| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.999508116084604, | |
| "eval_steps": 200, | |
| "global_step": 1016, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.901960784313726e-08, | |
| "logits/chosen": -2.0748214721679688, | |
| "logits/rejected": -2.1528916358947754, | |
| "logps/chosen": -78.83467102050781, | |
| "logps/rejected": -90.04127502441406, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.901960784313725e-07, | |
| "logits/chosen": -2.167158603668213, | |
| "logits/rejected": -2.065119743347168, | |
| "logps/chosen": -101.462646484375, | |
| "logps/rejected": -67.91659545898438, | |
| "loss": 0.6937, | |
| "rewards/accuracies": 0.3611111044883728, | |
| "rewards/chosen": -0.0003858842537738383, | |
| "rewards/margins": -0.001147754956036806, | |
| "rewards/rejected": 0.0007618705276399851, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 9.80392156862745e-07, | |
| "logits/chosen": -2.139626979827881, | |
| "logits/rejected": -2.0195488929748535, | |
| "logps/chosen": -109.9924545288086, | |
| "logps/rejected": -84.65155029296875, | |
| "loss": 0.695, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.002081229817122221, | |
| "rewards/margins": -0.0036208624951541424, | |
| "rewards/rejected": 0.0015396325616165996, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.4705882352941177e-06, | |
| "logits/chosen": -2.2819817066192627, | |
| "logits/rejected": -2.181386947631836, | |
| "logps/chosen": -100.8692398071289, | |
| "logps/rejected": -84.57305908203125, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.0019000970060005784, | |
| "rewards/margins": 0.00016770660295151174, | |
| "rewards/rejected": 0.0017323906067758799, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.96078431372549e-06, | |
| "logits/chosen": -2.274721622467041, | |
| "logits/rejected": -2.156294584274292, | |
| "logps/chosen": -105.41255950927734, | |
| "logps/rejected": -83.99583435058594, | |
| "loss": 0.6929, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.001408256939612329, | |
| "rewards/margins": 0.0004250857455190271, | |
| "rewards/rejected": 0.0009831712814047933, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.450980392156863e-06, | |
| "logits/chosen": -2.29875111579895, | |
| "logits/rejected": -2.20742130279541, | |
| "logps/chosen": -103.51786041259766, | |
| "logps/rejected": -77.90748596191406, | |
| "loss": 0.692, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.005386353936046362, | |
| "rewards/margins": 0.002320895902812481, | |
| "rewards/rejected": 0.0030654582660645247, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 2.9411764705882355e-06, | |
| "logits/chosen": -2.30989408493042, | |
| "logits/rejected": -2.233097553253174, | |
| "logps/chosen": -107.9452896118164, | |
| "logps/rejected": -90.55638122558594, | |
| "loss": 0.6912, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.008994068950414658, | |
| "rewards/margins": 0.003984561190009117, | |
| "rewards/rejected": 0.005009507294744253, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.431372549019608e-06, | |
| "logits/chosen": -2.2964837551116943, | |
| "logits/rejected": -2.192676544189453, | |
| "logps/chosen": -98.52997589111328, | |
| "logps/rejected": -73.2212142944336, | |
| "loss": 0.6907, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.011149758473038673, | |
| "rewards/margins": 0.004859681706875563, | |
| "rewards/rejected": 0.006290078163146973, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.92156862745098e-06, | |
| "logits/chosen": -2.147590398788452, | |
| "logits/rejected": -2.0242252349853516, | |
| "logps/chosen": -110.5780258178711, | |
| "logps/rejected": -85.19944763183594, | |
| "loss": 0.6903, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.018292848020792007, | |
| "rewards/margins": 0.005854339338839054, | |
| "rewards/rejected": 0.012438507750630379, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.411764705882353e-06, | |
| "logits/chosen": -2.103201389312744, | |
| "logits/rejected": -1.9731849431991577, | |
| "logps/chosen": -86.84721374511719, | |
| "logps/rejected": -63.73047637939453, | |
| "loss": 0.6889, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.02545500174164772, | |
| "rewards/margins": 0.008647125214338303, | |
| "rewards/rejected": 0.016807876527309418, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.901960784313726e-06, | |
| "logits/chosen": -2.071173906326294, | |
| "logits/rejected": -1.8952083587646484, | |
| "logps/chosen": -120.53657531738281, | |
| "logps/rejected": -84.6740951538086, | |
| "loss": 0.6804, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.051459141075611115, | |
| "rewards/margins": 0.02619524672627449, | |
| "rewards/rejected": 0.025263899937272072, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.9990549169459415e-06, | |
| "logits/chosen": -2.2821316719055176, | |
| "logits/rejected": -2.1455578804016113, | |
| "logps/chosen": -103.51020812988281, | |
| "logps/rejected": -78.41320037841797, | |
| "loss": 0.6767, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": 0.06272837519645691, | |
| "rewards/margins": 0.03374689072370529, | |
| "rewards/rejected": 0.028981482610106468, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.995216741642263e-06, | |
| "logits/chosen": -2.284080743789673, | |
| "logits/rejected": -2.210580825805664, | |
| "logps/chosen": -96.24214172363281, | |
| "logps/rejected": -77.19042205810547, | |
| "loss": 0.6812, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.08176051080226898, | |
| "rewards/margins": 0.02550395205616951, | |
| "rewards/rejected": 0.056256555020809174, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.988430936991089e-06, | |
| "logits/chosen": -2.2925381660461426, | |
| "logits/rejected": -2.149710178375244, | |
| "logps/chosen": -106.36643981933594, | |
| "logps/rejected": -80.01815795898438, | |
| "loss": 0.6647, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.11486251652240753, | |
| "rewards/margins": 0.06016199663281441, | |
| "rewards/rejected": 0.054700516164302826, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.978705519144525e-06, | |
| "logits/chosen": -2.1970953941345215, | |
| "logits/rejected": -2.0385475158691406, | |
| "logps/chosen": -118.03541564941406, | |
| "logps/rejected": -81.73417663574219, | |
| "loss": 0.6639, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.10544681549072266, | |
| "rewards/margins": 0.06229867413640022, | |
| "rewards/rejected": 0.04314813390374184, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.966051976854862e-06, | |
| "logits/chosen": -2.339622735977173, | |
| "logits/rejected": -2.2239785194396973, | |
| "logps/chosen": -90.04592895507812, | |
| "logps/rejected": -68.85932922363281, | |
| "loss": 0.6506, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.1409004032611847, | |
| "rewards/margins": 0.09264969825744629, | |
| "rewards/rejected": 0.048250701278448105, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.950485257902782e-06, | |
| "logits/chosen": -2.2208049297332764, | |
| "logits/rejected": -2.1466565132141113, | |
| "logps/chosen": -101.00809478759766, | |
| "logps/rejected": -76.30353546142578, | |
| "loss": 0.6618, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.13603894412517548, | |
| "rewards/margins": 0.0690341368317604, | |
| "rewards/rejected": 0.06700481474399567, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.932023751439358e-06, | |
| "logits/chosen": -2.286597728729248, | |
| "logits/rejected": -2.13307523727417, | |
| "logps/chosen": -110.24186706542969, | |
| "logps/rejected": -81.82392120361328, | |
| "loss": 0.6666, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.13935917615890503, | |
| "rewards/margins": 0.06269019842147827, | |
| "rewards/rejected": 0.07666899263858795, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.9106892662627395e-06, | |
| "logits/chosen": -2.357682466506958, | |
| "logits/rejected": -2.237700939178467, | |
| "logps/chosen": -104.79756927490234, | |
| "logps/rejected": -81.48443603515625, | |
| "loss": 0.6615, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.16287028789520264, | |
| "rewards/margins": 0.07204125821590424, | |
| "rewards/rejected": 0.0908290296792984, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.886507005055149e-06, | |
| "logits/chosen": -2.3158655166625977, | |
| "logits/rejected": -2.1448147296905518, | |
| "logps/chosen": -113.81204986572266, | |
| "logps/rejected": -80.2816390991211, | |
| "loss": 0.646, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.15469492971897125, | |
| "rewards/margins": 0.10541819036006927, | |
| "rewards/rejected": 0.049276743084192276, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.859505534610658e-06, | |
| "logits/chosen": -2.2734456062316895, | |
| "logits/rejected": -2.1840872764587402, | |
| "logps/chosen": -94.27523040771484, | |
| "logps/rejected": -76.13407897949219, | |
| "loss": 0.65, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.12800325453281403, | |
| "rewards/margins": 0.10445593297481537, | |
| "rewards/rejected": 0.023547304794192314, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_logits/chosen": -2.1420769691467285, | |
| "eval_logits/rejected": -2.046384811401367, | |
| "eval_logps/chosen": -98.08348083496094, | |
| "eval_logps/rejected": -76.25611877441406, | |
| "eval_loss": 0.656255841255188, | |
| "eval_rewards/accuracies": 0.650943398475647, | |
| "eval_rewards/chosen": 0.106992706656456, | |
| "eval_rewards/margins": 0.08928683400154114, | |
| "eval_rewards/rejected": 0.017705870792269707, | |
| "eval_runtime": 425.8967, | |
| "eval_samples_per_second": 0.981, | |
| "eval_steps_per_second": 0.124, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.829716752088893e-06, | |
| "logits/chosen": -2.2180724143981934, | |
| "logits/rejected": -2.1603446006774902, | |
| "logps/chosen": -85.03736877441406, | |
| "logps/rejected": -78.87088012695312, | |
| "loss": 0.6573, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.10240446031093597, | |
| "rewards/margins": 0.08565853536128998, | |
| "rewards/rejected": 0.016745951026678085, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.797175847334535e-06, | |
| "logits/chosen": -2.2404112815856934, | |
| "logits/rejected": -2.12317156791687, | |
| "logps/chosen": -107.25093841552734, | |
| "logps/rejected": -84.33776092529297, | |
| "loss": 0.6679, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.06942565739154816, | |
| "rewards/margins": 0.06278938055038452, | |
| "rewards/rejected": 0.006636275444179773, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.761921261307143e-06, | |
| "logits/chosen": -2.271301746368408, | |
| "logits/rejected": -2.1368188858032227, | |
| "logps/chosen": -102.20993041992188, | |
| "logps/rejected": -80.41031646728516, | |
| "loss": 0.656, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.06752457469701767, | |
| "rewards/margins": 0.09335462749004364, | |
| "rewards/rejected": -0.02583005093038082, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.723994640670377e-06, | |
| "logits/chosen": -2.2616968154907227, | |
| "logits/rejected": -2.090630292892456, | |
| "logps/chosen": -112.83064270019531, | |
| "logps/rejected": -78.7665023803711, | |
| "loss": 0.613, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": 0.1038433164358139, | |
| "rewards/margins": 0.18221423029899597, | |
| "rewards/rejected": -0.07837091386318207, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68344078859431e-06, | |
| "logits/chosen": -2.214836597442627, | |
| "logits/rejected": -2.1651015281677246, | |
| "logps/chosen": -94.0736312866211, | |
| "logps/rejected": -87.97615814208984, | |
| "loss": 0.6446, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.05433257669210434, | |
| "rewards/margins": 0.1173890084028244, | |
| "rewards/rejected": -0.06305641680955887, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.6403076118289006e-06, | |
| "logits/chosen": -2.191092014312744, | |
| "logits/rejected": -2.0465681552886963, | |
| "logps/chosen": -106.81085205078125, | |
| "logps/rejected": -80.91065216064453, | |
| "loss": 0.6325, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.04650095850229263, | |
| "rewards/margins": 0.14686846733093262, | |
| "rewards/rejected": -0.10036750882863998, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5946460641111776e-06, | |
| "logits/chosen": -2.3747944831848145, | |
| "logits/rejected": -2.1630542278289795, | |
| "logps/chosen": -103.29341125488281, | |
| "logps/rejected": -80.09381103515625, | |
| "loss": 0.6041, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.06812257319688797, | |
| "rewards/margins": 0.2125866860151291, | |
| "rewards/rejected": -0.14446412026882172, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.546510085972983e-06, | |
| "logits/chosen": -2.385852813720703, | |
| "logits/rejected": -2.2712764739990234, | |
| "logps/chosen": -115.0221176147461, | |
| "logps/rejected": -90.1436538696289, | |
| "loss": 0.6212, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.017763856798410416, | |
| "rewards/margins": 0.1770613044500351, | |
| "rewards/rejected": -0.15929746627807617, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.495956541020376e-06, | |
| "logits/chosen": -2.3901445865631104, | |
| "logits/rejected": -2.247413396835327, | |
| "logps/chosen": -119.44215393066406, | |
| "logps/rejected": -101.89396667480469, | |
| "loss": 0.613, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.03466514125466347, | |
| "rewards/margins": 0.2126956284046173, | |
| "rewards/rejected": -0.24736078083515167, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.443045148759978e-06, | |
| "logits/chosen": -2.3851912021636963, | |
| "logits/rejected": -2.246699094772339, | |
| "logps/chosen": -126.57054138183594, | |
| "logps/rejected": -93.43769836425781, | |
| "loss": 0.5898, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.06855957210063934, | |
| "rewards/margins": 0.2794524133205414, | |
| "rewards/rejected": -0.3480120301246643, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.3878384140516025e-06, | |
| "logits/chosen": -2.4334146976470947, | |
| "logits/rejected": -2.3179290294647217, | |
| "logps/chosen": -118.78206634521484, | |
| "logps/rejected": -107.4043197631836, | |
| "loss": 0.5792, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.15513427555561066, | |
| "rewards/margins": 0.30940642952919006, | |
| "rewards/rejected": -0.46454063057899475, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.330401553270522e-06, | |
| "logits/chosen": -2.34911847114563, | |
| "logits/rejected": -2.2519352436065674, | |
| "logps/chosen": -116.80616760253906, | |
| "logps/rejected": -108.0991439819336, | |
| "loss": 0.5869, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.32395830750465393, | |
| "rewards/margins": 0.30500033497810364, | |
| "rewards/rejected": -0.6289585828781128, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.2708024172665795e-06, | |
| "logits/chosen": -2.448741912841797, | |
| "logits/rejected": -2.2790164947509766, | |
| "logps/chosen": -119.88993072509766, | |
| "logps/rejected": -94.18651580810547, | |
| "loss": 0.5564, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.37532466650009155, | |
| "rewards/margins": 0.37958258390426636, | |
| "rewards/rejected": -0.7549072504043579, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.209111411211174e-06, | |
| "logits/chosen": -2.4606077671051025, | |
| "logits/rejected": -2.3158764839172363, | |
| "logps/chosen": -108.91439056396484, | |
| "logps/rejected": -94.19551086425781, | |
| "loss": 0.5416, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.3131277859210968, | |
| "rewards/margins": 0.4208511710166931, | |
| "rewards/rejected": -0.7339790463447571, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.145401411426788e-06, | |
| "logits/chosen": -2.4888482093811035, | |
| "logits/rejected": -2.3739964962005615, | |
| "logps/chosen": -106.79335021972656, | |
| "logps/rejected": -94.10002136230469, | |
| "loss": 0.5863, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.24795857071876526, | |
| "rewards/margins": 0.3726752996444702, | |
| "rewards/rejected": -0.6206337809562683, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.079747679297314e-06, | |
| "logits/chosen": -2.320113182067871, | |
| "logits/rejected": -2.214874744415283, | |
| "logps/chosen": -118.53324890136719, | |
| "logps/rejected": -100.85164642333984, | |
| "loss": 0.574, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.29153329133987427, | |
| "rewards/margins": 0.3650625944137573, | |
| "rewards/rejected": -0.6565959453582764, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.012227772360889e-06, | |
| "logits/chosen": -2.3389956951141357, | |
| "logits/rejected": -2.1818504333496094, | |
| "logps/chosen": -123.92159271240234, | |
| "logps/rejected": -110.42405700683594, | |
| "loss": 0.5134, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.2689405083656311, | |
| "rewards/margins": 0.5353353023529053, | |
| "rewards/rejected": -0.8042758703231812, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.942921452690245e-06, | |
| "logits/chosen": -2.369523525238037, | |
| "logits/rejected": -2.285311222076416, | |
| "logps/chosen": -132.2254180908203, | |
| "logps/rejected": -125.16448974609375, | |
| "loss": 0.6273, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.6799182891845703, | |
| "rewards/margins": 0.3109356760978699, | |
| "rewards/rejected": -0.9908539652824402, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.871910592668817e-06, | |
| "logits/chosen": -2.4459290504455566, | |
| "logits/rejected": -2.3258919715881348, | |
| "logps/chosen": -127.33780670166016, | |
| "logps/rejected": -129.97348022460938, | |
| "loss": 0.4728, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.7220331430435181, | |
| "rewards/margins": 0.7715710997581482, | |
| "rewards/rejected": -1.4936041831970215, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.799279078273921e-06, | |
| "logits/chosen": -2.3586697578430176, | |
| "logits/rejected": -2.190298080444336, | |
| "logps/chosen": -137.87445068359375, | |
| "logps/rejected": -125.60152435302734, | |
| "loss": 0.456, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.9262881278991699, | |
| "rewards/margins": 0.8406116366386414, | |
| "rewards/rejected": -1.766899824142456, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "eval_logits/chosen": -2.21016526222229, | |
| "eval_logits/rejected": -2.1202762126922607, | |
| "eval_logps/chosen": -142.6661376953125, | |
| "eval_logps/rejected": -139.3409881591797, | |
| "eval_loss": 0.5446351766586304, | |
| "eval_rewards/accuracies": 0.7216981053352356, | |
| "eval_rewards/chosen": -1.23048734664917, | |
| "eval_rewards/margins": 0.6443533301353455, | |
| "eval_rewards/rejected": -1.8748407363891602, | |
| "eval_runtime": 427.1638, | |
| "eval_samples_per_second": 0.979, | |
| "eval_steps_per_second": 0.124, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.725112709981249e-06, | |
| "logits/chosen": -2.2144548892974854, | |
| "logits/rejected": -2.0825304985046387, | |
| "logps/chosen": -153.91371154785156, | |
| "logps/rejected": -152.90274047851562, | |
| "loss": 0.524, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.2480976581573486, | |
| "rewards/margins": 0.7566108703613281, | |
| "rewards/rejected": -2.0047080516815186, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.649499101407737e-06, | |
| "logits/chosen": -2.3662781715393066, | |
| "logits/rejected": -2.2161378860473633, | |
| "logps/chosen": -157.3763427734375, | |
| "logps/rejected": -154.7397918701172, | |
| "loss": 0.5717, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.3450363874435425, | |
| "rewards/margins": 0.736282467842102, | |
| "rewards/rejected": -2.0813188552856445, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.5725275758125564e-06, | |
| "logits/chosen": -2.2823150157928467, | |
| "logits/rejected": -2.1387839317321777, | |
| "logps/chosen": -135.34524536132812, | |
| "logps/rejected": -134.94952392578125, | |
| "loss": 0.4512, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -1.2761428356170654, | |
| "rewards/margins": 1.0124304294586182, | |
| "rewards/rejected": -2.2885732650756836, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.494289060578478e-06, | |
| "logits/chosen": -2.3979570865631104, | |
| "logits/rejected": -2.3127408027648926, | |
| "logps/chosen": -148.89285278320312, | |
| "logps/rejected": -157.427978515625, | |
| "loss": 0.5081, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.634016990661621, | |
| "rewards/margins": 0.8799861669540405, | |
| "rewards/rejected": -2.514003276824951, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.414875979798272e-06, | |
| "logits/chosen": -2.349696636199951, | |
| "logits/rejected": -2.236372709274292, | |
| "logps/chosen": -163.39578247070312, | |
| "logps/rejected": -173.43084716796875, | |
| "loss": 0.4372, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.1473162174224854, | |
| "rewards/margins": 1.2105666399002075, | |
| "rewards/rejected": -2.3578829765319824, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3343821450930196e-06, | |
| "logits/chosen": -2.468815326690674, | |
| "logits/rejected": -2.319728374481201, | |
| "logps/chosen": -147.9761199951172, | |
| "logps/rejected": -140.23440551757812, | |
| "loss": 0.4893, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.0942740440368652, | |
| "rewards/margins": 1.0382341146469116, | |
| "rewards/rejected": -2.132507801055908, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.252902644791325e-06, | |
| "logits/chosen": -2.2242140769958496, | |
| "logits/rejected": -2.078382968902588, | |
| "logps/chosen": -138.49478149414062, | |
| "logps/rejected": -156.64273071289062, | |
| "loss": 0.4146, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -1.0762813091278076, | |
| "rewards/margins": 1.254016399383545, | |
| "rewards/rejected": -2.3302977085113525, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.170533731600339e-06, | |
| "logits/chosen": -2.334146022796631, | |
| "logits/rejected": -2.1843693256378174, | |
| "logps/chosen": -145.91741943359375, | |
| "logps/rejected": -150.4651336669922, | |
| "loss": 0.5102, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.1809916496276855, | |
| "rewards/margins": 1.0365092754364014, | |
| "rewards/rejected": -2.217501163482666, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.0873727089012816e-06, | |
| "logits/chosen": -2.4750094413757324, | |
| "logits/rejected": -2.3860108852386475, | |
| "logps/chosen": -168.5699005126953, | |
| "logps/rejected": -172.41029357910156, | |
| "loss": 0.522, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.693511962890625, | |
| "rewards/margins": 1.0143020153045654, | |
| "rewards/rejected": -2.7078135013580322, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.0035178158038026e-06, | |
| "logits/chosen": -2.261049747467041, | |
| "logits/rejected": -2.0733211040496826, | |
| "logps/chosen": -178.0288543701172, | |
| "logps/rejected": -181.95970153808594, | |
| "loss": 0.4784, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.8506263494491577, | |
| "rewards/margins": 1.296245813369751, | |
| "rewards/rejected": -3.146872043609619, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.919068111094937e-06, | |
| "logits/chosen": -2.3573222160339355, | |
| "logits/rejected": -2.2329063415527344, | |
| "logps/chosen": -169.11073303222656, | |
| "logps/rejected": -186.6231231689453, | |
| "loss": 0.6133, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -2.2830090522766113, | |
| "rewards/margins": 1.0279150009155273, | |
| "rewards/rejected": -3.3109240531921387, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.8341233562197895e-06, | |
| "logits/chosen": -2.3493080139160156, | |
| "logits/rejected": -2.2869277000427246, | |
| "logps/chosen": -156.89324951171875, | |
| "logps/rejected": -172.35809326171875, | |
| "loss": 0.4775, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.8899619579315186, | |
| "rewards/margins": 1.054958701133728, | |
| "rewards/rejected": -2.944920778274536, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7487838974321352e-06, | |
| "logits/chosen": -2.294079065322876, | |
| "logits/rejected": -2.1721999645233154, | |
| "logps/chosen": -166.351806640625, | |
| "logps/rejected": -181.2619171142578, | |
| "loss": 0.501, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.059480905532837, | |
| "rewards/margins": 1.2059848308563232, | |
| "rewards/rejected": -3.265465259552002, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.6631505472541997e-06, | |
| "logits/chosen": -2.284674644470215, | |
| "logits/rejected": -2.151689291000366, | |
| "logps/chosen": -170.77108764648438, | |
| "logps/rejected": -185.42703247070312, | |
| "loss": 0.4457, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.8532861471176147, | |
| "rewards/margins": 1.2717969417572021, | |
| "rewards/rejected": -3.1250832080841064, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.5773244653856173e-06, | |
| "logits/chosen": -2.262991428375244, | |
| "logits/rejected": -2.140122652053833, | |
| "logps/chosen": -171.88180541992188, | |
| "logps/rejected": -183.91680908203125, | |
| "loss": 0.4782, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.8343322277069092, | |
| "rewards/margins": 1.209377408027649, | |
| "rewards/rejected": -3.0437099933624268, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4914070392022717e-06, | |
| "logits/chosen": -2.287393093109131, | |
| "logits/rejected": -2.1789438724517822, | |
| "logps/chosen": -167.02993774414062, | |
| "logps/rejected": -178.21737670898438, | |
| "loss": 0.4847, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.4670352935791016, | |
| "rewards/margins": 1.1378867626190186, | |
| "rewards/rejected": -2.60492205619812, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.4054997639861778e-06, | |
| "logits/chosen": -2.2121901512145996, | |
| "logits/rejected": -2.039933443069458, | |
| "logps/chosen": -164.4762725830078, | |
| "logps/rejected": -171.06558227539062, | |
| "loss": 0.4581, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.45037043094635, | |
| "rewards/margins": 1.332580327987671, | |
| "rewards/rejected": -2.7829508781433105, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3197041230278905e-06, | |
| "logits/chosen": -2.336905002593994, | |
| "logits/rejected": -2.2310211658477783, | |
| "logps/chosen": -159.6435089111328, | |
| "logps/rejected": -185.4768829345703, | |
| "loss": 0.4182, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -1.4801210165023804, | |
| "rewards/margins": 1.5004690885543823, | |
| "rewards/rejected": -2.980590343475342, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.234121467743082e-06, | |
| "logits/chosen": -2.356192111968994, | |
| "logits/rejected": -2.269792079925537, | |
| "logps/chosen": -152.16677856445312, | |
| "logps/rejected": -166.55563354492188, | |
| "loss": 0.545, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.5818209648132324, | |
| "rewards/margins": 1.1744325160980225, | |
| "rewards/rejected": -2.756253719329834, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.148852897944905e-06, | |
| "logits/chosen": -2.4117207527160645, | |
| "logits/rejected": -2.2672557830810547, | |
| "logps/chosen": -149.95504760742188, | |
| "logps/rejected": -165.35263061523438, | |
| "loss": 0.4388, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.488303303718567, | |
| "rewards/margins": 1.227616786956787, | |
| "rewards/rejected": -2.7159199714660645, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "eval_logits/chosen": -2.1954238414764404, | |
| "eval_logits/rejected": -2.1017305850982666, | |
| "eval_logps/chosen": -161.69036865234375, | |
| "eval_logps/rejected": -173.2707977294922, | |
| "eval_loss": 0.5324557423591614, | |
| "eval_rewards/accuracies": 0.7122641801834106, | |
| "eval_rewards/chosen": -1.8012140989303589, | |
| "eval_rewards/margins": 1.091520071029663, | |
| "eval_rewards/rejected": -2.8927345275878906, | |
| "eval_runtime": 427.5006, | |
| "eval_samples_per_second": 0.978, | |
| "eval_steps_per_second": 0.124, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.063999142413574e-06, | |
| "logits/chosen": -2.3014285564422607, | |
| "logits/rejected": -2.1665964126586914, | |
| "logps/chosen": -164.73809814453125, | |
| "logps/rejected": -163.736083984375, | |
| "loss": 0.4451, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.6745765209197998, | |
| "rewards/margins": 1.155146598815918, | |
| "rewards/rejected": -2.8297228813171387, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9796604399042547e-06, | |
| "logits/chosen": -2.375365972518921, | |
| "logits/rejected": -2.2335057258605957, | |
| "logps/chosen": -170.6911163330078, | |
| "logps/rejected": -186.2073974609375, | |
| "loss": 0.4407, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.9759544134140015, | |
| "rewards/margins": 1.4642221927642822, | |
| "rewards/rejected": -3.440176486968994, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8959364207338216e-06, | |
| "logits/chosen": -2.3572235107421875, | |
| "logits/rejected": -2.1937854290008545, | |
| "logps/chosen": -148.90460205078125, | |
| "logps/rejected": -174.04818725585938, | |
| "loss": 0.5111, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.8767086267471313, | |
| "rewards/margins": 1.4870884418487549, | |
| "rewards/rejected": -3.363797426223755, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.8129259890863825e-06, | |
| "logits/chosen": -2.3194022178649902, | |
| "logits/rejected": -2.2175240516662598, | |
| "logps/chosen": -165.7982177734375, | |
| "logps/rejected": -185.64183044433594, | |
| "loss": 0.5652, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.662651777267456, | |
| "rewards/margins": 1.1808096170425415, | |
| "rewards/rejected": -2.843461513519287, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.7307272061765738e-06, | |
| "logits/chosen": -2.355379819869995, | |
| "logits/rejected": -2.256904125213623, | |
| "logps/chosen": -160.78778076171875, | |
| "logps/rejected": -178.40591430664062, | |
| "loss": 0.489, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -1.5761786699295044, | |
| "rewards/margins": 1.2927827835083008, | |
| "rewards/rejected": -2.8689613342285156, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.649437174408685e-06, | |
| "logits/chosen": -2.2938830852508545, | |
| "logits/rejected": -2.162132740020752, | |
| "logps/chosen": -150.61842346191406, | |
| "logps/rejected": -170.839111328125, | |
| "loss": 0.5404, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.1881531476974487, | |
| "rewards/margins": 1.283790111541748, | |
| "rewards/rejected": -2.4719433784484863, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.569151922668422e-06, | |
| "logits/chosen": -2.3966734409332275, | |
| "logits/rejected": -2.287233829498291, | |
| "logps/chosen": -138.367431640625, | |
| "logps/rejected": -159.68118286132812, | |
| "loss": 0.5196, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.2400791645050049, | |
| "rewards/margins": 1.232775330543518, | |
| "rewards/rejected": -2.4728543758392334, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4899662928828428e-06, | |
| "logits/chosen": -2.333634853363037, | |
| "logits/rejected": -2.157811403274536, | |
| "logps/chosen": -127.2032699584961, | |
| "logps/rejected": -149.32907104492188, | |
| "loss": 0.4387, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -1.0127990245819092, | |
| "rewards/margins": 1.4315626621246338, | |
| "rewards/rejected": -2.444361686706543, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.4119738279824507e-06, | |
| "logits/chosen": -2.272979259490967, | |
| "logits/rejected": -2.127248764038086, | |
| "logps/chosen": -131.46400451660156, | |
| "logps/rejected": -144.866455078125, | |
| "loss": 0.5303, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.0165386199951172, | |
| "rewards/margins": 1.0676348209381104, | |
| "rewards/rejected": -2.0841732025146484, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3352666613978152e-06, | |
| "logits/chosen": -2.192830801010132, | |
| "logits/rejected": -2.111144542694092, | |
| "logps/chosen": -134.23226928710938, | |
| "logps/rejected": -143.62905883789062, | |
| "loss": 0.5064, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.0449942350387573, | |
| "rewards/margins": 1.0384838581085205, | |
| "rewards/rejected": -2.0834782123565674, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2599354082212523e-06, | |
| "logits/chosen": -2.2571260929107666, | |
| "logits/rejected": -2.1350224018096924, | |
| "logps/chosen": -129.79287719726562, | |
| "logps/rejected": -138.2635498046875, | |
| "loss": 0.5149, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.9037138223648071, | |
| "rewards/margins": 0.9939008951187134, | |
| "rewards/rejected": -1.8976147174835205, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.186069058162127e-06, | |
| "logits/chosen": -2.3779895305633545, | |
| "logits/rejected": -2.2798473834991455, | |
| "logps/chosen": -136.55984497070312, | |
| "logps/rejected": -145.81918334960938, | |
| "loss": 0.5318, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.2205983400344849, | |
| "rewards/margins": 0.9511237144470215, | |
| "rewards/rejected": -2.171722173690796, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.113754870422254e-06, | |
| "logits/chosen": -2.4235808849334717, | |
| "logits/rejected": -2.3193821907043457, | |
| "logps/chosen": -142.02517700195312, | |
| "logps/rejected": -161.961669921875, | |
| "loss": 0.5042, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.2923016548156738, | |
| "rewards/margins": 0.9127277135848999, | |
| "rewards/rejected": -2.205029249191284, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.0430782706155545e-06, | |
| "logits/chosen": -2.435793399810791, | |
| "logits/rejected": -2.296560764312744, | |
| "logps/chosen": -145.57290649414062, | |
| "logps/rejected": -156.23214721679688, | |
| "loss": 0.4346, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.0421411991119385, | |
| "rewards/margins": 1.3148539066314697, | |
| "rewards/rejected": -2.356995105743408, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.741227498537615e-07, | |
| "logits/chosen": -2.4774973392486572, | |
| "logits/rejected": -2.3496251106262207, | |
| "logps/chosen": -133.50381469726562, | |
| "logps/rejected": -148.293701171875, | |
| "loss": 0.4295, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -1.2879221439361572, | |
| "rewards/margins": 1.1688247919082642, | |
| "rewards/rejected": -2.45674729347229, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.069697661173668e-07, | |
| "logits/chosen": -2.3304691314697266, | |
| "logits/rejected": -2.2422666549682617, | |
| "logps/chosen": -143.60433959960938, | |
| "logps/rejected": -161.01876831054688, | |
| "loss": 0.6187, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -1.4187496900558472, | |
| "rewards/margins": 0.990622878074646, | |
| "rewards/rejected": -2.4093728065490723, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.416986480283434e-07, | |
| "logits/chosen": -2.2641918659210205, | |
| "logits/rejected": -2.1294054985046387, | |
| "logps/chosen": -150.9503173828125, | |
| "logps/rejected": -169.94259643554688, | |
| "loss": 0.5414, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.6702228784561157, | |
| "rewards/margins": 1.2004982233047485, | |
| "rewards/rejected": -2.8707211017608643, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.783865011382876e-07, | |
| "logits/chosen": -2.318986415863037, | |
| "logits/rejected": -2.189507246017456, | |
| "logps/chosen": -157.97384643554688, | |
| "logps/rejected": -164.84359741210938, | |
| "loss": 0.4953, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.3389387130737305, | |
| "rewards/margins": 1.2802064418792725, | |
| "rewards/rejected": -2.619144916534424, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.171081168427205e-07, | |
| "logits/chosen": -2.3565828800201416, | |
| "logits/rejected": -2.193580150604248, | |
| "logps/chosen": -169.49929809570312, | |
| "logps/rejected": -177.23731994628906, | |
| "loss": 0.5052, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.3413608074188232, | |
| "rewards/margins": 1.271929144859314, | |
| "rewards/rejected": -2.6132900714874268, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.579358840291064e-07, | |
| "logits/chosen": -2.3053181171417236, | |
| "logits/rejected": -2.205723762512207, | |
| "logps/chosen": -155.66004943847656, | |
| "logps/rejected": -173.5889434814453, | |
| "loss": 0.6137, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.5923388004302979, | |
| "rewards/margins": 0.8797184228897095, | |
| "rewards/rejected": -2.4720568656921387, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "eval_logits/chosen": -2.19348406791687, | |
| "eval_logits/rejected": -2.0962369441986084, | |
| "eval_logps/chosen": -149.9387664794922, | |
| "eval_logps/rejected": -160.84133911132812, | |
| "eval_loss": 0.5198491215705872, | |
| "eval_rewards/accuracies": 0.7382075190544128, | |
| "eval_rewards/chosen": -1.448665976524353, | |
| "eval_rewards/margins": 1.0711843967437744, | |
| "eval_rewards/rejected": -2.519850254058838, | |
| "eval_runtime": 426.9839, | |
| "eval_samples_per_second": 0.979, | |
| "eval_steps_per_second": 0.124, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 6.00939703563006e-07, | |
| "logits/chosen": -2.3648667335510254, | |
| "logits/rejected": -2.2830593585968018, | |
| "logps/chosen": -136.9246826171875, | |
| "logps/rejected": -156.57559204101562, | |
| "loss": 0.5648, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.4630320072174072, | |
| "rewards/margins": 0.9899848699569702, | |
| "rewards/rejected": -2.453016757965088, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.461869057133412e-07, | |
| "logits/chosen": -2.408620595932007, | |
| "logits/rejected": -2.319190502166748, | |
| "logps/chosen": -145.22872924804688, | |
| "logps/rejected": -167.64926147460938, | |
| "loss": 0.487, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.306095838546753, | |
| "rewards/margins": 1.1592603921890259, | |
| "rewards/rejected": -2.4653563499450684, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.937421706143497e-07, | |
| "logits/chosen": -2.290102481842041, | |
| "logits/rejected": -2.158676862716675, | |
| "logps/chosen": -152.98867797851562, | |
| "logps/rejected": -159.16566467285156, | |
| "loss": 0.5218, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.3972251415252686, | |
| "rewards/margins": 1.101043462753296, | |
| "rewards/rejected": -2.4982683658599854, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.43667451858166e-07, | |
| "logits/chosen": -2.2902016639709473, | |
| "logits/rejected": -2.131404399871826, | |
| "logps/chosen": -134.6024627685547, | |
| "logps/rejected": -151.9069061279297, | |
| "loss": 0.4313, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -1.2463924884796143, | |
| "rewards/margins": 1.268364667892456, | |
| "rewards/rejected": -2.514756917953491, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.9602190330830484e-07, | |
| "logits/chosen": -2.2570953369140625, | |
| "logits/rejected": -2.157632827758789, | |
| "logps/chosen": -141.2190704345703, | |
| "logps/rejected": -162.86306762695312, | |
| "loss": 0.4876, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.3870246410369873, | |
| "rewards/margins": 1.1176416873931885, | |
| "rewards/rejected": -2.504666328430176, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.5086180922049295e-07, | |
| "logits/chosen": -2.394253969192505, | |
| "logits/rejected": -2.2592263221740723, | |
| "logps/chosen": -165.12005615234375, | |
| "logps/rejected": -166.54446411132812, | |
| "loss": 0.4366, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.3376173973083496, | |
| "rewards/margins": 1.2365379333496094, | |
| "rewards/rejected": -2.574155330657959, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.0824051775340895e-07, | |
| "logits/chosen": -2.424966335296631, | |
| "logits/rejected": -2.3199541568756104, | |
| "logps/chosen": -124.06752014160156, | |
| "logps/rejected": -140.36489868164062, | |
| "loss": 0.5696, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.2786959409713745, | |
| "rewards/margins": 0.8720223307609558, | |
| "rewards/rejected": -2.1507184505462646, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.6820837794786336e-07, | |
| "logits/chosen": -2.2567453384399414, | |
| "logits/rejected": -2.1820857524871826, | |
| "logps/chosen": -145.2144317626953, | |
| "logps/rejected": -155.57261657714844, | |
| "loss": 0.6788, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -1.4533874988555908, | |
| "rewards/margins": 0.4883599281311035, | |
| "rewards/rejected": -1.9417474269866943, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.3081268024887694e-07, | |
| "logits/chosen": -2.2515053749084473, | |
| "logits/rejected": -2.0878632068634033, | |
| "logps/chosen": -141.62442016601562, | |
| "logps/rejected": -151.8519287109375, | |
| "loss": 0.4185, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -1.220234990119934, | |
| "rewards/margins": 1.2837200164794922, | |
| "rewards/rejected": -2.5039544105529785, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.9609760064091044e-07, | |
| "logits/chosen": -2.343606472015381, | |
| "logits/rejected": -2.277740955352783, | |
| "logps/chosen": -144.82054138183594, | |
| "logps/rejected": -142.51901245117188, | |
| "loss": 0.5214, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.2351983785629272, | |
| "rewards/margins": 0.8959378004074097, | |
| "rewards/rejected": -2.131136417388916, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.6410414846224992e-07, | |
| "logits/chosen": -2.256499767303467, | |
| "logits/rejected": -2.1499085426330566, | |
| "logps/chosen": -139.65145874023438, | |
| "logps/rejected": -156.27642822265625, | |
| "loss": 0.429, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -1.2983112335205078, | |
| "rewards/margins": 1.2496817111968994, | |
| "rewards/rejected": -2.547992706298828, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.348701179601819e-07, | |
| "logits/chosen": -2.4174904823303223, | |
| "logits/rejected": -2.2869112491607666, | |
| "logps/chosen": -156.2935028076172, | |
| "logps/rejected": -166.49362182617188, | |
| "loss": 0.4412, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -1.221897840499878, | |
| "rewards/margins": 1.1692349910736084, | |
| "rewards/rejected": -2.3911328315734863, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.0843004364420151e-07, | |
| "logits/chosen": -2.2345454692840576, | |
| "logits/rejected": -2.1238150596618652, | |
| "logps/chosen": -157.7463836669922, | |
| "logps/rejected": -174.59848022460938, | |
| "loss": 0.5407, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.309340476989746, | |
| "rewards/margins": 0.9167596101760864, | |
| "rewards/rejected": -2.226100206375122, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.481515948997931e-08, | |
| "logits/chosen": -2.384162425994873, | |
| "logits/rejected": -2.2841312885284424, | |
| "logps/chosen": -161.34854125976562, | |
| "logps/rejected": -156.95523071289062, | |
| "loss": 0.6309, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.4349567890167236, | |
| "rewards/margins": 0.7904176115989685, | |
| "rewards/rejected": -2.225374460220337, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 6.4053362042297e-08, | |
| "logits/chosen": -2.284637928009033, | |
| "logits/rejected": -2.121169090270996, | |
| "logps/chosen": -147.6913299560547, | |
| "logps/rejected": -160.8138885498047, | |
| "loss": 0.4361, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.3165013790130615, | |
| "rewards/margins": 1.2154157161712646, | |
| "rewards/rejected": -2.5319173336029053, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.616917746052163e-08, | |
| "logits/chosen": -2.3731329441070557, | |
| "logits/rejected": -2.2411139011383057, | |
| "logps/chosen": -147.39578247070312, | |
| "logps/rejected": -155.5648651123047, | |
| "loss": 0.5548, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.4831774234771729, | |
| "rewards/margins": 1.0112624168395996, | |
| "rewards/rejected": -2.4944396018981934, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.118373254556412e-08, | |
| "logits/chosen": -2.4090075492858887, | |
| "logits/rejected": -2.2784786224365234, | |
| "logps/chosen": -138.59571838378906, | |
| "logps/rejected": -139.35055541992188, | |
| "loss": 0.4862, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.2963029146194458, | |
| "rewards/margins": 1.0239927768707275, | |
| "rewards/rejected": -2.320295810699463, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.9114729782535037e-08, | |
| "logits/chosen": -2.432577133178711, | |
| "logits/rejected": -2.3220748901367188, | |
| "logps/chosen": -136.9718017578125, | |
| "logps/rejected": -146.90240478515625, | |
| "loss": 0.5037, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.215503454208374, | |
| "rewards/margins": 0.9537515640258789, | |
| "rewards/rejected": -2.169254779815674, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 9.97642642858815e-09, | |
| "logits/chosen": -2.30542254447937, | |
| "logits/rejected": -2.1695542335510254, | |
| "logps/chosen": -152.14138793945312, | |
| "logps/rejected": -153.89279174804688, | |
| "loss": 0.5001, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.3888603448867798, | |
| "rewards/margins": 0.9819372296333313, | |
| "rewards/rejected": -2.3707973957061768, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.779617670651436e-09, | |
| "logits/chosen": -2.278373956680298, | |
| "logits/rejected": -2.17798113822937, | |
| "logps/chosen": -157.8939208984375, | |
| "logps/rejected": -168.04055786132812, | |
| "loss": 0.5866, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.575810194015503, | |
| "rewards/margins": 0.995488166809082, | |
| "rewards/rejected": -2.571298599243164, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "eval_logits/chosen": -2.192172050476074, | |
| "eval_logits/rejected": -2.095205545425415, | |
| "eval_logps/chosen": -147.71951293945312, | |
| "eval_logps/rejected": -158.54696655273438, | |
| "eval_loss": 0.5194684863090515, | |
| "eval_rewards/accuracies": 0.7358490824699402, | |
| "eval_rewards/chosen": -1.382088541984558, | |
| "eval_rewards/margins": 1.068930983543396, | |
| "eval_rewards/rejected": -2.451019525527954, | |
| "eval_runtime": 427.3583, | |
| "eval_samples_per_second": 0.978, | |
| "eval_steps_per_second": 0.124, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.316238729444201e-10, | |
| "logits/chosen": -2.2608113288879395, | |
| "logits/rejected": -2.101694345474243, | |
| "logps/chosen": -152.0623779296875, | |
| "logps/rejected": -162.7583770751953, | |
| "loss": 0.4821, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.5301237106323242, | |
| "rewards/margins": 1.2161496877670288, | |
| "rewards/rejected": -2.7462732791900635, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1016, | |
| "total_flos": 0.0, | |
| "train_loss": 0.5536013367607837, | |
| "train_runtime": 17040.3461, | |
| "train_samples_per_second": 0.477, | |
| "train_steps_per_second": 0.06 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1016, | |
| "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 | |
| } | |