| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9992156862745099, | |
| "eval_steps": 500, | |
| "global_step": 637, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.8125e-08, | |
| "logits/chosen": 0.0918828696012497, | |
| "logits/rejected": 0.26330235600471497, | |
| "logps/chosen": -458.63958740234375, | |
| "logps/rejected": -388.8646240234375, | |
| "loss": 0.4001, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.8125e-07, | |
| "logits/chosen": 0.07364361733198166, | |
| "logits/rejected": 0.15213775634765625, | |
| "logps/chosen": -318.695556640625, | |
| "logps/rejected": -323.5196533203125, | |
| "loss": 0.3624, | |
| "rewards/accuracies": 0.4166666567325592, | |
| "rewards/chosen": 5.298387259244919e-05, | |
| "rewards/margins": -6.156884046504274e-05, | |
| "rewards/rejected": 0.00011455271305749193, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.5625e-06, | |
| "logits/chosen": 0.10303022712469101, | |
| "logits/rejected": 0.1733068823814392, | |
| "logps/chosen": -370.1763916015625, | |
| "logps/rejected": -344.3110046386719, | |
| "loss": 0.3763, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -9.297926590079442e-05, | |
| "rewards/margins": -0.00011266962246736512, | |
| "rewards/rejected": 1.9690334738697857e-05, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.3437500000000002e-06, | |
| "logits/chosen": 0.10951884090900421, | |
| "logits/rejected": 0.171161487698555, | |
| "logps/chosen": -374.95770263671875, | |
| "logps/rejected": -342.6380615234375, | |
| "loss": 0.3916, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": -2.8160246074548922e-05, | |
| "rewards/margins": 1.7810778444982134e-05, | |
| "rewards/rejected": -4.5971013605594635e-05, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.125e-06, | |
| "logits/chosen": 0.1174466460943222, | |
| "logits/rejected": 0.18849265575408936, | |
| "logps/chosen": -321.7309875488281, | |
| "logps/rejected": -292.35772705078125, | |
| "loss": 0.363, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.00019040452025365084, | |
| "rewards/margins": 0.00019320171850267798, | |
| "rewards/rejected": -2.797205524984747e-06, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.90625e-06, | |
| "logits/chosen": 0.19317670166492462, | |
| "logits/rejected": 0.24143847823143005, | |
| "logps/chosen": -354.2140808105469, | |
| "logps/rejected": -362.29583740234375, | |
| "loss": 0.3632, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.00048213909030891955, | |
| "rewards/margins": 0.00036385521525517106, | |
| "rewards/rejected": 0.00011828396236523986, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.6875000000000004e-06, | |
| "logits/chosen": 0.18060320615768433, | |
| "logits/rejected": 0.22344954311847687, | |
| "logps/chosen": -357.6122131347656, | |
| "logps/rejected": -364.1398620605469, | |
| "loss": 0.3805, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.0008422869141213596, | |
| "rewards/margins": 0.001053076470270753, | |
| "rewards/rejected": -0.00021078942518215626, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.998647417232375e-06, | |
| "logits/chosen": 0.08928748220205307, | |
| "logits/rejected": 0.15246489644050598, | |
| "logps/chosen": -318.8173522949219, | |
| "logps/rejected": -304.9363708496094, | |
| "loss": 0.356, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.0005034763598814607, | |
| "rewards/margins": 0.0022861119359731674, | |
| "rewards/rejected": -0.0017826355760917068, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.990386933279973e-06, | |
| "logits/chosen": 0.12598779797554016, | |
| "logits/rejected": 0.20218820869922638, | |
| "logps/chosen": -341.4804382324219, | |
| "logps/rejected": -344.4537048339844, | |
| "loss": 0.3596, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.00020067775039933622, | |
| "rewards/margins": 0.0030791102908551693, | |
| "rewards/rejected": -0.003279787953943014, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.974642195009681e-06, | |
| "logits/chosen": 0.17095333337783813, | |
| "logits/rejected": 0.2281300127506256, | |
| "logps/chosen": -314.36920166015625, | |
| "logps/rejected": -312.678466796875, | |
| "loss": 0.369, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.002451724372804165, | |
| "rewards/margins": 0.004382356069982052, | |
| "rewards/rejected": -0.006834079977124929, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.951460519416228e-06, | |
| "logits/chosen": 0.14624568819999695, | |
| "logits/rejected": 0.19171926379203796, | |
| "logps/chosen": -362.2412414550781, | |
| "logps/rejected": -357.18170166015625, | |
| "loss": 0.3748, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.004997233394533396, | |
| "rewards/margins": 0.008778322488069534, | |
| "rewards/rejected": -0.013775557279586792, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.920911573406925e-06, | |
| "logits/chosen": 0.0949273556470871, | |
| "logits/rejected": 0.15069428086280823, | |
| "logps/chosen": -337.5642395019531, | |
| "logps/rejected": -346.2424621582031, | |
| "loss": 0.3603, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.006651742849498987, | |
| "rewards/margins": 0.01205554511398077, | |
| "rewards/rejected": -0.018707286566495895, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.883087164434672e-06, | |
| "logits/chosen": 0.08049245178699493, | |
| "logits/rejected": 0.17421886324882507, | |
| "logps/chosen": -359.105224609375, | |
| "logps/rejected": -364.32904052734375, | |
| "loss": 0.358, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.013828782364726067, | |
| "rewards/margins": 0.016524765640497208, | |
| "rewards/rejected": -0.030353549867868423, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8381009645929044e-06, | |
| "logits/chosen": 0.117076575756073, | |
| "logits/rejected": 0.1626085489988327, | |
| "logps/chosen": -359.46728515625, | |
| "logps/rejected": -418.33868408203125, | |
| "loss": 0.3424, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.03268348425626755, | |
| "rewards/margins": 0.038762401789426804, | |
| "rewards/rejected": -0.07144588232040405, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.786088169001671e-06, | |
| "logits/chosen": 0.12248921394348145, | |
| "logits/rejected": 0.203883558511734, | |
| "logps/chosen": -427.392333984375, | |
| "logps/rejected": -504.49908447265625, | |
| "loss": 0.3317, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.11223888397216797, | |
| "rewards/margins": 0.05915551260113716, | |
| "rewards/rejected": -0.17139439284801483, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.727205089511466e-06, | |
| "logits/chosen": 0.08418472111225128, | |
| "logits/rejected": 0.21436505019664764, | |
| "logps/chosen": -579.9471435546875, | |
| "logps/rejected": -638.8032836914062, | |
| "loss": 0.3334, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.20061063766479492, | |
| "rewards/margins": 0.07825453579425812, | |
| "rewards/rejected": -0.27886515855789185, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.661628684945851e-06, | |
| "logits/chosen": 0.13719815015792847, | |
| "logits/rejected": 0.22896163165569305, | |
| "logps/chosen": -483.87744140625, | |
| "logps/rejected": -555.54931640625, | |
| "loss": 0.3506, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.1468028724193573, | |
| "rewards/margins": 0.06705925613641739, | |
| "rewards/rejected": -0.21386213600635529, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5895560292946e-06, | |
| "logits/chosen": 0.15668627619743347, | |
| "logits/rejected": 0.24040012061595917, | |
| "logps/chosen": -489.44195556640625, | |
| "logps/rejected": -503.8895568847656, | |
| "loss": 0.3035, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.11202345043420792, | |
| "rewards/margins": 0.060164742171764374, | |
| "rewards/rejected": -0.1721881926059723, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.511203719455588e-06, | |
| "logits/chosen": 0.1728764772415161, | |
| "logits/rejected": 0.21227356791496277, | |
| "logps/chosen": -434.38555908203125, | |
| "logps/rejected": -474.145263671875, | |
| "loss": 0.3136, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.10123687982559204, | |
| "rewards/margins": 0.06594525277614594, | |
| "rewards/rejected": -0.16718213260173798, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.426807224305315e-06, | |
| "logits/chosen": 0.1319476068019867, | |
| "logits/rejected": 0.2203722447156906, | |
| "logps/chosen": -476.09088134765625, | |
| "logps/rejected": -508.6603088378906, | |
| "loss": 0.3452, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.1248413696885109, | |
| "rewards/margins": 0.052263855934143066, | |
| "rewards/rejected": -0.17710521817207336, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.336620177054269e-06, | |
| "logits/chosen": 0.1470222920179367, | |
| "logits/rejected": 0.2037900686264038, | |
| "logps/chosen": -433.387939453125, | |
| "logps/rejected": -526.0863037109375, | |
| "loss": 0.3211, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.10074210166931152, | |
| "rewards/margins": 0.07372565567493439, | |
| "rewards/rejected": -0.1744677573442459, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.240913613013785e-06, | |
| "logits/chosen": 0.13598838448524475, | |
| "logits/rejected": 0.24791710078716278, | |
| "logps/chosen": -504.983642578125, | |
| "logps/rejected": -547.6514282226562, | |
| "loss": 0.3223, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.13141490519046783, | |
| "rewards/margins": 0.06105287745594978, | |
| "rewards/rejected": -0.19246776401996613, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.139975155065109e-06, | |
| "logits/chosen": 0.10929516702890396, | |
| "logits/rejected": 0.2205590009689331, | |
| "logps/chosen": -466.38543701171875, | |
| "logps/rejected": -554.4016723632812, | |
| "loss": 0.3052, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.1270093470811844, | |
| "rewards/margins": 0.08514191210269928, | |
| "rewards/rejected": -0.21215125918388367, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.034108149278544e-06, | |
| "logits/chosen": 0.12504920363426208, | |
| "logits/rejected": 0.20017783343791962, | |
| "logps/chosen": -500.18218994140625, | |
| "logps/rejected": -552.7307739257812, | |
| "loss": 0.2987, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.13430464267730713, | |
| "rewards/margins": 0.07368321716785431, | |
| "rewards/rejected": -0.20798785984516144, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.923630753280358e-06, | |
| "logits/chosen": 0.14272692799568176, | |
| "logits/rejected": 0.20935645699501038, | |
| "logps/chosen": -516.2708740234375, | |
| "logps/rejected": -599.9803466796875, | |
| "loss": 0.2931, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.138849675655365, | |
| "rewards/margins": 0.1058221235871315, | |
| "rewards/rejected": -0.2446717917919159, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.80887498010715e-06, | |
| "logits/chosen": 0.20914337038993835, | |
| "logits/rejected": 0.22870346903800964, | |
| "logps/chosen": -463.67340087890625, | |
| "logps/rejected": -534.5090942382812, | |
| "loss": 0.3168, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.1403440535068512, | |
| "rewards/margins": 0.06889508664608002, | |
| "rewards/rejected": -0.20923912525177002, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.690185700421145e-06, | |
| "logits/chosen": 0.17314383387565613, | |
| "logits/rejected": 0.2541898787021637, | |
| "logps/chosen": -500.0283203125, | |
| "logps/rejected": -567.5086059570312, | |
| "loss": 0.3046, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.12471531331539154, | |
| "rewards/margins": 0.11497502028942108, | |
| "rewards/rejected": -0.23969034850597382, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.567919606085004e-06, | |
| "logits/chosen": 0.18270955979824066, | |
| "logits/rejected": 0.24541044235229492, | |
| "logps/chosen": -524.0221557617188, | |
| "logps/rejected": -606.4444580078125, | |
| "loss": 0.3055, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.13682256639003754, | |
| "rewards/margins": 0.10733524709939957, | |
| "rewards/rejected": -0.2441578209400177, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.442444138210883e-06, | |
| "logits/chosen": 0.19482332468032837, | |
| "logits/rejected": 0.29221558570861816, | |
| "logps/chosen": -545.2671508789062, | |
| "logps/rejected": -661.770263671875, | |
| "loss": 0.2962, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.15382714569568634, | |
| "rewards/margins": 0.1256464123725891, | |
| "rewards/rejected": -0.27947354316711426, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.314136382905234e-06, | |
| "logits/chosen": 0.2596381902694702, | |
| "logits/rejected": 0.29522743821144104, | |
| "logps/chosen": -531.8798217773438, | |
| "logps/rejected": -688.622802734375, | |
| "loss": 0.2707, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.17996947467327118, | |
| "rewards/margins": 0.1445850431919098, | |
| "rewards/rejected": -0.3245545029640198, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.1833819380279028e-06, | |
| "logits/chosen": 0.30931779742240906, | |
| "logits/rejected": 0.2789193391799927, | |
| "logps/chosen": -576.7262573242188, | |
| "logps/rejected": -676.7431640625, | |
| "loss": 0.3178, | |
| "rewards/accuracies": 0.6187499761581421, | |
| "rewards/chosen": -0.2274230420589447, | |
| "rewards/margins": 0.11135544627904892, | |
| "rewards/rejected": -0.3387784957885742, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.050573754371228e-06, | |
| "logits/chosen": 0.20585477352142334, | |
| "logits/rejected": 0.25459426641464233, | |
| "logps/chosen": -536.5878295898438, | |
| "logps/rejected": -603.1542358398438, | |
| "loss": 0.3075, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.21307428181171417, | |
| "rewards/margins": 0.07962417602539062, | |
| "rewards/rejected": -0.2926984429359436, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.916110954741667e-06, | |
| "logits/chosen": 0.19497337937355042, | |
| "logits/rejected": 0.39090901613235474, | |
| "logps/chosen": -572.049560546875, | |
| "logps/rejected": -697.1812133789062, | |
| "loss": 0.2871, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.2017597258090973, | |
| "rewards/margins": 0.134249746799469, | |
| "rewards/rejected": -0.3360094726085663, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7803976344929497e-06, | |
| "logits/chosen": 0.16042862832546234, | |
| "logits/rejected": 0.262906551361084, | |
| "logps/chosen": -574.7508544921875, | |
| "logps/rejected": -686.5535888671875, | |
| "loss": 0.3012, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.20231370627880096, | |
| "rewards/margins": 0.13932286202907562, | |
| "rewards/rejected": -0.3416365683078766, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.6438416471154277e-06, | |
| "logits/chosen": 0.23894529044628143, | |
| "logits/rejected": 0.24849145114421844, | |
| "logps/chosen": -529.4860229492188, | |
| "logps/rejected": -650.188232421875, | |
| "loss": 0.2966, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.20321831107139587, | |
| "rewards/margins": 0.11060591042041779, | |
| "rewards/rejected": -0.31382423639297485, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.5068533785312673e-06, | |
| "logits/chosen": 0.16733770072460175, | |
| "logits/rejected": 0.24236424267292023, | |
| "logps/chosen": -526.2505493164062, | |
| "logps/rejected": -613.4932861328125, | |
| "loss": 0.2909, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.2088821828365326, | |
| "rewards/margins": 0.10516791045665741, | |
| "rewards/rejected": -0.3140501081943512, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.369844513779026e-06, | |
| "logits/chosen": 0.16742342710494995, | |
| "logits/rejected": 0.2210749387741089, | |
| "logps/chosen": -558.5418701171875, | |
| "logps/rejected": -629.0260620117188, | |
| "loss": 0.3169, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.20471343398094177, | |
| "rewards/margins": 0.09914077073335648, | |
| "rewards/rejected": -0.30385422706604004, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2332267997940514e-06, | |
| "logits/chosen": 0.2225278615951538, | |
| "logits/rejected": 0.25043898820877075, | |
| "logps/chosen": -623.16748046875, | |
| "logps/rejected": -674.67724609375, | |
| "loss": 0.3545, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.24807922542095184, | |
| "rewards/margins": 0.08578108996152878, | |
| "rewards/rejected": -0.3338603377342224, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.097410808002869e-06, | |
| "logits/chosen": 0.17123399674892426, | |
| "logits/rejected": 0.24292032420635223, | |
| "logps/chosen": -598.9353637695312, | |
| "logps/rejected": -703.7257080078125, | |
| "loss": 0.2788, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.20804348587989807, | |
| "rewards/margins": 0.1404215544462204, | |
| "rewards/rejected": -0.3484650254249573, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.962804700450265e-06, | |
| "logits/chosen": 0.13856971263885498, | |
| "logits/rejected": 0.204025536775589, | |
| "logps/chosen": -580.52197265625, | |
| "logps/rejected": -670.6529541015625, | |
| "loss": 0.2891, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.22241008281707764, | |
| "rewards/margins": 0.10534980148077011, | |
| "rewards/rejected": -0.32775989174842834, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8298130031671974e-06, | |
| "logits/chosen": 0.15039893984794617, | |
| "logits/rejected": 0.18419519066810608, | |
| "logps/chosen": -529.6956787109375, | |
| "logps/rejected": -624.8928833007812, | |
| "loss": 0.2796, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.1940956562757492, | |
| "rewards/margins": 0.1203426718711853, | |
| "rewards/rejected": -0.3144383430480957, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.6988353904658495e-06, | |
| "logits/chosen": 0.17148879170417786, | |
| "logits/rejected": 0.24635398387908936, | |
| "logps/chosen": -560.6425170898438, | |
| "logps/rejected": -683.8626098632812, | |
| "loss": 0.2732, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.2084239274263382, | |
| "rewards/margins": 0.1307951807975769, | |
| "rewards/rejected": -0.33921903371810913, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5702654838153641e-06, | |
| "logits/chosen": 0.13374552130699158, | |
| "logits/rejected": 0.21898336708545685, | |
| "logps/chosen": -533.1779174804688, | |
| "logps/rejected": -647.463134765625, | |
| "loss": 0.2948, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.1993607133626938, | |
| "rewards/margins": 0.10736644268035889, | |
| "rewards/rejected": -0.30672717094421387, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4444896689079142e-06, | |
| "logits/chosen": 0.2023026943206787, | |
| "logits/rejected": 0.2337256669998169, | |
| "logps/chosen": -578.2020874023438, | |
| "logps/rejected": -652.1826171875, | |
| "loss": 0.3217, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.21291308104991913, | |
| "rewards/margins": 0.093999482691288, | |
| "rewards/rejected": -0.30691254138946533, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3218859344701634e-06, | |
| "logits/chosen": 0.10349559783935547, | |
| "logits/rejected": 0.16690309345722198, | |
| "logps/chosen": -569.8800048828125, | |
| "logps/rejected": -680.734619140625, | |
| "loss": 0.2608, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.20618288218975067, | |
| "rewards/margins": 0.1448386162519455, | |
| "rewards/rejected": -0.3510214686393738, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.2028227363097583e-06, | |
| "logits/chosen": 0.13843318819999695, | |
| "logits/rejected": 0.18968960642814636, | |
| "logps/chosen": -550.9771728515625, | |
| "logps/rejected": -633.4754638671875, | |
| "loss": 0.3117, | |
| "rewards/accuracies": 0.6187499761581421, | |
| "rewards/chosen": -0.21659895777702332, | |
| "rewards/margins": 0.11474967002868652, | |
| "rewards/rejected": -0.33134859800338745, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.0876578900107053e-06, | |
| "logits/chosen": 0.17231318354606628, | |
| "logits/rejected": 0.17545001208782196, | |
| "logps/chosen": -558.6343994140625, | |
| "logps/rejected": -641.2686767578125, | |
| "loss": 0.2869, | |
| "rewards/accuracies": 0.606249988079071, | |
| "rewards/chosen": -0.21609997749328613, | |
| "rewards/margins": 0.10281254351139069, | |
| "rewards/rejected": -0.3189125061035156, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.767374956053584e-07, | |
| "logits/chosen": 0.20297455787658691, | |
| "logits/rejected": 0.27692240476608276, | |
| "logps/chosen": -603.3773803710938, | |
| "logps/rejected": -657.2963256835938, | |
| "loss": 0.3629, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -0.24698205292224884, | |
| "rewards/margins": 0.058980755507946014, | |
| "rewards/rejected": -0.30596280097961426, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 8.703948974546592e-07, | |
| "logits/chosen": 0.17270350456237793, | |
| "logits/rejected": 0.20394925773143768, | |
| "logps/chosen": -522.0367431640625, | |
| "logps/rejected": -605.554443359375, | |
| "loss": 0.3139, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": -0.20573575794696808, | |
| "rewards/margins": 0.10728572309017181, | |
| "rewards/rejected": -0.3130214810371399, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.689496824624526e-07, | |
| "logits/chosen": 0.13458377122879028, | |
| "logits/rejected": 0.19917598366737366, | |
| "logps/chosen": -544.7473754882812, | |
| "logps/rejected": -663.064208984375, | |
| "loss": 0.2822, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.21486207842826843, | |
| "rewards/margins": 0.11823858320713043, | |
| "rewards/rejected": -0.3331006169319153, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 6.7270671963451e-07, | |
| "logits/chosen": 0.11218737065792084, | |
| "logits/rejected": 0.21531948447227478, | |
| "logps/chosen": -529.8823852539062, | |
| "logps/rejected": -646.2406005859375, | |
| "loss": 0.2889, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.202474907040596, | |
| "rewards/margins": 0.11822378635406494, | |
| "rewards/rejected": -0.32069867849349976, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.819552438686238e-07, | |
| "logits/chosen": 0.14948822557926178, | |
| "logits/rejected": 0.17894427478313446, | |
| "logps/chosen": -545.9841918945312, | |
| "logps/rejected": -637.9070434570312, | |
| "loss": 0.3393, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.2308504283428192, | |
| "rewards/margins": 0.08178289234638214, | |
| "rewards/rejected": -0.31263333559036255, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.969679867292276e-07, | |
| "logits/chosen": 0.15977367758750916, | |
| "logits/rejected": 0.14595797657966614, | |
| "logps/chosen": -558.384521484375, | |
| "logps/rejected": -640.0645751953125, | |
| "loss": 0.3019, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.23137719929218292, | |
| "rewards/margins": 0.09733609855175018, | |
| "rewards/rejected": -0.3287132680416107, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1800035681877765e-07, | |
| "logits/chosen": 0.1737111657857895, | |
| "logits/rejected": 0.2044694423675537, | |
| "logps/chosen": -557.0775146484375, | |
| "logps/rejected": -703.0406494140625, | |
| "loss": 0.2799, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.21399672329425812, | |
| "rewards/margins": 0.1301182061433792, | |
| "rewards/rejected": -0.34411489963531494, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.4528967220911287e-07, | |
| "logits/chosen": 0.11454256623983383, | |
| "logits/rejected": 0.22100505232810974, | |
| "logps/chosen": -569.0726318359375, | |
| "logps/rejected": -655.3294677734375, | |
| "loss": 0.3021, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.21696233749389648, | |
| "rewards/margins": 0.10485026985406876, | |
| "rewards/rejected": -0.32181257009506226, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.7905444723949765e-07, | |
| "logits/chosen": 0.08618897199630737, | |
| "logits/rejected": 0.19658343493938446, | |
| "logps/chosen": -588.386962890625, | |
| "logps/rejected": -682.5491333007812, | |
| "loss": 0.2921, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.22489456832408905, | |
| "rewards/margins": 0.09568282216787338, | |
| "rewards/rejected": -0.3205774426460266, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.1949373582475065e-07, | |
| "logits/chosen": 0.17197827994823456, | |
| "logits/rejected": 0.24235980212688446, | |
| "logps/chosen": -550.1292724609375, | |
| "logps/rejected": -628.2039794921875, | |
| "loss": 0.3376, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.22096483409404755, | |
| "rewards/margins": 0.08723219484090805, | |
| "rewards/rejected": -0.3081970512866974, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.667865332469379e-07, | |
| "logits/chosen": 0.1541893035173416, | |
| "logits/rejected": 0.2220904380083084, | |
| "logps/chosen": -562.9713745117188, | |
| "logps/rejected": -657.7110595703125, | |
| "loss": 0.2415, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.1808656007051468, | |
| "rewards/margins": 0.13864107429981232, | |
| "rewards/rejected": -0.3195067048072815, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.2109123822844653e-07, | |
| "logits/chosen": 0.16481325030326843, | |
| "logits/rejected": 0.20911450684070587, | |
| "logps/chosen": -529.6812744140625, | |
| "logps/rejected": -672.5157470703125, | |
| "loss": 0.3085, | |
| "rewards/accuracies": 0.6312500238418579, | |
| "rewards/chosen": -0.21774494647979736, | |
| "rewards/margins": 0.11994588375091553, | |
| "rewards/rejected": -0.3376908600330353, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 8.254517690300946e-08, | |
| "logits/chosen": 0.12228590250015259, | |
| "logits/rejected": 0.19594553112983704, | |
| "logps/chosen": -565.060546875, | |
| "logps/rejected": -686.70947265625, | |
| "loss": 0.2775, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.20458519458770752, | |
| "rewards/margins": 0.12763497233390808, | |
| "rewards/rejected": -0.3322201371192932, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.126419011529993e-08, | |
| "logits/chosen": 0.1007891520857811, | |
| "logits/rejected": 0.23624570667743683, | |
| "logps/chosen": -578.2725830078125, | |
| "logps/rejected": -646.430908203125, | |
| "loss": 0.2885, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.20723719894886017, | |
| "rewards/margins": 0.1008644700050354, | |
| "rewards/rejected": -0.3081016540527344, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.734228528934679e-08, | |
| "logits/chosen": 0.1077275276184082, | |
| "logits/rejected": 0.18885311484336853, | |
| "logps/chosen": -591.439208984375, | |
| "logps/rejected": -687.6322021484375, | |
| "loss": 0.2913, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.23343046009540558, | |
| "rewards/margins": 0.11967509984970093, | |
| "rewards/rejected": -0.3531055748462677, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.0851353912008644e-08, | |
| "logits/chosen": 0.1404169648885727, | |
| "logits/rejected": 0.17623500525951385, | |
| "logps/chosen": -562.1318969726562, | |
| "logps/rejected": -675.3463134765625, | |
| "loss": 0.3244, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": -0.22516414523124695, | |
| "rewards/margins": 0.1208973079919815, | |
| "rewards/rejected": -0.3460614085197449, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.8409554805329243e-09, | |
| "logits/chosen": 0.18373514711856842, | |
| "logits/rejected": 0.21246370673179626, | |
| "logps/chosen": -578.3257446289062, | |
| "logps/rejected": -651.350830078125, | |
| "loss": 0.3169, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.22300894558429718, | |
| "rewards/margins": 0.09293990582227707, | |
| "rewards/rejected": -0.31594881415367126, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 637, | |
| "total_flos": 0.0, | |
| "train_loss": 0.31719773806619866, | |
| "train_runtime": 5988.5761, | |
| "train_samples_per_second": 1.703, | |
| "train_steps_per_second": 0.106 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 637, | |
| "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 | |
| } | |