| { | |
| "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": 6.093040937403707, | |
| "learning_rate": 3.90625e-09, | |
| "logits/chosen": 5914.52099609375, | |
| "logits/rejected": 2785.021484375, | |
| "logps/chosen": -212.45889282226562, | |
| "logps/rejected": -98.59669494628906, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.007849293563579277, | |
| "grad_norm": 6.05072585015556, | |
| "learning_rate": 3.9062499999999997e-08, | |
| "logits/chosen": 4973.81005859375, | |
| "logits/rejected": 4328.2421875, | |
| "logps/chosen": -204.26458740234375, | |
| "logps/rejected": -179.73109436035156, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.5092592835426331, | |
| "rewards/chosen": 0.019286984577775, | |
| "rewards/margins": 0.04475558549165726, | |
| "rewards/rejected": -0.025468600913882256, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.015698587127158554, | |
| "grad_norm": 6.19573847618762, | |
| "learning_rate": 7.812499999999999e-08, | |
| "logits/chosen": 6084.08447265625, | |
| "logits/rejected": 4834.19189453125, | |
| "logps/chosen": -217.2179412841797, | |
| "logps/rejected": -196.71627807617188, | |
| "loss": 0.6932, | |
| "rewards/accuracies": 0.491666704416275, | |
| "rewards/chosen": -0.006735034286975861, | |
| "rewards/margins": -0.00402786722406745, | |
| "rewards/rejected": -0.002707168459892273, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.023547880690737835, | |
| "grad_norm": 5.478939628494671, | |
| "learning_rate": 1.1718749999999999e-07, | |
| "logits/chosen": 6084.3291015625, | |
| "logits/rejected": 5104.9287109375, | |
| "logps/chosen": -250.56930541992188, | |
| "logps/rejected": -209.365966796875, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.5916666388511658, | |
| "rewards/chosen": 0.0030434236396104097, | |
| "rewards/margins": 0.07141536474227905, | |
| "rewards/rejected": -0.06837192922830582, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03139717425431711, | |
| "grad_norm": 5.683384309053487, | |
| "learning_rate": 1.5624999999999999e-07, | |
| "logits/chosen": 5311.80126953125, | |
| "logits/rejected": 4346.90771484375, | |
| "logps/chosen": -212.0178985595703, | |
| "logps/rejected": -181.733642578125, | |
| "loss": 0.6928, | |
| "rewards/accuracies": 0.5916666984558105, | |
| "rewards/chosen": 0.06136460229754448, | |
| "rewards/margins": 0.10798362642526627, | |
| "rewards/rejected": -0.04661902040243149, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03924646781789639, | |
| "grad_norm": 5.773772011816319, | |
| "learning_rate": 1.9531249999999998e-07, | |
| "logits/chosen": 6425.0439453125, | |
| "logits/rejected": 5042.47119140625, | |
| "logps/chosen": -265.3039855957031, | |
| "logps/rejected": -206.7879180908203, | |
| "loss": 0.6924, | |
| "rewards/accuracies": 0.6666666865348816, | |
| "rewards/chosen": 0.32259494066238403, | |
| "rewards/margins": 0.4108748435974121, | |
| "rewards/rejected": -0.08827991783618927, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04709576138147567, | |
| "grad_norm": 5.549546838003912, | |
| "learning_rate": 2.3437499999999998e-07, | |
| "logits/chosen": 5484.3701171875, | |
| "logits/rejected": 4559.94970703125, | |
| "logps/chosen": -213.7654266357422, | |
| "logps/rejected": -209.1769256591797, | |
| "loss": 0.6918, | |
| "rewards/accuracies": 0.6666666865348816, | |
| "rewards/chosen": 0.23852241039276123, | |
| "rewards/margins": 0.6153593063354492, | |
| "rewards/rejected": -0.3768369257450104, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.054945054945054944, | |
| "grad_norm": 5.3363089121065395, | |
| "learning_rate": 2.734375e-07, | |
| "logits/chosen": 5194.4814453125, | |
| "logits/rejected": 4918.6982421875, | |
| "logps/chosen": -178.33897399902344, | |
| "logps/rejected": -177.4080047607422, | |
| "loss": 0.6911, | |
| "rewards/accuracies": 0.6750000715255737, | |
| "rewards/chosen": 0.14194367825984955, | |
| "rewards/margins": 0.8182166218757629, | |
| "rewards/rejected": -0.6762728691101074, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06279434850863422, | |
| "grad_norm": 5.589990285987142, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": 5774.57080078125, | |
| "logits/rejected": 5269.759765625, | |
| "logps/chosen": -196.86941528320312, | |
| "logps/rejected": -183.05673217773438, | |
| "loss": 0.6889, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": -0.3008244037628174, | |
| "rewards/margins": 1.1653871536254883, | |
| "rewards/rejected": -1.4662115573883057, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0706436420722135, | |
| "grad_norm": 6.3326385736616, | |
| "learning_rate": 3.5156249999999997e-07, | |
| "logits/chosen": 6037.1435546875, | |
| "logits/rejected": 5178.23046875, | |
| "logps/chosen": -220.5373077392578, | |
| "logps/rejected": -191.18222045898438, | |
| "loss": 0.6847, | |
| "rewards/accuracies": 0.7583332657814026, | |
| "rewards/chosen": -1.6551120281219482, | |
| "rewards/margins": 5.083296298980713, | |
| "rewards/rejected": -6.738408088684082, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07849293563579278, | |
| "grad_norm": 7.037637671287219, | |
| "learning_rate": 3.9062499999999997e-07, | |
| "logits/chosen": 5949.7509765625, | |
| "logits/rejected": 5727.8193359375, | |
| "logps/chosen": -216.74551391601562, | |
| "logps/rejected": -212.49813842773438, | |
| "loss": 0.684, | |
| "rewards/accuracies": 0.6583333611488342, | |
| "rewards/chosen": -6.514804840087891, | |
| "rewards/margins": 6.541618347167969, | |
| "rewards/rejected": -13.056424140930176, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08634222919937205, | |
| "grad_norm": 6.748245033085064, | |
| "learning_rate": 4.2968749999999996e-07, | |
| "logits/chosen": 6423.7314453125, | |
| "logits/rejected": 5241.818359375, | |
| "logps/chosen": -198.58094787597656, | |
| "logps/rejected": -202.92428588867188, | |
| "loss": 0.6778, | |
| "rewards/accuracies": 0.6750000715255737, | |
| "rewards/chosen": -15.466090202331543, | |
| "rewards/margins": 8.465994834899902, | |
| "rewards/rejected": -23.932086944580078, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09419152276295134, | |
| "grad_norm": 10.445551519317151, | |
| "learning_rate": 4.6874999999999996e-07, | |
| "logits/chosen": 6351.18212890625, | |
| "logits/rejected": 5273.6953125, | |
| "logps/chosen": -226.8378143310547, | |
| "logps/rejected": -234.71200561523438, | |
| "loss": 0.672, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -24.001096725463867, | |
| "rewards/margins": 12.545796394348145, | |
| "rewards/rejected": -36.54689025878906, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.10204081632653061, | |
| "grad_norm": 7.973080880427375, | |
| "learning_rate": 4.999962424962166e-07, | |
| "logits/chosen": 6361.275390625, | |
| "logits/rejected": 5860.3828125, | |
| "logps/chosen": -239.5140838623047, | |
| "logps/rejected": -246.1350555419922, | |
| "loss": 0.6659, | |
| "rewards/accuracies": 0.6833333373069763, | |
| "rewards/chosen": -29.705577850341797, | |
| "rewards/margins": 18.78491973876953, | |
| "rewards/rejected": -48.490501403808594, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.10989010989010989, | |
| "grad_norm": 9.656148828714871, | |
| "learning_rate": 4.998647417232375e-07, | |
| "logits/chosen": 6324.2783203125, | |
| "logits/rejected": 5568.7294921875, | |
| "logps/chosen": -233.6736602783203, | |
| "logps/rejected": -243.1141815185547, | |
| "loss": 0.6643, | |
| "rewards/accuracies": 0.6500000357627869, | |
| "rewards/chosen": -48.00621795654297, | |
| "rewards/margins": 17.51606559753418, | |
| "rewards/rejected": -65.52229309082031, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11773940345368916, | |
| "grad_norm": 12.68709283782261, | |
| "learning_rate": 4.995454786965036e-07, | |
| "logits/chosen": 6543.3310546875, | |
| "logits/rejected": 5417.107421875, | |
| "logps/chosen": -245.687744140625, | |
| "logps/rejected": -238.84017944335938, | |
| "loss": 0.6608, | |
| "rewards/accuracies": 0.6833333373069763, | |
| "rewards/chosen": -44.00338363647461, | |
| "rewards/margins": 22.5390682220459, | |
| "rewards/rejected": -66.5424575805664, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12558869701726844, | |
| "grad_norm": 8.336675270819248, | |
| "learning_rate": 4.990386933279972e-07, | |
| "logits/chosen": 6377.17236328125, | |
| "logits/rejected": 5662.22607421875, | |
| "logps/chosen": -243.30667114257812, | |
| "logps/rejected": -268.36431884765625, | |
| "loss": 0.6573, | |
| "rewards/accuracies": 0.6750000715255737, | |
| "rewards/chosen": -45.77846145629883, | |
| "rewards/margins": 25.528247833251953, | |
| "rewards/rejected": -71.30671691894531, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.13343799058084774, | |
| "grad_norm": 9.518393732214355, | |
| "learning_rate": 4.983447664444096e-07, | |
| "logits/chosen": 6593.54541015625, | |
| "logits/rejected": 5842.90625, | |
| "logps/chosen": -261.85162353515625, | |
| "logps/rejected": -269.01934814453125, | |
| "loss": 0.6572, | |
| "rewards/accuracies": 0.6416667103767395, | |
| "rewards/chosen": -52.23096466064453, | |
| "rewards/margins": 20.466854095458984, | |
| "rewards/rejected": -72.69782257080078, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.141287284144427, | |
| "grad_norm": 9.551095723742446, | |
| "learning_rate": 4.97464219500968e-07, | |
| "logits/chosen": 5738.0458984375, | |
| "logits/rejected": 4965.66162109375, | |
| "logps/chosen": -243.9608917236328, | |
| "logps/rejected": -260.8826599121094, | |
| "loss": 0.6523, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": -55.4331169128418, | |
| "rewards/margins": 27.288272857666016, | |
| "rewards/rejected": -82.72139739990234, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14913657770800628, | |
| "grad_norm": 9.729373391308025, | |
| "learning_rate": 4.963977141895843e-07, | |
| "logits/chosen": 5827.6171875, | |
| "logits/rejected": 4940.27880859375, | |
| "logps/chosen": -265.14093017578125, | |
| "logps/rejected": -304.3471374511719, | |
| "loss": 0.6414, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -61.3819694519043, | |
| "rewards/margins": 51.72563934326172, | |
| "rewards/rejected": -113.10758972167969, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15698587127158556, | |
| "grad_norm": 12.708150524514142, | |
| "learning_rate": 4.951460519416227e-07, | |
| "logits/chosen": 5715.02197265625, | |
| "logits/rejected": 5263.1376953125, | |
| "logps/chosen": -255.45626831054688, | |
| "logps/rejected": -308.5564880371094, | |
| "loss": 0.6432, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -72.91200256347656, | |
| "rewards/margins": 38.973487854003906, | |
| "rewards/rejected": -111.8854751586914, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16483516483516483, | |
| "grad_norm": 13.38335440738734, | |
| "learning_rate": 4.937101733256606e-07, | |
| "logits/chosen": 5213.759765625, | |
| "logits/rejected": 4587.78564453125, | |
| "logps/chosen": -221.2035675048828, | |
| "logps/rejected": -255.7327423095703, | |
| "loss": 0.645, | |
| "rewards/accuracies": 0.6666666269302368, | |
| "rewards/chosen": -67.2903060913086, | |
| "rewards/margins": 31.2614688873291, | |
| "rewards/rejected": -98.5517807006836, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.1726844583987441, | |
| "grad_norm": 13.23088850436655, | |
| "learning_rate": 4.920911573406924e-07, | |
| "logits/chosen": 6382.59326171875, | |
| "logits/rejected": 5369.0458984375, | |
| "logps/chosen": -254.50845336914062, | |
| "logps/rejected": -261.02984619140625, | |
| "loss": 0.6396, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -58.17182159423828, | |
| "rewards/margins": 38.86132049560547, | |
| "rewards/rejected": -97.03313446044922, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.18053375196232338, | |
| "grad_norm": 13.7002975170986, | |
| "learning_rate": 4.902902206053098e-07, | |
| "logits/chosen": 5725.72998046875, | |
| "logits/rejected": 5134.1142578125, | |
| "logps/chosen": -295.9504089355469, | |
| "logps/rejected": -334.36383056640625, | |
| "loss": 0.641, | |
| "rewards/accuracies": 0.6750000715255737, | |
| "rewards/chosen": -108.6116714477539, | |
| "rewards/margins": 45.984703063964844, | |
| "rewards/rejected": -154.59637451171875, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.18838304552590268, | |
| "grad_norm": 11.470848035692125, | |
| "learning_rate": 4.883087164434672e-07, | |
| "logits/chosen": 5131.9599609375, | |
| "logits/rejected": 4044.186279296875, | |
| "logps/chosen": -262.76409912109375, | |
| "logps/rejected": -300.7772521972656, | |
| "loss": 0.63, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -95.13509368896484, | |
| "rewards/margins": 50.95269012451172, | |
| "rewards/rejected": -146.08779907226562, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19623233908948196, | |
| "grad_norm": 16.695171401268386, | |
| "learning_rate": 4.861481338675183e-07, | |
| "logits/chosen": 6127.5693359375, | |
| "logits/rejected": 5402.2724609375, | |
| "logps/chosen": -271.9035949707031, | |
| "logps/rejected": -345.59136962890625, | |
| "loss": 0.6424, | |
| "rewards/accuracies": 0.7083333134651184, | |
| "rewards/chosen": -104.75971984863281, | |
| "rewards/margins": 56.324981689453125, | |
| "rewards/rejected": -161.08468627929688, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.20408163265306123, | |
| "grad_norm": 15.604975772107185, | |
| "learning_rate": 4.838100964592904e-07, | |
| "logits/chosen": 6368.3525390625, | |
| "logits/rejected": 5054.92529296875, | |
| "logps/chosen": -287.34759521484375, | |
| "logps/rejected": -301.80914306640625, | |
| "loss": 0.6438, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -87.629150390625, | |
| "rewards/margins": 47.95604705810547, | |
| "rewards/rejected": -135.585205078125, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2119309262166405, | |
| "grad_norm": 12.014162614621322, | |
| "learning_rate": 4.812963611500339e-07, | |
| "logits/chosen": 6167.998046875, | |
| "logits/rejected": 5935.1181640625, | |
| "logps/chosen": -288.6238708496094, | |
| "logps/rejected": -318.166015625, | |
| "loss": 0.6315, | |
| "rewards/accuracies": 0.7166667580604553, | |
| "rewards/chosen": -91.86804962158203, | |
| "rewards/margins": 37.66888427734375, | |
| "rewards/rejected": -129.5369415283203, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.21978021978021978, | |
| "grad_norm": 25.285219401191153, | |
| "learning_rate": 4.786088169001671e-07, | |
| "logits/chosen": 5262.55029296875, | |
| "logits/rejected": 4536.3759765625, | |
| "logps/chosen": -249.36483764648438, | |
| "logps/rejected": -323.6359558105469, | |
| "loss": 0.6274, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -86.21394348144531, | |
| "rewards/margins": 68.12052917480469, | |
| "rewards/rejected": -154.33448791503906, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22762951334379905, | |
| "grad_norm": 21.48499040413718, | |
| "learning_rate": 4.7574948327980567e-07, | |
| "logits/chosen": 7437.4541015625, | |
| "logits/rejected": 5381.4013671875, | |
| "logps/chosen": -332.65570068359375, | |
| "logps/rejected": -372.79583740234375, | |
| "loss": 0.6225, | |
| "rewards/accuracies": 0.7916666865348816, | |
| "rewards/chosen": -95.62727355957031, | |
| "rewards/margins": 94.65281677246094, | |
| "rewards/rejected": -190.28009033203125, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.23547880690737832, | |
| "grad_norm": 14.64094770923456, | |
| "learning_rate": 4.727205089511466e-07, | |
| "logits/chosen": 5407.33056640625, | |
| "logits/rejected": 5314.2958984375, | |
| "logps/chosen": -247.26754760742188, | |
| "logps/rejected": -310.8182678222656, | |
| "loss": 0.6323, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -79.79512023925781, | |
| "rewards/margins": 60.83668899536133, | |
| "rewards/rejected": -140.63180541992188, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.24332810047095763, | |
| "grad_norm": 15.88030348707004, | |
| "learning_rate": 4.6952417005384247e-07, | |
| "logits/chosen": 6043.89404296875, | |
| "logits/rejected": 5373.5927734375, | |
| "logps/chosen": -264.2408752441406, | |
| "logps/rejected": -301.0606994628906, | |
| "loss": 0.6422, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -86.66017150878906, | |
| "rewards/margins": 38.932220458984375, | |
| "rewards/rejected": -125.5923843383789, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.25117739403453687, | |
| "grad_norm": 12.942810394852572, | |
| "learning_rate": 4.661628684945851e-07, | |
| "logits/chosen": 5981.30615234375, | |
| "logits/rejected": 5134.4638671875, | |
| "logps/chosen": -321.04461669921875, | |
| "logps/rejected": -394.8025817871094, | |
| "loss": 0.6322, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -119.98759460449219, | |
| "rewards/margins": 72.79707336425781, | |
| "rewards/rejected": -192.78469848632812, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.25902668759811615, | |
| "grad_norm": 17.65657239960532, | |
| "learning_rate": 4.626391301421782e-07, | |
| "logits/chosen": 5767.1728515625, | |
| "logits/rejected": 5235.9375, | |
| "logps/chosen": -319.49798583984375, | |
| "logps/rejected": -344.1056213378906, | |
| "loss": 0.6455, | |
| "rewards/accuracies": 0.7416667342185974, | |
| "rewards/chosen": -125.0447769165039, | |
| "rewards/margins": 41.73804473876953, | |
| "rewards/rejected": -166.78280639648438, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2668759811616955, | |
| "grad_norm": 10.256221501911574, | |
| "learning_rate": 4.5895560292945996e-07, | |
| "logits/chosen": 6147.9560546875, | |
| "logits/rejected": 6276.4072265625, | |
| "logps/chosen": -274.11529541015625, | |
| "logps/rejected": -340.8870849609375, | |
| "loss": 0.651, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -80.25703430175781, | |
| "rewards/margins": 40.26253128051758, | |
| "rewards/rejected": -120.51956939697266, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.27472527472527475, | |
| "grad_norm": 23.69669876872275, | |
| "learning_rate": 4.5511505486349865e-07, | |
| "logits/chosen": 6537.46337890625, | |
| "logits/rejected": 5889.2177734375, | |
| "logps/chosen": -257.62890625, | |
| "logps/rejected": -332.9743957519531, | |
| "loss": 0.6256, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -62.803199768066406, | |
| "rewards/margins": 63.349143981933594, | |
| "rewards/rejected": -126.15234375, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.282574568288854, | |
| "grad_norm": 16.915648438332628, | |
| "learning_rate": 4.5112037194555876e-07, | |
| "logits/chosen": 6069.42578125, | |
| "logits/rejected": 5893.24951171875, | |
| "logps/chosen": -293.303955078125, | |
| "logps/rejected": -398.67724609375, | |
| "loss": 0.6222, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": -115.38047790527344, | |
| "rewards/margins": 84.01496887207031, | |
| "rewards/rejected": -199.39544677734375, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2904238618524333, | |
| "grad_norm": 12.555265198666554, | |
| "learning_rate": 4.4697455600239863e-07, | |
| "logits/chosen": 5456.2978515625, | |
| "logits/rejected": 5073.7060546875, | |
| "logps/chosen": -296.23944091796875, | |
| "logps/rejected": -330.85284423828125, | |
| "loss": 0.6434, | |
| "rewards/accuracies": 0.7083333134651184, | |
| "rewards/chosen": -111.81736755371094, | |
| "rewards/margins": 50.96443176269531, | |
| "rewards/rejected": -162.78179931640625, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.29827315541601257, | |
| "grad_norm": 14.362943259943494, | |
| "learning_rate": 4.426807224305315e-07, | |
| "logits/chosen": 6574.45458984375, | |
| "logits/rejected": 5350.2958984375, | |
| "logps/chosen": -315.9097595214844, | |
| "logps/rejected": -338.9637451171875, | |
| "loss": 0.6267, | |
| "rewards/accuracies": 0.7666667103767395, | |
| "rewards/chosen": -87.56874084472656, | |
| "rewards/margins": 68.38545227050781, | |
| "rewards/rejected": -155.95419311523438, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.30612244897959184, | |
| "grad_norm": 17.276156747689946, | |
| "learning_rate": 4.3824209785514326e-07, | |
| "logits/chosen": 6646.9892578125, | |
| "logits/rejected": 5010.45947265625, | |
| "logps/chosen": -303.3409118652344, | |
| "logps/rejected": -347.64373779296875, | |
| "loss": 0.6345, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": -87.84745788574219, | |
| "rewards/margins": 79.94316101074219, | |
| "rewards/rejected": -167.79061889648438, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3139717425431711, | |
| "grad_norm": 16.15627046135233, | |
| "learning_rate": 4.3366201770542687e-07, | |
| "logits/chosen": 5679.2353515625, | |
| "logits/rejected": 5516.23583984375, | |
| "logps/chosen": -289.3190612792969, | |
| "logps/rejected": -350.9610900878906, | |
| "loss": 0.637, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -98.53050994873047, | |
| "rewards/margins": 63.691078186035156, | |
| "rewards/rejected": -162.22158813476562, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3218210361067504, | |
| "grad_norm": 25.935620681417305, | |
| "learning_rate": 4.2894392370815567e-07, | |
| "logits/chosen": 6088.6064453125, | |
| "logits/rejected": 5395.84912109375, | |
| "logps/chosen": -339.7603759765625, | |
| "logps/rejected": -411.3926696777344, | |
| "loss": 0.6019, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -127.27352142333984, | |
| "rewards/margins": 75.66678619384766, | |
| "rewards/rejected": -202.9403076171875, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.32967032967032966, | |
| "grad_norm": 21.279196743760856, | |
| "learning_rate": 4.2409136130137845e-07, | |
| "logits/chosen": 5789.5654296875, | |
| "logits/rejected": 5153.6044921875, | |
| "logps/chosen": -319.99505615234375, | |
| "logps/rejected": -372.98162841796875, | |
| "loss": 0.6269, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -115.96516418457031, | |
| "rewards/margins": 75.77505493164062, | |
| "rewards/rejected": -191.74020385742188, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.33751962323390894, | |
| "grad_norm": 14.662428194905122, | |
| "learning_rate": 4.1910797697018017e-07, | |
| "logits/chosen": 5631.46337890625, | |
| "logits/rejected": 4653.8154296875, | |
| "logps/chosen": -277.2632141113281, | |
| "logps/rejected": -342.4476623535156, | |
| "loss": 0.624, | |
| "rewards/accuracies": 0.8583332896232605, | |
| "rewards/chosen": -97.5182113647461, | |
| "rewards/margins": 83.69635772705078, | |
| "rewards/rejected": -181.21456909179688, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3453689167974882, | |
| "grad_norm": 20.509381423565515, | |
| "learning_rate": 4.1399751550651084e-07, | |
| "logits/chosen": 5983.00341796875, | |
| "logits/rejected": 5909.9189453125, | |
| "logps/chosen": -294.4239807128906, | |
| "logps/rejected": -366.37548828125, | |
| "loss": 0.6294, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -112.0306396484375, | |
| "rewards/margins": 62.66124725341797, | |
| "rewards/rejected": -174.69189453125, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3532182103610675, | |
| "grad_norm": 11.57774703743698, | |
| "learning_rate": 4.087638171951401e-07, | |
| "logits/chosen": 6951.85302734375, | |
| "logits/rejected": 4954.27978515625, | |
| "logps/chosen": -307.88006591796875, | |
| "logps/rejected": -344.5763244628906, | |
| "loss": 0.6245, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -99.11131286621094, | |
| "rewards/margins": 82.47236633300781, | |
| "rewards/rejected": -181.5836639404297, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.36106750392464676, | |
| "grad_norm": 10.747587473405163, | |
| "learning_rate": 4.034108149278543e-07, | |
| "logits/chosen": 7119.5966796875, | |
| "logits/rejected": 5446.58544921875, | |
| "logps/chosen": -365.51141357421875, | |
| "logps/rejected": -376.9080810546875, | |
| "loss": 0.6136, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": -115.07914733886719, | |
| "rewards/margins": 71.62091064453125, | |
| "rewards/rejected": -186.7000732421875, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.36891679748822603, | |
| "grad_norm": 18.39563106747747, | |
| "learning_rate": 3.979425312480629e-07, | |
| "logits/chosen": 6053.51171875, | |
| "logits/rejected": 5261.99365234375, | |
| "logps/chosen": -349.2318420410156, | |
| "logps/rejected": -409.2492370605469, | |
| "loss": 0.634, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": -136.96707153320312, | |
| "rewards/margins": 69.71588134765625, | |
| "rewards/rejected": -206.68295288085938, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.37676609105180536, | |
| "grad_norm": 9.88717064834881, | |
| "learning_rate": 3.923630753280357e-07, | |
| "logits/chosen": 6531.0361328125, | |
| "logits/rejected": 5563.5673828125, | |
| "logps/chosen": -304.6759338378906, | |
| "logps/rejected": -353.3858642578125, | |
| "loss": 0.6225, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": -97.28072357177734, | |
| "rewards/margins": 81.7093734741211, | |
| "rewards/rejected": -178.9901123046875, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 18.872015793003488, | |
| "learning_rate": 3.866766398810424e-07, | |
| "logits/chosen": 6090.294921875, | |
| "logits/rejected": 5757.2998046875, | |
| "logps/chosen": -270.914794921875, | |
| "logps/rejected": -388.12066650390625, | |
| "loss": 0.6047, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -94.30628967285156, | |
| "rewards/margins": 91.7900619506836, | |
| "rewards/rejected": -186.0963592529297, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3924646781789639, | |
| "grad_norm": 19.921576157729877, | |
| "learning_rate": 3.8088749801071496e-07, | |
| "logits/chosen": 6461.3857421875, | |
| "logits/rejected": 4896.03759765625, | |
| "logps/chosen": -404.9927673339844, | |
| "logps/rejected": -488.85992431640625, | |
| "loss": 0.6363, | |
| "rewards/accuracies": 0.7583333849906921, | |
| "rewards/chosen": -183.0391082763672, | |
| "rewards/margins": 103.30455017089844, | |
| "rewards/rejected": -286.3436584472656, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.4003139717425432, | |
| "grad_norm": 14.976740215989906, | |
| "learning_rate": 3.75e-07, | |
| "logits/chosen": 5273.30029296875, | |
| "logits/rejected": 4571.634765625, | |
| "logps/chosen": -316.82159423828125, | |
| "logps/rejected": -388.461669921875, | |
| "loss": 0.6164, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -127.3386001586914, | |
| "rewards/margins": 94.75777435302734, | |
| "rewards/rejected": -222.0963897705078, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 15.310486602213532, | |
| "learning_rate": 3.6901857004211443e-07, | |
| "logits/chosen": 5530.29443359375, | |
| "logits/rejected": 5099.7265625, | |
| "logps/chosen": -315.60919189453125, | |
| "logps/rejected": -381.65789794921875, | |
| "loss": 0.6418, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -117.1432113647461, | |
| "rewards/margins": 68.7266845703125, | |
| "rewards/rejected": -185.86990356445312, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.41601255886970173, | |
| "grad_norm": 14.08983541941681, | |
| "learning_rate": 3.6294770291596076e-07, | |
| "logits/chosen": 6285.1962890625, | |
| "logits/rejected": 5097.83935546875, | |
| "logps/chosen": -322.7880554199219, | |
| "logps/rejected": -361.3253479003906, | |
| "loss": 0.6168, | |
| "rewards/accuracies": 0.6916666030883789, | |
| "rewards/chosen": -109.86531829833984, | |
| "rewards/margins": 52.69968795776367, | |
| "rewards/rejected": -162.5650177001953, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.423861852433281, | |
| "grad_norm": 20.022842230444766, | |
| "learning_rate": 3.5679196060850034e-07, | |
| "logits/chosen": 5982.83935546875, | |
| "logits/rejected": 5300.0537109375, | |
| "logps/chosen": -333.60455322265625, | |
| "logps/rejected": -389.5378723144531, | |
| "loss": 0.6173, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -122.7867431640625, | |
| "rewards/margins": 78.51233673095703, | |
| "rewards/rejected": -201.29908752441406, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4317111459968603, | |
| "grad_norm": 24.653995851898625, | |
| "learning_rate": 3.505559688866229e-07, | |
| "logits/chosen": 5769.158203125, | |
| "logits/rejected": 5321.18359375, | |
| "logps/chosen": -375.0007019042969, | |
| "logps/rejected": -469.31951904296875, | |
| "loss": 0.6294, | |
| "rewards/accuracies": 0.7500000596046448, | |
| "rewards/chosen": -160.26025390625, | |
| "rewards/margins": 83.9951171875, | |
| "rewards/rejected": -244.25537109375, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.43956043956043955, | |
| "grad_norm": 17.972565211019212, | |
| "learning_rate": 3.4424441382108826e-07, | |
| "logits/chosen": 5792.72998046875, | |
| "logits/rejected": 5407.20068359375, | |
| "logps/chosen": -351.23291015625, | |
| "logps/rejected": -403.1977233886719, | |
| "loss": 0.6269, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -147.9576416015625, | |
| "rewards/margins": 62.0308952331543, | |
| "rewards/rejected": -209.98855590820312, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4474097331240188, | |
| "grad_norm": 18.423331599938237, | |
| "learning_rate": 3.378620382651523e-07, | |
| "logits/chosen": 6100.2392578125, | |
| "logits/rejected": 5616.0419921875, | |
| "logps/chosen": -368.9530334472656, | |
| "logps/rejected": -418.500732421875, | |
| "loss": 0.6102, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -125.99314880371094, | |
| "rewards/margins": 68.73707580566406, | |
| "rewards/rejected": -194.730224609375, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.4552590266875981, | |
| "grad_norm": 40.58588385026, | |
| "learning_rate": 3.314136382905234e-07, | |
| "logits/chosen": 6104.7314453125, | |
| "logits/rejected": 5396.8330078125, | |
| "logps/chosen": -345.4382629394531, | |
| "logps/rejected": -464.142578125, | |
| "loss": 0.6198, | |
| "rewards/accuracies": 0.783333420753479, | |
| "rewards/chosen": -133.0227813720703, | |
| "rewards/margins": 117.80845642089844, | |
| "rewards/rejected": -250.8312530517578, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4631083202511774, | |
| "grad_norm": 21.906951775528675, | |
| "learning_rate": 3.249040595833274e-07, | |
| "logits/chosen": 6578.75537109375, | |
| "logits/rejected": 5453.22900390625, | |
| "logps/chosen": -414.60174560546875, | |
| "logps/rejected": -468.26995849609375, | |
| "loss": 0.6167, | |
| "rewards/accuracies": 0.7666667103767395, | |
| "rewards/chosen": -182.8759307861328, | |
| "rewards/margins": 106.53523254394531, | |
| "rewards/rejected": -289.41119384765625, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.47095761381475665, | |
| "grad_norm": 22.906733288517135, | |
| "learning_rate": 3.1833819380279023e-07, | |
| "logits/chosen": 6241.1650390625, | |
| "logits/rejected": 5260.3125, | |
| "logps/chosen": -361.357666015625, | |
| "logps/rejected": -456.5220642089844, | |
| "loss": 0.6132, | |
| "rewards/accuracies": 0.783333420753479, | |
| "rewards/chosen": -183.8093719482422, | |
| "rewards/margins": 83.0264663696289, | |
| "rewards/rejected": -266.8358154296875, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.478806907378336, | |
| "grad_norm": 12.469826515590201, | |
| "learning_rate": 3.11720974905373e-07, | |
| "logits/chosen": 6012.91552734375, | |
| "logits/rejected": 5177.1376953125, | |
| "logps/chosen": -345.945068359375, | |
| "logps/rejected": -407.5060729980469, | |
| "loss": 0.6138, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": -139.82691955566406, | |
| "rewards/margins": 82.2566146850586, | |
| "rewards/rejected": -222.0835418701172, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.48665620094191525, | |
| "grad_norm": 13.764094508128613, | |
| "learning_rate": 3.0505737543712275e-07, | |
| "logits/chosen": 5078.953125, | |
| "logits/rejected": 4125.95361328125, | |
| "logps/chosen": -322.87664794921875, | |
| "logps/rejected": -369.62127685546875, | |
| "loss": 0.6162, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -139.11798095703125, | |
| "rewards/margins": 69.40438079833984, | |
| "rewards/rejected": -208.52236938476562, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.4945054945054945, | |
| "grad_norm": 27.054436253186108, | |
| "learning_rate": 2.9835240279702513e-07, | |
| "logits/chosen": 6608.65478515625, | |
| "logits/rejected": 5554.52587890625, | |
| "logps/chosen": -362.4518127441406, | |
| "logps/rejected": -418.16998291015625, | |
| "loss": 0.6106, | |
| "rewards/accuracies": 0.7916666865348816, | |
| "rewards/chosen": -119.63581848144531, | |
| "rewards/margins": 104.2861328125, | |
| "rewards/rejected": -223.9219512939453, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.5023547880690737, | |
| "grad_norm": 17.671754927072275, | |
| "learning_rate": 2.9161109547416667e-07, | |
| "logits/chosen": 6268.4951171875, | |
| "logits/rejected": 5317.8037109375, | |
| "logps/chosen": -364.1328125, | |
| "logps/rejected": -429.2177734375, | |
| "loss": 0.6135, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -153.37124633789062, | |
| "rewards/margins": 61.33759689331055, | |
| "rewards/rejected": -214.70883178710938, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5102040816326531, | |
| "grad_norm": 14.137157591920442, | |
| "learning_rate": 2.848385192615339e-07, | |
| "logits/chosen": 5331.8037109375, | |
| "logits/rejected": 4288.7294921875, | |
| "logps/chosen": -347.09747314453125, | |
| "logps/rejected": -396.7239074707031, | |
| "loss": 0.6211, | |
| "rewards/accuracies": 0.8000000715255737, | |
| "rewards/chosen": -150.75814819335938, | |
| "rewards/margins": 77.23800659179688, | |
| "rewards/rejected": -227.9961395263672, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5180533751962323, | |
| "grad_norm": 16.88587503102184, | |
| "learning_rate": 2.780397634492949e-07, | |
| "logits/chosen": 5999.3125, | |
| "logits/rejected": 4691.3330078125, | |
| "logps/chosen": -358.98828125, | |
| "logps/rejected": -429.1502380371094, | |
| "loss": 0.6216, | |
| "rewards/accuracies": 0.8166666030883789, | |
| "rewards/chosen": -141.60177612304688, | |
| "rewards/margins": 99.60104370117188, | |
| "rewards/rejected": -241.2028045654297, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5259026687598116, | |
| "grad_norm": 22.96076824322748, | |
| "learning_rate": 2.71219937000424e-07, | |
| "logits/chosen": 6003.61962890625, | |
| "logits/rejected": 4884.7255859375, | |
| "logps/chosen": -345.4893493652344, | |
| "logps/rejected": -394.2846374511719, | |
| "loss": 0.6262, | |
| "rewards/accuracies": 0.7666667103767395, | |
| "rewards/chosen": -138.72991943359375, | |
| "rewards/margins": 66.7003402709961, | |
| "rewards/rejected": -205.4302520751953, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.533751962323391, | |
| "grad_norm": 22.813069309884035, | |
| "learning_rate": 2.6438416471154273e-07, | |
| "logits/chosen": 5843.70458984375, | |
| "logits/rejected": 4834.39599609375, | |
| "logps/chosen": -359.9562683105469, | |
| "logps/rejected": -401.135498046875, | |
| "loss": 0.6166, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": -146.29505920410156, | |
| "rewards/margins": 75.61378479003906, | |
| "rewards/rejected": -221.9088592529297, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5416012558869702, | |
| "grad_norm": 18.765764198829793, | |
| "learning_rate": 2.5753758336186326e-07, | |
| "logits/chosen": 5820.8349609375, | |
| "logits/rejected": 5283.55078125, | |
| "logps/chosen": -351.8568420410156, | |
| "logps/rejected": -444.1844787597656, | |
| "loss": 0.6021, | |
| "rewards/accuracies": 0.7916666269302368, | |
| "rewards/chosen": -142.75038146972656, | |
| "rewards/margins": 85.23744201660156, | |
| "rewards/rejected": -227.9878387451172, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5494505494505495, | |
| "grad_norm": 14.748661687164306, | |
| "learning_rate": 2.5068533785312666e-07, | |
| "logits/chosen": 5523.3955078125, | |
| "logits/rejected": 5268.107421875, | |
| "logps/chosen": -343.63934326171875, | |
| "logps/rejected": -420.27886962890625, | |
| "loss": 0.6369, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -154.3790283203125, | |
| "rewards/margins": 78.92584991455078, | |
| "rewards/rejected": -233.3048553466797, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5572998430141287, | |
| "grad_norm": 15.464054804569159, | |
| "learning_rate": 2.4383257734343794e-07, | |
| "logits/chosen": 5437.234375, | |
| "logits/rejected": 5426.69677734375, | |
| "logps/chosen": -344.1163330078125, | |
| "logps/rejected": -430.07904052734375, | |
| "loss": 0.608, | |
| "rewards/accuracies": 0.7500000596046448, | |
| "rewards/chosen": -150.20494079589844, | |
| "rewards/margins": 77.2534408569336, | |
| "rewards/rejected": -227.4583740234375, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.565149136577708, | |
| "grad_norm": 18.732066069845217, | |
| "learning_rate": 2.3698445137790258e-07, | |
| "logits/chosen": 5845.06640625, | |
| "logits/rejected": 4940.61865234375, | |
| "logps/chosen": -361.6242370605469, | |
| "logps/rejected": -430.0552673339844, | |
| "loss": 0.6127, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -142.62103271484375, | |
| "rewards/margins": 87.6690902709961, | |
| "rewards/rejected": -230.2901153564453, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5729984301412873, | |
| "grad_norm": 17.14795101097142, | |
| "learning_rate": 2.3014610601897157e-07, | |
| "logits/chosen": 6328.41552734375, | |
| "logits/rejected": 4775.5185546875, | |
| "logps/chosen": -369.47235107421875, | |
| "logps/rejected": -404.1590881347656, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": -144.32643127441406, | |
| "rewards/margins": 84.75894165039062, | |
| "rewards/rejected": -229.08535766601562, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5808477237048666, | |
| "grad_norm": 13.832807015732437, | |
| "learning_rate": 2.2332267997940513e-07, | |
| "logits/chosen": 5258.61572265625, | |
| "logits/rejected": 4415.40576171875, | |
| "logps/chosen": -324.30242919921875, | |
| "logps/rejected": -375.59405517578125, | |
| "loss": 0.6053, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": -131.87802124023438, | |
| "rewards/margins": 81.06999206542969, | |
| "rewards/rejected": -212.947998046875, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5886970172684458, | |
| "grad_norm": 15.482096718237434, | |
| "learning_rate": 2.1651930076075723e-07, | |
| "logits/chosen": 5748.50830078125, | |
| "logits/rejected": 5138.494140625, | |
| "logps/chosen": -329.36572265625, | |
| "logps/rejected": -381.31109619140625, | |
| "loss": 0.6253, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -148.93975830078125, | |
| "rewards/margins": 65.27706909179688, | |
| "rewards/rejected": -214.21682739257812, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5965463108320251, | |
| "grad_norm": 16.50457423188824, | |
| "learning_rate": 2.0974108080028692e-07, | |
| "logits/chosen": 6050.1806640625, | |
| "logits/rejected": 4695.55126953125, | |
| "logps/chosen": -332.60772705078125, | |
| "logps/rejected": -383.9396057128906, | |
| "loss": 0.615, | |
| "rewards/accuracies": 0.7666667103767395, | |
| "rewards/chosen": -129.35684204101562, | |
| "rewards/margins": 80.0522232055664, | |
| "rewards/rejected": -209.4090576171875, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.6043956043956044, | |
| "grad_norm": 16.19292968682205, | |
| "learning_rate": 2.0299311362918773e-07, | |
| "logits/chosen": 6270.7763671875, | |
| "logits/rejected": 5331.49755859375, | |
| "logps/chosen": -374.3070373535156, | |
| "logps/rejected": -449.4652404785156, | |
| "loss": 0.6274, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -144.57337951660156, | |
| "rewards/margins": 81.0317153930664, | |
| "rewards/rejected": -225.60513305664062, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.6122448979591837, | |
| "grad_norm": 13.854569491977498, | |
| "learning_rate": 1.962804700450265e-07, | |
| "logits/chosen": 6101.94970703125, | |
| "logits/rejected": 5776.4658203125, | |
| "logps/chosen": -351.2977294921875, | |
| "logps/rejected": -441.8768005371094, | |
| "loss": 0.6162, | |
| "rewards/accuracies": 0.73333340883255, | |
| "rewards/chosen": -135.1923828125, | |
| "rewards/margins": 69.29673767089844, | |
| "rewards/rejected": -204.4891357421875, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.6200941915227629, | |
| "grad_norm": 26.357276507505432, | |
| "learning_rate": 1.8960819430126334e-07, | |
| "logits/chosen": 5692.7373046875, | |
| "logits/rejected": 4988.34423828125, | |
| "logps/chosen": -352.75152587890625, | |
| "logps/rejected": -436.21380615234375, | |
| "loss": 0.606, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -154.62054443359375, | |
| "rewards/margins": 92.99554443359375, | |
| "rewards/rejected": -247.61605834960938, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6279434850863422, | |
| "grad_norm": 26.128991277323944, | |
| "learning_rate": 1.8298130031671972e-07, | |
| "logits/chosen": 5666.7470703125, | |
| "logits/rejected": 4928.1630859375, | |
| "logps/chosen": -374.0252380371094, | |
| "logps/rejected": -446.89398193359375, | |
| "loss": 0.6227, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -157.30239868164062, | |
| "rewards/margins": 76.89336395263672, | |
| "rewards/rejected": -234.1957550048828, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6357927786499215, | |
| "grad_norm": 21.20668683192694, | |
| "learning_rate": 1.7640476790784075e-07, | |
| "logits/chosen": 5246.98876953125, | |
| "logits/rejected": 4623.890625, | |
| "logps/chosen": -351.83685302734375, | |
| "logps/rejected": -468.95904541015625, | |
| "loss": 0.6118, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": -149.3814239501953, | |
| "rewards/margins": 98.28030395507812, | |
| "rewards/rejected": -247.6617431640625, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6436420722135008, | |
| "grad_norm": 20.646225897051636, | |
| "learning_rate": 1.6988353904658492e-07, | |
| "logits/chosen": 5737.23974609375, | |
| "logits/rejected": 4359.0126953125, | |
| "logps/chosen": -371.08319091796875, | |
| "logps/rejected": -399.9681396484375, | |
| "loss": 0.6117, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": -149.86865234375, | |
| "rewards/margins": 81.41114807128906, | |
| "rewards/rejected": -231.27981567382812, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.6514913657770801, | |
| "grad_norm": 39.77030169382563, | |
| "learning_rate": 1.634225141467513e-07, | |
| "logits/chosen": 5644.4755859375, | |
| "logits/rejected": 4988.80859375, | |
| "logps/chosen": -366.618408203125, | |
| "logps/rejected": -447.57489013671875, | |
| "loss": 0.6127, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -162.43235778808594, | |
| "rewards/margins": 92.68769836425781, | |
| "rewards/rejected": -255.1200408935547, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.6593406593406593, | |
| "grad_norm": 13.895052837155353, | |
| "learning_rate": 1.570265483815364e-07, | |
| "logits/chosen": 6220.3994140625, | |
| "logits/rejected": 5031.82080078125, | |
| "logps/chosen": -372.7491760253906, | |
| "logps/rejected": -434.67779541015625, | |
| "loss": 0.6189, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -145.41937255859375, | |
| "rewards/margins": 76.46521759033203, | |
| "rewards/rejected": -221.8845977783203, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6671899529042387, | |
| "grad_norm": 12.774810316086143, | |
| "learning_rate": 1.5070044803508691e-07, | |
| "logits/chosen": 5741.9931640625, | |
| "logits/rejected": 5091.0048828125, | |
| "logps/chosen": -347.2725524902344, | |
| "logps/rejected": -424.1080017089844, | |
| "loss": 0.5974, | |
| "rewards/accuracies": 0.7750000357627869, | |
| "rewards/chosen": -131.7950897216797, | |
| "rewards/margins": 91.85758209228516, | |
| "rewards/rejected": -223.6526641845703, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6750392464678179, | |
| "grad_norm": 17.871490937493935, | |
| "learning_rate": 1.444489668907914e-07, | |
| "logits/chosen": 6167.5400390625, | |
| "logits/rejected": 5200.32568359375, | |
| "logps/chosen": -374.5239562988281, | |
| "logps/rejected": -405.5271301269531, | |
| "loss": 0.6325, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -126.7532958984375, | |
| "rewards/margins": 76.09200286865234, | |
| "rewards/rejected": -202.84530639648438, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6828885400313972, | |
| "grad_norm": 17.06846349127134, | |
| "learning_rate": 1.3827680265902232e-07, | |
| "logits/chosen": 6157.00634765625, | |
| "logits/rejected": 5048.666015625, | |
| "logps/chosen": -344.4704895019531, | |
| "logps/rejected": -391.15289306640625, | |
| "loss": 0.6177, | |
| "rewards/accuracies": 0.7583333849906921, | |
| "rewards/chosen": -114.37376403808594, | |
| "rewards/margins": 76.09016418457031, | |
| "rewards/rejected": -190.4639434814453, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6907378335949764, | |
| "grad_norm": 15.528467832759812, | |
| "learning_rate": 1.3218859344701632e-07, | |
| "logits/chosen": 5415.63671875, | |
| "logits/rejected": 5133.5498046875, | |
| "logps/chosen": -312.1150817871094, | |
| "logps/rejected": -407.8897705078125, | |
| "loss": 0.6164, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": -101.6041259765625, | |
| "rewards/margins": 74.69647216796875, | |
| "rewards/rejected": -176.3006134033203, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6985871271585558, | |
| "grad_norm": 17.183055961017374, | |
| "learning_rate": 1.2618891427354172e-07, | |
| "logits/chosen": 6400.65478515625, | |
| "logits/rejected": 5158.58837890625, | |
| "logps/chosen": -360.47662353515625, | |
| "logps/rejected": -399.84051513671875, | |
| "loss": 0.6248, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -102.88056945800781, | |
| "rewards/margins": 85.1794204711914, | |
| "rewards/rejected": -188.0599822998047, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.706436420722135, | |
| "grad_norm": 14.954999382868223, | |
| "learning_rate": 1.202822736309758e-07, | |
| "logits/chosen": 5437.9140625, | |
| "logits/rejected": 4952.78515625, | |
| "logps/chosen": -315.7103576660156, | |
| "logps/rejected": -405.45086669921875, | |
| "loss": 0.6149, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -111.3611831665039, | |
| "rewards/margins": 83.65962982177734, | |
| "rewards/rejected": -195.02081298828125, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 17.482836351512447, | |
| "learning_rate": 1.1447311009737299e-07, | |
| "logits/chosen": 5352.5107421875, | |
| "logits/rejected": 5015.8212890625, | |
| "logps/chosen": -335.0443420410156, | |
| "logps/rejected": -422.0838317871094, | |
| "loss": 0.6238, | |
| "rewards/accuracies": 0.7916666269302368, | |
| "rewards/chosen": -126.8180160522461, | |
| "rewards/margins": 87.3373031616211, | |
| "rewards/rejected": -214.1553497314453, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.7221350078492935, | |
| "grad_norm": 14.726061760391564, | |
| "learning_rate": 1.0876578900107053e-07, | |
| "logits/chosen": 5908.712890625, | |
| "logits/rejected": 4864.37109375, | |
| "logps/chosen": -364.682861328125, | |
| "logps/rejected": -407.4881591796875, | |
| "loss": 0.6111, | |
| "rewards/accuracies": 0.8333333134651184, | |
| "rewards/chosen": -132.26129150390625, | |
| "rewards/margins": 77.08457946777344, | |
| "rewards/rejected": -209.34585571289062, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.7299843014128728, | |
| "grad_norm": 12.17491356029005, | |
| "learning_rate": 1.0316459914033793e-07, | |
| "logits/chosen": 5847.6396484375, | |
| "logits/rejected": 4304.47998046875, | |
| "logps/chosen": -374.629638671875, | |
| "logps/rejected": -411.969482421875, | |
| "loss": 0.6084, | |
| "rewards/accuracies": 0.783333420753479, | |
| "rewards/chosen": -138.6407012939453, | |
| "rewards/margins": 87.40455627441406, | |
| "rewards/rejected": -226.04525756835938, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7378335949764521, | |
| "grad_norm": 14.942393863065405, | |
| "learning_rate": 9.767374956053584e-08, | |
| "logits/chosen": 5655.4033203125, | |
| "logits/rejected": 4892.77978515625, | |
| "logps/chosen": -362.05084228515625, | |
| "logps/rejected": -447.89691162109375, | |
| "loss": 0.616, | |
| "rewards/accuracies": 0.7500000596046448, | |
| "rewards/chosen": -143.75718688964844, | |
| "rewards/margins": 99.15988159179688, | |
| "rewards/rejected": -242.91708374023438, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.7456828885400314, | |
| "grad_norm": 15.68275244303951, | |
| "learning_rate": 9.229736639120561e-08, | |
| "logits/chosen": 5820.70068359375, | |
| "logits/rejected": 5320.892578125, | |
| "logps/chosen": -365.70635986328125, | |
| "logps/rejected": -428.5652770996094, | |
| "loss": 0.631, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -147.90518188476562, | |
| "rewards/margins": 66.50559997558594, | |
| "rewards/rejected": -214.4107666015625, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7535321821036107, | |
| "grad_norm": 19.23996720411505, | |
| "learning_rate": 8.70394897454659e-08, | |
| "logits/chosen": 5713.681640625, | |
| "logits/rejected": 5020.9267578125, | |
| "logps/chosen": -342.4471130371094, | |
| "logps/rejected": -417.79901123046875, | |
| "loss": 0.6123, | |
| "rewards/accuracies": 0.8333333730697632, | |
| "rewards/chosen": -124.01869201660156, | |
| "rewards/margins": 91.303955078125, | |
| "rewards/rejected": -215.32266235351562, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7613814756671899, | |
| "grad_norm": 22.33397911127439, | |
| "learning_rate": 8.19040706840472e-08, | |
| "logits/chosen": 5806.43359375, | |
| "logits/rejected": 4807.3515625, | |
| "logps/chosen": -376.2249755859375, | |
| "logps/rejected": -434.6560974121094, | |
| "loss": 0.6119, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": -136.00497436523438, | |
| "rewards/margins": 91.96058654785156, | |
| "rewards/rejected": -227.965576171875, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 13.761008157112903, | |
| "learning_rate": 7.689496824624525e-08, | |
| "logits/chosen": 5534.2548828125, | |
| "logits/rejected": 4398.90087890625, | |
| "logps/chosen": -362.1374816894531, | |
| "logps/rejected": -422.9102478027344, | |
| "loss": 0.6029, | |
| "rewards/accuracies": 0.8083332777023315, | |
| "rewards/chosen": -133.32089233398438, | |
| "rewards/margins": 99.09576416015625, | |
| "rewards/rejected": -232.4166717529297, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7770800627943485, | |
| "grad_norm": 18.096681200370565, | |
| "learning_rate": 7.201594655002458e-08, | |
| "logits/chosen": 5845.0439453125, | |
| "logits/rejected": 4824.6376953125, | |
| "logps/chosen": -375.0849914550781, | |
| "logps/rejected": -435.990478515625, | |
| "loss": 0.599, | |
| "rewards/accuracies": 0.783333420753479, | |
| "rewards/chosen": -149.19004821777344, | |
| "rewards/margins": 92.73072814941406, | |
| "rewards/rejected": -241.92074584960938, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7849293563579278, | |
| "grad_norm": 19.80400883141873, | |
| "learning_rate": 6.727067196345099e-08, | |
| "logits/chosen": 5522.568359375, | |
| "logits/rejected": 4610.0341796875, | |
| "logps/chosen": -366.5689392089844, | |
| "logps/rejected": -410.90081787109375, | |
| "loss": 0.6154, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -152.8131103515625, | |
| "rewards/margins": 77.74247741699219, | |
| "rewards/rejected": -230.55557250976562, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.792778649921507, | |
| "grad_norm": 19.909366846280882, | |
| "learning_rate": 6.26627103495786e-08, | |
| "logits/chosen": 5702.2314453125, | |
| "logits/rejected": 4692.59130859375, | |
| "logps/chosen": -357.6479187011719, | |
| "logps/rejected": -417.7315368652344, | |
| "loss": 0.6253, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -142.938720703125, | |
| "rewards/margins": 83.35922241210938, | |
| "rewards/rejected": -226.2979736328125, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.8006279434850864, | |
| "grad_norm": 13.222506396766855, | |
| "learning_rate": 5.8195524386862374e-08, | |
| "logits/chosen": 5807.0205078125, | |
| "logits/rejected": 5114.4541015625, | |
| "logps/chosen": -379.9997863769531, | |
| "logps/rejected": -456.28125, | |
| "loss": 0.6175, | |
| "rewards/accuracies": 0.8166667222976685, | |
| "rewards/chosen": -131.8211669921875, | |
| "rewards/margins": 98.37938690185547, | |
| "rewards/rejected": -230.20053100585938, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.8084772370486656, | |
| "grad_norm": 14.840716751556135, | |
| "learning_rate": 5.38724709671092e-08, | |
| "logits/chosen": 6213.39697265625, | |
| "logits/rejected": 5793.4111328125, | |
| "logps/chosen": -374.4881286621094, | |
| "logps/rejected": -466.87982177734375, | |
| "loss": 0.6064, | |
| "rewards/accuracies": 0.7916666865348816, | |
| "rewards/chosen": -143.019775390625, | |
| "rewards/margins": 90.74076843261719, | |
| "rewards/rejected": -233.7605438232422, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.8163265306122449, | |
| "grad_norm": 15.902716058683295, | |
| "learning_rate": 4.969679867292276e-08, | |
| "logits/chosen": 5481.22265625, | |
| "logits/rejected": 4915.169921875, | |
| "logps/chosen": -365.6921691894531, | |
| "logps/rejected": -450.8426818847656, | |
| "loss": 0.6114, | |
| "rewards/accuracies": 0.7583333849906921, | |
| "rewards/chosen": -145.59046936035156, | |
| "rewards/margins": 96.06826782226562, | |
| "rewards/rejected": -241.6587371826172, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.8241758241758241, | |
| "grad_norm": 15.329902609184328, | |
| "learning_rate": 4.5671645336537416e-08, | |
| "logits/chosen": 5554.18017578125, | |
| "logits/rejected": 4997.1005859375, | |
| "logps/chosen": -381.65789794921875, | |
| "logps/rejected": -449.450927734375, | |
| "loss": 0.6103, | |
| "rewards/accuracies": 0.7916666865348816, | |
| "rewards/chosen": -144.17726135253906, | |
| "rewards/margins": 89.47932434082031, | |
| "rewards/rejected": -233.6565399169922, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.8320251177394035, | |
| "grad_norm": 55.197735270402006, | |
| "learning_rate": 4.180003568187776e-08, | |
| "logits/chosen": 6873.0361328125, | |
| "logits/rejected": 5345.8271484375, | |
| "logps/chosen": -410.3626403808594, | |
| "logps/rejected": -439.2491149902344, | |
| "loss": 0.6283, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -149.37527465820312, | |
| "rewards/margins": 69.83467102050781, | |
| "rewards/rejected": -219.20993041992188, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.8398744113029827, | |
| "grad_norm": 19.762608641409, | |
| "learning_rate": 3.8084879051612144e-08, | |
| "logits/chosen": 5707.0205078125, | |
| "logits/rejected": 5145.00927734375, | |
| "logps/chosen": -366.8685302734375, | |
| "logps/rejected": -423.092529296875, | |
| "loss": 0.6147, | |
| "rewards/accuracies": 0.7583333849906921, | |
| "rewards/chosen": -141.62551879882812, | |
| "rewards/margins": 88.30266571044922, | |
| "rewards/rejected": -229.9281768798828, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.847723704866562, | |
| "grad_norm": 16.196550037488134, | |
| "learning_rate": 3.452896722091128e-08, | |
| "logits/chosen": 6264.1298828125, | |
| "logits/rejected": 4766.61083984375, | |
| "logps/chosen": -405.56341552734375, | |
| "logps/rejected": -432.5325622558594, | |
| "loss": 0.6033, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -138.9940948486328, | |
| "rewards/margins": 92.55856323242188, | |
| "rewards/rejected": -231.55264282226562, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8555729984301413, | |
| "grad_norm": 20.291014842445787, | |
| "learning_rate": 3.11349722995527e-08, | |
| "logits/chosen": 6290.5830078125, | |
| "logits/rejected": 4671.265625, | |
| "logps/chosen": -373.1634826660156, | |
| "logps/rejected": -431.679931640625, | |
| "loss": 0.6037, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -145.68887329101562, | |
| "rewards/margins": 73.10652160644531, | |
| "rewards/rejected": -218.79537963867188, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8634222919937206, | |
| "grad_norm": 17.99288812895582, | |
| "learning_rate": 2.7905444723949762e-08, | |
| "logits/chosen": 6118.28271484375, | |
| "logits/rejected": 5002.5498046875, | |
| "logps/chosen": -382.5477294921875, | |
| "logps/rejected": -433.43798828125, | |
| "loss": 0.6151, | |
| "rewards/accuracies": 0.7750000953674316, | |
| "rewards/chosen": -144.01138305664062, | |
| "rewards/margins": 99.92918395996094, | |
| "rewards/rejected": -243.9405517578125, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.8712715855572999, | |
| "grad_norm": 18.552548907991827, | |
| "learning_rate": 2.484281134061142e-08, | |
| "logits/chosen": 6458.51123046875, | |
| "logits/rejected": 5143.70263671875, | |
| "logps/chosen": -413.1473693847656, | |
| "logps/rejected": -460.0634765625, | |
| "loss": 0.6059, | |
| "rewards/accuracies": 0.7916666865348816, | |
| "rewards/chosen": -148.53378295898438, | |
| "rewards/margins": 89.26164245605469, | |
| "rewards/rejected": -237.79541015625, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8791208791208791, | |
| "grad_norm": 39.56227188436796, | |
| "learning_rate": 2.194937358247506e-08, | |
| "logits/chosen": 6335.6669921875, | |
| "logits/rejected": 5074.86474609375, | |
| "logps/chosen": -396.36737060546875, | |
| "logps/rejected": -461.27423095703125, | |
| "loss": 0.608, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -151.22695922851562, | |
| "rewards/margins": 93.27647399902344, | |
| "rewards/rejected": -244.50341796875, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8869701726844584, | |
| "grad_norm": 18.148758631996266, | |
| "learning_rate": 1.9227305739481612e-08, | |
| "logits/chosen": 5765.23486328125, | |
| "logits/rejected": 4463.29541015625, | |
| "logps/chosen": -364.4913330078125, | |
| "logps/rejected": -408.61187744140625, | |
| "loss": 0.6017, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": -129.7092742919922, | |
| "rewards/margins": 98.6770248413086, | |
| "rewards/rejected": -228.3863067626953, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8948194662480377, | |
| "grad_norm": 16.994877433404973, | |
| "learning_rate": 1.6678653324693787e-08, | |
| "logits/chosen": 6387.3359375, | |
| "logits/rejected": 5097.41015625, | |
| "logps/chosen": -400.7387390136719, | |
| "logps/rejected": -451.8038024902344, | |
| "loss": 0.5946, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -143.11801147460938, | |
| "rewards/margins": 87.87274932861328, | |
| "rewards/rejected": -230.99075317382812, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.902668759811617, | |
| "grad_norm": 11.193775448088072, | |
| "learning_rate": 1.4305331537183384e-08, | |
| "logits/chosen": 5634.30419921875, | |
| "logits/rejected": 5102.05615234375, | |
| "logps/chosen": -367.156005859375, | |
| "logps/rejected": -442.4237365722656, | |
| "loss": 0.6015, | |
| "rewards/accuracies": 0.7166666984558105, | |
| "rewards/chosen": -140.172607421875, | |
| "rewards/margins": 82.76316833496094, | |
| "rewards/rejected": -222.935791015625, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.9105180533751962, | |
| "grad_norm": 18.50142880229726, | |
| "learning_rate": 1.2109123822844653e-08, | |
| "logits/chosen": 5791.82861328125, | |
| "logits/rejected": 4512.34228515625, | |
| "logps/chosen": -374.96600341796875, | |
| "logps/rejected": -425.4412536621094, | |
| "loss": 0.6118, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": -146.40921020507812, | |
| "rewards/margins": 83.09394836425781, | |
| "rewards/rejected": -229.50314331054688, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.9183673469387755, | |
| "grad_norm": 16.45322143829113, | |
| "learning_rate": 1.0091680534213387e-08, | |
| "logits/chosen": 6355.1123046875, | |
| "logits/rejected": 6031.12744140625, | |
| "logps/chosen": -378.3938903808594, | |
| "logps/rejected": -480.53863525390625, | |
| "loss": 0.6105, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -135.44552612304688, | |
| "rewards/margins": 98.34616088867188, | |
| "rewards/rejected": -233.7917022705078, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.9262166405023547, | |
| "grad_norm": 23.32993982299152, | |
| "learning_rate": 8.254517690300944e-09, | |
| "logits/chosen": 5549.52392578125, | |
| "logits/rejected": 4957.7138671875, | |
| "logps/chosen": -378.49566650390625, | |
| "logps/rejected": -448.3548889160156, | |
| "loss": 0.6038, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -138.54971313476562, | |
| "rewards/margins": 92.9834213256836, | |
| "rewards/rejected": -231.5331268310547, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.9340659340659341, | |
| "grad_norm": 11.697466694155857, | |
| "learning_rate": 6.599015837372907e-09, | |
| "logits/chosen": 6053.71044921875, | |
| "logits/rejected": 5201.91845703125, | |
| "logps/chosen": -410.1524963378906, | |
| "logps/rejected": -463.343994140625, | |
| "loss": 0.6132, | |
| "rewards/accuracies": 0.7166666388511658, | |
| "rewards/chosen": -161.26699829101562, | |
| "rewards/margins": 84.92450714111328, | |
| "rewards/rejected": -246.1914825439453, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.9419152276295133, | |
| "grad_norm": 29.00561816867174, | |
| "learning_rate": 5.126419011529992e-09, | |
| "logits/chosen": 6306.44140625, | |
| "logits/rejected": 5252.5625, | |
| "logps/chosen": -392.0524597167969, | |
| "logps/rejected": -466.8185119628906, | |
| "loss": 0.6043, | |
| "rewards/accuracies": 0.8333333134651184, | |
| "rewards/chosen": -136.3907470703125, | |
| "rewards/margins": 111.76222229003906, | |
| "rewards/rejected": -248.15292358398438, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9497645211930926, | |
| "grad_norm": 24.03652143626847, | |
| "learning_rate": 3.837833803870177e-09, | |
| "logits/chosen": 5867.791015625, | |
| "logits/rejected": 5074.435546875, | |
| "logps/chosen": -392.6157531738281, | |
| "logps/rejected": -466.23956298828125, | |
| "loss": 0.6164, | |
| "rewards/accuracies": 0.783333420753479, | |
| "rewards/chosen": -152.4302520751953, | |
| "rewards/margins": 95.60807037353516, | |
| "rewards/rejected": -248.03829956054688, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.957613814756672, | |
| "grad_norm": 23.781485972909927, | |
| "learning_rate": 2.734228528934679e-09, | |
| "logits/chosen": 7340.81494140625, | |
| "logits/rejected": 5272.85400390625, | |
| "logps/chosen": -449.55072021484375, | |
| "logps/rejected": -496.0115661621094, | |
| "loss": 0.6098, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -152.78236389160156, | |
| "rewards/margins": 98.0320053100586, | |
| "rewards/rejected": -250.8143768310547, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9654631083202512, | |
| "grad_norm": 25.628197989466248, | |
| "learning_rate": 1.8164324970625645e-09, | |
| "logits/chosen": 6506.375, | |
| "logits/rejected": 5016.78759765625, | |
| "logps/chosen": -400.7377624511719, | |
| "logps/rejected": -448.590576171875, | |
| "loss": 0.624, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -140.1560821533203, | |
| "rewards/margins": 95.43777465820312, | |
| "rewards/rejected": -235.59384155273438, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9733124018838305, | |
| "grad_norm": 19.957302011366814, | |
| "learning_rate": 1.0851353912008642e-09, | |
| "logits/chosen": 5557.27685546875, | |
| "logits/rejected": 5068.2919921875, | |
| "logps/chosen": -379.62493896484375, | |
| "logps/rejected": -472.33477783203125, | |
| "loss": 0.6141, | |
| "rewards/accuracies": 0.7333332896232605, | |
| "rewards/chosen": -148.0125732421875, | |
| "rewards/margins": 89.1643295288086, | |
| "rewards/rejected": -237.17691040039062, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.9811616954474097, | |
| "grad_norm": 15.727811449834665, | |
| "learning_rate": 5.408867486384471e-10, | |
| "logits/chosen": 5699.9697265625, | |
| "logits/rejected": 4726.44580078125, | |
| "logps/chosen": -367.1297912597656, | |
| "logps/rejected": -415.16668701171875, | |
| "loss": 0.6068, | |
| "rewards/accuracies": 0.7833332419395447, | |
| "rewards/chosen": -136.82162475585938, | |
| "rewards/margins": 88.91475677490234, | |
| "rewards/rejected": -225.7363739013672, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.989010989010989, | |
| "grad_norm": 18.48673738798219, | |
| "learning_rate": 1.840955480532924e-10, | |
| "logits/chosen": 5415.2001953125, | |
| "logits/rejected": 5022.5712890625, | |
| "logps/chosen": -370.6057434082031, | |
| "logps/rejected": -442.42633056640625, | |
| "loss": 0.6035, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": -138.44517517089844, | |
| "rewards/margins": 87.65608215332031, | |
| "rewards/rejected": -226.10122680664062, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9968602825745683, | |
| "grad_norm": 20.059405514745603, | |
| "learning_rate": 1.502990218302247e-11, | |
| "logits/chosen": 5661.7509765625, | |
| "logits/rejected": 4540.9052734375, | |
| "logps/chosen": -372.86297607421875, | |
| "logps/rejected": -427.48162841796875, | |
| "loss": 0.6166, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -148.38900756835938, | |
| "rewards/margins": 92.78841400146484, | |
| "rewards/rejected": -241.1774139404297, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1274, | |
| "total_flos": 0.0, | |
| "train_loss": 0.6278583559922558, | |
| "train_runtime": 14837.7437, | |
| "train_samples_per_second": 4.12, | |
| "train_steps_per_second": 0.086 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1274, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 6, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |