| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1274, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0007849293563579278, | |
| "grad_norm": 4023.4741988682163, | |
| "learning_rate": 2.34375e-09, | |
| "logits/chosen": 5881.4375, | |
| "logits/rejected": 2834.66162109375, | |
| "logps/chosen": -257.5969543457031, | |
| "logps/rejected": -120.09489440917969, | |
| "loss": 333.5628, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.007849293563579277, | |
| "grad_norm": 4170.182058416646, | |
| "learning_rate": 2.34375e-08, | |
| "logits/chosen": 4946.056640625, | |
| "logits/rejected": 4332.40380859375, | |
| "logps/chosen": -248.79656982421875, | |
| "logps/rejected": -214.2803497314453, | |
| "loss": 268.0952, | |
| "rewards/accuracies": 0.42592594027519226, | |
| "rewards/chosen": 0.019624345004558563, | |
| "rewards/margins": 0.007904333993792534, | |
| "rewards/rejected": 0.011720011942088604, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.015698587127158554, | |
| "grad_norm": 3836.4614841026223, | |
| "learning_rate": 4.6875e-08, | |
| "logits/chosen": 6022.55419921875, | |
| "logits/rejected": 4842.4208984375, | |
| "logps/chosen": -265.8888244628906, | |
| "logps/rejected": -232.03372192382812, | |
| "loss": 289.1787, | |
| "rewards/accuracies": 0.5000000596046448, | |
| "rewards/chosen": 0.42448392510414124, | |
| "rewards/margins": 0.0060923038981854916, | |
| "rewards/rejected": 0.4183916449546814, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.023547880690737835, | |
| "grad_norm": 3394.9727381463454, | |
| "learning_rate": 7.03125e-08, | |
| "logits/chosen": 6028.314453125, | |
| "logits/rejected": 5035.12939453125, | |
| "logps/chosen": -308.2351989746094, | |
| "logps/rejected": -260.7220764160156, | |
| "loss": 279.4081, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 2.3741536140441895, | |
| "rewards/margins": 0.05959435552358627, | |
| "rewards/rejected": 2.3145594596862793, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03139717425431711, | |
| "grad_norm": 2564.1472181771474, | |
| "learning_rate": 9.375e-08, | |
| "logits/chosen": 5227.39990234375, | |
| "logits/rejected": 4331.17822265625, | |
| "logps/chosen": -257.78082275390625, | |
| "logps/rejected": -218.92984008789062, | |
| "loss": 272.9862, | |
| "rewards/accuracies": 0.5833333730697632, | |
| "rewards/chosen": 6.2394118309021, | |
| "rewards/margins": 0.10523471981287003, | |
| "rewards/rejected": 6.134177207946777, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03924646781789639, | |
| "grad_norm": 2561.808932474401, | |
| "learning_rate": 1.1718749999999999e-07, | |
| "logits/chosen": 6324.8388671875, | |
| "logits/rejected": 4942.587890625, | |
| "logps/chosen": -316.49151611328125, | |
| "logps/rejected": -249.837646484375, | |
| "loss": 269.9292, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": 9.618066787719727, | |
| "rewards/margins": 0.9058569073677063, | |
| "rewards/rejected": 8.712209701538086, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04709576138147567, | |
| "grad_norm": 2720.8960721090184, | |
| "learning_rate": 1.40625e-07, | |
| "logits/chosen": 5320.58837890625, | |
| "logits/rejected": 4447.7490234375, | |
| "logps/chosen": -259.4016418457031, | |
| "logps/rejected": -251.9221649169922, | |
| "loss": 272.7126, | |
| "rewards/accuracies": 0.6083333492279053, | |
| "rewards/chosen": 10.525445938110352, | |
| "rewards/margins": 0.256784588098526, | |
| "rewards/rejected": 10.268662452697754, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.054945054945054944, | |
| "grad_norm": 1800.0733293997093, | |
| "learning_rate": 1.640625e-07, | |
| "logits/chosen": 5012.52783203125, | |
| "logits/rejected": 4752.578125, | |
| "logps/chosen": -219.9415740966797, | |
| "logps/rejected": -219.6404266357422, | |
| "loss": 249.871, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 13.316383361816406, | |
| "rewards/margins": 0.3037797212600708, | |
| "rewards/rejected": 13.012603759765625, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06279434850863422, | |
| "grad_norm": 1579.665603355307, | |
| "learning_rate": 1.875e-07, | |
| "logits/chosen": 5556.5517578125, | |
| "logits/rejected": 5045.2451171875, | |
| "logps/chosen": -240.605224609375, | |
| "logps/rejected": -228.12490844726562, | |
| "loss": 257.1457, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 15.373222351074219, | |
| "rewards/margins": 1.1857852935791016, | |
| "rewards/rejected": 14.187434196472168, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0706436420722135, | |
| "grad_norm": 1952.866456922638, | |
| "learning_rate": 2.109375e-07, | |
| "logits/chosen": 5717.4521484375, | |
| "logits/rejected": 4918.970703125, | |
| "logps/chosen": -268.8674011230469, | |
| "logps/rejected": -228.5010223388672, | |
| "loss": 261.2282, | |
| "rewards/accuracies": 0.5666667222976685, | |
| "rewards/chosen": 16.323701858520508, | |
| "rewards/margins": 1.3582711219787598, | |
| "rewards/rejected": 14.965429306030273, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07849293563579278, | |
| "grad_norm": 1725.9812958933148, | |
| "learning_rate": 2.3437499999999998e-07, | |
| "logits/chosen": 5592.4326171875, | |
| "logits/rejected": 5391.06982421875, | |
| "logps/chosen": -261.3314514160156, | |
| "logps/rejected": -251.45028686523438, | |
| "loss": 265.2925, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 16.94363784790039, | |
| "rewards/margins": 0.2056407928466797, | |
| "rewards/rejected": 16.73799705505371, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08634222919937205, | |
| "grad_norm": 1962.7700041862627, | |
| "learning_rate": 2.5781249999999997e-07, | |
| "logits/chosen": 5994.12060546875, | |
| "logits/rejected": 4923.7216796875, | |
| "logps/chosen": -236.5362548828125, | |
| "logps/rejected": -225.5522918701172, | |
| "loss": 260.0169, | |
| "rewards/accuracies": 0.5916666388511658, | |
| "rewards/chosen": 20.60161781311035, | |
| "rewards/margins": 2.5409743785858154, | |
| "rewards/rejected": 18.06064224243164, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09419152276295134, | |
| "grad_norm": 1774.512958199661, | |
| "learning_rate": 2.8125e-07, | |
| "logits/chosen": 5856.7294921875, | |
| "logits/rejected": 4897.4931640625, | |
| "logps/chosen": -257.4624328613281, | |
| "logps/rejected": -249.49014282226562, | |
| "loss": 257.9953, | |
| "rewards/accuracies": 0.5666667222976685, | |
| "rewards/chosen": 22.124418258666992, | |
| "rewards/margins": 2.0123040676116943, | |
| "rewards/rejected": 20.112112045288086, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.10204081632653061, | |
| "grad_norm": 1608.467487017438, | |
| "learning_rate": 2.999977454977299e-07, | |
| "logits/chosen": 5785.07177734375, | |
| "logits/rejected": 5351.7685546875, | |
| "logps/chosen": -268.048095703125, | |
| "logps/rejected": -252.2095947265625, | |
| "loss": 249.7454, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 22.05913543701172, | |
| "rewards/margins": 0.8380325436592102, | |
| "rewards/rejected": 21.22110366821289, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.10989010989010989, | |
| "grad_norm": 1609.8700616545043, | |
| "learning_rate": 2.999188450339425e-07, | |
| "logits/chosen": 5623.4873046875, | |
| "logits/rejected": 4953.0654296875, | |
| "logps/chosen": -238.1974334716797, | |
| "logps/rejected": -229.2379608154297, | |
| "loss": 247.3781, | |
| "rewards/accuracies": 0.5250000357627869, | |
| "rewards/chosen": 22.044784545898438, | |
| "rewards/margins": 0.9268926382064819, | |
| "rewards/rejected": 21.11789321899414, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11773940345368916, | |
| "grad_norm": 2203.775605805379, | |
| "learning_rate": 2.997272872179022e-07, | |
| "logits/chosen": 5708.5322265625, | |
| "logits/rejected": 4781.1630859375, | |
| "logps/chosen": -254.2611083984375, | |
| "logps/rejected": -220.6115264892578, | |
| "loss": 241.5003, | |
| "rewards/accuracies": 0.5666666626930237, | |
| "rewards/chosen": 23.610925674438477, | |
| "rewards/margins": 2.3330581188201904, | |
| "rewards/rejected": 21.277868270874023, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12558869701726844, | |
| "grad_norm": 1669.1024166476127, | |
| "learning_rate": 2.994232159967983e-07, | |
| "logits/chosen": 5665.7265625, | |
| "logits/rejected": 5053.70751953125, | |
| "logps/chosen": -248.9014434814453, | |
| "logps/rejected": -244.55520629882812, | |
| "loss": 243.1269, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 24.662750244140625, | |
| "rewards/margins": 3.0847768783569336, | |
| "rewards/rejected": 21.577974319458008, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.13343799058084774, | |
| "grad_norm": 1921.358714315976, | |
| "learning_rate": 2.990068598666458e-07, | |
| "logits/chosen": 5912.36669921875, | |
| "logits/rejected": 5278.53857421875, | |
| "logps/chosen": -263.9734191894531, | |
| "logps/rejected": -247.239013671875, | |
| "loss": 247.4817, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 25.82412338256836, | |
| "rewards/margins": 4.0570549964904785, | |
| "rewards/rejected": 21.76706886291504, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.141287284144427, | |
| "grad_norm": 1653.480058492674, | |
| "learning_rate": 2.984785317005808e-07, | |
| "logits/chosen": 5192.95166015625, | |
| "logits/rejected": 4544.33251953125, | |
| "logps/chosen": -239.09371948242188, | |
| "logps/rejected": -224.21298217773438, | |
| "loss": 235.4099, | |
| "rewards/accuracies": 0.5000000596046448, | |
| "rewards/chosen": 24.350610733032227, | |
| "rewards/margins": 1.2966423034667969, | |
| "rewards/rejected": 23.053966522216797, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14913657770800628, | |
| "grad_norm": 1675.237836462438, | |
| "learning_rate": 2.978386285137506e-07, | |
| "logits/chosen": 5245.845703125, | |
| "logits/rejected": 4518.92822265625, | |
| "logps/chosen": -256.43487548828125, | |
| "logps/rejected": -237.6322479248047, | |
| "loss": 248.2757, | |
| "rewards/accuracies": 0.5333333015441895, | |
| "rewards/chosen": 30.84979820251465, | |
| "rewards/margins": 7.020622253417969, | |
| "rewards/rejected": 23.829174041748047, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15698587127158556, | |
| "grad_norm": 1680.7108471261229, | |
| "learning_rate": 2.9708763116497365e-07, | |
| "logits/chosen": 5170.1884765625, | |
| "logits/rejected": 4779.9521484375, | |
| "logps/chosen": -226.2421112060547, | |
| "logps/rejected": -242.6790008544922, | |
| "loss": 245.4435, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 25.007875442504883, | |
| "rewards/margins": 4.2132248878479, | |
| "rewards/rejected": 20.794649124145508, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16483516483516483, | |
| "grad_norm": 1715.7554274864947, | |
| "learning_rate": 2.962261039953964e-07, | |
| "logits/chosen": 4647.2177734375, | |
| "logits/rejected": 4153.85205078125, | |
| "logps/chosen": -198.57461547851562, | |
| "logps/rejected": -198.82418823242188, | |
| "loss": 237.449, | |
| "rewards/accuracies": 0.5916666984558105, | |
| "rewards/chosen": 26.1383056640625, | |
| "rewards/margins": 2.5017244815826416, | |
| "rewards/rejected": 23.636577606201172, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.1726844583987441, | |
| "grad_norm": 1708.034439659388, | |
| "learning_rate": 2.952546944044155e-07, | |
| "logits/chosen": 5712.70458984375, | |
| "logits/rejected": 4872.09716796875, | |
| "logps/chosen": -250.9006805419922, | |
| "logps/rejected": -208.9671630859375, | |
| "loss": 242.5153, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 31.732202529907227, | |
| "rewards/margins": 7.158880710601807, | |
| "rewards/rejected": 24.573322296142578, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.18053375196232338, | |
| "grad_norm": 1640.0282387282753, | |
| "learning_rate": 2.941741323631859e-07, | |
| "logits/chosen": 5266.2744140625, | |
| "logits/rejected": 4779.90625, | |
| "logps/chosen": -237.95388793945312, | |
| "logps/rejected": -227.10824584960938, | |
| "loss": 242.998, | |
| "rewards/accuracies": 0.5666666626930237, | |
| "rewards/chosen": 32.3233528137207, | |
| "rewards/margins": 1.4093796014785767, | |
| "rewards/rejected": 30.913970947265625, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.18838304552590268, | |
| "grad_norm": 1593.5062293658036, | |
| "learning_rate": 2.929852298660803e-07, | |
| "logits/chosen": 4794.1279296875, | |
| "logits/rejected": 3836.48876953125, | |
| "logps/chosen": -211.5415496826172, | |
| "logps/rejected": -191.9940643310547, | |
| "loss": 242.3375, | |
| "rewards/accuracies": 0.6666666865348816, | |
| "rewards/chosen": 28.15199851989746, | |
| "rewards/margins": 7.219785213470459, | |
| "rewards/rejected": 20.932212829589844, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19623233908948196, | |
| "grad_norm": 1600.8546278975482, | |
| "learning_rate": 2.9168888032051097e-07, | |
| "logits/chosen": 5557.8974609375, | |
| "logits/rejected": 4967.7939453125, | |
| "logps/chosen": -210.0264434814453, | |
| "logps/rejected": -229.9081268310547, | |
| "loss": 222.3591, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 25.621007919311523, | |
| "rewards/margins": 1.1370152235031128, | |
| "rewards/rejected": 24.483993530273438, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.20408163265306123, | |
| "grad_norm": 7468.100775987575, | |
| "learning_rate": 2.9028605787557423e-07, | |
| "logits/chosen": 5713.48388671875, | |
| "logits/rejected": 4631.87060546875, | |
| "logps/chosen": -252.72000122070312, | |
| "logps/rejected": -211.0446014404297, | |
| "loss": 261.3835, | |
| "rewards/accuracies": 0.5666666626930237, | |
| "rewards/chosen": 32.03266143798828, | |
| "rewards/margins": 4.5531005859375, | |
| "rewards/rejected": 27.479562759399414, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2119309262166405, | |
| "grad_norm": 1515.290182279751, | |
| "learning_rate": 2.8877781669002034e-07, | |
| "logits/chosen": 5698.69970703125, | |
| "logits/rejected": 5504.67578125, | |
| "logps/chosen": -242.6372833251953, | |
| "logps/rejected": -234.9390869140625, | |
| "loss": 232.0639, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 31.871328353881836, | |
| "rewards/margins": -0.017942238599061966, | |
| "rewards/rejected": 31.889266967773438, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.21978021978021978, | |
| "grad_norm": 1819.80541525978, | |
| "learning_rate": 2.871652901401002e-07, | |
| "logits/chosen": 4851.04052734375, | |
| "logits/rejected": 4222.8994140625, | |
| "logps/chosen": -203.6181640625, | |
| "logps/rejected": -209.3738555908203, | |
| "loss": 231.3454, | |
| "rewards/accuracies": 0.5666667222976685, | |
| "rewards/chosen": 25.097145080566406, | |
| "rewards/margins": 0.06098451465368271, | |
| "rewards/rejected": 25.036161422729492, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22762951334379905, | |
| "grad_norm": 1875.5124653133516, | |
| "learning_rate": 2.854496899678834e-07, | |
| "logits/chosen": 6743.14990234375, | |
| "logits/rejected": 5003.833984375, | |
| "logps/chosen": -291.8956604003906, | |
| "logps/rejected": -228.0226287841797, | |
| "loss": 264.4877, | |
| "rewards/accuracies": 0.6166666746139526, | |
| "rewards/chosen": 32.894901275634766, | |
| "rewards/margins": 3.151154041290283, | |
| "rewards/rejected": 29.74374771118164, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.23547880690737832, | |
| "grad_norm": 1670.212642757904, | |
| "learning_rate": 2.8363230537068793e-07, | |
| "logits/chosen": 4895.9951171875, | |
| "logits/rejected": 4853.8779296875, | |
| "logps/chosen": -206.589111328125, | |
| "logps/rejected": -210.8595733642578, | |
| "loss": 244.6626, | |
| "rewards/accuracies": 0.45833340287208557, | |
| "rewards/chosen": 30.15146827697754, | |
| "rewards/margins": -2.91595458984375, | |
| "rewards/rejected": 33.067420959472656, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.24332810047095763, | |
| "grad_norm": 1609.4315518521369, | |
| "learning_rate": 2.817145020323055e-07, | |
| "logits/chosen": 5462.11181640625, | |
| "logits/rejected": 4891.4931640625, | |
| "logps/chosen": -225.6922149658203, | |
| "logps/rejected": -216.83224487304688, | |
| "loss": 232.9871, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 29.976367950439453, | |
| "rewards/margins": 5.142911911010742, | |
| "rewards/rejected": 24.833454132080078, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.25117739403453687, | |
| "grad_norm": 1702.9807268880163, | |
| "learning_rate": 2.7969772109675106e-07, | |
| "logits/chosen": 5495.5859375, | |
| "logits/rejected": 4778.9248046875, | |
| "logps/chosen": -246.85824584960938, | |
| "logps/rejected": -245.73159790039062, | |
| "loss": 256.5588, | |
| "rewards/accuracies": 0.5583333969116211, | |
| "rewards/chosen": 30.428979873657227, | |
| "rewards/margins": 4.235407829284668, | |
| "rewards/rejected": 26.19357681274414, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.25902668759811615, | |
| "grad_norm": 1341.9973700905261, | |
| "learning_rate": 2.775834780853069e-07, | |
| "logits/chosen": 5346.3876953125, | |
| "logits/rejected": 4866.0966796875, | |
| "logps/chosen": -239.4844512939453, | |
| "logps/rejected": -220.07534790039062, | |
| "loss": 241.643, | |
| "rewards/accuracies": 0.5416666865348816, | |
| "rewards/chosen": 30.613622665405273, | |
| "rewards/margins": 1.2968807220458984, | |
| "rewards/rejected": 29.31673812866211, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2668759811616955, | |
| "grad_norm": 1524.4179101852199, | |
| "learning_rate": 2.75373361757676e-07, | |
| "logits/chosen": 5587.9873046875, | |
| "logits/rejected": 5700.83642578125, | |
| "logps/chosen": -238.8589630126953, | |
| "logps/rejected": -268.36297607421875, | |
| "loss": 245.3856, | |
| "rewards/accuracies": 0.4583333134651184, | |
| "rewards/chosen": 29.224136352539062, | |
| "rewards/margins": -4.2943596839904785, | |
| "rewards/rejected": 33.51850128173828, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.27472527472527475, | |
| "grad_norm": 1629.7476503449539, | |
| "learning_rate": 2.730690329180992e-07, | |
| "logits/chosen": 5857.9404296875, | |
| "logits/rejected": 5350.62548828125, | |
| "logps/chosen": -241.11770629882812, | |
| "logps/rejected": -252.30581665039062, | |
| "loss": 251.1805, | |
| "rewards/accuracies": 0.533333420753479, | |
| "rewards/chosen": 31.790069580078125, | |
| "rewards/margins": 0.5973652005195618, | |
| "rewards/rejected": 31.192707061767578, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.282574568288854, | |
| "grad_norm": 2038.4672957771452, | |
| "learning_rate": 2.7067222316733527e-07, | |
| "logits/chosen": 5389.8154296875, | |
| "logits/rejected": 5303.53515625, | |
| "logps/chosen": -217.7191162109375, | |
| "logps/rejected": -245.5166473388672, | |
| "loss": 256.7446, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 25.96584701538086, | |
| "rewards/margins": -2.4905505180358887, | |
| "rewards/rejected": 28.456396102905273, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2904238618524333, | |
| "grad_norm": 1758.147691322977, | |
| "learning_rate": 2.6818473360143916e-07, | |
| "logits/chosen": 4950.9892578125, | |
| "logits/rejected": 4674.2900390625, | |
| "logps/chosen": -227.2458038330078, | |
| "logps/rejected": -207.7462921142578, | |
| "loss": 243.6053, | |
| "rewards/accuracies": 0.5833333730697632, | |
| "rewards/chosen": 31.656543731689453, | |
| "rewards/margins": 0.9851619005203247, | |
| "rewards/rejected": 30.6713809967041, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.29827315541601257, | |
| "grad_norm": 1498.3192440466703, | |
| "learning_rate": 2.6560843345831893e-07, | |
| "logits/chosen": 5929.0947265625, | |
| "logits/rejected": 4928.9697265625, | |
| "logps/chosen": -275.950927734375, | |
| "logps/rejected": -225.73373413085938, | |
| "loss": 243.0677, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 32.026023864746094, | |
| "rewards/margins": 1.8513380289077759, | |
| "rewards/rejected": 30.174686431884766, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.30612244897959184, | |
| "grad_norm": 1796.2377577638977, | |
| "learning_rate": 2.62945258713086e-07, | |
| "logits/chosen": 6053.13916015625, | |
| "logits/rejected": 4656.35595703125, | |
| "logps/chosen": -260.4866027832031, | |
| "logps/rejected": -217.1722412109375, | |
| "loss": 249.0038, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 31.03982162475586, | |
| "rewards/margins": 2.7203409671783447, | |
| "rewards/rejected": 28.31947898864746, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3139717425431711, | |
| "grad_norm": 1772.4830395099723, | |
| "learning_rate": 2.601972106232561e-07, | |
| "logits/chosen": 5182.9228515625, | |
| "logits/rejected": 5086.61865234375, | |
| "logps/chosen": -233.95730590820312, | |
| "logps/rejected": -235.7310791015625, | |
| "loss": 232.4267, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 40.660152435302734, | |
| "rewards/margins": -10.21926498413086, | |
| "rewards/rejected": 50.87942123413086, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3218210361067504, | |
| "grad_norm": 1931.2099299747044, | |
| "learning_rate": 2.5736635422489343e-07, | |
| "logits/chosen": 5618.06005859375, | |
| "logits/rejected": 5055.83203125, | |
| "logps/chosen": -257.80230712890625, | |
| "logps/rejected": -252.75057983398438, | |
| "loss": 242.3262, | |
| "rewards/accuracies": 0.5583332777023315, | |
| "rewards/chosen": 34.06652069091797, | |
| "rewards/margins": 4.192555904388428, | |
| "rewards/rejected": 29.873971939086914, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.32967032967032966, | |
| "grad_norm": 1863.6742216394089, | |
| "learning_rate": 2.5445481678082706e-07, | |
| "logits/chosen": 5285.50341796875, | |
| "logits/rejected": 4831.49755859375, | |
| "logps/chosen": -246.82455444335938, | |
| "logps/rejected": -220.78927612304688, | |
| "loss": 248.3231, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 39.06867218017578, | |
| "rewards/margins": 6.775533199310303, | |
| "rewards/rejected": 32.29314422607422, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.33751962323390894, | |
| "grad_norm": 1887.6941614294287, | |
| "learning_rate": 2.514647861821081e-07, | |
| "logits/chosen": 5075.509765625, | |
| "logits/rejected": 4261.77685546875, | |
| "logps/chosen": -220.1343536376953, | |
| "logps/rejected": -197.20022583007812, | |
| "loss": 248.9747, | |
| "rewards/accuracies": 0.5833333730697632, | |
| "rewards/chosen": 36.660560607910156, | |
| "rewards/margins": 8.8565092086792, | |
| "rewards/rejected": 27.804052352905273, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3453689167974882, | |
| "grad_norm": 1672.573803320896, | |
| "learning_rate": 2.483985093039065e-07, | |
| "logits/chosen": 5437.03271484375, | |
| "logits/rejected": 5404.0185546875, | |
| "logps/chosen": -222.5334930419922, | |
| "logps/rejected": -233.4775848388672, | |
| "loss": 245.6488, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 28.234506607055664, | |
| "rewards/margins": 1.496565818786621, | |
| "rewards/rejected": 26.737939834594727, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3532182103610675, | |
| "grad_norm": 2019.5173676452935, | |
| "learning_rate": 2.4525829031708404e-07, | |
| "logits/chosen": 6254.5849609375, | |
| "logits/rejected": 4548.39453125, | |
| "logps/chosen": -256.54510498046875, | |
| "logps/rejected": -199.13783264160156, | |
| "loss": 244.7201, | |
| "rewards/accuracies": 0.5916666984558105, | |
| "rewards/chosen": 30.51175308227539, | |
| "rewards/margins": 3.650390148162842, | |
| "rewards/rejected": 26.86136245727539, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.36106750392464676, | |
| "grad_norm": 1928.609840072256, | |
| "learning_rate": 2.420464889567126e-07, | |
| "logits/chosen": 6496.107421875, | |
| "logits/rejected": 5095.75439453125, | |
| "logps/chosen": -300.14495849609375, | |
| "logps/rejected": -231.71176147460938, | |
| "loss": 258.4863, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 32.566139221191406, | |
| "rewards/margins": -0.3735731244087219, | |
| "rewards/rejected": 32.93971633911133, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.36891679748822603, | |
| "grad_norm": 1732.6013846090254, | |
| "learning_rate": 2.3876551874883775e-07, | |
| "logits/chosen": 5562.2060546875, | |
| "logits/rejected": 4931.4013671875, | |
| "logps/chosen": -254.93942260742188, | |
| "logps/rejected": -242.7599639892578, | |
| "loss": 252.0907, | |
| "rewards/accuracies": 0.5166667103767395, | |
| "rewards/chosen": 32.846343994140625, | |
| "rewards/margins": -2.4924960136413574, | |
| "rewards/rejected": 35.338836669921875, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.37676609105180536, | |
| "grad_norm": 1795.3299642685095, | |
| "learning_rate": 2.354178451968214e-07, | |
| "logits/chosen": 5991.8564453125, | |
| "logits/rejected": 5220.8095703125, | |
| "logps/chosen": -249.9952392578125, | |
| "logps/rejected": -215.29141235351562, | |
| "loss": 230.2996, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": 29.81136703491211, | |
| "rewards/margins": -1.69051992893219, | |
| "rewards/rejected": 31.501888275146484, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 1803.5827886353418, | |
| "learning_rate": 2.3200598392862542e-07, | |
| "logits/chosen": 5602.65673828125, | |
| "logits/rejected": 5416.28955078125, | |
| "logps/chosen": -211.1875457763672, | |
| "logps/rejected": -241.5133819580078, | |
| "loss": 239.6045, | |
| "rewards/accuracies": 0.46666669845581055, | |
| "rewards/chosen": 29.753490447998047, | |
| "rewards/margins": -1.332529067993164, | |
| "rewards/rejected": 31.08602523803711, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3924646781789639, | |
| "grad_norm": 1658.6097442809094, | |
| "learning_rate": 2.2853249880642897e-07, | |
| "logits/chosen": 6143.66796875, | |
| "logits/rejected": 4787.69287109375, | |
| "logps/chosen": -264.48699951171875, | |
| "logps/rejected": -238.17813110351562, | |
| "loss": 257.8732, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": 34.00247573852539, | |
| "rewards/margins": 7.747855186462402, | |
| "rewards/rejected": 26.254619598388672, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.4003139717425432, | |
| "grad_norm": 1514.6883220373581, | |
| "learning_rate": 2.25e-07, | |
| "logits/chosen": 4914.34130859375, | |
| "logits/rejected": 4383.74365234375, | |
| "logps/chosen": -223.74484252929688, | |
| "logps/rejected": -202.2457733154297, | |
| "loss": 226.7803, | |
| "rewards/accuracies": 0.5833333730697632, | |
| "rewards/chosen": 33.11497116088867, | |
| "rewards/margins": 6.243075370788574, | |
| "rewards/rejected": 26.871898651123047, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 1665.9109440249433, | |
| "learning_rate": 2.2141114202526867e-07, | |
| "logits/chosen": 5225.7666015625, | |
| "logits/rejected": 4860.1201171875, | |
| "logps/chosen": -235.2001953125, | |
| "logps/rejected": -234.8746795654297, | |
| "loss": 236.1752, | |
| "rewards/accuracies": 0.5166667103767395, | |
| "rewards/chosen": 30.972644805908203, | |
| "rewards/margins": -3.2789902687072754, | |
| "rewards/rejected": 34.25163650512695, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.41601255886970173, | |
| "grad_norm": 1690.4709243018706, | |
| "learning_rate": 2.1776862174957646e-07, | |
| "logits/chosen": 5882.06787109375, | |
| "logits/rejected": 4868.77099609375, | |
| "logps/chosen": -251.9287109375, | |
| "logps/rejected": -235.5530242919922, | |
| "loss": 250.0482, | |
| "rewards/accuracies": 0.5916666388511658, | |
| "rewards/chosen": 32.3634033203125, | |
| "rewards/margins": 3.845641613006592, | |
| "rewards/rejected": 28.51776123046875, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.423861852433281, | |
| "grad_norm": 1870.423608333694, | |
| "learning_rate": 2.1407517636510023e-07, | |
| "logits/chosen": 5593.61669921875, | |
| "logits/rejected": 5044.35888671875, | |
| "logps/chosen": -246.6546173095703, | |
| "logps/rejected": -225.1569366455078, | |
| "loss": 254.2598, | |
| "rewards/accuracies": 0.4583333134651184, | |
| "rewards/chosen": 30.72903823852539, | |
| "rewards/margins": 0.4885719418525696, | |
| "rewards/rejected": 30.240467071533203, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4317111459968603, | |
| "grad_norm": 1890.648426124477, | |
| "learning_rate": 2.1033358133197376e-07, | |
| "logits/chosen": 5424.3427734375, | |
| "logits/rejected": 5095.46923828125, | |
| "logps/chosen": -251.3126983642578, | |
| "logps/rejected": -264.8800964355469, | |
| "loss": 247.7755, | |
| "rewards/accuracies": 0.5583333969116211, | |
| "rewards/chosen": 30.0239315032959, | |
| "rewards/margins": 3.7207419872283936, | |
| "rewards/rejected": 26.303192138671875, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.43956043956043955, | |
| "grad_norm": 1808.8716538905564, | |
| "learning_rate": 2.0654664829265295e-07, | |
| "logits/chosen": 5456.46044921875, | |
| "logits/rejected": 5125.83642578125, | |
| "logps/chosen": -240.5121612548828, | |
| "logps/rejected": -231.93063354492188, | |
| "loss": 241.07, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 35.61545181274414, | |
| "rewards/margins": 0.7679373025894165, | |
| "rewards/rejected": 34.84751892089844, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4474097331240188, | |
| "grad_norm": 1616.196740293062, | |
| "learning_rate": 2.027172229590914e-07, | |
| "logits/chosen": 5779.5791015625, | |
| "logits/rejected": 5372.1708984375, | |
| "logps/chosen": -282.7325439453125, | |
| "logps/rejected": -262.868896484375, | |
| "loss": 233.9623, | |
| "rewards/accuracies": 0.5666666626930237, | |
| "rewards/chosen": 31.83243751525879, | |
| "rewards/margins": 2.7603840827941895, | |
| "rewards/rejected": 29.07205581665039, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.4552590266875981, | |
| "grad_norm": 1706.4244996250075, | |
| "learning_rate": 1.9884818297431403e-07, | |
| "logits/chosen": 5726.626953125, | |
| "logits/rejected": 5229.80322265625, | |
| "logps/chosen": -249.5810089111328, | |
| "logps/rejected": -252.15927124023438, | |
| "loss": 259.9974, | |
| "rewards/accuracies": 0.5666667222976685, | |
| "rewards/chosen": 37.66581726074219, | |
| "rewards/margins": 4.974664211273193, | |
| "rewards/rejected": 32.6911506652832, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4631083202511774, | |
| "grad_norm": 1838.0616973374604, | |
| "learning_rate": 1.9494243574999645e-07, | |
| "logits/chosen": 6241.70166015625, | |
| "logits/rejected": 5314.6220703125, | |
| "logps/chosen": -271.5558776855469, | |
| "logps/rejected": -215.36904907226562, | |
| "loss": 257.9668, | |
| "rewards/accuracies": 0.5583333373069763, | |
| "rewards/chosen": 34.5975341796875, | |
| "rewards/margins": 4.126400947570801, | |
| "rewards/rejected": 30.47113037109375, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.47095761381475665, | |
| "grad_norm": 1775.81515330381, | |
| "learning_rate": 1.9100291628167416e-07, | |
| "logits/chosen": 5888.79833984375, | |
| "logits/rejected": 5069.1513671875, | |
| "logps/chosen": -213.1555938720703, | |
| "logps/rejected": -223.5966796875, | |
| "loss": 233.7554, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 39.33544921875, | |
| "rewards/margins": 7.968379974365234, | |
| "rewards/rejected": 31.3670711517334, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.478806907378336, | |
| "grad_norm": 1552.3179583078756, | |
| "learning_rate": 1.870325849432238e-07, | |
| "logits/chosen": 5687.79052734375, | |
| "logits/rejected": 5030.01220703125, | |
| "logps/chosen": -241.0684356689453, | |
| "logps/rejected": -219.44580078125, | |
| "loss": 237.8548, | |
| "rewards/accuracies": 0.5583333373069763, | |
| "rewards/chosen": 31.7730655670166, | |
| "rewards/margins": 0.7271839380264282, | |
| "rewards/rejected": 31.045883178710938, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.48665620094191525, | |
| "grad_norm": 1711.0322266359663, | |
| "learning_rate": 1.8303442526227364e-07, | |
| "logits/chosen": 4855.2119140625, | |
| "logits/rejected": 4031.395263671875, | |
| "logps/chosen": -214.16616821289062, | |
| "logps/rejected": -187.9397735595703, | |
| "loss": 239.5005, | |
| "rewards/accuracies": 0.6083333492279053, | |
| "rewards/chosen": 33.65977478027344, | |
| "rewards/margins": 8.022300720214844, | |
| "rewards/rejected": 25.637475967407227, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.4945054945054945, | |
| "grad_norm": 1687.5431903848942, | |
| "learning_rate": 1.7901144167821507e-07, | |
| "logits/chosen": 6347.2724609375, | |
| "logits/rejected": 5476.73681640625, | |
| "logps/chosen": -282.24456787109375, | |
| "logps/rejected": -229.3341827392578, | |
| "loss": 249.354, | |
| "rewards/accuracies": 0.5166666507720947, | |
| "rewards/chosen": 34.467002868652344, | |
| "rewards/margins": -2.233703851699829, | |
| "rewards/rejected": 36.700706481933594, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.5023547880690737, | |
| "grad_norm": 1828.9272635886298, | |
| "learning_rate": 1.749666572845e-07, | |
| "logits/chosen": 6022.80517578125, | |
| "logits/rejected": 5191.70458984375, | |
| "logps/chosen": -245.8931884765625, | |
| "logps/rejected": -247.98025512695312, | |
| "loss": 238.2031, | |
| "rewards/accuracies": 0.5916666388511658, | |
| "rewards/chosen": 32.79724884033203, | |
| "rewards/margins": 2.936703681945801, | |
| "rewards/rejected": 29.860538482666016, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5102040816326531, | |
| "grad_norm": 1722.7656447865402, | |
| "learning_rate": 1.7090311155692034e-07, | |
| "logits/chosen": 5180.5146484375, | |
| "logits/rejected": 4280.607421875, | |
| "logps/chosen": -226.1409454345703, | |
| "logps/rejected": -195.9656982421875, | |
| "loss": 246.688, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": 31.397594451904297, | |
| "rewards/margins": 0.26145073771476746, | |
| "rewards/rejected": 31.136144638061523, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5180533751962323, | |
| "grad_norm": 1817.8924102466333, | |
| "learning_rate": 1.6682385806957695e-07, | |
| "logits/chosen": 5812.03369140625, | |
| "logits/rejected": 4713.640625, | |
| "logps/chosen": -250.41696166992188, | |
| "logps/rejected": -219.3455047607422, | |
| "loss": 245.3282, | |
| "rewards/accuracies": 0.5583333373069763, | |
| "rewards/chosen": 39.90130615234375, | |
| "rewards/margins": 6.975987434387207, | |
| "rewards/rejected": 32.925315856933594, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5259026687598116, | |
| "grad_norm": 1789.4848747476276, | |
| "learning_rate": 1.627319622002544e-07, | |
| "logits/chosen": 5810.23828125, | |
| "logits/rejected": 4819.130859375, | |
| "logps/chosen": -239.8245849609375, | |
| "logps/rejected": -218.93972778320312, | |
| "loss": 239.8734, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": 36.38592529296875, | |
| "rewards/margins": 4.236663341522217, | |
| "rewards/rejected": 32.149261474609375, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.533751962323391, | |
| "grad_norm": 1877.8981802256612, | |
| "learning_rate": 1.5863049882692565e-07, | |
| "logits/chosen": 5654.3876953125, | |
| "logits/rejected": 4770.8037109375, | |
| "logps/chosen": -246.2259063720703, | |
| "logps/rejected": -207.6519775390625, | |
| "loss": 244.4937, | |
| "rewards/accuracies": 0.5916666984558105, | |
| "rewards/chosen": 31.601959228515625, | |
| "rewards/margins": 4.248345375061035, | |
| "rewards/rejected": 27.353612899780273, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5416012558869702, | |
| "grad_norm": 1695.7674557595915, | |
| "learning_rate": 1.5452255001711795e-07, | |
| "logits/chosen": 5584.14501953125, | |
| "logits/rejected": 5180.7001953125, | |
| "logps/chosen": -239.70040893554688, | |
| "logps/rejected": -246.1269989013672, | |
| "loss": 248.5505, | |
| "rewards/accuracies": 0.5750000476837158, | |
| "rewards/chosen": 35.38867950439453, | |
| "rewards/margins": 4.056863307952881, | |
| "rewards/rejected": 31.331817626953125, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5494505494505495, | |
| "grad_norm": 1442.702974237139, | |
| "learning_rate": 1.50411202711876e-07, | |
| "logits/chosen": 5305.8359375, | |
| "logits/rejected": 5153.296875, | |
| "logps/chosen": -217.58108520507812, | |
| "logps/rejected": -214.3724822998047, | |
| "loss": 230.0359, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 32.53514862060547, | |
| "rewards/margins": 1.4560534954071045, | |
| "rewards/rejected": 31.0791015625, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5572998430141287, | |
| "grad_norm": 1739.7088494921825, | |
| "learning_rate": 1.4629954640606275e-07, | |
| "logits/chosen": 5288.1630859375, | |
| "logits/rejected": 5336.8583984375, | |
| "logps/chosen": -219.3708953857422, | |
| "logps/rejected": -231.584716796875, | |
| "loss": 232.7259, | |
| "rewards/accuracies": 0.5333333015441895, | |
| "rewards/chosen": 29.874980926513672, | |
| "rewards/margins": -2.6295037269592285, | |
| "rewards/rejected": 32.504486083984375, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.565149136577708, | |
| "grad_norm": 1827.5993338334715, | |
| "learning_rate": 1.4219067082674153e-07, | |
| "logits/chosen": 5675.33642578125, | |
| "logits/rejected": 4951.52587890625, | |
| "logps/chosen": -247.85867309570312, | |
| "logps/rejected": -226.98532104492188, | |
| "loss": 245.2734, | |
| "rewards/accuracies": 0.5833333730697632, | |
| "rewards/chosen": 36.906105041503906, | |
| "rewards/margins": 5.772969722747803, | |
| "rewards/rejected": 31.13313865661621, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5729984301412873, | |
| "grad_norm": 1792.0730055062295, | |
| "learning_rate": 1.3808766361138294e-07, | |
| "logits/chosen": 6143.78466796875, | |
| "logits/rejected": 4770.439453125, | |
| "logps/chosen": -255.16848754882812, | |
| "logps/rejected": -201.65060424804688, | |
| "loss": 245.6433, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 39.89057922363281, | |
| "rewards/margins": 1.2564189434051514, | |
| "rewards/rejected": 38.634159088134766, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5808477237048666, | |
| "grad_norm": 1566.9621023185236, | |
| "learning_rate": 1.3399360798764307e-07, | |
| "logits/chosen": 5103.2958984375, | |
| "logits/rejected": 4384.28369140625, | |
| "logps/chosen": -217.14840698242188, | |
| "logps/rejected": -186.41259765625, | |
| "loss": 225.6119, | |
| "rewards/accuracies": 0.5750000476837158, | |
| "rewards/chosen": 30.859649658203125, | |
| "rewards/margins": -0.28063496947288513, | |
| "rewards/rejected": 31.14028549194336, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5886970172684458, | |
| "grad_norm": 1843.2051918557183, | |
| "learning_rate": 1.2991158045645435e-07, | |
| "logits/chosen": 5532.2783203125, | |
| "logits/rejected": 5077.6689453125, | |
| "logps/chosen": -205.27560424804688, | |
| "logps/rejected": -190.50119018554688, | |
| "loss": 240.2889, | |
| "rewards/accuracies": 0.5583333969116211, | |
| "rewards/chosen": 34.98659133911133, | |
| "rewards/margins": 2.9378249645233154, | |
| "rewards/rejected": 32.04876708984375, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5965463108320251, | |
| "grad_norm": 1680.704944409726, | |
| "learning_rate": 1.2584464848017214e-07, | |
| "logits/chosen": 5844.38671875, | |
| "logits/rejected": 4681.17236328125, | |
| "logps/chosen": -229.312255859375, | |
| "logps/rejected": -200.0790252685547, | |
| "loss": 239.1803, | |
| "rewards/accuracies": 0.6750000715255737, | |
| "rewards/chosen": 34.04833221435547, | |
| "rewards/margins": -3.792478561401367, | |
| "rewards/rejected": 37.8408088684082, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.6043956043956044, | |
| "grad_norm": 1687.1339052837984, | |
| "learning_rate": 1.2179586817751263e-07, | |
| "logits/chosen": 6057.12060546875, | |
| "logits/rejected": 5267.056640625, | |
| "logps/chosen": -257.5902404785156, | |
| "logps/rejected": -250.58474731445312, | |
| "loss": 237.518, | |
| "rewards/accuracies": 0.5833333730697632, | |
| "rewards/chosen": 37.05469512939453, | |
| "rewards/margins": 2.841217517852783, | |
| "rewards/rejected": 34.213478088378906, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.6122448979591837, | |
| "grad_norm": 1590.9656817472576, | |
| "learning_rate": 1.177682820270159e-07, | |
| "logits/chosen": 5934.1083984375, | |
| "logits/rejected": 5675.3408203125, | |
| "logps/chosen": -240.993896484375, | |
| "logps/rejected": -267.52471923828125, | |
| "loss": 242.7917, | |
| "rewards/accuracies": 0.4833333492279053, | |
| "rewards/chosen": 42.1386833190918, | |
| "rewards/margins": -4.899778842926025, | |
| "rewards/rejected": 47.0384635925293, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.6200941915227629, | |
| "grad_norm": 1685.3500270449556, | |
| "learning_rate": 1.13764916580758e-07, | |
| "logits/chosen": 5502.9443359375, | |
| "logits/rejected": 4947.611328125, | |
| "logps/chosen": -221.35421752929688, | |
| "logps/rejected": -213.71725463867188, | |
| "loss": 234.2771, | |
| "rewards/accuracies": 0.5166667103767395, | |
| "rewards/chosen": 33.09716796875, | |
| "rewards/margins": -8.531310081481934, | |
| "rewards/rejected": 41.628475189208984, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6279434850863422, | |
| "grad_norm": 1770.3595364831465, | |
| "learning_rate": 1.0978878019003183e-07, | |
| "logits/chosen": 5524.69921875, | |
| "logits/rejected": 4883.42333984375, | |
| "logps/chosen": -240.7305450439453, | |
| "logps/rejected": -235.4129180908203, | |
| "loss": 253.7563, | |
| "rewards/accuracies": 0.5833333134651184, | |
| "rewards/chosen": 33.00588607788086, | |
| "rewards/margins": 1.9016128778457642, | |
| "rewards/rejected": 31.10427474975586, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6357927786499215, | |
| "grad_norm": 1872.5809465409866, | |
| "learning_rate": 1.0584286074470446e-07, | |
| "logits/chosen": 5082.62841796875, | |
| "logits/rejected": 4645.69384765625, | |
| "logps/chosen": -225.14041137695312, | |
| "logps/rejected": -244.2375030517578, | |
| "loss": 229.0616, | |
| "rewards/accuracies": 0.5500000715255737, | |
| "rewards/chosen": 48.689491271972656, | |
| "rewards/margins": 19.15165138244629, | |
| "rewards/rejected": 29.537841796875, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6436420722135008, | |
| "grad_norm": 1845.608987792677, | |
| "learning_rate": 1.0193012342795095e-07, | |
| "logits/chosen": 5539.86181640625, | |
| "logits/rejected": 4341.95556640625, | |
| "logps/chosen": -246.3572235107422, | |
| "logps/rejected": -188.48760986328125, | |
| "loss": 248.0096, | |
| "rewards/accuracies": 0.6166667342185974, | |
| "rewards/chosen": 36.71552276611328, | |
| "rewards/margins": 9.378055572509766, | |
| "rewards/rejected": 27.337467193603516, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.6514913657770801, | |
| "grad_norm": 1575.3273152954032, | |
| "learning_rate": 9.805350848805078e-08, | |
| "logits/chosen": 5490.1826171875, | |
| "logits/rejected": 4940.8955078125, | |
| "logps/chosen": -228.04708862304688, | |
| "logps/rejected": -214.98373413085938, | |
| "loss": 231.8448, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 35.40738296508789, | |
| "rewards/margins": 2.6058733463287354, | |
| "rewards/rejected": 32.80150604248047, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.6593406593406593, | |
| "grad_norm": 1580.4247014532164, | |
| "learning_rate": 9.421592902892184e-08, | |
| "logits/chosen": 6016.4072265625, | |
| "logits/rejected": 4991.06005859375, | |
| "logps/chosen": -249.4292755126953, | |
| "logps/rejected": -234.61740112304688, | |
| "loss": 247.4855, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 36.10396194458008, | |
| "rewards/margins": -3.4738247394561768, | |
| "rewards/rejected": 39.577781677246094, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6671899529042387, | |
| "grad_norm": 1865.8968721112744, | |
| "learning_rate": 9.042026882105215e-08, | |
| "logits/chosen": 5550.4228515625, | |
| "logits/rejected": 5054.94580078125, | |
| "logps/chosen": -238.27517700195312, | |
| "logps/rejected": -221.6426239013672, | |
| "loss": 238.2114, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": 34.266990661621094, | |
| "rewards/margins": 9.951603889465332, | |
| "rewards/rejected": 24.315387725830078, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6750392464678179, | |
| "grad_norm": 1813.7081724731129, | |
| "learning_rate": 8.666938013447484e-08, | |
| "logits/chosen": 6002.4140625, | |
| "logits/rejected": 5152.6474609375, | |
| "logps/chosen": -268.81427001953125, | |
| "logps/rejected": -223.5155792236328, | |
| "loss": 252.0845, | |
| "rewards/accuracies": 0.5500000715255737, | |
| "rewards/chosen": 34.054100036621094, | |
| "rewards/margins": 3.2645421028137207, | |
| "rewards/rejected": 30.7895565032959, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6828885400313972, | |
| "grad_norm": 1672.5038074416736, | |
| "learning_rate": 8.29660815954134e-08, | |
| "logits/chosen": 5968.4775390625, | |
| "logits/rejected": 4994.8232421875, | |
| "logps/chosen": -252.988525390625, | |
| "logps/rejected": -221.22079467773438, | |
| "loss": 219.2042, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 36.42396926879883, | |
| "rewards/margins": 3.4356703758239746, | |
| "rewards/rejected": 32.9882926940918, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6907378335949764, | |
| "grad_norm": 1598.733157082823, | |
| "learning_rate": 7.931315606820979e-08, | |
| "logits/chosen": 5265.46728515625, | |
| "logits/rejected": 5068.302734375, | |
| "logps/chosen": -228.0703582763672, | |
| "logps/rejected": -253.70602416992188, | |
| "loss": 251.9328, | |
| "rewards/accuracies": 0.4583333432674408, | |
| "rewards/chosen": 30.682369232177734, | |
| "rewards/margins": -3.919257640838623, | |
| "rewards/rejected": 34.601627349853516, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6985871271585558, | |
| "grad_norm": 1822.545090596171, | |
| "learning_rate": 7.571334856412503e-08, | |
| "logits/chosen": 6199.7998046875, | |
| "logits/rejected": 5089.50830078125, | |
| "logps/chosen": -277.77056884765625, | |
| "logps/rejected": -230.9704132080078, | |
| "loss": 244.6672, | |
| "rewards/accuracies": 0.5583333373069763, | |
| "rewards/chosen": 34.757911682128906, | |
| "rewards/margins": 3.6986000537872314, | |
| "rewards/rejected": 31.059314727783203, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.706436420722135, | |
| "grad_norm": 1452.1470256152888, | |
| "learning_rate": 7.216936417858549e-08, | |
| "logits/chosen": 5261.8720703125, | |
| "logits/rejected": 4918.20166015625, | |
| "logps/chosen": -222.0632781982422, | |
| "logps/rejected": -230.07382202148438, | |
| "loss": 213.7731, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 35.97819519042969, | |
| "rewards/margins": 1.6305198669433594, | |
| "rewards/rejected": 34.34767150878906, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 1712.6491776060527, | |
| "learning_rate": 6.868386605842379e-08, | |
| "logits/chosen": 5177.64501953125, | |
| "logits/rejected": 4977.2197265625, | |
| "logps/chosen": -223.8202362060547, | |
| "logps/rejected": -225.1481170654297, | |
| "loss": 250.5905, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 33.74716567993164, | |
| "rewards/margins": 0.9199082255363464, | |
| "rewards/rejected": 32.82725524902344, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.7221350078492935, | |
| "grad_norm": 1607.9423450639895, | |
| "learning_rate": 6.525947340064232e-08, | |
| "logits/chosen": 5736.06982421875, | |
| "logits/rejected": 4804.224609375, | |
| "logps/chosen": -250.36328125, | |
| "logps/rejected": -215.31600952148438, | |
| "loss": 252.9119, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 36.358436584472656, | |
| "rewards/margins": 6.72072696685791, | |
| "rewards/rejected": 29.637704849243164, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.7299843014128728, | |
| "grad_norm": 1792.4930975317452, | |
| "learning_rate": 6.189875948420276e-08, | |
| "logits/chosen": 5656.0654296875, | |
| "logits/rejected": 4335.3193359375, | |
| "logps/chosen": -253.36782836914062, | |
| "logps/rejected": -201.13198852539062, | |
| "loss": 250.1941, | |
| "rewards/accuracies": 0.5500000715255737, | |
| "rewards/chosen": 34.58575439453125, | |
| "rewards/margins": 2.7708325386047363, | |
| "rewards/rejected": 31.81492042541504, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7378335949764521, | |
| "grad_norm": 1642.8478335358122, | |
| "learning_rate": 5.86042497363215e-08, | |
| "logits/chosen": 5481.2041015625, | |
| "logits/rejected": 4845.47216796875, | |
| "logps/chosen": -234.171630859375, | |
| "logps/rejected": -221.5301055908203, | |
| "loss": 239.7153, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 30.694355010986328, | |
| "rewards/margins": 5.063797473907471, | |
| "rewards/rejected": 25.63055419921875, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.7456828885400314, | |
| "grad_norm": 1569.6687776333497, | |
| "learning_rate": 5.537841983472337e-08, | |
| "logits/chosen": 5657.04443359375, | |
| "logits/rejected": 5260.193359375, | |
| "logps/chosen": -233.35964965820312, | |
| "logps/rejected": -231.0517120361328, | |
| "loss": 219.1115, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 30.20047950744629, | |
| "rewards/margins": -1.5409538745880127, | |
| "rewards/rejected": 31.741436004638672, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7535321821036107, | |
| "grad_norm": 1802.594897053482, | |
| "learning_rate": 5.222369384727954e-08, | |
| "logits/chosen": 5512.94921875, | |
| "logits/rejected": 4971.03759765625, | |
| "logps/chosen": -231.75552368164062, | |
| "logps/rejected": -217.78457641601562, | |
| "loss": 248.6497, | |
| "rewards/accuracies": 0.6416666507720947, | |
| "rewards/chosen": 32.25802230834961, | |
| "rewards/margins": 3.7199158668518066, | |
| "rewards/rejected": 28.538110733032227, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7613814756671899, | |
| "grad_norm": 1698.3575427695469, | |
| "learning_rate": 4.914244241042832e-08, | |
| "logits/chosen": 5621.8330078125, | |
| "logits/rejected": 4765.7197265625, | |
| "logps/chosen": -254.90676879882812, | |
| "logps/rejected": -220.02420043945312, | |
| "loss": 241.1903, | |
| "rewards/accuracies": 0.6666666865348816, | |
| "rewards/chosen": 34.224639892578125, | |
| "rewards/margins": 7.9143571853637695, | |
| "rewards/rejected": 26.31027603149414, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 1883.942692921232, | |
| "learning_rate": 4.613698094774715e-08, | |
| "logits/chosen": 5332.82275390625, | |
| "logits/rejected": 4352.8369140625, | |
| "logps/chosen": -242.8639678955078, | |
| "logps/rejected": -205.5134735107422, | |
| "loss": 245.9166, | |
| "rewards/accuracies": 0.6000000834465027, | |
| "rewards/chosen": 33.77801513671875, | |
| "rewards/margins": 0.6043466329574585, | |
| "rewards/rejected": 33.173667907714844, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7770800627943485, | |
| "grad_norm": 1843.9512527418883, | |
| "learning_rate": 4.3209567930014746e-08, | |
| "logits/chosen": 5655.3056640625, | |
| "logits/rejected": 4790.775390625, | |
| "logps/chosen": -238.4309539794922, | |
| "logps/rejected": -207.7635040283203, | |
| "loss": 238.3098, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 29.676456451416016, | |
| "rewards/margins": 1.1553412675857544, | |
| "rewards/rejected": 28.5211124420166, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7849293563579278, | |
| "grad_norm": 1668.660950027523, | |
| "learning_rate": 4.036240317807059e-08, | |
| "logits/chosen": 5348.1474609375, | |
| "logits/rejected": 4566.65185546875, | |
| "logps/chosen": -226.5524444580078, | |
| "logps/rejected": -193.11289978027344, | |
| "loss": 266.3675, | |
| "rewards/accuracies": 0.5416666865348816, | |
| "rewards/chosen": 33.76911163330078, | |
| "rewards/margins": 1.206241250038147, | |
| "rewards/rejected": 32.56287384033203, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.792778649921507, | |
| "grad_norm": 1737.8131017899611, | |
| "learning_rate": 3.759762620974716e-08, | |
| "logits/chosen": 5526.9853515625, | |
| "logits/rejected": 4676.6064453125, | |
| "logps/chosen": -225.6645965576172, | |
| "logps/rejected": -203.42306518554688, | |
| "loss": 244.4621, | |
| "rewards/accuracies": 0.6083333492279053, | |
| "rewards/chosen": 33.750736236572266, | |
| "rewards/margins": 1.6586647033691406, | |
| "rewards/rejected": 32.09206771850586, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.8006279434850864, | |
| "grad_norm": 1597.1576733390832, | |
| "learning_rate": 3.4917314632117423e-08, | |
| "logits/chosen": 5630.4970703125, | |
| "logits/rejected": 5054.865234375, | |
| "logps/chosen": -261.37945556640625, | |
| "logps/rejected": -239.4257354736328, | |
| "loss": 234.6974, | |
| "rewards/accuracies": 0.5416667461395264, | |
| "rewards/chosen": 36.51713943481445, | |
| "rewards/margins": 2.3422932624816895, | |
| "rewards/rejected": 34.17484664916992, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.8084772370486656, | |
| "grad_norm": 1815.3732207546507, | |
| "learning_rate": 3.232348258026552e-08, | |
| "logits/chosen": 5996.50927734375, | |
| "logits/rejected": 5708.70947265625, | |
| "logps/chosen": -242.7357635498047, | |
| "logps/rejected": -245.0090789794922, | |
| "loss": 249.3523, | |
| "rewards/accuracies": 0.5416666865348816, | |
| "rewards/chosen": 35.06444549560547, | |
| "rewards/margins": 4.963265419006348, | |
| "rewards/rejected": 30.101181030273438, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.8163265306122449, | |
| "grad_norm": 1755.3258775145534, | |
| "learning_rate": 2.9818079203753654e-08, | |
| "logits/chosen": 5332.3662109375, | |
| "logits/rejected": 4912.1435546875, | |
| "logps/chosen": -231.0524139404297, | |
| "logps/rejected": -222.0134735107422, | |
| "loss": 233.431, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 34.19084167480469, | |
| "rewards/margins": 2.630544662475586, | |
| "rewards/rejected": 31.5602970123291, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.8241758241758241, | |
| "grad_norm": 1900.5440276236159, | |
| "learning_rate": 2.740298720192245e-08, | |
| "logits/chosen": 5384.9541015625, | |
| "logits/rejected": 4951.54541015625, | |
| "logps/chosen": -247.54208374023438, | |
| "logps/rejected": -229.23617553710938, | |
| "loss": 250.7183, | |
| "rewards/accuracies": 0.4583333432674408, | |
| "rewards/chosen": 33.991050720214844, | |
| "rewards/margins": -2.496851682662964, | |
| "rewards/rejected": 36.48790740966797, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.8320251177394035, | |
| "grad_norm": 1847.1557629487122, | |
| "learning_rate": 2.5080021409126655e-08, | |
| "logits/chosen": 6656.6982421875, | |
| "logits/rejected": 5280.37255859375, | |
| "logps/chosen": -272.25396728515625, | |
| "logps/rejected": -231.11996459960938, | |
| "loss": 247.2176, | |
| "rewards/accuracies": 0.5499999523162842, | |
| "rewards/chosen": 35.6686897277832, | |
| "rewards/margins": 2.253685235977173, | |
| "rewards/rejected": 33.41500473022461, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.8398744113029827, | |
| "grad_norm": 1736.7009211885536, | |
| "learning_rate": 2.2850927430967287e-08, | |
| "logits/chosen": 5553.9951171875, | |
| "logits/rejected": 5138.5400390625, | |
| "logps/chosen": -235.717529296875, | |
| "logps/rejected": -203.86489868164062, | |
| "loss": 238.1888, | |
| "rewards/accuracies": 0.5666667222976685, | |
| "rewards/chosen": 37.174827575683594, | |
| "rewards/margins": 3.8044097423553467, | |
| "rewards/rejected": 33.370418548583984, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.847723704866562, | |
| "grad_norm": 1704.3014296513616, | |
| "learning_rate": 2.0717380332546768e-08, | |
| "logits/chosen": 6091.14794921875, | |
| "logits/rejected": 4771.44482421875, | |
| "logps/chosen": -275.9549865722656, | |
| "logps/rejected": -210.57046508789062, | |
| "loss": 249.6065, | |
| "rewards/accuracies": 0.6083333492279053, | |
| "rewards/chosen": 36.53284454345703, | |
| "rewards/margins": 5.862208366394043, | |
| "rewards/rejected": 30.670639038085938, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8555729984301413, | |
| "grad_norm": 1716.8055083764252, | |
| "learning_rate": 1.8680983379731618e-08, | |
| "logits/chosen": 6181.90234375, | |
| "logits/rejected": 4690.3525390625, | |
| "logps/chosen": -237.79922485351562, | |
| "logps/rejected": -221.51358032226562, | |
| "loss": 232.9238, | |
| "rewards/accuracies": 0.6166666746139526, | |
| "rewards/chosen": 35.99847412109375, | |
| "rewards/margins": 7.843270301818848, | |
| "rewards/rejected": 28.155202865600586, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8634222919937206, | |
| "grad_norm": 1647.678912403429, | |
| "learning_rate": 1.6743266834369856e-08, | |
| "logits/chosen": 5940.05029296875, | |
| "logits/rejected": 4961.7763671875, | |
| "logps/chosen": -247.6129608154297, | |
| "logps/rejected": -199.68138122558594, | |
| "loss": 229.1985, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 35.60984802246094, | |
| "rewards/margins": 3.182173490524292, | |
| "rewards/rejected": 32.42768096923828, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.8712715855572999, | |
| "grad_norm": 1668.2635457420981, | |
| "learning_rate": 1.4905686804366853e-08, | |
| "logits/chosen": 6316.9775390625, | |
| "logits/rejected": 5142.0263671875, | |
| "logps/chosen": -273.46905517578125, | |
| "logps/rejected": -232.58212280273438, | |
| "loss": 253.3951, | |
| "rewards/accuracies": 0.5750000476837158, | |
| "rewards/chosen": 35.805320739746094, | |
| "rewards/margins": 5.666674613952637, | |
| "rewards/rejected": 30.138647079467773, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8791208791208791, | |
| "grad_norm": 1536.5689365344047, | |
| "learning_rate": 1.3169624149485037e-08, | |
| "logits/chosen": 6162.1533203125, | |
| "logits/rejected": 5069.86083984375, | |
| "logps/chosen": -253.84664916992188, | |
| "logps/rejected": -226.19973754882812, | |
| "loss": 239.2838, | |
| "rewards/accuracies": 0.5916666388511658, | |
| "rewards/chosen": 37.4914436340332, | |
| "rewards/margins": 4.20772647857666, | |
| "rewards/rejected": 33.283714294433594, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8869701726844584, | |
| "grad_norm": 1654.6894966459954, | |
| "learning_rate": 1.1536383443688968e-08, | |
| "logits/chosen": 5599.70849609375, | |
| "logits/rejected": 4468.1572265625, | |
| "logps/chosen": -242.8928680419922, | |
| "logps/rejected": -189.45855712890625, | |
| "loss": 244.5037, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 35.67714309692383, | |
| "rewards/margins": 0.5334984660148621, | |
| "rewards/rejected": 35.14364242553711, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8948194662480377, | |
| "grad_norm": 1684.4104185811714, | |
| "learning_rate": 1.0007191994816272e-08, | |
| "logits/chosen": 6150.08837890625, | |
| "logits/rejected": 5064.6552734375, | |
| "logps/chosen": -268.3207092285156, | |
| "logps/rejected": -230.5198974609375, | |
| "loss": 229.5084, | |
| "rewards/accuracies": 0.5583333373069763, | |
| "rewards/chosen": 37.118709564208984, | |
| "rewards/margins": 6.031697750091553, | |
| "rewards/rejected": 31.08701515197754, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.902668759811617, | |
| "grad_norm": 1744.2331724560595, | |
| "learning_rate": 8.583198922310031e-09, | |
| "logits/chosen": 5450.5439453125, | |
| "logits/rejected": 5070.1357421875, | |
| "logps/chosen": -235.63320922851562, | |
| "logps/rejected": -230.82260131835938, | |
| "loss": 248.6159, | |
| "rewards/accuracies": 0.5166667103767395, | |
| "rewards/chosen": 31.562170028686523, | |
| "rewards/margins": -1.0857709646224976, | |
| "rewards/rejected": 32.6479377746582, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.9105180533751962, | |
| "grad_norm": 1622.709500273534, | |
| "learning_rate": 7.2654742937067916e-09, | |
| "logits/chosen": 5609.70361328125, | |
| "logits/rejected": 4495.67236328125, | |
| "logps/chosen": -235.7538299560547, | |
| "logps/rejected": -203.3125762939453, | |
| "loss": 223.5192, | |
| "rewards/accuracies": 0.5416666865348816, | |
| "rewards/chosen": 34.924049377441406, | |
| "rewards/margins": 4.218123435974121, | |
| "rewards/rejected": 30.7059268951416, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.9183673469387755, | |
| "grad_norm": 1728.0725738021401, | |
| "learning_rate": 6.0550083205280325e-09, | |
| "logits/chosen": 6132.37890625, | |
| "logits/rejected": 5953.0400390625, | |
| "logps/chosen": -251.34353637695312, | |
| "logps/rejected": -256.0439758300781, | |
| "loss": 242.0468, | |
| "rewards/accuracies": 0.5583333373069763, | |
| "rewards/chosen": 40.78349685668945, | |
| "rewards/margins": -6.980599880218506, | |
| "rewards/rejected": 47.76409912109375, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.9262166405023547, | |
| "grad_norm": 1802.5836418725428, | |
| "learning_rate": 4.952710614180567e-09, | |
| "logits/chosen": 5429.2578125, | |
| "logits/rejected": 4973.4560546875, | |
| "logps/chosen": -246.8389129638672, | |
| "logps/rejected": -225.2523651123047, | |
| "loss": 235.8803, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": 32.11265563964844, | |
| "rewards/margins": -1.4480606317520142, | |
| "rewards/rejected": 33.56071472167969, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.9340659340659341, | |
| "grad_norm": 1611.1865695761599, | |
| "learning_rate": 3.959409502423744e-09, | |
| "logits/chosen": 5891.9716796875, | |
| "logits/rejected": 5187.6328125, | |
| "logps/chosen": -256.4589538574219, | |
| "logps/rejected": -226.36105346679688, | |
| "loss": 231.6442, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 35.85973358154297, | |
| "rewards/margins": -1.6157242059707642, | |
| "rewards/rejected": 37.47545623779297, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.9419152276295133, | |
| "grad_norm": 1746.3400057259223, | |
| "learning_rate": 3.075851406917995e-09, | |
| "logits/chosen": 6081.0791015625, | |
| "logits/rejected": 5227.61181640625, | |
| "logps/chosen": -262.1074523925781, | |
| "logps/rejected": -227.6859130859375, | |
| "loss": 245.4175, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 42.64065933227539, | |
| "rewards/margins": 12.17762279510498, | |
| "rewards/rejected": 30.463031768798828, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9497645211930926, | |
| "grad_norm": 1567.572615006209, | |
| "learning_rate": 2.3027002823221064e-09, | |
| "logits/chosen": 5693.5810546875, | |
| "logits/rejected": 5021.91845703125, | |
| "logps/chosen": -245.7609405517578, | |
| "logps/rejected": -225.04501342773438, | |
| "loss": 258.8579, | |
| "rewards/accuracies": 0.5916666388511658, | |
| "rewards/chosen": 34.04198455810547, | |
| "rewards/margins": 6.770219326019287, | |
| "rewards/rejected": 27.271770477294922, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.957613814756672, | |
| "grad_norm": 1666.3147747639978, | |
| "learning_rate": 1.6405371173608073e-09, | |
| "logits/chosen": 7103.5, | |
| "logits/rejected": 5285.4990234375, | |
| "logps/chosen": -304.9707336425781, | |
| "logps/rejected": -253.6177520751953, | |
| "loss": 256.5355, | |
| "rewards/accuracies": 0.6166666746139526, | |
| "rewards/chosen": 41.44480514526367, | |
| "rewards/margins": 9.763620376586914, | |
| "rewards/rejected": 31.681188583374023, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9654631083202512, | |
| "grad_norm": 1808.4108039408945, | |
| "learning_rate": 1.0898594982375387e-09, | |
| "logits/chosen": 6319.11767578125, | |
| "logits/rejected": 5047.662109375, | |
| "logps/chosen": -267.2971496582031, | |
| "logps/rejected": -221.73751831054688, | |
| "loss": 245.7213, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 41.058284759521484, | |
| "rewards/margins": -2.2646431922912598, | |
| "rewards/rejected": 43.32292938232422, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9733124018838305, | |
| "grad_norm": 1776.4568830478072, | |
| "learning_rate": 6.510812347205185e-10, | |
| "logits/chosen": 5447.41796875, | |
| "logits/rejected": 5032.95947265625, | |
| "logps/chosen": -237.95974731445312, | |
| "logps/rejected": -242.78213500976562, | |
| "loss": 252.7457, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 34.99381637573242, | |
| "rewards/margins": 1.5597941875457764, | |
| "rewards/rejected": 33.434024810791016, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.9811616954474097, | |
| "grad_norm": 1726.2131253075509, | |
| "learning_rate": 3.245320491830683e-10, | |
| "logits/chosen": 5544.4306640625, | |
| "logits/rejected": 4712.3466796875, | |
| "logps/chosen": -236.2392578125, | |
| "logps/rejected": -195.9231414794922, | |
| "loss": 235.8322, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": 34.199363708496094, | |
| "rewards/margins": 6.769803047180176, | |
| "rewards/rejected": 27.4295597076416, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.989010989010989, | |
| "grad_norm": 1631.723364046293, | |
| "learning_rate": 1.1045732883197545e-10, | |
| "logits/chosen": 5248.7646484375, | |
| "logits/rejected": 5007.3310546875, | |
| "logps/chosen": -237.17166137695312, | |
| "logps/rejected": -224.19943237304688, | |
| "loss": 247.723, | |
| "rewards/accuracies": 0.49166664481163025, | |
| "rewards/chosen": 29.48110008239746, | |
| "rewards/margins": -4.025880336761475, | |
| "rewards/rejected": 33.506980895996094, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9968602825745683, | |
| "grad_norm": 1809.917889750108, | |
| "learning_rate": 9.017941309813482e-12, | |
| "logits/chosen": 5500.5634765625, | |
| "logits/rejected": 4507.23779296875, | |
| "logps/chosen": -229.42147827148438, | |
| "logps/rejected": -194.06491088867188, | |
| "loss": 248.3556, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 32.052024841308594, | |
| "rewards/margins": -2.0413379669189453, | |
| "rewards/rejected": 34.09336471557617, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1274, | |
| "total_flos": 0.0, | |
| "train_loss": 245.25187364515367, | |
| "train_runtime": 13775.8625, | |
| "train_samples_per_second": 4.438, | |
| "train_steps_per_second": 0.092 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1274, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 2000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": false, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 6, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |