| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9996190476190476, | |
| "eval_steps": 500, | |
| "global_step": 656, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 7.575757575757576e-08, | |
| "logits/chosen": 0.040165986865758896, | |
| "logits/rejected": 0.1715753823518753, | |
| "logps/chosen": -294.844482421875, | |
| "logps/rejected": -361.2099914550781, | |
| "loss": 0.3581, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.575757575757576e-07, | |
| "logits/chosen": 0.08021458983421326, | |
| "logits/rejected": 0.320384681224823, | |
| "logps/chosen": -393.2468566894531, | |
| "logps/rejected": -318.70526123046875, | |
| "loss": 0.3399, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -5.44932481716387e-05, | |
| "rewards/margins": -7.010095578152686e-05, | |
| "rewards/rejected": 1.560769487696234e-05, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.5151515151515152e-06, | |
| "logits/chosen": 0.07484304904937744, | |
| "logits/rejected": 0.23277099430561066, | |
| "logps/chosen": -332.4987487792969, | |
| "logps/rejected": -281.78729248046875, | |
| "loss": 0.3453, | |
| "rewards/accuracies": 0.3687500059604645, | |
| "rewards/chosen": 1.3394804909694358e-06, | |
| "rewards/margins": -0.0002027603331953287, | |
| "rewards/rejected": 0.00020409980788826942, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.2727272727272728e-06, | |
| "logits/chosen": 0.07151266187429428, | |
| "logits/rejected": 0.25990238785743713, | |
| "logps/chosen": -358.3196716308594, | |
| "logps/rejected": -292.2561950683594, | |
| "loss": 0.3517, | |
| "rewards/accuracies": 0.543749988079071, | |
| "rewards/chosen": 0.0001912448788061738, | |
| "rewards/margins": 0.00028789255884476006, | |
| "rewards/rejected": -9.664769459050149e-05, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.0303030303030305e-06, | |
| "logits/chosen": 0.12050364166498184, | |
| "logits/rejected": 0.23899266123771667, | |
| "logps/chosen": -346.99310302734375, | |
| "logps/rejected": -286.31658935546875, | |
| "loss": 0.3416, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.0006401558639481664, | |
| "rewards/margins": 0.0007750070071779191, | |
| "rewards/rejected": -0.00013485117233358324, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 3.7878787878787882e-06, | |
| "logits/chosen": 0.07284825295209885, | |
| "logits/rejected": 0.3108685612678528, | |
| "logps/chosen": -343.5631408691406, | |
| "logps/rejected": -287.16876220703125, | |
| "loss": 0.3439, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": 0.0013991177547723055, | |
| "rewards/margins": 0.0013889471301808953, | |
| "rewards/rejected": 1.0170697350986302e-05, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.5454545454545455e-06, | |
| "logits/chosen": 0.07572797685861588, | |
| "logits/rejected": 0.2742985785007477, | |
| "logps/chosen": -371.0773620605469, | |
| "logps/rejected": -303.0706481933594, | |
| "loss": 0.3306, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.0029931231401860714, | |
| "rewards/margins": 0.004650537855923176, | |
| "rewards/rejected": -0.0016574144829064608, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999432965739786e-06, | |
| "logits/chosen": 0.06861326098442078, | |
| "logits/rejected": 0.25036171078681946, | |
| "logps/chosen": -321.8224792480469, | |
| "logps/rejected": -301.9892272949219, | |
| "loss": 0.3315, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.005247811786830425, | |
| "rewards/margins": 0.0075549171306192875, | |
| "rewards/rejected": -0.0023071051109582186, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9930567839810125e-06, | |
| "logits/chosen": 0.0997760072350502, | |
| "logits/rejected": 0.26540613174438477, | |
| "logps/chosen": -360.35430908203125, | |
| "logps/rejected": -307.5221862792969, | |
| "loss": 0.3184, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.006557592656463385, | |
| "rewards/margins": 0.01593288779258728, | |
| "rewards/rejected": -0.009375295601785183, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.979613761906212e-06, | |
| "logits/chosen": 0.11980845779180527, | |
| "logits/rejected": 0.2542170584201813, | |
| "logps/chosen": -325.69989013671875, | |
| "logps/rejected": -303.8279724121094, | |
| "loss": 0.3053, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.0012294445186853409, | |
| "rewards/margins": 0.02537659928202629, | |
| "rewards/rejected": -0.02414715476334095, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.959142005221991e-06, | |
| "logits/chosen": 0.15030920505523682, | |
| "logits/rejected": 0.24269947409629822, | |
| "logps/chosen": -357.73846435546875, | |
| "logps/rejected": -372.84625244140625, | |
| "loss": 0.2998, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.03152700141072273, | |
| "rewards/margins": 0.0535690113902092, | |
| "rewards/rejected": -0.08509601652622223, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.931699543346854e-06, | |
| "logits/chosen": 0.10781173408031464, | |
| "logits/rejected": 0.28030428290367126, | |
| "logps/chosen": -428.8057556152344, | |
| "logps/rejected": -428.8355407714844, | |
| "loss": 0.2724, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.09748424589633942, | |
| "rewards/margins": 0.0674622505903244, | |
| "rewards/rejected": -0.16494649648666382, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.897364164920515e-06, | |
| "logits/chosen": 0.1421918272972107, | |
| "logits/rejected": 0.2562108039855957, | |
| "logps/chosen": -584.6065673828125, | |
| "logps/rejected": -657.9326782226562, | |
| "loss": 0.2475, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.21280452609062195, | |
| "rewards/margins": 0.1321365237236023, | |
| "rewards/rejected": -0.34494107961654663, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8562331973035396e-06, | |
| "logits/chosen": 0.1893201768398285, | |
| "logits/rejected": 0.2765055298805237, | |
| "logps/chosen": -574.9005737304688, | |
| "logps/rejected": -700.53125, | |
| "loss": 0.2567, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.23022952675819397, | |
| "rewards/margins": 0.17663846909999847, | |
| "rewards/rejected": -0.40686798095703125, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.808423230692374e-06, | |
| "logits/chosen": 0.15862275660037994, | |
| "logits/rejected": 0.3002353310585022, | |
| "logps/chosen": -557.1515502929688, | |
| "logps/rejected": -636.1343994140625, | |
| "loss": 0.2578, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.2007971704006195, | |
| "rewards/margins": 0.1420392543077469, | |
| "rewards/rejected": -0.3428364396095276, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.754069787631761e-06, | |
| "logits/chosen": 0.20283110439777374, | |
| "logits/rejected": 0.2838110327720642, | |
| "logps/chosen": -539.7517700195312, | |
| "logps/rejected": -632.3270263671875, | |
| "loss": 0.2807, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.22490541636943817, | |
| "rewards/margins": 0.13436779379844666, | |
| "rewards/rejected": -0.359273225069046, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.693326938861367e-06, | |
| "logits/chosen": 0.16693079471588135, | |
| "logits/rejected": 0.3414613604545593, | |
| "logps/chosen": -519.4967651367188, | |
| "logps/rejected": -644.0454711914062, | |
| "loss": 0.2302, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.1742098033428192, | |
| "rewards/margins": 0.19720463454723358, | |
| "rewards/rejected": -0.3714144825935364, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.626366866585528e-06, | |
| "logits/chosen": 0.15613974630832672, | |
| "logits/rejected": 0.27570822834968567, | |
| "logps/chosen": -551.5475463867188, | |
| "logps/rejected": -641.4393310546875, | |
| "loss": 0.2342, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.18619480729103088, | |
| "rewards/margins": 0.17629162967205048, | |
| "rewards/rejected": -0.36248645186424255, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.553379376404085e-06, | |
| "logits/chosen": 0.15913400053977966, | |
| "logits/rejected": 0.28868401050567627, | |
| "logps/chosen": -575.120361328125, | |
| "logps/rejected": -672.7054443359375, | |
| "loss": 0.2586, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.23076090216636658, | |
| "rewards/margins": 0.1531594842672348, | |
| "rewards/rejected": -0.3839203715324402, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.474571359287791e-06, | |
| "logits/chosen": 0.2370852530002594, | |
| "logits/rejected": 0.313865065574646, | |
| "logps/chosen": -578.9676513671875, | |
| "logps/rejected": -646.5762329101562, | |
| "loss": 0.2759, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.22910158336162567, | |
| "rewards/margins": 0.14492908120155334, | |
| "rewards/rejected": -0.3740306496620178, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.3901662051233755e-06, | |
| "logits/chosen": 0.156154602766037, | |
| "logits/rejected": 0.28848981857299805, | |
| "logps/chosen": -529.0174560546875, | |
| "logps/rejected": -681.0127563476562, | |
| "loss": 0.2411, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.21960540115833282, | |
| "rewards/margins": 0.1676800698041916, | |
| "rewards/rejected": -0.3872854709625244, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.30040316949064e-06, | |
| "logits/chosen": 0.13694807887077332, | |
| "logits/rejected": 0.3102852702140808, | |
| "logps/chosen": -582.1616821289062, | |
| "logps/rejected": -681.3240966796875, | |
| "loss": 0.2166, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.24012689292430878, | |
| "rewards/margins": 0.18257644772529602, | |
| "rewards/rejected": -0.422703355550766, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.205536695466524e-06, | |
| "logits/chosen": 0.15840545296669006, | |
| "logits/rejected": 0.30105945467948914, | |
| "logps/chosen": -578.1398315429688, | |
| "logps/rejected": -766.7618408203125, | |
| "loss": 0.2259, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.2511864900588989, | |
| "rewards/margins": 0.22828085720539093, | |
| "rewards/rejected": -0.47946733236312866, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.105835692378557e-06, | |
| "logits/chosen": 0.14648036658763885, | |
| "logits/rejected": 0.21232767403125763, | |
| "logps/chosen": -583.7035522460938, | |
| "logps/rejected": -690.9400634765625, | |
| "loss": 0.2757, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.2628551125526428, | |
| "rewards/margins": 0.15015873312950134, | |
| "rewards/rejected": -0.41301384568214417, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.001582773552153e-06, | |
| "logits/chosen": 0.13691949844360352, | |
| "logits/rejected": 0.22931316494941711, | |
| "logps/chosen": -621.2189331054688, | |
| "logps/rejected": -718.5030517578125, | |
| "loss": 0.2518, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.24823768436908722, | |
| "rewards/margins": 0.16897296905517578, | |
| "rewards/rejected": -0.4172106683254242, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.893073455212438e-06, | |
| "logits/chosen": 0.15518508851528168, | |
| "logits/rejected": 0.27118122577667236, | |
| "logps/chosen": -649.8485107421875, | |
| "logps/rejected": -743.3021240234375, | |
| "loss": 0.2446, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.27570152282714844, | |
| "rewards/margins": 0.17269738018512726, | |
| "rewards/rejected": -0.4483988881111145, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3.7806153188114027e-06, | |
| "logits/chosen": 0.14696967601776123, | |
| "logits/rejected": 0.26466238498687744, | |
| "logps/chosen": -585.2598876953125, | |
| "logps/rejected": -646.9161376953125, | |
| "loss": 0.2236, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.22980618476867676, | |
| "rewards/margins": 0.16329681873321533, | |
| "rewards/rejected": -0.3931030035018921, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.6645271391548542e-06, | |
| "logits/chosen": 0.1697523295879364, | |
| "logits/rejected": 0.29494693875312805, | |
| "logps/chosen": -613.8264770507812, | |
| "logps/rejected": -749.4378051757812, | |
| "loss": 0.228, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.2746888995170593, | |
| "rewards/margins": 0.2020600140094757, | |
| "rewards/rejected": -0.47674888372421265, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.5451379808006014e-06, | |
| "logits/chosen": 0.13133810460567474, | |
| "logits/rejected": 0.2865068316459656, | |
| "logps/chosen": -590.5033569335938, | |
| "logps/rejected": -722.9710083007812, | |
| "loss": 0.2252, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.2536476254463196, | |
| "rewards/margins": 0.177708238363266, | |
| "rewards/rejected": -0.43135586380958557, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4227862652892106e-06, | |
| "logits/chosen": 0.16941356658935547, | |
| "logits/rejected": 0.30433687567710876, | |
| "logps/chosen": -602.0320434570312, | |
| "logps/rejected": -731.9866333007812, | |
| "loss": 0.2518, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.262613981962204, | |
| "rewards/margins": 0.17090369760990143, | |
| "rewards/rejected": -0.4335176944732666, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2978188118513814e-06, | |
| "logits/chosen": 0.1636120229959488, | |
| "logits/rejected": 0.2873205840587616, | |
| "logps/chosen": -595.1837768554688, | |
| "logps/rejected": -764.756103515625, | |
| "loss": 0.218, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.25328361988067627, | |
| "rewards/margins": 0.21258826553821564, | |
| "rewards/rejected": -0.4658718705177307, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.1705898543111576e-06, | |
| "logits/chosen": 0.128164142370224, | |
| "logits/rejected": 0.29025566577911377, | |
| "logps/chosen": -615.720947265625, | |
| "logps/rejected": -729.6194458007812, | |
| "loss": 0.2587, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.2794512212276459, | |
| "rewards/margins": 0.1740822196006775, | |
| "rewards/rejected": -0.45353350043296814, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.041460036971664e-06, | |
| "logits/chosen": 0.15139839053153992, | |
| "logits/rejected": 0.3216419517993927, | |
| "logps/chosen": -633.3115234375, | |
| "logps/rejected": -791.7572631835938, | |
| "loss": 0.2391, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3004799485206604, | |
| "rewards/margins": 0.18833482265472412, | |
| "rewards/rejected": -0.4888147711753845, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.910795392329649e-06, | |
| "logits/chosen": 0.21717897057533264, | |
| "logits/rejected": 0.3525051474571228, | |
| "logps/chosen": -616.4713134765625, | |
| "logps/rejected": -754.4976806640625, | |
| "loss": 0.2592, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.2672343850135803, | |
| "rewards/margins": 0.2101636379957199, | |
| "rewards/rejected": -0.47739800810813904, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.7789663035166035e-06, | |
| "logits/chosen": 0.2043914794921875, | |
| "logits/rejected": 0.34597498178482056, | |
| "logps/chosen": -658.1500244140625, | |
| "logps/rejected": -802.0507202148438, | |
| "loss": 0.2329, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.2817768454551697, | |
| "rewards/margins": 0.2174309492111206, | |
| "rewards/rejected": -0.4992077946662903, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.6463464544075344e-06, | |
| "logits/chosen": 0.16284213960170746, | |
| "logits/rejected": 0.28806525468826294, | |
| "logps/chosen": -627.6129760742188, | |
| "logps/rejected": -686.5743408203125, | |
| "loss": 0.2372, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.2634487748146057, | |
| "rewards/margins": 0.15015827119350433, | |
| "rewards/rejected": -0.41360706090927124, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.513311770373421e-06, | |
| "logits/chosen": 0.14228633046150208, | |
| "logits/rejected": 0.26139333844184875, | |
| "logps/chosen": -625.71875, | |
| "logps/rejected": -751.2933349609375, | |
| "loss": 0.2357, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.26470065116882324, | |
| "rewards/margins": 0.1807432472705841, | |
| "rewards/rejected": -0.44544392824172974, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.380239352679908e-06, | |
| "logits/chosen": 0.14897385239601135, | |
| "logits/rejected": 0.2588959336280823, | |
| "logps/chosen": -641.2374267578125, | |
| "logps/rejected": -802.7953491210938, | |
| "loss": 0.2166, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.30122682452201843, | |
| "rewards/margins": 0.1971724033355713, | |
| "rewards/rejected": -0.49839919805526733, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.247506409552795e-06, | |
| "logits/chosen": 0.14793309569358826, | |
| "logits/rejected": 0.3080625534057617, | |
| "logps/chosen": -668.0740966796875, | |
| "logps/rejected": -816.8853759765625, | |
| "loss": 0.2162, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.3042776584625244, | |
| "rewards/margins": 0.20680758357048035, | |
| "rewards/rejected": -0.5110852122306824, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1154891869403436e-06, | |
| "logits/chosen": 0.1291504055261612, | |
| "logits/rejected": 0.3194652497768402, | |
| "logps/chosen": -662.1779174804688, | |
| "logps/rejected": -797.3854370117188, | |
| "loss": 0.2264, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.33526724576950073, | |
| "rewards/margins": 0.19740521907806396, | |
| "rewards/rejected": -0.5326724648475647, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9845619020032552e-06, | |
| "logits/chosen": 0.16021743416786194, | |
| "logits/rejected": 0.3056327700614929, | |
| "logps/chosen": -644.52734375, | |
| "logps/rejected": -788.876953125, | |
| "loss": 0.2151, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.3085401952266693, | |
| "rewards/margins": 0.211081862449646, | |
| "rewards/rejected": -0.5196221470832825, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.8550956823554708e-06, | |
| "logits/chosen": 0.15647678077220917, | |
| "logits/rejected": 0.3057996332645416, | |
| "logps/chosen": -702.3175048828125, | |
| "logps/rejected": -853.0869140625, | |
| "loss": 0.2227, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.3388286232948303, | |
| "rewards/margins": 0.2211228311061859, | |
| "rewards/rejected": -0.5599514842033386, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.7274575140626318e-06, | |
| "logits/chosen": 0.14826330542564392, | |
| "logits/rejected": 0.32510313391685486, | |
| "logps/chosen": -669.1925659179688, | |
| "logps/rejected": -814.4791870117188, | |
| "loss": 0.2161, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.308666467666626, | |
| "rewards/margins": 0.18964755535125732, | |
| "rewards/rejected": -0.4983140528202057, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.6020092013802002e-06, | |
| "logits/chosen": 0.2039145529270172, | |
| "logits/rejected": 0.27107498049736023, | |
| "logps/chosen": -734.4779052734375, | |
| "logps/rejected": -801.9241333007812, | |
| "loss": 0.2567, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.3436730206012726, | |
| "rewards/margins": 0.15554025769233704, | |
| "rewards/rejected": -0.499213308095932, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4791063411799938e-06, | |
| "logits/chosen": 0.1372929960489273, | |
| "logits/rejected": 0.2705134451389313, | |
| "logps/chosen": -655.2030029296875, | |
| "logps/rejected": -761.6964111328125, | |
| "loss": 0.2425, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.2905011475086212, | |
| "rewards/margins": 0.1854308694601059, | |
| "rewards/rejected": -0.4759320318698883, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.3590973149722103e-06, | |
| "logits/chosen": 0.16004619002342224, | |
| "logits/rejected": 0.3311876654624939, | |
| "logps/chosen": -623.554443359375, | |
| "logps/rejected": -741.9662475585938, | |
| "loss": 0.2205, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.2771863043308258, | |
| "rewards/margins": 0.18584686517715454, | |
| "rewards/rejected": -0.46303310990333557, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2423223013801946e-06, | |
| "logits/chosen": 0.15525056421756744, | |
| "logits/rejected": 0.226671501994133, | |
| "logps/chosen": -620.3182373046875, | |
| "logps/rejected": -753.6258544921875, | |
| "loss": 0.2323, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.28617820143699646, | |
| "rewards/margins": 0.17914626002311707, | |
| "rewards/rejected": -0.4653244912624359, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.1291123118671665e-06, | |
| "logits/chosen": 0.1403166949748993, | |
| "logits/rejected": 0.21838609874248505, | |
| "logps/chosen": -640.9824829101562, | |
| "logps/rejected": -792.9696044921875, | |
| "loss": 0.23, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.281258225440979, | |
| "rewards/margins": 0.21520480513572693, | |
| "rewards/rejected": -0.4964630603790283, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.019788252448267e-06, | |
| "logits/chosen": 0.12808464467525482, | |
| "logits/rejected": 0.2763553857803345, | |
| "logps/chosen": -644.160888671875, | |
| "logps/rejected": -773.2025146484375, | |
| "loss": 0.2376, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.31234392523765564, | |
| "rewards/margins": 0.19615283608436584, | |
| "rewards/rejected": -0.5084967017173767, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.146600140475945e-07, | |
| "logits/chosen": 0.1601841151714325, | |
| "logits/rejected": 0.22859685122966766, | |
| "logps/chosen": -651.9871826171875, | |
| "logps/rejected": -752.2882080078125, | |
| "loss": 0.2404, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.3149036467075348, | |
| "rewards/margins": 0.15418918430805206, | |
| "rewards/rejected": -0.46909284591674805, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.140255940787059e-07, | |
| "logits/chosen": 0.1696256846189499, | |
| "logits/rejected": 0.2493252456188202, | |
| "logps/chosen": -615.3719482421875, | |
| "logps/rejected": -736.1841430664062, | |
| "loss": 0.2331, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.2796221375465393, | |
| "rewards/margins": 0.1891065537929535, | |
| "rewards/rejected": -0.4687287211418152, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.181702517385789e-07, | |
| "logits/chosen": 0.13488708436489105, | |
| "logits/rejected": 0.24344106018543243, | |
| "logps/chosen": -632.91064453125, | |
| "logps/rejected": -779.7465209960938, | |
| "loss": 0.2493, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.30229493975639343, | |
| "rewards/margins": 0.19277289509773254, | |
| "rewards/rejected": -0.49506789445877075, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.273656994094232e-07, | |
| "logits/chosen": 0.078878253698349, | |
| "logits/rejected": 0.32078155875205994, | |
| "logps/chosen": -629.8251342773438, | |
| "logps/rejected": -792.5658569335938, | |
| "loss": 0.2593, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.29794004559516907, | |
| "rewards/margins": 0.21232935786247253, | |
| "rewards/rejected": -0.5102694034576416, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.418693324604082e-07, | |
| "logits/chosen": 0.17152568697929382, | |
| "logits/rejected": 0.2723831534385681, | |
| "logps/chosen": -686.8225708007812, | |
| "logps/rejected": -765.6764526367188, | |
| "loss": 0.2106, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.2921481132507324, | |
| "rewards/margins": 0.19477225840091705, | |
| "rewards/rejected": -0.48692041635513306, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.619234996325314e-07, | |
| "logits/chosen": 0.16223089396953583, | |
| "logits/rejected": 0.22344419360160828, | |
| "logps/chosen": -644.6939086914062, | |
| "logps/rejected": -796.0565185546875, | |
| "loss": 0.2251, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.29831987619400024, | |
| "rewards/margins": 0.2159034013748169, | |
| "rewards/rejected": -0.5142232179641724, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.877548160747768e-07, | |
| "logits/chosen": 0.14092543721199036, | |
| "logits/rejected": 0.26907533407211304, | |
| "logps/chosen": -665.1543579101562, | |
| "logps/rejected": -785.69091796875, | |
| "loss": 0.2293, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.3221438229084015, | |
| "rewards/margins": 0.16273698210716248, | |
| "rewards/rejected": -0.4848807752132416, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.195735209788528e-07, | |
| "logits/chosen": 0.12745890021324158, | |
| "logits/rejected": 0.16941148042678833, | |
| "logps/chosen": -604.4722290039062, | |
| "logps/rejected": -726.4368286132812, | |
| "loss": 0.228, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.2934458255767822, | |
| "rewards/margins": 0.16752712428569794, | |
| "rewards/rejected": -0.46097296476364136, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.5757288163336806e-07, | |
| "logits/chosen": 0.14064130187034607, | |
| "logits/rejected": 0.2834423780441284, | |
| "logps/chosen": -670.5109252929688, | |
| "logps/rejected": -754.9119873046875, | |
| "loss": 0.2279, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.29777050018310547, | |
| "rewards/margins": 0.1834956705570221, | |
| "rewards/rejected": -0.4812661111354828, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.019286455866981e-07, | |
| "logits/chosen": 0.16343867778778076, | |
| "logits/rejected": 0.23200741410255432, | |
| "logps/chosen": -653.3126220703125, | |
| "logps/rejected": -788.0684204101562, | |
| "loss": 0.2129, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.30145275592803955, | |
| "rewards/margins": 0.18838538229465485, | |
| "rewards/rejected": -0.4898381233215332, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.5279854247146703e-07, | |
| "logits/chosen": 0.0768275260925293, | |
| "logits/rejected": 0.261466920375824, | |
| "logps/chosen": -653.1698608398438, | |
| "logps/rejected": -781.791259765625, | |
| "loss": 0.2128, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.28429535031318665, | |
| "rewards/margins": 0.22277021408081055, | |
| "rewards/rejected": -0.5070655941963196, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.1032183690276754e-07, | |
| "logits/chosen": 0.160492941737175, | |
| "logits/rejected": 0.28358930349349976, | |
| "logps/chosen": -716.923583984375, | |
| "logps/rejected": -817.1744995117188, | |
| "loss": 0.2252, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.3216014504432678, | |
| "rewards/margins": 0.18884047865867615, | |
| "rewards/rejected": -0.5104418992996216, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.46189337174788e-08, | |
| "logits/chosen": 0.138215571641922, | |
| "logits/rejected": 0.28554660081863403, | |
| "logps/chosen": -634.0874633789062, | |
| "logps/rejected": -762.0772094726562, | |
| "loss": 0.204, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2693154215812683, | |
| "rewards/margins": 0.21418070793151855, | |
| "rewards/rejected": -0.48349618911743164, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 4.579103667367385e-08, | |
| "logits/chosen": 0.13104286789894104, | |
| "logits/rejected": 0.280353844165802, | |
| "logps/chosen": -666.0810546875, | |
| "logps/rejected": -752.5096435546875, | |
| "loss": 0.229, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.3018819987773895, | |
| "rewards/margins": 0.1678927093744278, | |
| "rewards/rejected": -0.4697747230529785, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.3919861577572924e-08, | |
| "logits/chosen": 0.15655803680419922, | |
| "logits/rejected": 0.32044172286987305, | |
| "logps/chosen": -707.17138671875, | |
| "logps/rejected": -806.7648315429688, | |
| "loss": 0.229, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.30425772070884705, | |
| "rewards/margins": 0.21442052721977234, | |
| "rewards/rejected": -0.5186783075332642, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.067404651211808e-09, | |
| "logits/chosen": 0.146881565451622, | |
| "logits/rejected": 0.22085854411125183, | |
| "logps/chosen": -632.6345825195312, | |
| "logps/rejected": -776.0171508789062, | |
| "loss": 0.2122, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.2965379059314728, | |
| "rewards/margins": 0.19161197543144226, | |
| "rewards/rejected": -0.48814982175827026, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.2757667974155896e-09, | |
| "logits/chosen": 0.18307064473628998, | |
| "logits/rejected": 0.2208767831325531, | |
| "logps/chosen": -612.895751953125, | |
| "logps/rejected": -768.5109252929688, | |
| "loss": 0.2581, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.2931486666202545, | |
| "rewards/margins": 0.18815748393535614, | |
| "rewards/rejected": -0.4813062250614166, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 656, | |
| "total_flos": 0.0, | |
| "train_loss": 0.20043009792159244, | |
| "train_runtime": 6753.4712, | |
| "train_samples_per_second": 3.11, | |
| "train_steps_per_second": 0.097 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 656, | |
| "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 | |
| } | |