| { | |
| "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": 3445.5462870980487, | |
| "learning_rate": 2.34375e-09, | |
| "logits/chosen": 5881.4375, | |
| "logits/rejected": 2834.66162109375, | |
| "logps/chosen": -257.5969543457031, | |
| "logps/rejected": -120.09489440917969, | |
| "loss": 218.9473, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.007849293563579277, | |
| "grad_norm": 3743.042645442377, | |
| "learning_rate": 2.34375e-08, | |
| "logits/chosen": 4945.83349609375, | |
| "logits/rejected": 4332.380859375, | |
| "logps/chosen": -248.83306884765625, | |
| "logps/rejected": -214.26800537109375, | |
| "loss": 256.8144, | |
| "rewards/accuracies": 0.4074074327945709, | |
| "rewards/chosen": -0.01683972030878067, | |
| "rewards/margins": -0.04092198982834816, | |
| "rewards/rejected": 0.02408226579427719, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.015698587127158554, | |
| "grad_norm": 3515.078596585681, | |
| "learning_rate": 4.6875e-08, | |
| "logits/chosen": 6022.4619140625, | |
| "logits/rejected": 4842.2509765625, | |
| "logps/chosen": -265.91229248046875, | |
| "logps/rejected": -232.0332489013672, | |
| "loss": 255.3669, | |
| "rewards/accuracies": 0.4750000536441803, | |
| "rewards/chosen": 0.4009794592857361, | |
| "rewards/margins": -0.017850089818239212, | |
| "rewards/rejected": 0.41882961988449097, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.023547880690737835, | |
| "grad_norm": 3233.4999259564747, | |
| "learning_rate": 7.03125e-08, | |
| "logits/chosen": 6028.0654296875, | |
| "logits/rejected": 5035.0439453125, | |
| "logps/chosen": -308.24267578125, | |
| "logps/rejected": -260.7057189941406, | |
| "loss": 249.3583, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 2.3666536808013916, | |
| "rewards/margins": 0.03575681522488594, | |
| "rewards/rejected": 2.3308968544006348, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03139717425431711, | |
| "grad_norm": 2020.1797676584106, | |
| "learning_rate": 9.375e-08, | |
| "logits/chosen": 5226.27392578125, | |
| "logits/rejected": 4330.328125, | |
| "logps/chosen": -257.849365234375, | |
| "logps/rejected": -218.84963989257812, | |
| "loss": 240.6838, | |
| "rewards/accuracies": 0.5166666507720947, | |
| "rewards/chosen": 6.1708879470825195, | |
| "rewards/margins": -0.0434657521545887, | |
| "rewards/rejected": 6.214354038238525, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03924646781789639, | |
| "grad_norm": 2561.508350324481, | |
| "learning_rate": 1.1718749999999999e-07, | |
| "logits/chosen": 6320.7529296875, | |
| "logits/rejected": 4940.04443359375, | |
| "logps/chosen": -316.8150329589844, | |
| "logps/rejected": -249.85879516601562, | |
| "loss": 246.5142, | |
| "rewards/accuracies": 0.5833333730697632, | |
| "rewards/chosen": 9.29455852508545, | |
| "rewards/margins": 0.6035072207450867, | |
| "rewards/rejected": 8.69105052947998, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04709576138147567, | |
| "grad_norm": 1957.3156990028594, | |
| "learning_rate": 1.40625e-07, | |
| "logits/chosen": 5318.0908203125, | |
| "logits/rejected": 4446.0439453125, | |
| "logps/chosen": -259.66485595703125, | |
| "logps/rejected": -251.6824188232422, | |
| "loss": 243.6371, | |
| "rewards/accuracies": 0.4916667342185974, | |
| "rewards/chosen": 10.262237548828125, | |
| "rewards/margins": -0.2461708039045334, | |
| "rewards/rejected": 10.508408546447754, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.054945054945054944, | |
| "grad_norm": 1619.4009475572605, | |
| "learning_rate": 1.640625e-07, | |
| "logits/chosen": 5004.0126953125, | |
| "logits/rejected": 4745.84765625, | |
| "logps/chosen": -220.51748657226562, | |
| "logps/rejected": -219.388916015625, | |
| "loss": 224.5808, | |
| "rewards/accuracies": 0.46666663885116577, | |
| "rewards/chosen": 12.74049186706543, | |
| "rewards/margins": -0.5236075520515442, | |
| "rewards/rejected": 13.26409912109375, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06279434850863422, | |
| "grad_norm": 1748.2419337698434, | |
| "learning_rate": 1.875e-07, | |
| "logits/chosen": 5535.58935546875, | |
| "logits/rejected": 5032.51123046875, | |
| "logps/chosen": -241.40966796875, | |
| "logps/rejected": -227.45913696289062, | |
| "loss": 234.7487, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": 14.568753242492676, | |
| "rewards/margins": -0.284468412399292, | |
| "rewards/rejected": 14.85322093963623, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0706436420722135, | |
| "grad_norm": 1855.7074043624723, | |
| "learning_rate": 2.109375e-07, | |
| "logits/chosen": 5690.1923828125, | |
| "logits/rejected": 4902.64501953125, | |
| "logps/chosen": -270.3674621582031, | |
| "logps/rejected": -227.33267211914062, | |
| "loss": 228.8738, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 14.823643684387207, | |
| "rewards/margins": -1.3101537227630615, | |
| "rewards/rejected": 16.13379669189453, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07849293563579278, | |
| "grad_norm": 1811.9491268478862, | |
| "learning_rate": 2.3437499999999998e-07, | |
| "logits/chosen": 5561.8994140625, | |
| "logits/rejected": 5368.55712890625, | |
| "logps/chosen": -263.51702880859375, | |
| "logps/rejected": -250.44369506835938, | |
| "loss": 249.9085, | |
| "rewards/accuracies": 0.36666667461395264, | |
| "rewards/chosen": 14.758064270019531, | |
| "rewards/margins": -2.986525535583496, | |
| "rewards/rejected": 17.744590759277344, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08634222919937205, | |
| "grad_norm": 1708.7219279256387, | |
| "learning_rate": 2.5781249999999997e-07, | |
| "logits/chosen": 5959.92041015625, | |
| "logits/rejected": 4902.26025390625, | |
| "logps/chosen": -238.0492706298828, | |
| "logps/rejected": -223.87939453125, | |
| "loss": 223.299, | |
| "rewards/accuracies": 0.46666663885116577, | |
| "rewards/chosen": 19.08856773376465, | |
| "rewards/margins": -0.6449778079986572, | |
| "rewards/rejected": 19.73354721069336, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09419152276295134, | |
| "grad_norm": 1688.5582849868945, | |
| "learning_rate": 2.8125e-07, | |
| "logits/chosen": 5817.5576171875, | |
| "logits/rejected": 4869.8154296875, | |
| "logps/chosen": -258.73681640625, | |
| "logps/rejected": -247.53610229492188, | |
| "loss": 247.3887, | |
| "rewards/accuracies": 0.5083333849906921, | |
| "rewards/chosen": 20.850032806396484, | |
| "rewards/margins": -1.2161338329315186, | |
| "rewards/rejected": 22.066165924072266, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.10204081632653061, | |
| "grad_norm": 1542.8490886062796, | |
| "learning_rate": 2.999977454977299e-07, | |
| "logits/chosen": 5724.7861328125, | |
| "logits/rejected": 5303.6083984375, | |
| "logps/chosen": -270.79779052734375, | |
| "logps/rejected": -251.43814086914062, | |
| "loss": 224.9548, | |
| "rewards/accuracies": 0.44166669249534607, | |
| "rewards/chosen": 19.309473037719727, | |
| "rewards/margins": -2.6830902099609375, | |
| "rewards/rejected": 21.992563247680664, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.10989010989010989, | |
| "grad_norm": 1454.7566776751667, | |
| "learning_rate": 2.999188450339425e-07, | |
| "logits/chosen": 5554.083984375, | |
| "logits/rejected": 4900.0087890625, | |
| "logps/chosen": -239.4879150390625, | |
| "logps/rejected": -227.8543243408203, | |
| "loss": 244.2376, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 20.754297256469727, | |
| "rewards/margins": -1.7472045421600342, | |
| "rewards/rejected": 22.501501083374023, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11773940345368916, | |
| "grad_norm": 1991.9493197417385, | |
| "learning_rate": 2.997272872179022e-07, | |
| "logits/chosen": 5627.8154296875, | |
| "logits/rejected": 4723.08154296875, | |
| "logps/chosen": -256.41790771484375, | |
| "logps/rejected": -219.00839233398438, | |
| "loss": 221.5181, | |
| "rewards/accuracies": 0.5000000596046448, | |
| "rewards/chosen": 21.454105377197266, | |
| "rewards/margins": -1.4268842935562134, | |
| "rewards/rejected": 22.880990982055664, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12558869701726844, | |
| "grad_norm": 1589.525779755758, | |
| "learning_rate": 2.994232159967983e-07, | |
| "logits/chosen": 5580.58203125, | |
| "logits/rejected": 4987.36181640625, | |
| "logps/chosen": -249.8968505859375, | |
| "logps/rejected": -242.6744842529297, | |
| "loss": 219.9084, | |
| "rewards/accuracies": 0.5166666507720947, | |
| "rewards/chosen": 23.667348861694336, | |
| "rewards/margins": 0.20866379141807556, | |
| "rewards/rejected": 23.45868492126465, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.13343799058084774, | |
| "grad_norm": 1358.5700842246065, | |
| "learning_rate": 2.990068598666458e-07, | |
| "logits/chosen": 5814.65087890625, | |
| "logits/rejected": 5200.2861328125, | |
| "logps/chosen": -266.10748291015625, | |
| "logps/rejected": -245.8904571533203, | |
| "loss": 226.6567, | |
| "rewards/accuracies": 0.5166666507720947, | |
| "rewards/chosen": 23.689990997314453, | |
| "rewards/margins": 0.5743306279182434, | |
| "rewards/rejected": 23.115659713745117, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.141287284144427, | |
| "grad_norm": 1487.5984728822925, | |
| "learning_rate": 2.984785317005808e-07, | |
| "logits/chosen": 5106.20947265625, | |
| "logits/rejected": 4475.1318359375, | |
| "logps/chosen": -240.2284698486328, | |
| "logps/rejected": -223.2233428955078, | |
| "loss": 231.2222, | |
| "rewards/accuracies": 0.4583333432674408, | |
| "rewards/chosen": 23.215864181518555, | |
| "rewards/margins": -0.8277734518051147, | |
| "rewards/rejected": 24.043638229370117, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14913657770800628, | |
| "grad_norm": 1479.2229495545964, | |
| "learning_rate": 2.978386285137506e-07, | |
| "logits/chosen": 5158.18408203125, | |
| "logits/rejected": 4452.48876953125, | |
| "logps/chosen": -259.5328063964844, | |
| "logps/rejected": -235.11923217773438, | |
| "loss": 222.1397, | |
| "rewards/accuracies": 0.40833336114883423, | |
| "rewards/chosen": 27.751850128173828, | |
| "rewards/margins": 1.4096729755401611, | |
| "rewards/rejected": 26.342174530029297, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15698587127158556, | |
| "grad_norm": 1703.3616590250404, | |
| "learning_rate": 2.9708763116497365e-07, | |
| "logits/chosen": 5071.6259765625, | |
| "logits/rejected": 4691.37109375, | |
| "logps/chosen": -227.9725341796875, | |
| "logps/rejected": -241.0368194580078, | |
| "loss": 245.3326, | |
| "rewards/accuracies": 0.4333333373069763, | |
| "rewards/chosen": 23.277462005615234, | |
| "rewards/margins": 0.8406739234924316, | |
| "rewards/rejected": 22.43678855895996, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16483516483516483, | |
| "grad_norm": 1426.565402046009, | |
| "learning_rate": 2.962261039953964e-07, | |
| "logits/chosen": 4549.0439453125, | |
| "logits/rejected": 4077.908203125, | |
| "logps/chosen": -199.64437866210938, | |
| "logps/rejected": -196.94285583496094, | |
| "loss": 206.0972, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 25.06855583190918, | |
| "rewards/margins": -0.44935113191604614, | |
| "rewards/rejected": 25.517908096313477, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.1726844583987441, | |
| "grad_norm": 1601.210211554247, | |
| "learning_rate": 2.952546944044155e-07, | |
| "logits/chosen": 5617.42041015625, | |
| "logits/rejected": 4799.43359375, | |
| "logps/chosen": -252.7371063232422, | |
| "logps/rejected": -206.9757843017578, | |
| "loss": 210.9757, | |
| "rewards/accuracies": 0.4500000476837158, | |
| "rewards/chosen": 29.895782470703125, | |
| "rewards/margins": 3.331066131591797, | |
| "rewards/rejected": 26.56471824645996, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.18053375196232338, | |
| "grad_norm": 1556.5281793301856, | |
| "learning_rate": 2.941741323631859e-07, | |
| "logits/chosen": 5181.3984375, | |
| "logits/rejected": 4707.39892578125, | |
| "logps/chosen": -239.87460327148438, | |
| "logps/rejected": -225.77249145507812, | |
| "loss": 239.8674, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 30.402624130249023, | |
| "rewards/margins": -1.8471183776855469, | |
| "rewards/rejected": 32.2497444152832, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.18838304552590268, | |
| "grad_norm": 1394.8473002635717, | |
| "learning_rate": 2.929852298660803e-07, | |
| "logits/chosen": 4699.716796875, | |
| "logits/rejected": 3764.293701171875, | |
| "logps/chosen": -213.49160766601562, | |
| "logps/rejected": -189.942138671875, | |
| "loss": 221.9249, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 26.20192527770996, | |
| "rewards/margins": 3.2178139686584473, | |
| "rewards/rejected": 22.984111785888672, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19623233908948196, | |
| "grad_norm": 1470.3402415239893, | |
| "learning_rate": 2.9168888032051097e-07, | |
| "logits/chosen": 5438.083984375, | |
| "logits/rejected": 4869.716796875, | |
| "logps/chosen": -210.9588165283203, | |
| "logps/rejected": -227.9866180419922, | |
| "loss": 212.291, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 24.68861961364746, | |
| "rewards/margins": -1.7168676853179932, | |
| "rewards/rejected": 26.405487060546875, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.20408163265306123, | |
| "grad_norm": 10034.65497008399, | |
| "learning_rate": 2.9028605787557423e-07, | |
| "logits/chosen": 5617.7451171875, | |
| "logits/rejected": 4562.69287109375, | |
| "logps/chosen": -255.1033935546875, | |
| "logps/rejected": -209.45535278320312, | |
| "loss": 228.7449, | |
| "rewards/accuracies": 0.5083333849906921, | |
| "rewards/chosen": 29.64927101135254, | |
| "rewards/margins": 0.5804499387741089, | |
| "rewards/rejected": 29.068822860717773, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2119309262166405, | |
| "grad_norm": 1368.8248000894046, | |
| "learning_rate": 2.8877781669002034e-07, | |
| "logits/chosen": 5615.40966796875, | |
| "logits/rejected": 5430.4384765625, | |
| "logps/chosen": -244.25747680664062, | |
| "logps/rejected": -234.9336395263672, | |
| "loss": 222.7808, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 30.25113296508789, | |
| "rewards/margins": -1.6435935497283936, | |
| "rewards/rejected": 31.894723892211914, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.21978021978021978, | |
| "grad_norm": 1584.564072575878, | |
| "learning_rate": 2.871652901401002e-07, | |
| "logits/chosen": 4762.8251953125, | |
| "logits/rejected": 4148.6337890625, | |
| "logps/chosen": -205.24063110351562, | |
| "logps/rejected": -207.39407348632812, | |
| "loss": 222.7514, | |
| "rewards/accuracies": 0.4999999403953552, | |
| "rewards/chosen": 23.47471046447754, | |
| "rewards/margins": -3.5412209033966064, | |
| "rewards/rejected": 27.01593017578125, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22762951334379905, | |
| "grad_norm": 1607.027073077983, | |
| "learning_rate": 2.854496899678834e-07, | |
| "logits/chosen": 6597.9111328125, | |
| "logits/rejected": 4897.9716796875, | |
| "logps/chosen": -295.0384521484375, | |
| "logps/rejected": -226.37924194335938, | |
| "loss": 228.4457, | |
| "rewards/accuracies": 0.5250000357627869, | |
| "rewards/chosen": 29.752086639404297, | |
| "rewards/margins": -1.6350635290145874, | |
| "rewards/rejected": 31.38715171813965, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.23547880690737832, | |
| "grad_norm": 1545.016523783741, | |
| "learning_rate": 2.8363230537068793e-07, | |
| "logits/chosen": 4787.65771484375, | |
| "logits/rejected": 4753.82177734375, | |
| "logps/chosen": -208.2709503173828, | |
| "logps/rejected": -209.3594512939453, | |
| "loss": 236.415, | |
| "rewards/accuracies": 0.43333330750465393, | |
| "rewards/chosen": 28.469614028930664, | |
| "rewards/margins": -6.097928047180176, | |
| "rewards/rejected": 34.567543029785156, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.24332810047095763, | |
| "grad_norm": 1629.3860078147109, | |
| "learning_rate": 2.817145020323055e-07, | |
| "logits/chosen": 5360.5908203125, | |
| "logits/rejected": 4794.95166015625, | |
| "logps/chosen": -227.51614379882812, | |
| "logps/rejected": -215.46432495117188, | |
| "loss": 218.1309, | |
| "rewards/accuracies": 0.4916667342185974, | |
| "rewards/chosen": 28.152446746826172, | |
| "rewards/margins": 1.9510828256607056, | |
| "rewards/rejected": 26.201364517211914, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.25117739403453687, | |
| "grad_norm": 1502.6973146567518, | |
| "learning_rate": 2.7969772109675106e-07, | |
| "logits/chosen": 5386.66796875, | |
| "logits/rejected": 4683.4365234375, | |
| "logps/chosen": -248.57504272460938, | |
| "logps/rejected": -244.3946533203125, | |
| "loss": 235.1897, | |
| "rewards/accuracies": 0.533333420753479, | |
| "rewards/chosen": 28.712182998657227, | |
| "rewards/margins": 1.181710958480835, | |
| "rewards/rejected": 27.530475616455078, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.25902668759811615, | |
| "grad_norm": 1194.597501148511, | |
| "learning_rate": 2.775834780853069e-07, | |
| "logits/chosen": 5262.4208984375, | |
| "logits/rejected": 4791.0087890625, | |
| "logps/chosen": -241.16259765625, | |
| "logps/rejected": -218.7141571044922, | |
| "loss": 213.4267, | |
| "rewards/accuracies": 0.5166666507720947, | |
| "rewards/chosen": 28.93545913696289, | |
| "rewards/margins": -1.7424633502960205, | |
| "rewards/rejected": 30.677921295166016, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2668759811616955, | |
| "grad_norm": 1440.9429803318897, | |
| "learning_rate": 2.75373361757676e-07, | |
| "logits/chosen": 5499.55078125, | |
| "logits/rejected": 5618.97900390625, | |
| "logps/chosen": -241.36984252929688, | |
| "logps/rejected": -266.84698486328125, | |
| "loss": 219.2353, | |
| "rewards/accuracies": 0.3916666507720947, | |
| "rewards/chosen": 26.7132625579834, | |
| "rewards/margins": -8.321182250976562, | |
| "rewards/rejected": 35.034446716308594, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.27472527472527475, | |
| "grad_norm": 1664.3454100749987, | |
| "learning_rate": 2.730690329180992e-07, | |
| "logits/chosen": 5780.2802734375, | |
| "logits/rejected": 5284.71337890625, | |
| "logps/chosen": -243.7447509765625, | |
| "logps/rejected": -250.7445831298828, | |
| "loss": 245.1434, | |
| "rewards/accuracies": 0.4833333492279053, | |
| "rewards/chosen": 29.163005828857422, | |
| "rewards/margins": -3.5909409523010254, | |
| "rewards/rejected": 32.75394821166992, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.282574568288854, | |
| "grad_norm": 1752.9607648456033, | |
| "learning_rate": 2.7067222316733527e-07, | |
| "logits/chosen": 5304.2255859375, | |
| "logits/rejected": 5228.21142578125, | |
| "logps/chosen": -219.45101928710938, | |
| "logps/rejected": -242.72299194335938, | |
| "loss": 233.5434, | |
| "rewards/accuracies": 0.4166666865348816, | |
| "rewards/chosen": 24.233936309814453, | |
| "rewards/margins": -7.016136169433594, | |
| "rewards/rejected": 31.250070571899414, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2904238618524333, | |
| "grad_norm": 1518.5089221978192, | |
| "learning_rate": 2.6818473360143916e-07, | |
| "logits/chosen": 4872.24609375, | |
| "logits/rejected": 4605.54541015625, | |
| "logps/chosen": -228.77365112304688, | |
| "logps/rejected": -205.58590698242188, | |
| "loss": 219.2235, | |
| "rewards/accuracies": 0.5666667222976685, | |
| "rewards/chosen": 30.12870216369629, | |
| "rewards/margins": -2.703049421310425, | |
| "rewards/rejected": 32.831748962402344, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.29827315541601257, | |
| "grad_norm": 1422.2449613954848, | |
| "learning_rate": 2.6560843345831893e-07, | |
| "logits/chosen": 5828.52490234375, | |
| "logits/rejected": 4848.03271484375, | |
| "logps/chosen": -278.51641845703125, | |
| "logps/rejected": -224.0612030029297, | |
| "loss": 215.0398, | |
| "rewards/accuracies": 0.46666669845581055, | |
| "rewards/chosen": 29.460494995117188, | |
| "rewards/margins": -2.3867201805114746, | |
| "rewards/rejected": 31.847219467163086, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.30612244897959184, | |
| "grad_norm": 1513.0023102032137, | |
| "learning_rate": 2.62945258713086e-07, | |
| "logits/chosen": 5936.6494140625, | |
| "logits/rejected": 4572.04296875, | |
| "logps/chosen": -263.99676513671875, | |
| "logps/rejected": -214.51528930664062, | |
| "loss": 214.3006, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 27.529659271240234, | |
| "rewards/margins": -3.4467697143554688, | |
| "rewards/rejected": 30.976430892944336, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3139717425431711, | |
| "grad_norm": 1522.81787328639, | |
| "learning_rate": 2.601972106232561e-07, | |
| "logits/chosen": 5100.27392578125, | |
| "logits/rejected": 5009.63916015625, | |
| "logps/chosen": -234.50619506835938, | |
| "logps/rejected": -232.6081085205078, | |
| "loss": 221.4692, | |
| "rewards/accuracies": 0.4166666865348816, | |
| "rewards/chosen": 40.11125946044922, | |
| "rewards/margins": -13.891138076782227, | |
| "rewards/rejected": 54.00239944458008, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3218210361067504, | |
| "grad_norm": 1578.4118458950654, | |
| "learning_rate": 2.5736635422489343e-07, | |
| "logits/chosen": 5519.7861328125, | |
| "logits/rejected": 4965.2998046875, | |
| "logps/chosen": -260.37860107421875, | |
| "logps/rejected": -250.5183868408203, | |
| "loss": 233.8612, | |
| "rewards/accuracies": 0.4750000536441803, | |
| "rewards/chosen": 31.49020004272461, | |
| "rewards/margins": -0.6159704327583313, | |
| "rewards/rejected": 32.10617446899414, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.32967032967032966, | |
| "grad_norm": 1493.8297723956584, | |
| "learning_rate": 2.5445481678082706e-07, | |
| "logits/chosen": 5175.6259765625, | |
| "logits/rejected": 4738.25, | |
| "logps/chosen": -249.3583984375, | |
| "logps/rejected": -219.2287139892578, | |
| "loss": 230.7716, | |
| "rewards/accuracies": 0.5250000357627869, | |
| "rewards/chosen": 36.534812927246094, | |
| "rewards/margins": 2.681107997894287, | |
| "rewards/rejected": 33.85370635986328, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.33751962323390894, | |
| "grad_norm": 1514.8304800611984, | |
| "learning_rate": 2.514647861821081e-07, | |
| "logits/chosen": 4972.7841796875, | |
| "logits/rejected": 4177.3369140625, | |
| "logps/chosen": -222.5880126953125, | |
| "logps/rejected": -194.66665649414062, | |
| "loss": 221.0255, | |
| "rewards/accuracies": 0.5166666507720947, | |
| "rewards/chosen": 34.20689010620117, | |
| "rewards/margins": 3.869258403778076, | |
| "rewards/rejected": 30.337635040283203, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3453689167974882, | |
| "grad_norm": 1581.2232727588582, | |
| "learning_rate": 2.483985093039065e-07, | |
| "logits/chosen": 5326.6650390625, | |
| "logits/rejected": 5294.6982421875, | |
| "logps/chosen": -224.5026092529297, | |
| "logps/rejected": -231.0164031982422, | |
| "loss": 208.8783, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 26.265392303466797, | |
| "rewards/margins": -2.933743953704834, | |
| "rewards/rejected": 29.19913673400879, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3532182103610675, | |
| "grad_norm": 1455.5345936132253, | |
| "learning_rate": 2.4525829031708404e-07, | |
| "logits/chosen": 6133.8583984375, | |
| "logits/rejected": 4469.1396484375, | |
| "logps/chosen": -258.8287353515625, | |
| "logps/rejected": -196.3058319091797, | |
| "loss": 212.8841, | |
| "rewards/accuracies": 0.5500000715255737, | |
| "rewards/chosen": 28.228174209594727, | |
| "rewards/margins": -1.4651987552642822, | |
| "rewards/rejected": 29.693374633789062, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.36106750392464676, | |
| "grad_norm": 1496.194428040794, | |
| "learning_rate": 2.420464889567126e-07, | |
| "logits/chosen": 6362.32373046875, | |
| "logits/rejected": 5006.7490234375, | |
| "logps/chosen": -303.68841552734375, | |
| "logps/rejected": -229.64584350585938, | |
| "loss": 224.1689, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 29.022680282592773, | |
| "rewards/margins": -5.982969760894775, | |
| "rewards/rejected": 35.00564956665039, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.36891679748822603, | |
| "grad_norm": 1488.9159850847502, | |
| "learning_rate": 2.3876551874883775e-07, | |
| "logits/chosen": 5458.3955078125, | |
| "logits/rejected": 4845.20947265625, | |
| "logps/chosen": -257.0389099121094, | |
| "logps/rejected": -241.0809783935547, | |
| "loss": 221.2525, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 30.74686050415039, | |
| "rewards/margins": -6.270951747894287, | |
| "rewards/rejected": 37.0178108215332, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.37676609105180536, | |
| "grad_norm": 1438.7300656314928, | |
| "learning_rate": 2.354178451968214e-07, | |
| "logits/chosen": 5881.9755859375, | |
| "logits/rejected": 5136.19873046875, | |
| "logps/chosen": -252.09130859375, | |
| "logps/rejected": -212.52554321289062, | |
| "loss": 211.1514, | |
| "rewards/accuracies": 0.4166666865348816, | |
| "rewards/chosen": 27.715307235717773, | |
| "rewards/margins": -6.552438259124756, | |
| "rewards/rejected": 34.26774597167969, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 1715.7802912315544, | |
| "learning_rate": 2.3200598392862542e-07, | |
| "logits/chosen": 5506.41552734375, | |
| "logits/rejected": 5338.0859375, | |
| "logps/chosen": -214.28176879882812, | |
| "logps/rejected": -239.4601287841797, | |
| "loss": 222.2657, | |
| "rewards/accuracies": 0.40833330154418945, | |
| "rewards/chosen": 26.659282684326172, | |
| "rewards/margins": -6.479990482330322, | |
| "rewards/rejected": 33.13927459716797, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3924646781789639, | |
| "grad_norm": 1562.5025176018555, | |
| "learning_rate": 2.2853249880642897e-07, | |
| "logits/chosen": 6032.9599609375, | |
| "logits/rejected": 4701.92333984375, | |
| "logps/chosen": -266.0277404785156, | |
| "logps/rejected": -235.2860107421875, | |
| "loss": 219.3719, | |
| "rewards/accuracies": 0.5749999284744263, | |
| "rewards/chosen": 32.461727142333984, | |
| "rewards/margins": 3.31502103805542, | |
| "rewards/rejected": 29.146709442138672, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.4003139717425432, | |
| "grad_norm": 1457.4417923972553, | |
| "learning_rate": 2.25e-07, | |
| "logits/chosen": 4814.98876953125, | |
| "logits/rejected": 4304.96435546875, | |
| "logps/chosen": -227.20654296875, | |
| "logps/rejected": -199.1092529296875, | |
| "loss": 208.9939, | |
| "rewards/accuracies": 0.46666669845581055, | |
| "rewards/chosen": 29.653301239013672, | |
| "rewards/margins": -0.3551321029663086, | |
| "rewards/rejected": 30.008432388305664, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 1715.8781467997637, | |
| "learning_rate": 2.2141114202526867e-07, | |
| "logits/chosen": 5116.51611328125, | |
| "logits/rejected": 4759.4267578125, | |
| "logps/chosen": -237.9807891845703, | |
| "logps/rejected": -233.20166015625, | |
| "loss": 230.6836, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 28.192028045654297, | |
| "rewards/margins": -7.732627868652344, | |
| "rewards/rejected": 35.924652099609375, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.41601255886970173, | |
| "grad_norm": 1346.5147969202928, | |
| "learning_rate": 2.1776862174957646e-07, | |
| "logits/chosen": 5779.31201171875, | |
| "logits/rejected": 4786.1865234375, | |
| "logps/chosen": -255.0662078857422, | |
| "logps/rejected": -233.3073272705078, | |
| "loss": 218.3051, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 29.225900650024414, | |
| "rewards/margins": -1.5375829935073853, | |
| "rewards/rejected": 30.76348304748535, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.423861852433281, | |
| "grad_norm": 1499.0750246406994, | |
| "learning_rate": 2.1407517636510023e-07, | |
| "logits/chosen": 5495.806640625, | |
| "logits/rejected": 4962.28125, | |
| "logps/chosen": -248.8963623046875, | |
| "logps/rejected": -222.57217407226562, | |
| "loss": 211.9601, | |
| "rewards/accuracies": 0.3583333194255829, | |
| "rewards/chosen": 28.48727798461914, | |
| "rewards/margins": -4.337940692901611, | |
| "rewards/rejected": 32.82521438598633, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4317111459968603, | |
| "grad_norm": 1565.657112583857, | |
| "learning_rate": 2.1033358133197376e-07, | |
| "logits/chosen": 5328.36083984375, | |
| "logits/rejected": 5014.83544921875, | |
| "logps/chosen": -253.7369384765625, | |
| "logps/rejected": -261.7619934082031, | |
| "loss": 240.5277, | |
| "rewards/accuracies": 0.5416666865348816, | |
| "rewards/chosen": 27.599685668945312, | |
| "rewards/margins": -1.8215970993041992, | |
| "rewards/rejected": 29.421283721923828, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.43956043956043955, | |
| "grad_norm": 1748.3947802252596, | |
| "learning_rate": 2.0654664829265295e-07, | |
| "logits/chosen": 5355.08447265625, | |
| "logits/rejected": 5035.33056640625, | |
| "logps/chosen": -242.00680541992188, | |
| "logps/rejected": -229.71401977539062, | |
| "loss": 227.4089, | |
| "rewards/accuracies": 0.46666663885116577, | |
| "rewards/chosen": 34.12080764770508, | |
| "rewards/margins": -2.9433603286743164, | |
| "rewards/rejected": 37.064170837402344, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4474097331240188, | |
| "grad_norm": 1471.762281982405, | |
| "learning_rate": 2.027172229590914e-07, | |
| "logits/chosen": 5684.34228515625, | |
| "logits/rejected": 5284.1064453125, | |
| "logps/chosen": -285.0470275878906, | |
| "logps/rejected": -261.35931396484375, | |
| "loss": 228.5544, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 29.517953872680664, | |
| "rewards/margins": -1.0636838674545288, | |
| "rewards/rejected": 30.58163833618164, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.4552590266875981, | |
| "grad_norm": 1545.430358602671, | |
| "learning_rate": 1.9884818297431403e-07, | |
| "logits/chosen": 5618.81494140625, | |
| "logits/rejected": 5143.9365234375, | |
| "logps/chosen": -252.5173797607422, | |
| "logps/rejected": -249.45333862304688, | |
| "loss": 233.2178, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 34.72943878173828, | |
| "rewards/margins": -0.6676195859909058, | |
| "rewards/rejected": 35.397064208984375, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4631083202511774, | |
| "grad_norm": 1449.8785968100756, | |
| "learning_rate": 1.9494243574999645e-07, | |
| "logits/chosen": 6114.61328125, | |
| "logits/rejected": 5220.7265625, | |
| "logps/chosen": -274.8421936035156, | |
| "logps/rejected": -212.7829132080078, | |
| "loss": 220.9877, | |
| "rewards/accuracies": 0.46666669845581055, | |
| "rewards/chosen": 31.31123924255371, | |
| "rewards/margins": -1.746040940284729, | |
| "rewards/rejected": 33.057281494140625, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.47095761381475665, | |
| "grad_norm": 1810.9063175978736, | |
| "learning_rate": 1.9100291628167416e-07, | |
| "logits/chosen": 5782.0751953125, | |
| "logits/rejected": 4984.40576171875, | |
| "logps/chosen": -215.29296875, | |
| "logps/rejected": -221.63916015625, | |
| "loss": 218.7514, | |
| "rewards/accuracies": 0.46666669845581055, | |
| "rewards/chosen": 37.19805908203125, | |
| "rewards/margins": 3.8734607696533203, | |
| "rewards/rejected": 33.32460403442383, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.478806907378336, | |
| "grad_norm": 1479.0626949919067, | |
| "learning_rate": 1.870325849432238e-07, | |
| "logits/chosen": 5598.708984375, | |
| "logits/rejected": 4959.85498046875, | |
| "logps/chosen": -244.06192016601562, | |
| "logps/rejected": -216.4823455810547, | |
| "loss": 212.3216, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 28.77957534790039, | |
| "rewards/margins": -5.229770183563232, | |
| "rewards/rejected": 34.00934600830078, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.48665620094191525, | |
| "grad_norm": 1394.4572327557778, | |
| "learning_rate": 1.8303442526227364e-07, | |
| "logits/chosen": 4767.6767578125, | |
| "logits/rejected": 3963.44140625, | |
| "logps/chosen": -216.5692596435547, | |
| "logps/rejected": -186.02931213378906, | |
| "loss": 213.0455, | |
| "rewards/accuracies": 0.5416666865348816, | |
| "rewards/chosen": 31.256662368774414, | |
| "rewards/margins": 3.708723783493042, | |
| "rewards/rejected": 27.54793930053711, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.4945054945054945, | |
| "grad_norm": 1583.553645114108, | |
| "learning_rate": 1.7901144167821507e-07, | |
| "logits/chosen": 6239.07763671875, | |
| "logits/rejected": 5390.0087890625, | |
| "logps/chosen": -285.5887145996094, | |
| "logps/rejected": -225.98764038085938, | |
| "loss": 223.9324, | |
| "rewards/accuracies": 0.4500000476837158, | |
| "rewards/chosen": 31.122844696044922, | |
| "rewards/margins": -8.924400329589844, | |
| "rewards/rejected": 40.0472412109375, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.5023547880690737, | |
| "grad_norm": 1633.1580137437745, | |
| "learning_rate": 1.749666572845e-07, | |
| "logits/chosen": 5919.0166015625, | |
| "logits/rejected": 5109.03564453125, | |
| "logps/chosen": -248.50192260742188, | |
| "logps/rejected": -246.29367065429688, | |
| "loss": 222.4357, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 30.188512802124023, | |
| "rewards/margins": -1.358596682548523, | |
| "rewards/rejected": 31.5471134185791, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5102040816326531, | |
| "grad_norm": 1498.7274129972973, | |
| "learning_rate": 1.7090311155692034e-07, | |
| "logits/chosen": 5090.4228515625, | |
| "logits/rejected": 4211.20751953125, | |
| "logps/chosen": -228.23873901367188, | |
| "logps/rejected": -193.20361328125, | |
| "loss": 220.7539, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": 29.299793243408203, | |
| "rewards/margins": -4.598435401916504, | |
| "rewards/rejected": 33.898231506347656, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5180533751962323, | |
| "grad_norm": 1532.8820265157945, | |
| "learning_rate": 1.6682385806957695e-07, | |
| "logits/chosen": 5702.642578125, | |
| "logits/rejected": 4641.26611328125, | |
| "logps/chosen": -253.339599609375, | |
| "logps/rejected": -215.67916870117188, | |
| "loss": 216.7053, | |
| "rewards/accuracies": 0.4583333432674408, | |
| "rewards/chosen": 36.97866439819336, | |
| "rewards/margins": 0.3870201110839844, | |
| "rewards/rejected": 36.591651916503906, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5259026687598116, | |
| "grad_norm": 1382.4944477550252, | |
| "learning_rate": 1.627319622002544e-07, | |
| "logits/chosen": 5711.2265625, | |
| "logits/rejected": 4745.8408203125, | |
| "logps/chosen": -242.14077758789062, | |
| "logps/rejected": -216.5430908203125, | |
| "loss": 211.9106, | |
| "rewards/accuracies": 0.5416666865348816, | |
| "rewards/chosen": 34.0696907043457, | |
| "rewards/margins": -0.47615766525268555, | |
| "rewards/rejected": 34.54584884643555, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.533751962323391, | |
| "grad_norm": 1488.8151992606488, | |
| "learning_rate": 1.5863049882692565e-07, | |
| "logits/chosen": 5558.71240234375, | |
| "logits/rejected": 4694.978515625, | |
| "logps/chosen": -247.7496795654297, | |
| "logps/rejected": -205.45669555664062, | |
| "loss": 218.2415, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 30.078182220458984, | |
| "rewards/margins": 0.5293010473251343, | |
| "rewards/rejected": 29.54888343811035, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5416012558869702, | |
| "grad_norm": 1416.664927181213, | |
| "learning_rate": 1.5452255001711795e-07, | |
| "logits/chosen": 5479.2841796875, | |
| "logits/rejected": 5094.64111328125, | |
| "logps/chosen": -242.07577514648438, | |
| "logps/rejected": -244.10787963867188, | |
| "loss": 219.0931, | |
| "rewards/accuracies": 0.44166669249534607, | |
| "rewards/chosen": 33.013328552246094, | |
| "rewards/margins": -0.33762702345848083, | |
| "rewards/rejected": 33.350955963134766, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5494505494505495, | |
| "grad_norm": 1292.241154100283, | |
| "learning_rate": 1.50411202711876e-07, | |
| "logits/chosen": 5211.81787109375, | |
| "logits/rejected": 5068.29296875, | |
| "logps/chosen": -219.5565643310547, | |
| "logps/rejected": -212.0447235107422, | |
| "loss": 220.4337, | |
| "rewards/accuracies": 0.4583333432674408, | |
| "rewards/chosen": 30.55966567993164, | |
| "rewards/margins": -2.847227096557617, | |
| "rewards/rejected": 33.40689468383789, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5572998430141287, | |
| "grad_norm": 1570.569319491539, | |
| "learning_rate": 1.4629954640606275e-07, | |
| "logits/chosen": 5200.46728515625, | |
| "logits/rejected": 5258.4677734375, | |
| "logps/chosen": -221.57528686523438, | |
| "logps/rejected": -229.99609375, | |
| "loss": 212.8724, | |
| "rewards/accuracies": 0.44166669249534607, | |
| "rewards/chosen": 27.67055892944336, | |
| "rewards/margins": -6.422524452209473, | |
| "rewards/rejected": 34.09308624267578, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.565149136577708, | |
| "grad_norm": 1618.5577381254768, | |
| "learning_rate": 1.4219067082674153e-07, | |
| "logits/chosen": 5572.21240234375, | |
| "logits/rejected": 4871.40771484375, | |
| "logps/chosen": -251.16268920898438, | |
| "logps/rejected": -224.77078247070312, | |
| "loss": 219.4929, | |
| "rewards/accuracies": 0.46666669845581055, | |
| "rewards/chosen": 33.602088928222656, | |
| "rewards/margins": 0.2543960511684418, | |
| "rewards/rejected": 33.347694396972656, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5729984301412873, | |
| "grad_norm": 1452.566213144231, | |
| "learning_rate": 1.3808766361138294e-07, | |
| "logits/chosen": 6028.9140625, | |
| "logits/rejected": 4685.0654296875, | |
| "logps/chosen": -257.2322692871094, | |
| "logps/rejected": -200.23098754882812, | |
| "loss": 223.5138, | |
| "rewards/accuracies": 0.5083333849906921, | |
| "rewards/chosen": 37.826812744140625, | |
| "rewards/margins": -2.2269558906555176, | |
| "rewards/rejected": 40.05376434326172, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5808477237048666, | |
| "grad_norm": 1214.1682692680406, | |
| "learning_rate": 1.3399360798764307e-07, | |
| "logits/chosen": 4993.73681640625, | |
| "logits/rejected": 4297.19775390625, | |
| "logps/chosen": -219.44515991210938, | |
| "logps/rejected": -184.39517211914062, | |
| "loss": 212.41, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 28.56288719177246, | |
| "rewards/margins": -4.594831943511963, | |
| "rewards/rejected": 33.157718658447266, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5886970172684458, | |
| "grad_norm": 1605.6119803296422, | |
| "learning_rate": 1.2991158045645435e-07, | |
| "logits/chosen": 5418.52734375, | |
| "logits/rejected": 4984.91015625, | |
| "logps/chosen": -207.006591796875, | |
| "logps/rejected": -188.98345947265625, | |
| "loss": 225.4038, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 33.255592346191406, | |
| "rewards/margins": -0.3109052777290344, | |
| "rewards/rejected": 33.566497802734375, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5965463108320251, | |
| "grad_norm": 1457.6337011407554, | |
| "learning_rate": 1.2584464848017214e-07, | |
| "logits/chosen": 5738.86279296875, | |
| "logits/rejected": 4608.76708984375, | |
| "logps/chosen": -231.95437622070312, | |
| "logps/rejected": -198.1880340576172, | |
| "loss": 217.1632, | |
| "rewards/accuracies": 0.5833333730697632, | |
| "rewards/chosen": 31.406179428100586, | |
| "rewards/margins": -8.325632095336914, | |
| "rewards/rejected": 39.73180389404297, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.6043956043956044, | |
| "grad_norm": 1411.671276803978, | |
| "learning_rate": 1.2179586817751263e-07, | |
| "logits/chosen": 5957.5966796875, | |
| "logits/rejected": 5184.8095703125, | |
| "logps/chosen": -260.0323181152344, | |
| "logps/rejected": -248.357177734375, | |
| "loss": 221.0459, | |
| "rewards/accuracies": 0.49166664481163025, | |
| "rewards/chosen": 34.612640380859375, | |
| "rewards/margins": -1.8284162282943726, | |
| "rewards/rejected": 36.4410514831543, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.6122448979591837, | |
| "grad_norm": 1492.2015814411402, | |
| "learning_rate": 1.177682820270159e-07, | |
| "logits/chosen": 5829.42919921875, | |
| "logits/rejected": 5579.2265625, | |
| "logps/chosen": -242.4963836669922, | |
| "logps/rejected": -265.2953796386719, | |
| "loss": 237.0832, | |
| "rewards/accuracies": 0.46666663885116577, | |
| "rewards/chosen": 40.636192321777344, | |
| "rewards/margins": -8.631589889526367, | |
| "rewards/rejected": 49.26778030395508, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.6200941915227629, | |
| "grad_norm": 1495.9097339616183, | |
| "learning_rate": 1.13764916580758e-07, | |
| "logits/chosen": 5397.736328125, | |
| "logits/rejected": 4867.71435546875, | |
| "logps/chosen": -222.5872802734375, | |
| "logps/rejected": -210.5218963623047, | |
| "loss": 219.6537, | |
| "rewards/accuracies": 0.4416666626930237, | |
| "rewards/chosen": 31.86411476135254, | |
| "rewards/margins": -12.959714889526367, | |
| "rewards/rejected": 44.82383346557617, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6279434850863422, | |
| "grad_norm": 1574.6099507214267, | |
| "learning_rate": 1.0978878019003183e-07, | |
| "logits/chosen": 5425.0205078125, | |
| "logits/rejected": 4799.3193359375, | |
| "logps/chosen": -242.7035675048828, | |
| "logps/rejected": -233.2936248779297, | |
| "loss": 234.5304, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 31.032894134521484, | |
| "rewards/margins": -2.1906774044036865, | |
| "rewards/rejected": 33.223567962646484, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6357927786499215, | |
| "grad_norm": 1422.1375890886202, | |
| "learning_rate": 1.0584286074470446e-07, | |
| "logits/chosen": 4989.46435546875, | |
| "logits/rejected": 4564.41357421875, | |
| "logps/chosen": -227.3849639892578, | |
| "logps/rejected": -241.45614624023438, | |
| "loss": 220.5149, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 46.44493865966797, | |
| "rewards/margins": 14.12573528289795, | |
| "rewards/rejected": 32.3192024230957, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6436420722135008, | |
| "grad_norm": 1515.7323047240088, | |
| "learning_rate": 1.0193012342795095e-07, | |
| "logits/chosen": 5426.66943359375, | |
| "logits/rejected": 4256.740234375, | |
| "logps/chosen": -250.27249145507812, | |
| "logps/rejected": -186.4129180908203, | |
| "loss": 215.0192, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 32.800254821777344, | |
| "rewards/margins": 3.388108491897583, | |
| "rewards/rejected": 29.412139892578125, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.6514913657770801, | |
| "grad_norm": 1592.21195695231, | |
| "learning_rate": 9.805350848805078e-08, | |
| "logits/chosen": 5394.11376953125, | |
| "logits/rejected": 4852.3466796875, | |
| "logps/chosen": -229.39566040039062, | |
| "logps/rejected": -211.81704711914062, | |
| "loss": 211.0008, | |
| "rewards/accuracies": 0.5166666507720947, | |
| "rewards/chosen": 34.058807373046875, | |
| "rewards/margins": -1.909393310546875, | |
| "rewards/rejected": 35.968196868896484, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.6593406593406593, | |
| "grad_norm": 1463.4619536600578, | |
| "learning_rate": 9.421592902892184e-08, | |
| "logits/chosen": 5901.42578125, | |
| "logits/rejected": 4903.45703125, | |
| "logps/chosen": -251.6103515625, | |
| "logps/rejected": -232.90286254882812, | |
| "loss": 229.612, | |
| "rewards/accuracies": 0.48333340883255005, | |
| "rewards/chosen": 33.922889709472656, | |
| "rewards/margins": -7.369454383850098, | |
| "rewards/rejected": 41.2923469543457, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6671899529042387, | |
| "grad_norm": 1430.6404166752445, | |
| "learning_rate": 9.042026882105215e-08, | |
| "logits/chosen": 5449.1259765625, | |
| "logits/rejected": 4962.32080078125, | |
| "logps/chosen": -239.6713409423828, | |
| "logps/rejected": -219.03256225585938, | |
| "loss": 204.1345, | |
| "rewards/accuracies": 0.5166667103767395, | |
| "rewards/chosen": 32.870826721191406, | |
| "rewards/margins": 5.945416450500488, | |
| "rewards/rejected": 26.925411224365234, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6750392464678179, | |
| "grad_norm": 1439.8377626509975, | |
| "learning_rate": 8.666938013447484e-08, | |
| "logits/chosen": 5891.3916015625, | |
| "logits/rejected": 5061.30859375, | |
| "logps/chosen": -271.90179443359375, | |
| "logps/rejected": -222.03848266601562, | |
| "loss": 213.3483, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 30.966625213623047, | |
| "rewards/margins": -1.30001699924469, | |
| "rewards/rejected": 32.26664352416992, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6828885400313972, | |
| "grad_norm": 1504.0267806705758, | |
| "learning_rate": 8.29660815954134e-08, | |
| "logits/chosen": 5863.2724609375, | |
| "logits/rejected": 4913.24365234375, | |
| "logps/chosen": -254.78286743164062, | |
| "logps/rejected": -219.35617065429688, | |
| "loss": 206.4855, | |
| "rewards/accuracies": 0.5416666865348816, | |
| "rewards/chosen": 34.62961196899414, | |
| "rewards/margins": -0.22334370017051697, | |
| "rewards/rejected": 34.85295867919922, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6907378335949764, | |
| "grad_norm": 1554.680163596608, | |
| "learning_rate": 7.931315606820979e-08, | |
| "logits/chosen": 5168.1904296875, | |
| "logits/rejected": 4985.6689453125, | |
| "logps/chosen": -230.7018585205078, | |
| "logps/rejected": -251.09347534179688, | |
| "loss": 244.2562, | |
| "rewards/accuracies": 0.38333335518836975, | |
| "rewards/chosen": 28.050884246826172, | |
| "rewards/margins": -9.163290977478027, | |
| "rewards/rejected": 37.21417236328125, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6985871271585558, | |
| "grad_norm": 1627.1399603581747, | |
| "learning_rate": 7.571334856412503e-08, | |
| "logits/chosen": 6097.19873046875, | |
| "logits/rejected": 5008.0068359375, | |
| "logps/chosen": -280.9400329589844, | |
| "logps/rejected": -228.6689910888672, | |
| "loss": 212.1211, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": 31.588485717773438, | |
| "rewards/margins": -1.7722686529159546, | |
| "rewards/rejected": 33.360755920410156, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.706436420722135, | |
| "grad_norm": 1331.7748787838575, | |
| "learning_rate": 7.216936417858549e-08, | |
| "logits/chosen": 5167.65185546875, | |
| "logits/rejected": 4843.95654296875, | |
| "logps/chosen": -224.27139282226562, | |
| "logps/rejected": -226.9228057861328, | |
| "loss": 201.9469, | |
| "rewards/accuracies": 0.46666669845581055, | |
| "rewards/chosen": 33.77007293701172, | |
| "rewards/margins": -3.728670835494995, | |
| "rewards/rejected": 37.49874496459961, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 1622.6770770641656, | |
| "learning_rate": 6.868386605842379e-08, | |
| "logits/chosen": 5079.0634765625, | |
| "logits/rejected": 4892.828125, | |
| "logps/chosen": -226.36300659179688, | |
| "logps/rejected": -222.3538360595703, | |
| "loss": 236.4008, | |
| "rewards/accuracies": 0.3999999761581421, | |
| "rewards/chosen": 31.204381942749023, | |
| "rewards/margins": -4.417146682739258, | |
| "rewards/rejected": 35.62152862548828, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.7221350078492935, | |
| "grad_norm": 1481.2201052242087, | |
| "learning_rate": 6.525947340064232e-08, | |
| "logits/chosen": 5629.57568359375, | |
| "logits/rejected": 4716.466796875, | |
| "logps/chosen": -252.6859588623047, | |
| "logps/rejected": -213.24630737304688, | |
| "loss": 225.3293, | |
| "rewards/accuracies": 0.4999999403953552, | |
| "rewards/chosen": 34.035762786865234, | |
| "rewards/margins": 2.328340530395508, | |
| "rewards/rejected": 31.707422256469727, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.7299843014128728, | |
| "grad_norm": 1562.1540360815463, | |
| "learning_rate": 6.189875948420276e-08, | |
| "logits/chosen": 5545.7412109375, | |
| "logits/rejected": 4263.5537109375, | |
| "logps/chosen": -255.6289520263672, | |
| "logps/rejected": -198.6146697998047, | |
| "loss": 213.2609, | |
| "rewards/accuracies": 0.5166667103767395, | |
| "rewards/chosen": 32.32459259033203, | |
| "rewards/margins": -2.0076165199279785, | |
| "rewards/rejected": 34.33220672607422, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7378335949764521, | |
| "grad_norm": 1570.1907610799915, | |
| "learning_rate": 5.86042497363215e-08, | |
| "logits/chosen": 5369.9609375, | |
| "logits/rejected": 4753.33935546875, | |
| "logps/chosen": -236.205078125, | |
| "logps/rejected": -218.98220825195312, | |
| "loss": 215.4455, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 28.66092300415039, | |
| "rewards/margins": 0.48245278000831604, | |
| "rewards/rejected": 28.178470611572266, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.7456828885400314, | |
| "grad_norm": 1414.0851920251173, | |
| "learning_rate": 5.537841983472337e-08, | |
| "logits/chosen": 5544.67529296875, | |
| "logits/rejected": 5164.2119140625, | |
| "logps/chosen": -235.9311981201172, | |
| "logps/rejected": -229.0620574951172, | |
| "loss": 206.7006, | |
| "rewards/accuracies": 0.4583333134651184, | |
| "rewards/chosen": 27.62892723083496, | |
| "rewards/margins": -6.102116107940674, | |
| "rewards/rejected": 33.731048583984375, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7535321821036107, | |
| "grad_norm": 1499.0661724069257, | |
| "learning_rate": 5.222369384727954e-08, | |
| "logits/chosen": 5398.3662109375, | |
| "logits/rejected": 4874.31298828125, | |
| "logps/chosen": -234.80752563476562, | |
| "logps/rejected": -215.5985565185547, | |
| "loss": 215.2125, | |
| "rewards/accuracies": 0.5583332777023315, | |
| "rewards/chosen": 29.206045150756836, | |
| "rewards/margins": -1.5180606842041016, | |
| "rewards/rejected": 30.724105834960938, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7613814756671899, | |
| "grad_norm": 1471.3861308766175, | |
| "learning_rate": 4.914244241042832e-08, | |
| "logits/chosen": 5525.1630859375, | |
| "logits/rejected": 4682.8212890625, | |
| "logps/chosen": -257.53668212890625, | |
| "logps/rejected": -217.9530029296875, | |
| "loss": 217.309, | |
| "rewards/accuracies": 0.5250000357627869, | |
| "rewards/chosen": 31.594741821289062, | |
| "rewards/margins": 3.213296890258789, | |
| "rewards/rejected": 28.381444931030273, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 1648.6086810447048, | |
| "learning_rate": 4.613698094774715e-08, | |
| "logits/chosen": 5233.3359375, | |
| "logits/rejected": 4278.90234375, | |
| "logps/chosen": -245.3806610107422, | |
| "logps/rejected": -203.1794891357422, | |
| "loss": 221.9357, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 31.261327743530273, | |
| "rewards/margins": -4.2463178634643555, | |
| "rewards/rejected": 35.50764465332031, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7770800627943485, | |
| "grad_norm": 1598.714891089062, | |
| "learning_rate": 4.3209567930014746e-08, | |
| "logits/chosen": 5549.24072265625, | |
| "logits/rejected": 4706.05859375, | |
| "logps/chosen": -240.37057495117188, | |
| "logps/rejected": -205.52987670898438, | |
| "loss": 207.145, | |
| "rewards/accuracies": 0.4833333492279053, | |
| "rewards/chosen": 27.736812591552734, | |
| "rewards/margins": -3.017923355102539, | |
| "rewards/rejected": 30.75473976135254, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7849293563579278, | |
| "grad_norm": 1441.385725076965, | |
| "learning_rate": 4.036240317807059e-08, | |
| "logits/chosen": 5246.560546875, | |
| "logits/rejected": 4487.947265625, | |
| "logps/chosen": -228.5640106201172, | |
| "logps/rejected": -191.21078491210938, | |
| "loss": 229.8262, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 31.757516860961914, | |
| "rewards/margins": -2.707479476928711, | |
| "rewards/rejected": 34.46499252319336, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.792778649921507, | |
| "grad_norm": 1381.7771337206264, | |
| "learning_rate": 3.759762620974716e-08, | |
| "logits/chosen": 5418.3876953125, | |
| "logits/rejected": 4595.1962890625, | |
| "logps/chosen": -228.2830047607422, | |
| "logps/rejected": -201.64437866210938, | |
| "loss": 224.1323, | |
| "rewards/accuracies": 0.5416666269302368, | |
| "rewards/chosen": 31.132349014282227, | |
| "rewards/margins": -2.7384214401245117, | |
| "rewards/rejected": 33.870765686035156, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.8006279434850864, | |
| "grad_norm": 1325.1631767221224, | |
| "learning_rate": 3.4917314632117423e-08, | |
| "logits/chosen": 5521.35546875, | |
| "logits/rejected": 4965.1337890625, | |
| "logps/chosen": -264.6299133300781, | |
| "logps/rejected": -236.8341064453125, | |
| "loss": 214.5822, | |
| "rewards/accuracies": 0.4416666626930237, | |
| "rewards/chosen": 33.266685485839844, | |
| "rewards/margins": -3.4997456073760986, | |
| "rewards/rejected": 36.76642990112305, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.8084772370486656, | |
| "grad_norm": 1681.8613198125684, | |
| "learning_rate": 3.232348258026552e-08, | |
| "logits/chosen": 5885.1669921875, | |
| "logits/rejected": 5606.17822265625, | |
| "logps/chosen": -245.198486328125, | |
| "logps/rejected": -242.989501953125, | |
| "loss": 236.5168, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": 32.60173416137695, | |
| "rewards/margins": 0.4809885025024414, | |
| "rewards/rejected": 32.120750427246094, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.8163265306122449, | |
| "grad_norm": 1470.0498171387167, | |
| "learning_rate": 2.9818079203753654e-08, | |
| "logits/chosen": 5227.4228515625, | |
| "logits/rejected": 4828.2060546875, | |
| "logps/chosen": -233.7130584716797, | |
| "logps/rejected": -220.1761016845703, | |
| "loss": 216.1246, | |
| "rewards/accuracies": 0.4750000536441803, | |
| "rewards/chosen": 31.530200958251953, | |
| "rewards/margins": -1.8674514293670654, | |
| "rewards/rejected": 33.39765167236328, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.8241758241758241, | |
| "grad_norm": 1663.4960088294424, | |
| "learning_rate": 2.740298720192245e-08, | |
| "logits/chosen": 5276.015625, | |
| "logits/rejected": 4871.2431640625, | |
| "logps/chosen": -249.91268920898438, | |
| "logps/rejected": -225.966064453125, | |
| "loss": 222.9517, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 31.620433807373047, | |
| "rewards/margins": -8.137569427490234, | |
| "rewards/rejected": 39.75800323486328, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.8320251177394035, | |
| "grad_norm": 1476.4353999914413, | |
| "learning_rate": 2.5080021409126655e-08, | |
| "logits/chosen": 6538.1298828125, | |
| "logits/rejected": 5190.63671875, | |
| "logps/chosen": -274.5542297363281, | |
| "logps/rejected": -228.97250366210938, | |
| "loss": 222.4528, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": 33.368431091308594, | |
| "rewards/margins": -2.194060802459717, | |
| "rewards/rejected": 35.56249237060547, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.8398744113029827, | |
| "grad_norm": 1688.8054924459227, | |
| "learning_rate": 2.2850927430967287e-08, | |
| "logits/chosen": 5456.7177734375, | |
| "logits/rejected": 5063.908203125, | |
| "logps/chosen": -238.12759399414062, | |
| "logps/rejected": -200.9252471923828, | |
| "loss": 221.0992, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 34.7647705078125, | |
| "rewards/margins": -1.5452799797058105, | |
| "rewards/rejected": 36.31005096435547, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.847723704866562, | |
| "grad_norm": 1450.9816306288544, | |
| "learning_rate": 2.0717380332546768e-08, | |
| "logits/chosen": 5984.65283203125, | |
| "logits/rejected": 4696.62060546875, | |
| "logps/chosen": -279.5147399902344, | |
| "logps/rejected": -208.4358367919922, | |
| "loss": 223.3968, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 32.97307205200195, | |
| "rewards/margins": 0.16781501471996307, | |
| "rewards/rejected": 32.805259704589844, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8555729984301413, | |
| "grad_norm": 1484.5016593055886, | |
| "learning_rate": 1.8680983379731618e-08, | |
| "logits/chosen": 6087.83984375, | |
| "logits/rejected": 4614.283203125, | |
| "logps/chosen": -238.899658203125, | |
| "logps/rejected": -219.7810821533203, | |
| "loss": 218.1701, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 34.898014068603516, | |
| "rewards/margins": 5.0102949142456055, | |
| "rewards/rejected": 29.887714385986328, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8634222919937206, | |
| "grad_norm": 1413.0323509395673, | |
| "learning_rate": 1.6743266834369856e-08, | |
| "logits/chosen": 5834.3720703125, | |
| "logits/rejected": 4881.9150390625, | |
| "logps/chosen": -249.7890167236328, | |
| "logps/rejected": -196.9021453857422, | |
| "loss": 213.0929, | |
| "rewards/accuracies": 0.5583333373069763, | |
| "rewards/chosen": 33.433753967285156, | |
| "rewards/margins": -1.7731311321258545, | |
| "rewards/rejected": 35.206886291503906, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.8712715855572999, | |
| "grad_norm": 1509.8990940019671, | |
| "learning_rate": 1.4905686804366853e-08, | |
| "logits/chosen": 6213.298828125, | |
| "logits/rejected": 5062.87353515625, | |
| "logps/chosen": -275.6719970703125, | |
| "logps/rejected": -230.1014862060547, | |
| "loss": 231.1685, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 33.60237503051758, | |
| "rewards/margins": 0.9831056594848633, | |
| "rewards/rejected": 32.61927032470703, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8791208791208791, | |
| "grad_norm": 1646.7616066608398, | |
| "learning_rate": 1.3169624149485037e-08, | |
| "logits/chosen": 6053.14111328125, | |
| "logits/rejected": 4984.8291015625, | |
| "logps/chosen": -255.73403930664062, | |
| "logps/rejected": -224.060302734375, | |
| "loss": 215.0052, | |
| "rewards/accuracies": 0.5000000596046448, | |
| "rewards/chosen": 35.60407257080078, | |
| "rewards/margins": 0.18094520270824432, | |
| "rewards/rejected": 35.423126220703125, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8869701726844584, | |
| "grad_norm": 1451.8870738094931, | |
| "learning_rate": 1.1536383443688968e-08, | |
| "logits/chosen": 5491.31494140625, | |
| "logits/rejected": 4393.64306640625, | |
| "logps/chosen": -246.4952392578125, | |
| "logps/rejected": -187.03012084960938, | |
| "loss": 207.4564, | |
| "rewards/accuracies": 0.49166664481163025, | |
| "rewards/chosen": 32.07477951049805, | |
| "rewards/margins": -5.4972968101501465, | |
| "rewards/rejected": 37.57207489013672, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8948194662480377, | |
| "grad_norm": 1505.1660971801934, | |
| "learning_rate": 1.0007191994816272e-08, | |
| "logits/chosen": 6031.93212890625, | |
| "logits/rejected": 4980.4267578125, | |
| "logps/chosen": -271.4355773925781, | |
| "logps/rejected": -228.0433349609375, | |
| "loss": 210.7531, | |
| "rewards/accuracies": 0.5250000357627869, | |
| "rewards/chosen": 34.003868103027344, | |
| "rewards/margins": 0.44027748703956604, | |
| "rewards/rejected": 33.56359100341797, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.902668759811617, | |
| "grad_norm": 1415.140222754107, | |
| "learning_rate": 8.583198922310031e-09, | |
| "logits/chosen": 5347.54150390625, | |
| "logits/rejected": 4987.744140625, | |
| "logps/chosen": -237.6715850830078, | |
| "logps/rejected": -227.91390991210938, | |
| "loss": 221.575, | |
| "rewards/accuracies": 0.4166666865348816, | |
| "rewards/chosen": 29.523784637451172, | |
| "rewards/margins": -6.032816410064697, | |
| "rewards/rejected": 35.556602478027344, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.9105180533751962, | |
| "grad_norm": 1568.6298540055077, | |
| "learning_rate": 7.2654742937067916e-09, | |
| "logits/chosen": 5496.115234375, | |
| "logits/rejected": 4413.4248046875, | |
| "logps/chosen": -238.0878143310547, | |
| "logps/rejected": -201.83120727539062, | |
| "loss": 209.612, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": 32.59004592895508, | |
| "rewards/margins": 0.4027647078037262, | |
| "rewards/rejected": 32.187278747558594, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.9183673469387755, | |
| "grad_norm": 1627.658195500155, | |
| "learning_rate": 6.0550083205280325e-09, | |
| "logits/chosen": 6022.40966796875, | |
| "logits/rejected": 5845.90478515625, | |
| "logps/chosen": -252.76663208007812, | |
| "logps/rejected": -253.51553344726562, | |
| "loss": 228.8431, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": 39.36040496826172, | |
| "rewards/margins": -10.932153701782227, | |
| "rewards/rejected": 50.29255294799805, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.9262166405023547, | |
| "grad_norm": 1449.8280670316117, | |
| "learning_rate": 4.952710614180567e-09, | |
| "logits/chosen": 5332.541015625, | |
| "logits/rejected": 4890.25146484375, | |
| "logps/chosen": -249.4139404296875, | |
| "logps/rejected": -223.42172241210938, | |
| "loss": 213.2137, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 29.5376033782959, | |
| "rewards/margins": -5.8537397384643555, | |
| "rewards/rejected": 35.3913459777832, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.9340659340659341, | |
| "grad_norm": 1398.2589825717653, | |
| "learning_rate": 3.959409502423744e-09, | |
| "logits/chosen": 5784.1748046875, | |
| "logits/rejected": 5105.06396484375, | |
| "logps/chosen": -258.03179931640625, | |
| "logps/rejected": -223.38906860351562, | |
| "loss": 216.4434, | |
| "rewards/accuracies": 0.49166664481163025, | |
| "rewards/chosen": 34.286903381347656, | |
| "rewards/margins": -6.160577774047852, | |
| "rewards/rejected": 40.44747543334961, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.9419152276295133, | |
| "grad_norm": 1421.2492677796333, | |
| "learning_rate": 3.075851406917995e-09, | |
| "logits/chosen": 5959.9599609375, | |
| "logits/rejected": 5138.84765625, | |
| "logps/chosen": -264.71771240234375, | |
| "logps/rejected": -224.459716796875, | |
| "loss": 209.8964, | |
| "rewards/accuracies": 0.5083333253860474, | |
| "rewards/chosen": 40.030338287353516, | |
| "rewards/margins": 6.341104030609131, | |
| "rewards/rejected": 33.689231872558594, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9497645211930926, | |
| "grad_norm": 1740.3335038015473, | |
| "learning_rate": 2.3027002823221064e-09, | |
| "logits/chosen": 5583.9404296875, | |
| "logits/rejected": 4929.333984375, | |
| "logps/chosen": -249.03201293945312, | |
| "logps/rejected": -222.62588500976562, | |
| "loss": 238.4498, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 30.770893096923828, | |
| "rewards/margins": 1.0800079107284546, | |
| "rewards/rejected": 29.69088363647461, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.957613814756672, | |
| "grad_norm": 1578.4060588481516, | |
| "learning_rate": 1.6405371173608073e-09, | |
| "logits/chosen": 6972.1318359375, | |
| "logits/rejected": 5199.83544921875, | |
| "logps/chosen": -308.28448486328125, | |
| "logps/rejected": -251.09619140625, | |
| "loss": 230.6452, | |
| "rewards/accuracies": 0.5166667103767395, | |
| "rewards/chosen": 38.131099700927734, | |
| "rewards/margins": 3.9283604621887207, | |
| "rewards/rejected": 34.20273971557617, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9654631083202512, | |
| "grad_norm": 1432.5912879172274, | |
| "learning_rate": 1.0898594982375387e-09, | |
| "logits/chosen": 6196.76513671875, | |
| "logits/rejected": 4968.3154296875, | |
| "logps/chosen": -270.8320617675781, | |
| "logps/rejected": -219.25363159179688, | |
| "loss": 216.4258, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": 37.52337646484375, | |
| "rewards/margins": -8.283441543579102, | |
| "rewards/rejected": 45.80681228637695, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9733124018838305, | |
| "grad_norm": 1548.5421134823648, | |
| "learning_rate": 6.510812347205185e-10, | |
| "logits/chosen": 5352.60009765625, | |
| "logits/rejected": 4941.98046875, | |
| "logps/chosen": -239.3649139404297, | |
| "logps/rejected": -241.218994140625, | |
| "loss": 221.0179, | |
| "rewards/accuracies": 0.5000000596046448, | |
| "rewards/chosen": 33.58864974975586, | |
| "rewards/margins": -1.408493995666504, | |
| "rewards/rejected": 34.99714279174805, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.9811616954474097, | |
| "grad_norm": 1479.098581026341, | |
| "learning_rate": 3.245320491830683e-10, | |
| "logits/chosen": 5452.64794921875, | |
| "logits/rejected": 4636.62158203125, | |
| "logps/chosen": -238.23037719726562, | |
| "logps/rejected": -194.11471557617188, | |
| "loss": 213.899, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 32.20820617675781, | |
| "rewards/margins": 2.9702253341674805, | |
| "rewards/rejected": 29.237985610961914, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.989010989010989, | |
| "grad_norm": 1398.0707627189024, | |
| "learning_rate": 1.1045732883197545e-10, | |
| "logits/chosen": 5141.85546875, | |
| "logits/rejected": 4927.9990234375, | |
| "logps/chosen": -239.6378173828125, | |
| "logps/rejected": -221.8780975341797, | |
| "loss": 221.4562, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 27.01497459411621, | |
| "rewards/margins": -8.813344955444336, | |
| "rewards/rejected": 35.82831573486328, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9968602825745683, | |
| "grad_norm": 1555.3207879595661, | |
| "learning_rate": 9.017941309813482e-12, | |
| "logits/chosen": 5397.27392578125, | |
| "logits/rejected": 4430.40087890625, | |
| "logps/chosen": -232.0002899169922, | |
| "logps/rejected": -192.43386840820312, | |
| "loss": 228.9403, | |
| "rewards/accuracies": 0.4416666626930237, | |
| "rewards/chosen": 29.473217010498047, | |
| "rewards/margins": -6.251194000244141, | |
| "rewards/rejected": 35.72441482543945, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1274, | |
| "total_flos": 0.0, | |
| "train_loss": 223.05648856410053, | |
| "train_runtime": 13773.8983, | |
| "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 | |
| } | |