejected": -304.53387451171875, | |
| "loss": 0.2098, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": -0.002512121107429266, | |
| "rewards/margins": 0.00019210236496292055, | |
| "rewards/rejected": -0.002704223617911339, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.27288615703582764, | |
| "logits/rejected": 0.30947965383529663, | |
| "logps/chosen": -304.9461975097656, | |
| "logps/rejected": -314.1529235839844, | |
| "loss": 0.2097, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": -0.003562493249773979, | |
| "rewards/margins": 0.0003294651978649199, | |
| "rewards/rejected": -0.003891958389431238, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.98936170212766e-06, | |
| "logits/chosen": 0.2627570331096649, | |
| "logits/rejected": 0.34792831540107727, | |
| "logps/chosen": -272.2464294433594, | |
| "logps/rejected": -296.91107177734375, | |
| "loss": 0.218, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": -0.0038285471964627504, | |
| "rewards/margins": 0.0007261269493028522, | |
| "rewards/rejected": -0.0045546735636889935, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": 0.27352815866470337, | |
| "logits/rejected": 0.30977898836135864, | |
| "logps/chosen": -315.76251220703125, | |
| "logps/rejected": -325.41546630859375, | |
| "loss": 0.2011, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": -0.0019420146709308028, | |
| "rewards/margins": 0.0003213664167560637, | |
| "rewards/rejected": -0.0022633811458945274, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.521276595744681e-06, | |
| "logits/chosen": 0.30782410502433777, | |
| "logits/rejected": 0.26858752965927124, | |
| "logps/chosen": -291.3421630859375, | |
| "logps/rejected": -320.3433837890625, | |
| "loss": 0.2131, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.00032241392182186246, | |
| "rewards/margins": 0.0007240624399855733, | |
| "rewards/rejected": -0.00040164851816371083, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": 0.24440982937812805, | |
| "logits/rejected": 0.2718171179294586, | |
| "logps/chosen": -296.61517333984375, | |
| "logps/rejected": -307.93157958984375, | |
| "loss": 0.2069, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": -0.00043370993807911873, | |
| "rewards/margins": 0.00036561707383953035, | |
| "rewards/rejected": -0.0007993269828148186, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.999982660399688e-06, | |
| "logits/chosen": 0.17186453938484192, | |
| "logits/rejected": 0.16024716198444366, | |
| "logps/chosen": -247.7923126220703, | |
| "logps/rejected": -246.61996459960938, | |
| "loss": 0.2256, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": -0.00252458150498569, | |
| "rewards/margins": 0.00020101522386539727, | |
| "rewards/rejected": -0.0027255965396761894, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.99937579964398e-06, | |
| "logits/chosen": 0.14718958735466003, | |
| "logits/rejected": 0.217782661318779, | |
| "logps/chosen": -305.3340759277344, | |
| "logps/rejected": -323.8966064453125, | |
| "loss": 0.1991, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": -0.0030492418445646763, | |
| "rewards/margins": 3.36905250151176e-05, | |
| "rewards/rejected": -0.0030829322058707476, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9979021993870645e-06, | |
| "logits/chosen": 0.18054236471652985, | |
| "logits/rejected": 0.1431611180305481, | |
| "logps/chosen": -289.236328125, | |
| "logps/rejected": -295.9798278808594, | |
| "loss": 0.2109, | |
| "rewards/accuracies": 0.21875, | |
| "rewards/chosen": -0.0029132370837032795, | |
| "rewards/margins": -0.0005081830895505846, | |
| "rewards/rejected": -0.0024050544016063213, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.995562370647553e-06, | |
| "logits/chosen": 0.17394112050533295, | |
| "logits/rejected": 0.17676064372062683, | |
| "logps/chosen": -295.6396179199219, | |
| "logps/rejected": -312.4654235839844, | |
| "loss": 0.201, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": -0.0027714716270565987, | |
| "rewards/margins": 0.0016848895465955138, | |
| "rewards/rejected": -0.004456360824406147, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.992357124836838e-06, | |
| "logits/chosen": 0.1840275675058365, | |
| "logits/rejected": 0.1552029550075531, | |
| "logps/chosen": -324.6024169921875, | |
| "logps/rejected": -344.926513671875, | |
| "loss": 0.2192, | |
| "rewards/accuracies": 0.34375, | |
| "rewards/chosen": -0.004003914073109627, | |
| "rewards/margins": 0.0011075122747570276, | |
| "rewards/rejected": -0.005111426115036011, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882875734777044e-06, | |
| "logits/chosen": 0.15174627304077148, | |
| "logits/rejected": 0.15548208355903625, | |
| "logps/chosen": -301.86090087890625, | |
| "logps/rejected": -324.1502380371094, | |
| "loss": 0.2128, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": -0.00383182754740119, | |
| "rewards/margins": 0.00154831912368536, | |
| "rewards/rejected": -0.00538014667108655, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.983355127818882e-06, | |
| "logits/chosen": 0.1338830292224884, | |
| "logits/rejected": 0.126266211271286, | |
| "logps/chosen": -244.6050262451172, | |
| "logps/rejected": -272.59906005859375, | |
| "loss": 0.2059, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": -0.002967983018606901, | |
| "rewards/margins": 0.00032762493356131017, | |
| "rewards/rejected": -0.00329560786485672, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977561498345639e-06, | |
| "logits/chosen": 0.14223454892635345, | |
| "logits/rejected": 0.11773631721735, | |
| "logps/chosen": -312.8759460449219, | |
| "logps/rejected": -332.5674743652344, | |
| "loss": 0.2178, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": -0.002887751441448927, | |
| "rewards/margins": 0.002338848076760769, | |
| "rewards/rejected": -0.005226599983870983, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.970908694186624e-06, | |
| "logits/chosen": 0.054258234798908234, | |
| "logits/rejected": 0.049981165677309036, | |
| "logps/chosen": -252.06396484375, | |
| "logps/rejected": -297.46844482421875, | |
| "loss": 0.2038, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.003387606469914317, | |
| "rewards/margins": 0.004537281580269337, | |
| "rewards/rejected": -0.007924887351691723, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.9633990224171305e-06, | |
| "logits/chosen": -0.030563954263925552, | |
| "logits/rejected": 0.0610717348754406, | |
| "logps/chosen": -299.67572021484375, | |
| "logps/rejected": -325.629638671875, | |
| "loss": 0.2046, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.015065389685332775, | |
| "rewards/margins": 0.004181383643299341, | |
| "rewards/rejected": -0.019246775656938553, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.955035087259046e-06, | |
| "logits/chosen": 0.021658485755324364, | |
| "logits/rejected": 0.0002574294921942055, | |
| "logps/chosen": -279.9824523925781, | |
| "logps/rejected": -308.2466125488281, | |
| "loss": 0.2082, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": -0.0099684102460742, | |
| "rewards/margins": 0.003934068605303764, | |
| "rewards/rejected": -0.013902479782700539, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945819789177756e-06, | |
| "logits/chosen": -0.02307768724858761, | |
| "logits/rejected": 0.027991438284516335, | |
| "logps/chosen": -297.05120849609375, | |
| "logps/rejected": -306.86993408203125, | |
| "loss": 0.214, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": -0.003176984842866659, | |
| "rewards/margins": 0.0040995096787810326, | |
| "rewards/rejected": -0.00727649312466383, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.935756323876306e-06, | |
| "logits/chosen": -0.04134337604045868, | |
| "logits/rejected": -0.06087446212768555, | |
| "logps/chosen": -291.1628723144531, | |
| "logps/rejected": -304.87701416015625, | |
| "loss": 0.2055, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.0040726205334067345, | |
| "rewards/margins": 0.005008176900446415, | |
| "rewards/rejected": -0.009080798365175724, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924848181187199e-06, | |
| "logits/chosen": -0.08453729003667831, | |
| "logits/rejected": -0.09256991744041443, | |
| "logps/chosen": -261.996337890625, | |
| "logps/rejected": -302.02056884765625, | |
| "loss": 0.204, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.00689683947712183, | |
| "rewards/margins": 0.011717710644006729, | |
| "rewards/rejected": -0.018614551052451134, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.913099143862173e-06, | |
| "logits/chosen": -0.1567036509513855, | |
| "logits/rejected": -0.11430159956216812, | |
| "logps/chosen": -293.12091064453125, | |
| "logps/rejected": -317.5729064941406, | |
| "loss": 0.1985, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.014261825010180473, | |
| "rewards/margins": 0.005913619417697191, | |
| "rewards/rejected": -0.020175445824861526, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900513286260416e-06, | |
| "logits/chosen": -0.1313917338848114, | |
| "logits/rejected": -0.13553771376609802, | |
| "logps/chosen": -293.6689758300781, | |
| "logps/rejected": -307.853271484375, | |
| "loss": 0.2022, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": -0.00804301630705595, | |
| "rewards/margins": 0.005196637008339167, | |
| "rewards/rejected": -0.013239651918411255, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.887094972935645e-06, | |
| "logits/chosen": -0.11833436787128448, | |
| "logits/rejected": -0.11948303878307343, | |
| "logps/chosen": -305.84649658203125, | |
| "logps/rejected": -327.93115234375, | |
| "loss": 0.199, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.005626929923892021, | |
| "rewards/margins": 0.005307743325829506, | |
| "rewards/rejected": 0.0003191869764123112, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.87284885712256e-06, | |
| "logits/chosen": -0.1287965327501297, | |
| "logits/rejected": -0.07743237912654877, | |
| "logps/chosen": -305.2966613769531, | |
| "logps/rejected": -322.216064453125, | |
| "loss": 0.2042, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.013340242207050323, | |
| "rewards/margins": 0.006324528716504574, | |
| "rewards/rejected": 0.007015712559223175, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.857779879123181e-06, | |
| "logits/chosen": -0.13824257254600525, | |
| "logits/rejected": -0.17385557293891907, | |
| "logps/chosen": -271.77142333984375, | |
| "logps/rejected": -285.82611083984375, | |
| "loss": 0.2166, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.012355504557490349, | |
| "rewards/margins": 0.004869457334280014, | |
| "rewards/rejected": 0.0074860467575490475, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.841893264593643e-06, | |
| "logits/chosen": -0.1971556842327118, | |
| "logits/rejected": -0.14368289709091187, | |
| "logps/chosen": -289.67315673828125, | |
| "logps/rejected": -294.14239501953125, | |
| "loss": 0.2142, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.014088312163949013, | |
| "rewards/margins": 0.0018969494849443436, | |
| "rewards/rejected": 0.01219136081635952, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.825194522732023e-06, | |
| "logits/chosen": -0.15685835480690002, | |
| "logits/rejected": -0.14415477216243744, | |
| "logps/chosen": -258.82476806640625, | |
| "logps/rejected": -279.218505859375, | |
| "loss": 0.2035, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.012283021584153175, | |
| "rewards/margins": 0.007540999446064234, | |
| "rewards/rejected": 0.004742022603750229, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807689444367853e-06, | |
| "logits/chosen": -0.10798895359039307, | |
| "logits/rejected": -0.06372461467981339, | |
| "logps/chosen": -314.35968017578125, | |
| "logps/rejected": -325.0350036621094, | |
| "loss": 0.2124, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.015810634940862656, | |
| "rewards/margins": 0.0051138391718268394, | |
| "rewards/rejected": 0.01069679670035839, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.78938409995396e-06, | |
| "logits/chosen": -0.12151198089122772, | |
| "logits/rejected": -0.1099950298666954, | |
| "logps/chosen": -265.03253173828125, | |
| "logps/rejected": -293.6728210449219, | |
| "loss": 0.2046, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.008226810023188591, | |
| "rewards/margins": 0.003223727224394679, | |
| "rewards/rejected": 0.005003082565963268, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770284837461342e-06, | |
| "logits/chosen": -0.09874553233385086, | |
| "logits/rejected": -0.10907171666622162, | |
| "logps/chosen": -299.97259521484375, | |
| "logps/rejected": -309.26495361328125, | |
| "loss": 0.2047, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.009101228788495064, | |
| "rewards/margins": 0.0020746844820678234, | |
| "rewards/rejected": 0.00702654430642724, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7503982801778015e-06, | |
| "logits/chosen": -0.09576672315597534, | |
| "logits/rejected": -0.08475270867347717, | |
| "logps/chosen": -258.3625793457031, | |
| "logps/rejected": -288.2427673339844, | |
| "loss": 0.2207, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.013314478099346161, | |
| "rewards/margins": 0.005582061596214771, | |
| "rewards/rejected": 0.00773241650313139, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.729731324411104e-06, | |
| "logits/chosen": -0.11287762224674225, | |
| "logits/rejected": -0.1189860925078392, | |
| "logps/chosen": -269.3243713378906, | |
| "logps/rejected": -290.4499206542969, | |
| "loss": 0.2115, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.006840028800070286, | |
| "rewards/margins": 0.003011752385646105, | |
| "rewards/rejected": 0.0038282759487628937, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.7082911370974645e-06, | |
| "logits/chosen": -0.06270084530115128, | |
| "logits/rejected": -0.11760212481021881, | |
| "logps/chosen": -266.041259765625, | |
| "logps/rejected": -278.55743408203125, | |
| "loss": 0.219, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.014174575917422771, | |
| "rewards/margins": 0.0036397804506123066, | |
| "rewards/rejected": 0.010534795001149178, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68608515331618e-06, | |
| "logits/chosen": -0.08950970321893692, | |
| "logits/rejected": -0.12192481756210327, | |
| "logps/chosen": -220.92153930664062, | |
| "logps/rejected": -227.0800018310547, | |
| "loss": 0.2105, | |
| "rewards/accuracies": 0.2562499940395355, | |
| "rewards/chosen": 0.00980322528630495, | |
| "rewards/margins": 0.003905829042196274, | |
| "rewards/rejected": 0.005897396244108677, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.663121073711269e-06, | |
| "logits/chosen": -0.14287583529949188, | |
| "logits/rejected": -0.11475469172000885, | |
| "logps/chosen": -249.2185821533203, | |
| "logps/rejected": -269.6126708984375, | |
| "loss": 0.224, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.009408132173120975, | |
| "rewards/margins": 0.007567667867988348, | |
| "rewards/rejected": 0.0018404644215479493, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.63940686182103e-06, | |
| "logits/chosen": -0.13410678505897522, | |
| "logits/rejected": -0.1070268377661705, | |
| "logps/chosen": -276.46856689453125, | |
| "logps/rejected": -310.9843444824219, | |
| "loss": 0.2096, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.0070691583678126335, | |
| "rewards/margins": 0.010405855253338814, | |
| "rewards/rejected": -0.003336697118356824, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.614950741316425e-06, | |
| "logits/chosen": -0.21566160023212433, | |
| "logits/rejected": -0.18181584775447845, | |
| "logps/chosen": -219.1405487060547, | |
| "logps/rejected": -248.4638671875, | |
| "loss": 0.1972, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": 0.007403376046568155, | |
| "rewards/margins": 0.004462541546672583, | |
| "rewards/rejected": 0.0029408347327262163, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.589761193149254e-06, | |
| "logits/chosen": -0.17594993114471436, | |
| "logits/rejected": -0.1787882298231125, | |
| "logps/chosen": -248.833251953125, | |
| "logps/rejected": -265.1255798339844, | |
| "loss": 0.2058, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.005866335239261389, | |
| "rewards/margins": 0.008035833016037941, | |
| "rewards/rejected": -0.002169497776776552, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.563846952611112e-06, | |
| "logits/chosen": -0.2459220141172409, | |
| "logits/rejected": -0.21481938660144806, | |
| "logps/chosen": -290.0213623046875, | |
| "logps/rejected": -324.1817321777344, | |
| "loss": 0.1908, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.0015872556250542402, | |
| "rewards/margins": 0.011338635347783566, | |
| "rewards/rejected": -0.009751381352543831, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.537217006304141e-06, | |
| "logits/chosen": -0.22694924473762512, | |
| "logits/rejected": -0.22461485862731934, | |
| "logps/chosen": -272.0402526855469, | |
| "logps/rejected": -286.1138916015625, | |
| "loss": 0.2185, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": -1.9139843061566353e-05, | |
| "rewards/margins": 0.00025601504603400826, | |
| "rewards/rejected": -0.00027515500551089644, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.50988058902464e-06, | |
| "logits/chosen": -0.2873474061489105, | |
| "logits/rejected": -0.24353663623332977, | |
| "logps/chosen": -319.0919189453125, | |
| "logps/rejected": -352.68621826171875, | |
| "loss": 0.2102, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.011865551583468914, | |
| "rewards/margins": 0.006343546323478222, | |
| "rewards/rejected": 0.005522005259990692, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481847180560593e-06, | |
| "logits/chosen": -0.2149961292743683, | |
| "logits/rejected": -0.22789709270000458, | |
| "logps/chosen": -286.41851806640625, | |
| "logps/rejected": -297.98248291015625, | |
| "loss": 0.2049, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.010984813794493675, | |
| "rewards/margins": 0.0014541022246703506, | |
| "rewards/rejected": 0.009530710987746716, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.453126502404253e-06, | |
| "logits/chosen": -0.2241373062133789, | |
| "logits/rejected": -0.1860576570034027, | |
| "logps/chosen": -279.4507141113281, | |
| "logps/rejected": -281.85516357421875, | |
| "loss": 0.2199, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.020039092749357224, | |
| "rewards/margins": 0.0005937496898695827, | |
| "rewards/rejected": 0.019445344805717468, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.423728514380892e-06, | |
| "logits/chosen": -0.18391680717468262, | |
| "logits/rejected": -0.162066251039505, | |
| "logps/chosen": -269.0604248046875, | |
| "logps/rejected": -299.13055419921875, | |
| "loss": 0.1967, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.025296321138739586, | |
| "rewards/margins": 0.006020676344633102, | |
| "rewards/rejected": 0.019275646656751633, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.393663411194918e-06, | |
| "logits/chosen": -0.15143278241157532, | |
| "logits/rejected": -0.1272958219051361, | |
| "logps/chosen": -275.3654479980469, | |
| "logps/rejected": -332.71636962890625, | |
| "loss": 0.2102, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.024300305172801018, | |
| "rewards/margins": 0.006145091727375984, | |
| "rewards/rejected": 0.018155213445425034, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.362941618894523e-06, | |
| "logits/chosen": -0.24861934781074524, | |
| "logits/rejected": -0.21499958634376526, | |
| "logps/chosen": -242.86181640625, | |
| "logps/rejected": -256.13409423828125, | |
| "loss": 0.2059, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": 0.01787170208990574, | |
| "rewards/margins": 0.002406838582828641, | |
| "rewards/rejected": 0.015464862808585167, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.331573791256116e-06, | |
| "logits/chosen": -0.24243931472301483, | |
| "logits/rejected": -0.20876605808734894, | |
| "logps/chosen": -222.46432495117188, | |
| "logps/rejected": -240.7364044189453, | |
| "loss": 0.2187, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.02379140630364418, | |
| "rewards/margins": 0.004867316223680973, | |
| "rewards/rejected": 0.018924091011285782, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.299570806089786e-06, | |
| "logits/chosen": -0.25369709730148315, | |
| "logits/rejected": -0.26770374178886414, | |
| "logps/chosen": -262.30889892578125, | |
| "logps/rejected": -301.822998046875, | |
| "loss": 0.2092, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.024355897679924965, | |
| "rewards/margins": 0.008627799339592457, | |
| "rewards/rejected": 0.015728099271655083, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.266943761467057e-06, | |
| "logits/chosen": -0.278990775346756, | |
| "logits/rejected": -0.23303751647472382, | |
| "logps/chosen": -281.37811279296875, | |
| "logps/rejected": -289.0556945800781, | |
| "loss": 0.1924, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.028277212753891945, | |
| "rewards/margins": 0.003026631660759449, | |
| "rewards/rejected": 0.02525058388710022, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.233703971872287e-06, | |
| "logits/chosen": -0.2651432752609253, | |
| "logits/rejected": -0.23803620040416718, | |
| "logps/chosen": -274.52972412109375, | |
| "logps/rejected": -284.91864013671875, | |
| "loss": 0.2075, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.029297566041350365, | |
| "rewards/margins": 0.0042782137170434, | |
| "rewards/rejected": 0.02501935325562954, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.1998629642789925e-06, | |
| "logits/chosen": -0.2681610584259033, | |
| "logits/rejected": -0.2098967581987381, | |
| "logps/chosen": -273.29522705078125, | |
| "logps/rejected": -307.06207275390625, | |
| "loss": 0.2109, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": 0.026482801884412766, | |
| "rewards/margins": 0.008065291680395603, | |
| "rewards/rejected": 0.018417512997984886, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.165432474152505e-06, | |
| "logits/chosen": -0.2867113947868347, | |
| "logits/rejected": -0.3026007413864136, | |
| "logps/chosen": -258.7665100097656, | |
| "logps/rejected": -276.9542541503906, | |
| "loss": 0.1981, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.03178761154413223, | |
| "rewards/margins": 0.0078035155311226845, | |
| "rewards/rejected": 0.023984093219041824, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.130424441380308e-06, | |
| "logits/chosen": -0.3068264424800873, | |
| "logits/rejected": -0.2953132390975952, | |
| "logps/chosen": -278.3927001953125, | |
| "logps/rejected": -294.95611572265625, | |
| "loss": 0.2182, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.024625886231660843, | |
| "rewards/margins": 0.0047270385548472404, | |
| "rewards/rejected": 0.019898848608136177, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.09485100613151e-06, | |
| "logits/chosen": -0.3508884310722351, | |
| "logits/rejected": -0.3666234314441681, | |
| "logps/chosen": -255.5622100830078, | |
| "logps/rejected": -267.3009033203125, | |
| "loss": 0.2199, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.025408515706658363, | |
| "rewards/margins": 0.0033288276754319668, | |
| "rewards/rejected": 0.02207968756556511, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.058724504646834e-06, | |
| "logits/chosen": -0.27611714601516724, | |
| "logits/rejected": -0.2742993235588074, | |
| "logps/chosen": -255.03402709960938, | |
| "logps/rejected": -279.2400207519531, | |
| "loss": 0.1928, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.026569899171590805, | |
| "rewards/margins": 0.006919047329574823, | |
| "rewards/rejected": 0.01965085044503212, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.022057464960632e-06, | |
| "logits/chosen": -0.29631996154785156, | |
| "logits/rejected": -0.3354335427284241, | |
| "logps/chosen": -224.5742645263672, | |
| "logps/rejected": -251.43563842773438, | |
| "loss": 0.2081, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.02301851287484169, | |
| "rewards/margins": 0.00754903256893158, | |
| "rewards/rejected": 0.015469479374587536, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.984862602556383e-06, | |
| "logits/chosen": -0.3194465637207031, | |
| "logits/rejected": -0.2741653025150299, | |
| "logps/chosen": -292.7205505371094, | |
| "logps/rejected": -280.698486328125, | |
| "loss": 0.2113, | |
| "rewards/accuracies": 0.24375000596046448, | |
| "rewards/chosen": 0.01744149625301361, | |
| "rewards/margins": -0.002081270096823573, | |
| "rewards/rejected": 0.019522765651345253, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.947152815957187e-06, | |
| "logits/chosen": -0.3113805949687958, | |
| "logits/rejected": -0.27869972586631775, | |
| "logps/chosen": -267.34503173828125, | |
| "logps/rejected": -260.38397216796875, | |
| "loss": 0.2177, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.03015447035431862, | |
| "rewards/margins": 0.002841535024344921, | |
| "rewards/rejected": 0.027312934398651123, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.908941182252785e-06, | |
| "logits/chosen": -0.2623198926448822, | |
| "logits/rejected": -0.2412450760602951, | |
| "logps/chosen": -268.50933837890625, | |
| "logps/rejected": -307.6174621582031, | |
| "loss": 0.1982, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.025938820093870163, | |
| "rewards/margins": 0.005723121576011181, | |
| "rewards/rejected": 0.020215699449181557, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.8702409525646535e-06, | |
| "logits/chosen": -0.26187169551849365, | |
| "logits/rejected": -0.2671493887901306, | |
| "logps/chosen": -234.5527801513672, | |
| "logps/rejected": -276.95465087890625, | |
| "loss": 0.2124, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.02780374325811863, | |
| "rewards/margins": 0.00728409830480814, | |
| "rewards/rejected": 0.020519645884633064, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.8310655474507495e-06, | |
| "logits/chosen": -0.31342610716819763, | |
| "logits/rejected": -0.3103486895561218, | |
| "logps/chosen": -272.54730224609375, | |
| "logps/rejected": -289.4356384277344, | |
| "loss": 0.1874, | |
| "rewards/accuracies": 0.34375, | |
| "rewards/chosen": 0.025291483849287033, | |
| "rewards/margins": 0.0070581152103841305, | |
| "rewards/rejected": 0.018233366310596466, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7914285522515002e-06, | |
| "logits/chosen": -0.31440383195877075, | |
| "logits/rejected": -0.32669904828071594, | |
| "logps/chosen": -278.309814453125, | |
| "logps/rejected": -316.8228759765625, | |
| "loss": 0.202, | |
| "rewards/accuracies": 0.40625, | |
| "rewards/chosen": 0.032375235110521317, | |
| "rewards/margins": 0.007661235518753529, | |
| "rewards/rejected": 0.024713998660445213, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.751343712378639e-06, | |
| "logits/chosen": -0.3353944420814514, | |
| "logits/rejected": -0.3212040364742279, | |
| "logps/chosen": -254.7311553955078, | |
| "logps/rejected": -263.9468994140625, | |
| "loss": 0.2265, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.020177539438009262, | |
| "rewards/margins": 0.003813125193119049, | |
| "rewards/rejected": 0.016364412382245064, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.710824928548546e-06, | |
| "logits/chosen": -0.4110103249549866, | |
| "logits/rejected": -0.38202887773513794, | |
| "logps/chosen": -304.6473693847656, | |
| "logps/rejected": -348.3372802734375, | |
| "loss": 0.204, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.028538638725876808, | |
| "rewards/margins": 0.008010391145944595, | |
| "rewards/rejected": 0.020528245717287064, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.6698862519617225e-06, | |
| "logits/chosen": -0.37583234906196594, | |
| "logits/rejected": -0.39775121212005615, | |
| "logps/chosen": -250.30337524414062, | |
| "logps/rejected": -283.32464599609375, | |
| "loss": 0.198, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.023627078160643578, | |
| "rewards/margins": 0.008884016424417496, | |
| "rewards/rejected": 0.014743062667548656, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6285418794300793e-06, | |
| "logits/chosen": -0.40414610505104065, | |
| "logits/rejected": -0.37960129976272583, | |
| "logps/chosen": -300.35296630859375, | |
| "logps/rejected": -304.61834716796875, | |
| "loss": 0.215, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.021693145856261253, | |
| "rewards/margins": 0.002025413792580366, | |
| "rewards/rejected": 0.01966773346066475, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.5868061484537365e-06, | |
| "logits/chosen": -0.3469582498073578, | |
| "logits/rejected": -0.2765125632286072, | |
| "logps/chosen": -248.47525024414062, | |
| "logps/rejected": -262.3398132324219, | |
| "loss": 0.2054, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.023630034178495407, | |
| "rewards/margins": 0.006072540767490864, | |
| "rewards/rejected": 0.01755749061703682, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.5446935322490285e-06, | |
| "logits/chosen": -0.31192153692245483, | |
| "logits/rejected": -0.3198978304862976, | |
| "logps/chosen": -264.4431457519531, | |
| "logps/rejected": -276.5923156738281, | |
| "loss": 0.1997, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.019144730642437935, | |
| "rewards/margins": 0.007585918065160513, | |
| "rewards/rejected": 0.011558813974261284, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.502218634729447e-06, | |
| "logits/chosen": -0.3657092750072479, | |
| "logits/rejected": -0.36896997690200806, | |
| "logps/chosen": -285.9904479980469, | |
| "logps/rejected": -304.4721984863281, | |
| "loss": 0.2131, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.01876096799969673, | |
| "rewards/margins": 0.0027052960358560085, | |
| "rewards/rejected": 0.016055669635534286, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.459396185441265e-06, | |
| "logits/chosen": -0.3460562825202942, | |
| "logits/rejected": -0.32295316457748413, | |
| "logps/chosen": -223.621337890625, | |
| "logps/rejected": -230.530029296875, | |
| "loss": 0.2041, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.020683612674474716, | |
| "rewards/margins": 0.004844160284847021, | |
| "rewards/rejected": 0.015839453786611557, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4162410344555834e-06, | |
| "logits/chosen": -0.2728729844093323, | |
| "logits/rejected": -0.27788490056991577, | |
| "logps/chosen": -268.2927551269531, | |
| "logps/rejected": -274.38690185546875, | |
| "loss": 0.2068, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.019065286964178085, | |
| "rewards/margins": 0.0036661080084741116, | |
| "rewards/rejected": 0.015399177558720112, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3727681472185937e-06, | |
| "logits/chosen": -0.3043738603591919, | |
| "logits/rejected": -0.2903715968132019, | |
| "logps/chosen": -255.2548370361328, | |
| "logps/rejected": -280.83477783203125, | |
| "loss": 0.209, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": 0.01805216260254383, | |
| "rewards/margins": 0.002792063634842634, | |
| "rewards/rejected": 0.01526009850203991, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3289925993618217e-06, | |
| "logits/chosen": -0.3053690195083618, | |
| "logits/rejected": -0.3325154483318329, | |
| "logps/chosen": -249.2985382080078, | |
| "logps/rejected": -285.81744384765625, | |
| "loss": 0.209, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.02784855104982853, | |
| "rewards/margins": 0.010397961363196373, | |
| "rewards/rejected": 0.017450589686632156, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2849295714741643e-06, | |
| "logits/chosen": -0.3305487334728241, | |
| "logits/rejected": -0.3348706364631653, | |
| "logps/chosen": -256.72357177734375, | |
| "logps/rejected": -275.69354248046875, | |
| "loss": 0.195, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.027434032410383224, | |
| "rewards/margins": 0.0030927606858313084, | |
| "rewards/rejected": 0.024341274052858353, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2405943438375287e-06, | |
| "logits/chosen": -0.3463178873062134, | |
| "logits/rejected": -0.3138378858566284, | |
| "logps/chosen": -270.30596923828125, | |
| "logps/rejected": -297.59344482421875, | |
| "loss": 0.2149, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.02239718660712242, | |
| "rewards/margins": 0.007217752281576395, | |
| "rewards/rejected": 0.015179434791207314, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.1960022911279036e-06, | |
| "logits/chosen": -0.3470946252346039, | |
| "logits/rejected": -0.2967691421508789, | |
| "logps/chosen": -262.6204528808594, | |
| "logps/rejected": -301.16021728515625, | |
| "loss": 0.2051, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": 0.02056020125746727, | |
| "rewards/margins": 0.0049241771921515465, | |
| "rewards/rejected": 0.015636026859283447, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.1511688770836844e-06, | |
| "logits/chosen": -0.3337462842464447, | |
| "logits/rejected": -0.3448733985424042, | |
| "logps/chosen": -290.02923583984375, | |
| "logps/rejected": -320.1620178222656, | |
| "loss": 0.2, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.02383309230208397, | |
| "rewards/margins": 0.01055764127522707, | |
| "rewards/rejected": 0.013275451958179474, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.1061096491431307e-06, | |
| "logits/chosen": -0.35830608010292053, | |
| "logits/rejected": -0.353718101978302, | |
| "logps/chosen": -252.27413940429688, | |
| "logps/rejected": -262.99957275390625, | |
| "loss": 0.207, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.015391258522868156, | |
| "rewards/margins": 0.0036770054139196873, | |
| "rewards/rejected": 0.011714254505932331, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.0608402330527796e-06, | |
| "logits/chosen": -0.30209046602249146, | |
| "logits/rejected": -0.3192787766456604, | |
| "logps/chosen": -286.3800354003906, | |
| "logps/rejected": -281.1745910644531, | |
| "loss": 0.2099, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.0164722241461277, | |
| "rewards/margins": 0.0019333992386236787, | |
| "rewards/rejected": 0.014538826420903206, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.0153763274487176e-06, | |
| "logits/chosen": -0.2973942458629608, | |
| "logits/rejected": -0.3040563464164734, | |
| "logps/chosen": -284.22100830078125, | |
| "logps/rejected": -326.08551025390625, | |
| "loss": 0.2, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.019569557160139084, | |
| "rewards/margins": 0.011648133397102356, | |
| "rewards/rejected": 0.007921423763036728, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9697336984125683e-06, | |
| "logits/chosen": -0.3307463824748993, | |
| "logits/rejected": -0.2803489863872528, | |
| "logps/chosen": -278.6820068359375, | |
| "logps/rejected": -318.23333740234375, | |
| "loss": 0.2145, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.021947836503386497, | |
| "rewards/margins": 0.008288668468594551, | |
| "rewards/rejected": 0.013659171760082245, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.923928174004094e-06, | |
| "logits/chosen": -0.2774783968925476, | |
| "logits/rejected": -0.287009060382843, | |
| "logps/chosen": -306.74951171875, | |
| "logps/rejected": -352.70037841796875, | |
| "loss": 0.1975, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.021842315793037415, | |
| "rewards/margins": 0.012494521215558052, | |
| "rewards/rejected": 0.009347794577479362, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.8779756387723036e-06, | |
| "logits/chosen": -0.3110584616661072, | |
| "logits/rejected": -0.2976732850074768, | |
| "logps/chosen": -263.1468505859375, | |
| "logps/rejected": -299.36834716796875, | |
| "loss": 0.2162, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.02124103158712387, | |
| "rewards/margins": 0.007206873502582312, | |
| "rewards/rejected": 0.014034159481525421, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.831892028246968e-06, | |
| "logits/chosen": -0.2773077189922333, | |
| "logits/rejected": -0.27867287397384644, | |
| "logps/chosen": -244.27871704101562, | |
| "logps/rejected": -311.6878967285156, | |
| "loss": 0.2025, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.026071447879076004, | |
| "rewards/margins": 0.012237906455993652, | |
| "rewards/rejected": 0.013833543285727501, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7856933234124617e-06, | |
| "logits/chosen": -0.30510154366493225, | |
| "logits/rejected": -0.3271741569042206, | |
| "logps/chosen": -284.21990966796875, | |
| "logps/rejected": -316.27313232421875, | |
| "loss": 0.2202, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.022279905155301094, | |
| "rewards/margins": 0.007268127985298634, | |
| "rewards/rejected": 0.015011775307357311, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7393955451658387e-06, | |
| "logits/chosen": -0.36560115218162537, | |
| "logits/rejected": -0.30673733353614807, | |
| "logps/chosen": -299.2345275878906, | |
| "logps/rejected": -282.7044982910156, | |
| "loss": 0.2174, | |
| "rewards/accuracies": 0.2562499940395355, | |
| "rewards/chosen": 0.01567717455327511, | |
| "rewards/margins": -0.0012121463660150766, | |
| "rewards/rejected": 0.01688932254910469, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.6930147487610667e-06, | |
| "logits/chosen": -0.33714741468429565, | |
| "logits/rejected": -0.28691935539245605, | |
| "logps/chosen": -266.865966796875, | |
| "logps/rejected": -300.762939453125, | |
| "loss": 0.1991, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.0294143445789814, | |
| "rewards/margins": 0.008245664648711681, | |
| "rewards/rejected": 0.021168678998947144, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.6465670182413487e-06, | |
| "logits/chosen": -0.2915731370449066, | |
| "logits/rejected": -0.24867181479930878, | |
| "logps/chosen": -282.2441101074219, | |
| "logps/rejected": -286.1959228515625, | |
| "loss": 0.1965, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.025241781026124954, | |
| "rewards/margins": 0.0037514586001634598, | |
| "rewards/rejected": 0.021490320563316345, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.6000684608614594e-06, | |
| "logits/chosen": -0.28083381056785583, | |
| "logits/rejected": -0.29559266567230225, | |
| "logps/chosen": -283.65484619140625, | |
| "logps/rejected": -310.78387451171875, | |
| "loss": 0.2098, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.028087038546800613, | |
| "rewards/margins": 0.0046967873349785805, | |
| "rewards/rejected": 0.023390252143144608, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.5535352015020338e-06, | |
| "logits/chosen": -0.2830182909965515, | |
| "logits/rejected": -0.28061428666114807, | |
| "logps/chosen": -273.9687194824219, | |
| "logps/rejected": -298.3106689453125, | |
| "loss": 0.2072, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.025423675775527954, | |
| "rewards/margins": 0.006900913082063198, | |
| "rewards/rejected": 0.018522759899497032, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.506983377077741e-06, | |
| "logits/chosen": -0.2505617141723633, | |
| "logits/rejected": -0.2675876021385193, | |
| "logps/chosen": -269.5409851074219, | |
| "logps/rejected": -285.0895080566406, | |
| "loss": 0.205, | |
| "rewards/accuracies": 0.34375, | |
| "rewards/chosen": 0.026411527767777443, | |
| "rewards/margins": 0.0052161491475999355, | |
| "rewards/rejected": 0.02119537815451622, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.460429130941289e-06, | |
| "logits/chosen": -0.2733067572116852, | |
| "logits/rejected": -0.24096527695655823, | |
| "logps/chosen": -231.34054565429688, | |
| "logps/rejected": -242.85177612304688, | |
| "loss": 0.2126, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": 0.0189601369202137, | |
| "rewards/margins": 0.005297792609781027, | |
| "rewards/rejected": 0.013662343844771385, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.413888607285192e-06, | |
| "logits/chosen": -0.330772340297699, | |
| "logits/rejected": -0.27838239073753357, | |
| "logps/chosen": -248.7228546142578, | |
| "logps/rejected": -279.42071533203125, | |
| "loss": 0.2178, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.020268592983484268, | |
| "rewards/margins": 0.005627959035336971, | |
| "rewards/rejected": 0.014640633948147297, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.367377945543249e-06, | |
| "logits/chosen": -0.32522693276405334, | |
| "logits/rejected": -0.26629459857940674, | |
| "logps/chosen": -245.46865844726562, | |
| "logps/rejected": -254.28073120117188, | |
| "loss": 0.2134, | |
| "rewards/accuracies": 0.24375000596046448, | |
| "rewards/chosen": 0.021207790821790695, | |
| "rewards/margins": 0.0022269729524850845, | |
| "rewards/rejected": 0.01898081973195076, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.320913274793676e-06, | |
| "logits/chosen": -0.3246033787727356, | |
| "logits/rejected": -0.28526002168655396, | |
| "logps/chosen": -275.1258239746094, | |
| "logps/rejected": -282.76788330078125, | |
| "loss": 0.214, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.029394343495368958, | |
| "rewards/margins": 0.003130835946649313, | |
| "rewards/rejected": 0.026263505220413208, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.27451070816582e-06, | |
| "logits/chosen": -0.31318122148513794, | |
| "logits/rejected": -0.3022027015686035, | |
| "logps/chosen": -289.7471923828125, | |
| "logps/rejected": -316.6593322753906, | |
| "loss": 0.2139, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.03029504418373108, | |
| "rewards/margins": 0.008487190119922161, | |
| "rewards/rejected": 0.021807854995131493, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.228186337252414e-06, | |
| "logits/chosen": -0.32163310050964355, | |
| "logits/rejected": -0.2794838547706604, | |
| "logps/chosen": -237.8488311767578, | |
| "logps/rejected": -261.3009948730469, | |
| "loss": 0.2022, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.024872692301869392, | |
| "rewards/margins": 0.006985250860452652, | |
| "rewards/rejected": 0.01788743957877159, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1819562265292946e-06, | |
| "logits/chosen": -0.33526480197906494, | |
| "logits/rejected": -0.3139267563819885, | |
| "logps/chosen": -314.11761474609375, | |
| "logps/rejected": -324.87353515625, | |
| "loss": 0.2031, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.026857828721404076, | |
| "rewards/margins": 0.0052116261795163155, | |
| "rewards/rejected": 0.021646203473210335, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1358364077845236e-06, | |
| "logits/chosen": -0.3513597846031189, | |
| "logits/rejected": -0.3594262897968292, | |
| "logps/chosen": -252.52978515625, | |
| "logps/rejected": -294.7911376953125, | |
| "loss": 0.2068, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.03454085439443588, | |
| "rewards/margins": 0.009111504070460796, | |
| "rewards/rejected": 0.02542934939265251, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.089842874558849e-06, | |
| "logits/chosen": -0.3378306031227112, | |
| "logits/rejected": -0.35515913367271423, | |
| "logps/chosen": -246.19320678710938, | |
| "logps/rejected": -299.27508544921875, | |
| "loss": 0.2069, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.03665949031710625, | |
| "rewards/margins": 0.01151054073125124, | |
| "rewards/rejected": 0.02514895237982273, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0439915765994242e-06, | |
| "logits/chosen": -0.3291921317577362, | |
| "logits/rejected": -0.3076435923576355, | |
| "logps/chosen": -245.3810577392578, | |
| "logps/rejected": -282.0654296875, | |
| "loss": 0.1944, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.03457210958003998, | |
| "rewards/margins": 0.012660647742450237, | |
| "rewards/rejected": 0.02191145531833172, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9982984143287186e-06, | |
| "logits/chosen": -0.4073921740055084, | |
| "logits/rejected": -0.4076351523399353, | |
| "logps/chosen": -243.7863311767578, | |
| "logps/rejected": -278.9535217285156, | |
| "loss": 0.2071, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.031270626932382584, | |
| "rewards/margins": 0.011814272962510586, | |
| "rewards/rejected": 0.019456351175904274, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.95277923333053e-06, | |
| "logits/chosen": -0.3861219882965088, | |
| "logits/rejected": -0.3469892144203186, | |
| "logps/chosen": -302.31243896484375, | |
| "logps/rejected": -358.0201721191406, | |
| "loss": 0.2139, | |
| "rewards/accuracies": 0.3812499940395355, | |
| "rewards/chosen": 0.033745743334293365, | |
| "rewards/margins": 0.011395677924156189, | |
| "rewards/rejected": 0.022350067272782326, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.9074498188550156e-06, | |
| "logits/chosen": -0.382877379655838, | |
| "logits/rejected": -0.3579499423503876, | |
| "logps/chosen": -255.16983032226562, | |
| "logps/rejected": -281.951904296875, | |
| "loss": 0.1989, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.026954079046845436, | |
| "rewards/margins": 0.005889051593840122, | |
| "rewards/rejected": 0.02106502838432789, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.862325890344643e-06, | |
| "logits/chosen": -0.41709408164024353, | |
| "logits/rejected": -0.3790467381477356, | |
| "logps/chosen": -237.56845092773438, | |
| "logps/rejected": -253.5933074951172, | |
| "loss": 0.2093, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.02573922835290432, | |
| "rewards/margins": 0.006860650144517422, | |
| "rewards/rejected": 0.018878577277064323, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.817423095982972e-06, | |
| "logits/chosen": -0.4551427364349365, | |
| "logits/rejected": -0.4378645420074463, | |
| "logps/chosen": -259.9010925292969, | |
| "logps/rejected": -296.59539794921875, | |
| "loss": 0.1942, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.02989533171057701, | |
| "rewards/margins": 0.01016165129840374, | |
| "rewards/rejected": 0.01973368227481842, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.7727570072681293e-06, | |
| "logits/chosen": -0.34378260374069214, | |
| "logits/rejected": -0.34789830446243286, | |
| "logps/chosen": -238.48300170898438, | |
| "logps/rejected": -260.9052429199219, | |
| "loss": 0.211, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.027008891105651855, | |
| "rewards/margins": 0.00940448883920908, | |
| "rewards/rejected": 0.01760440319776535, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.7283431136128961e-06, | |
| "logits/chosen": -0.4389713406562805, | |
| "logits/rejected": -0.4431864619255066, | |
| "logps/chosen": -286.2460632324219, | |
| "logps/rejected": -335.5301208496094, | |
| "loss": 0.2002, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.024310097098350525, | |
| "rewards/margins": 0.011298440396785736, | |
| "rewards/rejected": 0.013011658564209938, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6841968169732478e-06, | |
| "logits/chosen": -0.44720926880836487, | |
| "logits/rejected": -0.40391507744789124, | |
| "logps/chosen": -310.9598083496094, | |
| "logps/rejected": -295.41510009765625, | |
| "loss": 0.2271, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.024556024000048637, | |
| "rewards/margins": 0.003975218627601862, | |
| "rewards/rejected": 0.020580807700753212, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6403334265072284e-06, | |
| "logits/chosen": -0.4026651382446289, | |
| "logits/rejected": -0.4319339692592621, | |
| "logps/chosen": -235.80569458007812, | |
| "logps/rejected": -272.6563415527344, | |
| "loss": 0.195, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.02504395879805088, | |
| "rewards/margins": 0.005421781446784735, | |
| "rewards/rejected": 0.019622178748250008, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5967681532660066e-06, | |
| "logits/chosen": -0.4050057530403137, | |
| "logits/rejected": -0.3732849657535553, | |
| "logps/chosen": -269.2541809082031, | |
| "logps/rejected": -283.1937561035156, | |
| "loss": 0.2061, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.025783170014619827, | |
| "rewards/margins": 0.005010620225220919, | |
| "rewards/rejected": 0.020772550255060196, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5535161049189463e-06, | |
| "logits/chosen": -0.38786306977272034, | |
| "logits/rejected": -0.41273707151412964, | |
| "logps/chosen": -281.177734375, | |
| "logps/rejected": -291.16522216796875, | |
| "loss": 0.1982, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.024268019944429398, | |
| "rewards/margins": 0.006826480384916067, | |
| "rewards/rejected": 0.017441539093852043, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5105922805145356e-06, | |
| "logits/chosen": -0.3743259310722351, | |
| "logits/rejected": -0.42252597212791443, | |
| "logps/chosen": -276.8372497558594, | |
| "logps/rejected": -302.574951171875, | |
| "loss": 0.2033, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.027395809069275856, | |
| "rewards/margins": 0.0073099941946566105, | |
| "rewards/rejected": 0.020085815340280533, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4680115652789823e-06, | |
| "logits/chosen": -0.411508172750473, | |
| "logits/rejected": -0.40416446328163147, | |
| "logps/chosen": -303.2101745605469, | |
| "logps/rejected": -330.8990173339844, | |
| "loss": 0.1967, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.030878880992531776, | |
| "rewards/margins": 0.006534040905535221, | |
| "rewards/rejected": 0.02434483729302883, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.4257887254542767e-06, | |
| "logits/chosen": -0.4163438677787781, | |
| "logits/rejected": -0.4173736572265625, | |
| "logps/chosen": -227.8070068359375, | |
| "logps/rejected": -288.9392395019531, | |
| "loss": 0.1975, | |
| "rewards/accuracies": 0.39375001192092896, | |
| "rewards/chosen": 0.029939154163002968, | |
| "rewards/margins": 0.019917789846658707, | |
| "rewards/rejected": 0.010021360591053963, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.3839384031775227e-06, | |
| "logits/chosen": -0.4054839611053467, | |
| "logits/rejected": -0.37493056058883667, | |
| "logps/chosen": -272.3634033203125, | |
| "logps/rejected": -300.8335266113281, | |
| "loss": 0.1959, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.025880863890051842, | |
| "rewards/margins": 0.006764904595911503, | |
| "rewards/rejected": 0.019115960225462914, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.342475111403298e-06, | |
| "logits/chosen": -0.43276605010032654, | |
| "logits/rejected": -0.3933637738227844, | |
| "logps/chosen": -277.875244140625, | |
| "logps/rejected": -315.5643310546875, | |
| "loss": 0.2086, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.02479017712175846, | |
| "rewards/margins": 0.009227531030774117, | |
| "rewards/rejected": 0.015562644228339195, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3014132288708209e-06, | |
| "logits/chosen": -0.4051785469055176, | |
| "logits/rejected": -0.4154541492462158, | |
| "logps/chosen": -284.35870361328125, | |
| "logps/rejected": -343.5641174316406, | |
| "loss": 0.1946, | |
| "rewards/accuracies": 0.39375001192092896, | |
| "rewards/chosen": 0.030164774507284164, | |
| "rewards/margins": 0.018032994121313095, | |
| "rewards/rejected": 0.012131785042583942, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2607669951176549e-06, | |
| "logits/chosen": -0.39242345094680786, | |
| "logits/rejected": -0.4176657795906067, | |
| "logps/chosen": -265.1470947265625, | |
| "logps/rejected": -283.01898193359375, | |
| "loss": 0.2152, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.01843050867319107, | |
| "rewards/margins": 0.00565576646476984, | |
| "rewards/rejected": 0.012774743139743805, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2205505055416891e-06, | |
| "logits/chosen": -0.46165138483047485, | |
| "logits/rejected": -0.4150448739528656, | |
| "logps/chosen": -259.72576904296875, | |
| "logps/rejected": -290.30096435546875, | |
| "loss": 0.2172, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.019215788692235947, | |
| "rewards/margins": 0.008153979666531086, | |
| "rewards/rejected": 0.011061809957027435, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1807777065131002e-06, | |
| "logits/chosen": -0.447805255651474, | |
| "logits/rejected": -0.4743974208831787, | |
| "logps/chosen": -299.7063903808594, | |
| "logps/rejected": -316.0007629394531, | |
| "loss": 0.2131, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.022526143118739128, | |
| "rewards/margins": 0.008774956688284874, | |
| "rewards/rejected": 0.013751191087067127, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1414623905380012e-06, | |
| "logits/chosen": -0.4054747521877289, | |
| "logits/rejected": -0.3948088586330414, | |
| "logps/chosen": -269.71209716796875, | |
| "logps/rejected": -296.8516540527344, | |
| "loss": 0.1979, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.024300457909703255, | |
| "rewards/margins": 0.012143954634666443, | |
| "rewards/rejected": 0.012156503275036812, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.1026181914754388e-06, | |
| "logits/chosen": -0.44318923354148865, | |
| "logits/rejected": -0.4359446167945862, | |
| "logps/chosen": -267.4984436035156, | |
| "logps/rejected": -290.05694580078125, | |
| "loss": 0.2111, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.019330361858010292, | |
| "rewards/margins": 0.010160105302929878, | |
| "rewards/rejected": 0.00917025562375784, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.0642585798094136e-06, | |
| "logits/chosen": -0.40256720781326294, | |
| "logits/rejected": -0.4148158133029938, | |
| "logps/chosen": -269.02618408203125, | |
| "logps/rejected": -296.11041259765625, | |
| "loss": 0.2057, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.019753821194171906, | |
| "rewards/margins": 0.003918430767953396, | |
| "rewards/rejected": 0.015835389494895935, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.0263968579775522e-06, | |
| "logits/chosen": -0.43118053674697876, | |
| "logits/rejected": -0.4353235363960266, | |
| "logps/chosen": -219.4114227294922, | |
| "logps/rejected": -289.5980529785156, | |
| "loss": 0.1959, | |
| "rewards/accuracies": 0.34375, | |
| "rewards/chosen": 0.01978078857064247, | |
| "rewards/margins": 0.01855427958071232, | |
| "rewards/rejected": 0.0012265086406841874, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.89046155758058e-07, | |
| "logits/chosen": -0.40550509095191956, | |
| "logits/rejected": -0.34553685784339905, | |
| "logps/chosen": -281.02508544921875, | |
| "logps/rejected": -293.77801513671875, | |
| "loss": 0.2007, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.015590362250804901, | |
| "rewards/margins": 0.009258143603801727, | |
| "rewards/rejected": 0.0063322195783257484, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.52219425716534e-07, | |
| "logits/chosen": -0.4825696349143982, | |
| "logits/rejected": -0.46453267335891724, | |
| "logps/chosen": -277.46527099609375, | |
| "logps/rejected": -294.99041748046875, | |
| "loss": 0.2056, | |
| "rewards/accuracies": 0.28125, | |
| "rewards/chosen": 0.020664995536208153, | |
| "rewards/margins": 0.008111684583127499, | |
| "rewards/rejected": 0.01255331002175808, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.15929438714262e-07, | |
| "logits/chosen": -0.4495898187160492, | |
| "logits/rejected": -0.4320181906223297, | |
| "logps/chosen": -319.02703857421875, | |
| "logps/rejected": -341.0379943847656, | |
| "loss": 0.1993, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.020045015960931778, | |
| "rewards/margins": 0.009719369933009148, | |
| "rewards/rejected": 0.01032564602792263, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.801887794794911e-07, | |
| "logits/chosen": -0.40276139974594116, | |
| "logits/rejected": -0.43361392617225647, | |
| "logps/chosen": -240.11416625976562, | |
| "logps/rejected": -250.44827270507812, | |
| "loss": 0.2006, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.018377160653471947, | |
| "rewards/margins": 0.01084678340703249, | |
| "rewards/rejected": 0.007530376315116882, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.450098422432787e-07, | |
| "logits/chosen": -0.4492071270942688, | |
| "logits/rejected": -0.4732417166233063, | |
| "logps/chosen": -304.6590270996094, | |
| "logps/rejected": -320.5320739746094, | |
| "loss": 0.2072, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.01209027785807848, | |
| "rewards/margins": 0.009711054153740406, | |
| "rewards/rejected": 0.0023792237043380737, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.104048264413858e-07, | |
| "logits/chosen": -0.3988429605960846, | |
| "logits/rejected": -0.39577528834342957, | |
| "logps/chosen": -239.76278686523438, | |
| "logps/rejected": -274.7440490722656, | |
| "loss": 0.2002, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.01936698704957962, | |
| "rewards/margins": 0.015584215521812439, | |
| "rewards/rejected": 0.0037827729247510433, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.763857324837321e-07, | |
| "logits/chosen": -0.39769840240478516, | |
| "logits/rejected": -0.4869154095649719, | |
| "logps/chosen": -306.97052001953125, | |
| "logps/rejected": -341.172607421875, | |
| "loss": 0.2137, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.01544914674013853, | |
| "rewards/margins": 0.01086854375898838, | |
| "rewards/rejected": 0.004580603446811438, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.429643575928605e-07, | |
| "logits/chosen": -0.4462651312351227, | |
| "logits/rejected": -0.45454901456832886, | |
| "logps/chosen": -272.3761901855469, | |
| "logps/rejected": -300.822509765625, | |
| "loss": 0.2174, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.012530229985713959, | |
| "rewards/margins": 0.0049475752748548985, | |
| "rewards/rejected": 0.007582654245197773, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.101522917128709e-07, | |
| "logits/chosen": -0.40284043550491333, | |
| "logits/rejected": -0.43488210439682007, | |
| "logps/chosen": -252.0594024658203, | |
| "logps/rejected": -309.368408203125, | |
| "loss": 0.2068, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.022649750113487244, | |
| "rewards/margins": 0.019699372351169586, | |
| "rewards/rejected": 0.002950376598164439, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.779609134902312e-07, | |
| "logits/chosen": -0.45037612318992615, | |
| "logits/rejected": -0.4128038287162781, | |
| "logps/chosen": -235.04953002929688, | |
| "logps/rejected": -276.83453369140625, | |
| "loss": 0.2024, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.012530568055808544, | |
| "rewards/margins": 0.013141656294465065, | |
| "rewards/rejected": -0.0006110899848863482, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.464013863278629e-07, | |
| "logits/chosen": -0.41907772421836853, | |
| "logits/rejected": -0.43223705887794495, | |
| "logps/chosen": -335.57958984375, | |
| "logps/rejected": -339.4792785644531, | |
| "loss": 0.2068, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.008910799399018288, | |
| "rewards/margins": 0.006074207369238138, | |
| "rewards/rejected": 0.002836592961102724, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.154846545138696e-07, | |
| "logits/chosen": -0.4513935148715973, | |
| "logits/rejected": -0.4283982217311859, | |
| "logps/chosen": -293.88079833984375, | |
| "logps/rejected": -323.3922119140625, | |
| "loss": 0.2166, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.019896145910024643, | |
| "rewards/margins": 0.011627765372395515, | |
| "rewards/rejected": 0.008268382400274277, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.852214394262515e-07, | |
| "logits/chosen": -0.4485366940498352, | |
| "logits/rejected": -0.4402342736721039, | |
| "logps/chosen": -282.20526123046875, | |
| "logps/rejected": -319.2931213378906, | |
| "loss": 0.2056, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.013388392515480518, | |
| "rewards/margins": 0.012794624082744122, | |
| "rewards/rejected": 0.000593768258113414, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.556222358149191e-07, | |
| "logits/chosen": -0.46508869528770447, | |
| "logits/rejected": -0.44511333107948303, | |
| "logps/chosen": -306.74212646484375, | |
| "logps/rejected": -336.5767517089844, | |
| "loss": 0.1951, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.014937457628548145, | |
| "rewards/margins": 0.012279443442821503, | |
| "rewards/rejected": 0.0026580151170492172, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.266973081622992e-07, | |
| "logits/chosen": -0.4336974024772644, | |
| "logits/rejected": -0.4386768937110901, | |
| "logps/chosen": -276.2127990722656, | |
| "logps/rejected": -314.35894775390625, | |
| "loss": 0.2138, | |
| "rewards/accuracies": 0.34375, | |
| "rewards/chosen": 0.01599228009581566, | |
| "rewards/margins": 0.015695402398705482, | |
| "rewards/rejected": 0.00029687798814848065, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.984566871237942e-07, | |
| "logits/chosen": -0.42907437682151794, | |
| "logits/rejected": -0.45513230562210083, | |
| "logps/chosen": -255.0353546142578, | |
| "logps/rejected": -301.3641662597656, | |
| "loss": 0.195, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.011285373941063881, | |
| "rewards/margins": 0.015537778846919537, | |
| "rewards/rejected": -0.004252408631145954, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.709101660493251e-07, | |
| "logits/chosen": -0.4831358790397644, | |
| "logits/rejected": -0.4467880129814148, | |
| "logps/chosen": -278.4053955078125, | |
| "logps/rejected": -315.4570617675781, | |
| "loss": 0.1998, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.014852491207420826, | |
| "rewards/margins": 0.015064602717757225, | |
| "rewards/rejected": -0.00021211188868619502, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.440672975871743e-07, | |
| "logits/chosen": -0.46845632791519165, | |
| "logits/rejected": -0.43739065527915955, | |
| "logps/chosen": -276.8359375, | |
| "logps/rejected": -319.9267883300781, | |
| "loss": 0.2153, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.011290190741419792, | |
| "rewards/margins": 0.008315078914165497, | |
| "rewards/rejected": 0.002975111361593008, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1793739037129134e-07, | |
| "logits/chosen": -0.41762232780456543, | |
| "logits/rejected": -0.3986124098300934, | |
| "logps/chosen": -287.2295227050781, | |
| "logps/rejected": -293.7361755371094, | |
| "loss": 0.2257, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": 0.011349224485456944, | |
| "rewards/margins": 0.0031309097539633512, | |
| "rewards/rejected": 0.008218314498662949, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.9252950579322405e-07, | |
| "logits/chosen": -0.4659145772457123, | |
| "logits/rejected": -0.5180898904800415, | |
| "logps/chosen": -250.5138397216797, | |
| "logps/rejected": -294.9307861328125, | |
| "loss": 0.203, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.015681695193052292, | |
| "rewards/margins": 0.01762702874839306, | |
| "rewards/rejected": -0.0019453342538326979, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.6785245485978864e-07, | |
| "logits/chosen": -0.46653518080711365, | |
| "logits/rejected": -0.4424509108066559, | |
| "logps/chosen": -245.3058624267578, | |
| "logps/rejected": -291.9484558105469, | |
| "loss": 0.1848, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.020799651741981506, | |
| "rewards/margins": 0.01163767371326685, | |
| "rewards/rejected": 0.009161979891359806, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.43914795137566e-07, | |
| "logits/chosen": -0.5105794668197632, | |
| "logits/rejected": -0.46027788519859314, | |
| "logps/chosen": -267.0513610839844, | |
| "logps/rejected": -316.2474365234375, | |
| "loss": 0.2101, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.025547105818986893, | |
| "rewards/margins": 0.014689281582832336, | |
| "rewards/rejected": 0.010857820510864258, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.207248277852901e-07, | |
| "logits/chosen": -0.4583914279937744, | |
| "logits/rejected": -0.4020233154296875, | |
| "logps/chosen": -307.8725891113281, | |
| "logps/rejected": -330.9596252441406, | |
| "loss": 0.2184, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.005914419423788786, | |
| "rewards/margins": 0.0035868945997208357, | |
| "rewards/rejected": 0.0023275248240679502, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.9829059467515074e-07, | |
| "logits/chosen": -0.47686338424682617, | |
| "logits/rejected": -0.4638807773590088, | |
| "logps/chosen": -276.5890197753906, | |
| "logps/rejected": -311.7373352050781, | |
| "loss": 0.2044, | |
| "rewards/accuracies": 0.34375, | |
| "rewards/chosen": 0.009661246091127396, | |
| "rewards/margins": 0.008756840601563454, | |
| "rewards/rejected": 0.0009044073522090912, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.766198756040153e-07, | |
| "logits/chosen": -0.5123279094696045, | |
| "logits/rejected": -0.5023345351219177, | |
| "logps/chosen": -325.7563171386719, | |
| "logps/rejected": -341.7509765625, | |
| "loss": 0.2056, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 0.01731071248650551, | |
| "rewards/margins": 0.012032909318804741, | |
| "rewards/rejected": 0.005277802236378193, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.5572018559553155e-07, | |
| "logits/chosen": -0.42395853996276855, | |
| "logits/rejected": -0.40241509675979614, | |
| "logps/chosen": -260.11376953125, | |
| "logps/rejected": -274.5872802734375, | |
| "loss": 0.2063, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.007420125417411327, | |
| "rewards/margins": 0.003118096850812435, | |
| "rewards/rejected": 0.0043020290322601795, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3559877229404864e-07, | |
| "logits/chosen": -0.4700084626674652, | |
| "logits/rejected": -0.48075324296951294, | |
| "logps/chosen": -251.65200805664062, | |
| "logps/rejected": -273.92889404296875, | |
| "loss": 0.202, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.01748984307050705, | |
| "rewards/margins": 0.006852374877780676, | |
| "rewards/rejected": 0.010637464001774788, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.1626261345126576e-07, | |
| "logits/chosen": -0.4547249674797058, | |
| "logits/rejected": -0.49483567476272583, | |
| "logps/chosen": -257.3490905761719, | |
| "logps/rejected": -283.01068115234375, | |
| "loss": 0.2061, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.012025897391140461, | |
| "rewards/margins": 0.010915794409811497, | |
| "rewards/rejected": 0.0011101035634055734, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.9771841450646505e-07, | |
| "logits/chosen": -0.425775945186615, | |
| "logits/rejected": -0.39094358682632446, | |
| "logps/chosen": -268.5893859863281, | |
| "logps/rejected": -291.09295654296875, | |
| "loss": 0.2173, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.01688522659242153, | |
| "rewards/margins": 0.008994406089186668, | |
| "rewards/rejected": 0.007890822365880013, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.7997260626118758e-07, | |
| "logits/chosen": -0.4601113796234131, | |
| "logits/rejected": -0.5219780802726746, | |
| "logps/chosen": -286.64410400390625, | |
| "logps/rejected": -286.85223388671875, | |
| "loss": 0.2119, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.012106504291296005, | |
| "rewards/margins": 0.0008594643441028893, | |
| "rewards/rejected": 0.011247038841247559, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.6303134264914365e-07, | |
| "logits/chosen": -0.4205317497253418, | |
| "logits/rejected": -0.40168771147727966, | |
| "logps/chosen": -293.42626953125, | |
| "logps/rejected": -321.1906433105469, | |
| "loss": 0.2134, | |
| "rewards/accuracies": 0.34375, | |
| "rewards/chosen": 0.0118024330586195, | |
| "rewards/margins": 0.013113128952682018, | |
| "rewards/rejected": -0.0013106951955705881, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.469004986021355e-07, | |
| "logits/chosen": -0.4995250105857849, | |
| "logits/rejected": -0.4347379803657532, | |
| "logps/chosen": -279.1140441894531, | |
| "logps/rejected": -305.504638671875, | |
| "loss": 0.1933, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.014054737985134125, | |
| "rewards/margins": 0.010226726531982422, | |
| "rewards/rejected": 0.003828010056167841, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.315856680127367e-07, | |
| "logits/chosen": -0.42301422357559204, | |
| "logits/rejected": -0.4993847906589508, | |
| "logps/chosen": -293.3585510253906, | |
| "logps/rejected": -314.04095458984375, | |
| "loss": 0.2044, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.01272254716604948, | |
| "rewards/margins": 0.008608581498265266, | |
| "rewards/rejected": 0.004113964736461639, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1709216179442817e-07, | |
| "logits/chosen": -0.4361719489097595, | |
| "logits/rejected": -0.4560726284980774, | |
| "logps/chosen": -283.5252990722656, | |
| "logps/rejected": -320.2547607421875, | |
| "loss": 0.2074, | |
| "rewards/accuracies": 0.3187499940395355, | |
| "rewards/chosen": 0.008978287689387798, | |
| "rewards/margins": 0.010601527988910675, | |
| "rewards/rejected": -0.001623238087631762, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0342500603986421e-07, | |
| "logits/chosen": -0.409717857837677, | |
| "logits/rejected": -0.43887776136398315, | |
| "logps/chosen": -274.9674072265625, | |
| "logps/rejected": -320.1651916503906, | |
| "loss": 0.2081, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.014241678640246391, | |
| "rewards/margins": 0.013103106990456581, | |
| "rewards/rejected": 0.0011385709512978792, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.058894027791643e-08, | |
| "logits/chosen": -0.49651774764060974, | |
| "logits/rejected": -0.4656089246273041, | |
| "logps/chosen": -239.1632843017578, | |
| "logps/rejected": -252.4584503173828, | |
| "loss": 0.2061, | |
| "rewards/accuracies": 0.23749999701976776, | |
| "rewards/chosen": 0.015118755400180817, | |
| "rewards/margins": 0.006344223860651255, | |
| "rewards/rejected": 0.0087745301425457, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.858841583008592e-08, | |
| "logits/chosen": -0.42512744665145874, | |
| "logits/rejected": -0.46020737290382385, | |
| "logps/chosen": -251.6322784423828, | |
| "logps/rejected": -235.462158203125, | |
| "loss": 0.1982, | |
| "rewards/accuracies": 0.23125000298023224, | |
| "rewards/chosen": 0.009195510298013687, | |
| "rewards/margins": -0.0015735321212559938, | |
| "rewards/rejected": 0.010769042186439037, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.742759426686313e-08, | |
| "logits/chosen": -0.3904110789299011, | |
| "logits/rejected": -0.4210638999938965, | |
| "logps/chosen": -246.4021759033203, | |
| "logps/rejected": -279.08160400390625, | |
| "loss": 0.1995, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.020919088274240494, | |
| "rewards/margins": 0.016136836260557175, | |
| "rewards/rejected": 0.004782252479344606, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.7110345964571104e-08, | |
| "logits/chosen": -0.5139604806900024, | |
| "logits/rejected": -0.4221692681312561, | |
| "logps/chosen": -285.17803955078125, | |
| "logps/rejected": -315.17071533203125, | |
| "loss": 0.1886, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": 0.01321653462946415, | |
| "rewards/margins": 0.010121813975274563, | |
| "rewards/rejected": 0.003094720421358943, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.764024876318357e-08, | |
| "logits/chosen": -0.48494213819503784, | |
| "logits/rejected": -0.4467412531375885, | |
| "logps/chosen": -319.4140319824219, | |
| "logps/rejected": -333.9644775390625, | |
| "loss": 0.2053, | |
| "rewards/accuracies": 0.29374998807907104, | |
| "rewards/chosen": 0.00665741041302681, | |
| "rewards/margins": 0.0072568124160170555, | |
| "rewards/rejected": -0.0005994020029902458, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.902058672559633e-08, | |
| "logits/chosen": -0.4710802435874939, | |
| "logits/rejected": -0.44002991914749146, | |
| "logps/chosen": -266.371337890625, | |
| "logps/rejected": -304.83526611328125, | |
| "loss": 0.2004, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.013240337371826172, | |
| "rewards/margins": 0.00819102581590414, | |
| "rewards/rejected": 0.005049312487244606, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.125434899876933e-08, | |
| "logits/chosen": -0.46560755372047424, | |
| "logits/rejected": -0.4170166552066803, | |
| "logps/chosen": -273.154052734375, | |
| "logps/rejected": -292.3708801269531, | |
| "loss": 0.2181, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.012563738040626049, | |
| "rewards/margins": 0.008571961894631386, | |
| "rewards/rejected": 0.003991776145994663, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.4344228777145873e-08, | |
| "logits/chosen": -0.4589017927646637, | |
| "logits/rejected": -0.46233731508255005, | |
| "logps/chosen": -330.4123840332031, | |
| "logps/rejected": -345.12005615234375, | |
| "loss": 0.2102, | |
| "rewards/accuracies": 0.35624998807907104, | |
| "rewards/chosen": 0.016749653965234756, | |
| "rewards/margins": 0.009503444656729698, | |
| "rewards/rejected": 0.0072462051175534725, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.829262236869772e-08, | |
| "logits/chosen": -0.4653870463371277, | |
| "logits/rejected": -0.4221324324607849, | |
| "logps/chosen": -295.13922119140625, | |
| "logps/rejected": -343.47979736328125, | |
| "loss": 0.2124, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.014039866626262665, | |
| "rewards/margins": 0.011893939226865768, | |
| "rewards/rejected": 0.0021459278650581837, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.3101628363929586e-08, | |
| "logits/chosen": -0.4033223092556, | |
| "logits/rejected": -0.4376245439052582, | |
| "logps/chosen": -233.1123504638672, | |
| "logps/rejected": -269.4231262207031, | |
| "loss": 0.2073, | |
| "rewards/accuracies": 0.26875001192092896, | |
| "rewards/chosen": 0.012142864055931568, | |
| "rewards/margins": 0.008770270273089409, | |
| "rewards/rejected": 0.003372594015672803, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.773046908123195e-09, | |
| "logits/chosen": -0.41063612699508667, | |
| "logits/rejected": -0.4108348488807678, | |
| "logps/chosen": -305.08514404296875, | |
| "logps/rejected": -316.0191955566406, | |
| "loss": 0.1879, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.011063008569180965, | |
| "rewards/margins": 0.007611425127834082, | |
| "rewards/rejected": 0.0034515834413468838, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.308379077080817e-09, | |
| "logits/chosen": -0.4463173449039459, | |
| "logits/rejected": -0.427543967962265, | |
| "logps/chosen": -271.1603088378906, | |
| "logps/rejected": -276.7561950683594, | |
| "loss": 0.2193, | |
| "rewards/accuracies": 0.33125001192092896, | |
| "rewards/chosen": 0.013324853964149952, | |
| "rewards/margins": 0.00910397619009018, | |
| "rewards/rejected": 0.00422087823972106, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.7088263565760996e-09, | |
| "logits/chosen": -0.4519821107387543, | |
| "logits/rejected": -0.45307645201683044, | |
| "logps/chosen": -291.6419982910156, | |
| "logps/rejected": -338.31207275390625, | |
| "loss": 0.1975, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.009216098114848137, | |
| "rewards/margins": 0.009984901174902916, | |
| "rewards/rejected": -0.00076880311826244, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.752902257023633e-10, | |
| "logits/chosen": -0.4056398272514343, | |
| "logits/rejected": -0.39859509468078613, | |
| "logps/chosen": -206.9446258544922, | |
| "logps/rejected": -251.4867401123047, | |
| "loss": 0.1943, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.016167674213647842, | |
| "rewards/margins": 0.014986155554652214, | |
| "rewards/rejected": 0.0011815171455964446, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.083718442532189e-10, | |
| "logits/chosen": -0.47832536697387695, | |
| "logits/rejected": -0.42318111658096313, | |
| "logps/chosen": -300.47515869140625, | |
| "logps/rejected": -325.100830078125, | |
| "loss": 0.217, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.00793883390724659, | |
| "rewards/margins": 0.004473513457924128, | |
| "rewards/rejected": 0.0034653202164918184, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1875, | |
| "total_flos": 0.0, | |
| "train_loss": 0.20694573613802592, | |
| "train_runtime": 16535.0328, | |
| "train_samples_per_second": 1.814, | |
| "train_steps_per_second": 0.113 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1875, | |
| "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 | |
| } | |