|
{ |
|
"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": 76.09110305026772, |
|
"learning_rate": 3.90625e-09, |
|
"logits/chosen": 5881.4375, |
|
"logits/rejected": 2834.66162109375, |
|
"logps/chosen": -257.5969543457031, |
|
"logps/rejected": -120.09489440917969, |
|
"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": 84.53465777823584, |
|
"learning_rate": 3.9062499999999997e-08, |
|
"logits/chosen": 4946.009765625, |
|
"logits/rejected": 4332.37451171875, |
|
"logps/chosen": -248.8301544189453, |
|
"logps/rejected": -214.27517700195312, |
|
"loss": 0.6939, |
|
"rewards/accuracies": 0.43518519401550293, |
|
"rewards/chosen": -0.013952851295471191, |
|
"rewards/margins": -0.030882200226187706, |
|
"rewards/rejected": 0.016929350793361664, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.015698587127158554, |
|
"grad_norm": 70.05213184547209, |
|
"learning_rate": 7.812499999999999e-08, |
|
"logits/chosen": 6023.31396484375, |
|
"logits/rejected": 4842.8310546875, |
|
"logps/chosen": -266.2083435058594, |
|
"logps/rejected": -232.37954711914062, |
|
"loss": 0.6938, |
|
"rewards/accuracies": 0.5000000596046448, |
|
"rewards/chosen": 0.10495556890964508, |
|
"rewards/margins": 0.032393090426921844, |
|
"rewards/rejected": 0.07256247103214264, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.023547880690737835, |
|
"grad_norm": 69.00995049951652, |
|
"learning_rate": 1.1718749999999999e-07, |
|
"logits/chosen": 6032.76123046875, |
|
"logits/rejected": 5038.65087890625, |
|
"logps/chosen": -310.43328857421875, |
|
"logps/rejected": -262.861572265625, |
|
"loss": 0.6926, |
|
"rewards/accuracies": 0.5416666865348816, |
|
"rewards/chosen": 0.17605653405189514, |
|
"rewards/margins": 0.0010276169050484896, |
|
"rewards/rejected": 0.17502892017364502, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.03139717425431711, |
|
"grad_norm": 69.7664006852952, |
|
"learning_rate": 1.5624999999999999e-07, |
|
"logits/chosen": 5242.107421875, |
|
"logits/rejected": 4344.2041015625, |
|
"logps/chosen": -263.4407043457031, |
|
"logps/rejected": -224.73001098632812, |
|
"loss": 0.6854, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": 0.5795542001724243, |
|
"rewards/margins": 0.2455345094203949, |
|
"rewards/rejected": 0.33401963114738464, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.03924646781789639, |
|
"grad_norm": 75.39254669427922, |
|
"learning_rate": 1.9531249999999998e-07, |
|
"logits/chosen": 6364.07373046875, |
|
"logits/rejected": 4975.279296875, |
|
"logps/chosen": -324.3768615722656, |
|
"logps/rejected": -257.8438415527344, |
|
"loss": 0.6764, |
|
"rewards/accuracies": 0.7083333134651184, |
|
"rewards/chosen": 1.7327311038970947, |
|
"rewards/margins": 1.0267372131347656, |
|
"rewards/rejected": 0.7059938907623291, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.04709576138147567, |
|
"grad_norm": 71.83951557562865, |
|
"learning_rate": 2.3437499999999998e-07, |
|
"logits/chosen": 5382.095703125, |
|
"logits/rejected": 4503.6728515625, |
|
"logps/chosen": -268.0951843261719, |
|
"logps/rejected": -261.60919189453125, |
|
"loss": 0.6609, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 1.8319120407104492, |
|
"rewards/margins": 1.2502626180648804, |
|
"rewards/rejected": 0.5816493630409241, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.054945054945054944, |
|
"grad_norm": 58.133090222213596, |
|
"learning_rate": 2.734375e-07, |
|
"logits/chosen": 5104.4208984375, |
|
"logits/rejected": 4838.51025390625, |
|
"logps/chosen": -231.4933624267578, |
|
"logps/rejected": -232.9977264404297, |
|
"loss": 0.6474, |
|
"rewards/accuracies": 0.675000011920929, |
|
"rewards/chosen": 1.7646211385726929, |
|
"rewards/margins": 2.1093406677246094, |
|
"rewards/rejected": -0.3447194993495941, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.06279434850863422, |
|
"grad_norm": 57.9049780596852, |
|
"learning_rate": 3.1249999999999997e-07, |
|
"logits/chosen": 5690.080078125, |
|
"logits/rejected": 5160.9189453125, |
|
"logps/chosen": -255.88723754882812, |
|
"logps/rejected": -245.4872283935547, |
|
"loss": 0.6237, |
|
"rewards/accuracies": 0.6666666269302368, |
|
"rewards/chosen": 0.0911921039223671, |
|
"rewards/margins": 3.2660727500915527, |
|
"rewards/rejected": -3.1748805046081543, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.0706436420722135, |
|
"grad_norm": 60.22967129442178, |
|
"learning_rate": 3.5156249999999997e-07, |
|
"logits/chosen": 5870.26171875, |
|
"logits/rejected": 5052.40869140625, |
|
"logps/chosen": -284.24462890625, |
|
"logps/rejected": -248.9542236328125, |
|
"loss": 0.5984, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": 0.9464850425720215, |
|
"rewards/margins": 6.434256076812744, |
|
"rewards/rejected": -5.487771034240723, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.07849293563579278, |
|
"grad_norm": 69.2909935844847, |
|
"learning_rate": 3.9062499999999997e-07, |
|
"logits/chosen": 5756.40869140625, |
|
"logits/rejected": 5542.99267578125, |
|
"logps/chosen": -278.2003173828125, |
|
"logps/rejected": -275.8293151855469, |
|
"loss": 0.61, |
|
"rewards/accuracies": 0.7083333730697632, |
|
"rewards/chosen": 0.07480049133300781, |
|
"rewards/margins": 7.715832710266113, |
|
"rewards/rejected": -7.641031742095947, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.08634222919937205, |
|
"grad_norm": 63.59587178292999, |
|
"learning_rate": 4.2968749999999996e-07, |
|
"logits/chosen": 6190.86376953125, |
|
"logits/rejected": 5089.04150390625, |
|
"logps/chosen": -257.26153564453125, |
|
"logps/rejected": -252.33932495117188, |
|
"loss": 0.5655, |
|
"rewards/accuracies": 0.7166666984558105, |
|
"rewards/chosen": -0.12369728088378906, |
|
"rewards/margins": 8.602740287780762, |
|
"rewards/rejected": -8.726436614990234, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.09419152276295134, |
|
"grad_norm": 75.09104325187393, |
|
"learning_rate": 4.6874999999999996e-07, |
|
"logits/chosen": 6041.5400390625, |
|
"logits/rejected": 5059.2177734375, |
|
"logps/chosen": -280.3619079589844, |
|
"logps/rejected": -283.30877685546875, |
|
"loss": 0.5634, |
|
"rewards/accuracies": 0.7083333730697632, |
|
"rewards/chosen": -0.7750555276870728, |
|
"rewards/margins": 12.931465148925781, |
|
"rewards/rejected": -13.706521987915039, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.10204081632653061, |
|
"grad_norm": 56.6049419522278, |
|
"learning_rate": 4.999962424962166e-07, |
|
"logits/chosen": 6000.89306640625, |
|
"logits/rejected": 5544.8837890625, |
|
"logps/chosen": -286.98211669921875, |
|
"logps/rejected": -278.4353332519531, |
|
"loss": 0.5788, |
|
"rewards/accuracies": 0.6916667222976685, |
|
"rewards/chosen": 3.1251087188720703, |
|
"rewards/margins": 8.129714012145996, |
|
"rewards/rejected": -5.004604816436768, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.10989010989010989, |
|
"grad_norm": 66.17159473008851, |
|
"learning_rate": 4.998647417232375e-07, |
|
"logits/chosen": 5857.5234375, |
|
"logits/rejected": 5142.9794921875, |
|
"logps/chosen": -259.30584716796875, |
|
"logps/rejected": -257.5000915527344, |
|
"loss": 0.5949, |
|
"rewards/accuracies": 0.6666666269302368, |
|
"rewards/chosen": 0.936366081237793, |
|
"rewards/margins": 8.080657005310059, |
|
"rewards/rejected": -7.144291877746582, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.11773940345368916, |
|
"grad_norm": 149.49556516257127, |
|
"learning_rate": 4.995454786965036e-07, |
|
"logits/chosen": 5945.76806640625, |
|
"logits/rejected": 4966.1328125, |
|
"logps/chosen": -276.73760986328125, |
|
"logps/rejected": -253.0986785888672, |
|
"loss": 0.5912, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": 1.1343941688537598, |
|
"rewards/margins": 12.343668937683105, |
|
"rewards/rejected": -11.20927619934082, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.12558869701726844, |
|
"grad_norm": 59.806353223597256, |
|
"learning_rate": 4.990386933279972e-07, |
|
"logits/chosen": 5893.48095703125, |
|
"logits/rejected": 5249.22216796875, |
|
"logps/chosen": -283.9483642578125, |
|
"logps/rejected": -286.0248107910156, |
|
"loss": 0.5564, |
|
"rewards/accuracies": 0.7166666984558105, |
|
"rewards/chosen": -10.384180068969727, |
|
"rewards/margins": 9.5074462890625, |
|
"rewards/rejected": -19.891626358032227, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.13343799058084774, |
|
"grad_norm": 62.325385969713366, |
|
"learning_rate": 4.983447664444096e-07, |
|
"logits/chosen": 6131.96435546875, |
|
"logits/rejected": 5474.5771484375, |
|
"logps/chosen": -294.34442138671875, |
|
"logps/rejected": -284.6779479980469, |
|
"loss": 0.5598, |
|
"rewards/accuracies": 0.675000011920929, |
|
"rewards/chosen": -4.54689884185791, |
|
"rewards/margins": 11.124954223632812, |
|
"rewards/rejected": -15.671854972839355, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.141287284144427, |
|
"grad_norm": 70.02419866191639, |
|
"learning_rate": 4.97464219500968e-07, |
|
"logits/chosen": 5411.71435546875, |
|
"logits/rejected": 4724.818359375, |
|
"logps/chosen": -262.61627197265625, |
|
"logps/rejected": -257.47509765625, |
|
"loss": 0.5817, |
|
"rewards/accuracies": 0.6666666269302368, |
|
"rewards/chosen": 0.8280799984931946, |
|
"rewards/margins": 11.036195755004883, |
|
"rewards/rejected": -10.20811653137207, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.14913657770800628, |
|
"grad_norm": 61.31639433830006, |
|
"learning_rate": 4.963977141895843e-07, |
|
"logits/chosen": 5467.083984375, |
|
"logits/rejected": 4708.22216796875, |
|
"logps/chosen": -287.8260803222656, |
|
"logps/rejected": -282.094482421875, |
|
"loss": 0.5171, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -0.541407585144043, |
|
"rewards/margins": 20.091650009155273, |
|
"rewards/rejected": -20.633060455322266, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.15698587127158556, |
|
"grad_norm": 89.05267401282032, |
|
"learning_rate": 4.951460519416227e-07, |
|
"logits/chosen": 5396.1455078125, |
|
"logits/rejected": 4986.21142578125, |
|
"logps/chosen": -257.17950439453125, |
|
"logps/rejected": -285.96527099609375, |
|
"loss": 0.545, |
|
"rewards/accuracies": 0.7166666984558105, |
|
"rewards/chosen": -5.9294867515563965, |
|
"rewards/margins": 16.56216812133789, |
|
"rewards/rejected": -22.491649627685547, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.16483516483516483, |
|
"grad_norm": 71.11721989569415, |
|
"learning_rate": 4.937101733256606e-07, |
|
"logits/chosen": 4882.0869140625, |
|
"logits/rejected": 4334.1650390625, |
|
"logps/chosen": -225.06906127929688, |
|
"logps/rejected": -236.1047821044922, |
|
"loss": 0.5541, |
|
"rewards/accuracies": 0.6916667222976685, |
|
"rewards/chosen": -0.35613110661506653, |
|
"rewards/margins": 13.287859916687012, |
|
"rewards/rejected": -13.643989562988281, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.1726844583987441, |
|
"grad_norm": 95.02411341600269, |
|
"learning_rate": 4.920911573406924e-07, |
|
"logits/chosen": 5941.76708984375, |
|
"logits/rejected": 5054.4560546875, |
|
"logps/chosen": -286.0159912109375, |
|
"logps/rejected": -250.0142059326172, |
|
"loss": 0.5552, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -3.383108615875244, |
|
"rewards/margins": 13.090596199035645, |
|
"rewards/rejected": -16.473705291748047, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.18053375196232338, |
|
"grad_norm": 61.44548957102422, |
|
"learning_rate": 4.902902206053098e-07, |
|
"logits/chosen": 5438.53564453125, |
|
"logits/rejected": 4923.58203125, |
|
"logps/chosen": -280.6929626464844, |
|
"logps/rejected": -282.5475769042969, |
|
"loss": 0.5384, |
|
"rewards/accuracies": 0.6916666626930237, |
|
"rewards/chosen": -10.415714263916016, |
|
"rewards/margins": 14.109655380249023, |
|
"rewards/rejected": -24.525371551513672, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.18838304552590268, |
|
"grad_norm": 57.61246221015275, |
|
"learning_rate": 4.883087164434672e-07, |
|
"logits/chosen": 4929.82470703125, |
|
"logits/rejected": 3957.67041015625, |
|
"logps/chosen": -247.64553833007812, |
|
"logps/rejected": -235.91378784179688, |
|
"loss": 0.5196, |
|
"rewards/accuracies": 0.73333340883255, |
|
"rewards/chosen": -7.9520063400268555, |
|
"rewards/margins": 15.035487174987793, |
|
"rewards/rejected": -22.98749351501465, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.19623233908948196, |
|
"grad_norm": 81.52336264216343, |
|
"learning_rate": 4.861481338675183e-07, |
|
"logits/chosen": 5786.1494140625, |
|
"logits/rejected": 5151.611328125, |
|
"logps/chosen": -243.9603729248047, |
|
"logps/rejected": -279.42626953125, |
|
"loss": 0.5503, |
|
"rewards/accuracies": 0.6666666865348816, |
|
"rewards/chosen": -8.31295394897461, |
|
"rewards/margins": 16.721200942993164, |
|
"rewards/rejected": -25.03415298461914, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.20408163265306123, |
|
"grad_norm": 161.8757450979854, |
|
"learning_rate": 4.838100964592904e-07, |
|
"logits/chosen": 5958.8291015625, |
|
"logits/rejected": 4793.7353515625, |
|
"logps/chosen": -300.9515686035156, |
|
"logps/rejected": -272.099609375, |
|
"loss": 0.5609, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -16.198909759521484, |
|
"rewards/margins": 17.37653160095215, |
|
"rewards/rejected": -33.575443267822266, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.2119309262166405, |
|
"grad_norm": 55.67188711424705, |
|
"learning_rate": 4.812963611500339e-07, |
|
"logits/chosen": 5856.1484375, |
|
"logits/rejected": 5652.234375, |
|
"logps/chosen": -285.68646240234375, |
|
"logps/rejected": -292.1693115234375, |
|
"loss": 0.4969, |
|
"rewards/accuracies": 0.6916667222976685, |
|
"rewards/chosen": -11.177817344665527, |
|
"rewards/margins": 14.163126945495605, |
|
"rewards/rejected": -25.3409423828125, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.21978021978021978, |
|
"grad_norm": 85.12978558309311, |
|
"learning_rate": 4.786088169001671e-07, |
|
"logits/chosen": 5007.3935546875, |
|
"logits/rejected": 4349.2841796875, |
|
"logps/chosen": -232.29440307617188, |
|
"logps/rejected": -257.35736083984375, |
|
"loss": 0.502, |
|
"rewards/accuracies": 0.7833333611488342, |
|
"rewards/chosen": -3.5790748596191406, |
|
"rewards/margins": 19.368267059326172, |
|
"rewards/rejected": -22.947338104248047, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.22762951334379905, |
|
"grad_norm": 66.34271443416326, |
|
"learning_rate": 4.7574948327980567e-07, |
|
"logits/chosen": 6927.55078125, |
|
"logits/rejected": 5120.35498046875, |
|
"logps/chosen": -338.6285705566406, |
|
"logps/rejected": -294.0506896972656, |
|
"loss": 0.508, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -13.838041305541992, |
|
"rewards/margins": 22.446279525756836, |
|
"rewards/rejected": -36.28432083129883, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.23547880690737832, |
|
"grad_norm": 70.84013248396022, |
|
"learning_rate": 4.727205089511466e-07, |
|
"logits/chosen": 5050.2119140625, |
|
"logits/rejected": 4979.74609375, |
|
"logps/chosen": -250.53823852539062, |
|
"logps/rejected": -271.61199951171875, |
|
"loss": 0.5375, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -13.79767894744873, |
|
"rewards/margins": 13.887298583984375, |
|
"rewards/rejected": -27.684978485107422, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.24332810047095763, |
|
"grad_norm": 63.39626059992431, |
|
"learning_rate": 4.6952417005384247e-07, |
|
"logits/chosen": 5686.9599609375, |
|
"logits/rejected": 5094.78076171875, |
|
"logps/chosen": -264.87811279296875, |
|
"logps/rejected": -264.01177978515625, |
|
"loss": 0.5307, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -9.209539413452148, |
|
"rewards/margins": 13.136590957641602, |
|
"rewards/rejected": -22.34613037109375, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.25117739403453687, |
|
"grad_norm": 51.01706515176811, |
|
"learning_rate": 4.661628684945851e-07, |
|
"logits/chosen": 5712.5009765625, |
|
"logits/rejected": 4960.869140625, |
|
"logps/chosen": -284.8694763183594, |
|
"logps/rejected": -297.58978271484375, |
|
"loss": 0.5275, |
|
"rewards/accuracies": 0.7000000476837158, |
|
"rewards/chosen": -7.582206726074219, |
|
"rewards/margins": 18.082399368286133, |
|
"rewards/rejected": -25.664602279663086, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.25902668759811615, |
|
"grad_norm": 67.84592644137479, |
|
"learning_rate": 4.626391301421782e-07, |
|
"logits/chosen": 5515.0732421875, |
|
"logits/rejected": 5026.42431640625, |
|
"logps/chosen": -276.51904296875, |
|
"logps/rejected": -269.2738342285156, |
|
"loss": 0.5399, |
|
"rewards/accuracies": 0.7000000476837158, |
|
"rewards/chosen": -6.420998573303223, |
|
"rewards/margins": 13.4607572555542, |
|
"rewards/rejected": -19.881755828857422, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.2668759811616955, |
|
"grad_norm": 60.95792670490128, |
|
"learning_rate": 4.5895560292945996e-07, |
|
"logits/chosen": 5796.21826171875, |
|
"logits/rejected": 5875.0517578125, |
|
"logps/chosen": -270.3171081542969, |
|
"logps/rejected": -319.8937072753906, |
|
"loss": 0.5257, |
|
"rewards/accuracies": 0.7000000476837158, |
|
"rewards/chosen": -2.2340104579925537, |
|
"rewards/margins": 15.778246879577637, |
|
"rewards/rejected": -18.012258529663086, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.27472527472527475, |
|
"grad_norm": 84.59524350470046, |
|
"learning_rate": 4.5511505486349865e-07, |
|
"logits/chosen": 6091.1845703125, |
|
"logits/rejected": 5517.7958984375, |
|
"logps/chosen": -275.9291076660156, |
|
"logps/rejected": -308.3497314453125, |
|
"loss": 0.5434, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -3.021336078643799, |
|
"rewards/margins": 21.829858779907227, |
|
"rewards/rejected": -24.851192474365234, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.282574568288854, |
|
"grad_norm": 77.73130050367033, |
|
"learning_rate": 4.5112037194555876e-07, |
|
"logits/chosen": 5645.1474609375, |
|
"logits/rejected": 5508.9814453125, |
|
"logps/chosen": -256.37933349609375, |
|
"logps/rejected": -305.8360900878906, |
|
"loss": 0.5344, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -12.694351196289062, |
|
"rewards/margins": 19.168682098388672, |
|
"rewards/rejected": -31.863033294677734, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.2904238618524333, |
|
"grad_norm": 112.03101565670647, |
|
"learning_rate": 4.4697455600239863e-07, |
|
"logits/chosen": 5141.33544921875, |
|
"logits/rejected": 4822.3193359375, |
|
"logps/chosen": -271.9143371582031, |
|
"logps/rejected": -263.94171142578125, |
|
"loss": 0.559, |
|
"rewards/accuracies": 0.7083333134651184, |
|
"rewards/chosen": -13.011955261230469, |
|
"rewards/margins": 12.512086868286133, |
|
"rewards/rejected": -25.524044036865234, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.29827315541601257, |
|
"grad_norm": 63.51891587893552, |
|
"learning_rate": 4.426807224305315e-07, |
|
"logits/chosen": 6163.4716796875, |
|
"logits/rejected": 5088.28125, |
|
"logps/chosen": -318.40228271484375, |
|
"logps/rejected": -287.8067321777344, |
|
"loss": 0.5253, |
|
"rewards/accuracies": 0.824999988079071, |
|
"rewards/chosen": -10.425338745117188, |
|
"rewards/margins": 21.47298812866211, |
|
"rewards/rejected": -31.898324966430664, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.30612244897959184, |
|
"grad_norm": 75.92756479977929, |
|
"learning_rate": 4.3824209785514326e-07, |
|
"logits/chosen": 6272.45703125, |
|
"logits/rejected": 4800.2626953125, |
|
"logps/chosen": -301.3458251953125, |
|
"logps/rejected": -278.79302978515625, |
|
"loss": 0.5135, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -9.819374084472656, |
|
"rewards/margins": 23.481924057006836, |
|
"rewards/rejected": -33.301300048828125, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.3139717425431711, |
|
"grad_norm": 66.47374556483102, |
|
"learning_rate": 4.3366201770542687e-07, |
|
"logits/chosen": 5328.0849609375, |
|
"logits/rejected": 5181.458984375, |
|
"logps/chosen": -295.9270324707031, |
|
"logps/rejected": -328.08013916015625, |
|
"loss": 0.519, |
|
"rewards/accuracies": 0.7583333849906921, |
|
"rewards/chosen": -21.309566497802734, |
|
"rewards/margins": 20.160045623779297, |
|
"rewards/rejected": -41.46961212158203, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.3218210361067504, |
|
"grad_norm": 123.73907215954236, |
|
"learning_rate": 4.2894392370815567e-07, |
|
"logits/chosen": 5759.8017578125, |
|
"logits/rejected": 5182.2548828125, |
|
"logps/chosen": -308.05609130859375, |
|
"logps/rejected": -322.4450988769531, |
|
"loss": 0.4596, |
|
"rewards/accuracies": 0.7916666269302368, |
|
"rewards/chosen": -16.187286376953125, |
|
"rewards/margins": 23.63322639465332, |
|
"rewards/rejected": -39.820518493652344, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.32967032967032966, |
|
"grad_norm": 79.2606182693223, |
|
"learning_rate": 4.2409136130137845e-07, |
|
"logits/chosen": 5431.6875, |
|
"logits/rejected": 4942.8408203125, |
|
"logps/chosen": -294.1439514160156, |
|
"logps/rejected": -284.1300964355469, |
|
"loss": 0.5317, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -8.250704765319824, |
|
"rewards/margins": 22.796977996826172, |
|
"rewards/rejected": -31.047687530517578, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.33751962323390894, |
|
"grad_norm": 68.54405429145365, |
|
"learning_rate": 4.1910797697018017e-07, |
|
"logits/chosen": 5246.7197265625, |
|
"logits/rejected": 4415.2548828125, |
|
"logps/chosen": -265.8824157714844, |
|
"logps/rejected": -262.2781677246094, |
|
"loss": 0.4959, |
|
"rewards/accuracies": 0.8333333134651184, |
|
"rewards/chosen": -9.087538719177246, |
|
"rewards/margins": 28.186351776123047, |
|
"rewards/rejected": -37.27389144897461, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.3453689167974882, |
|
"grad_norm": 83.24939415391015, |
|
"learning_rate": 4.1399751550651084e-07, |
|
"logits/chosen": 5628.4384765625, |
|
"logits/rejected": 5584.71240234375, |
|
"logps/chosen": -259.50360107421875, |
|
"logps/rejected": -286.0181884765625, |
|
"loss": 0.5124, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -8.73564624786377, |
|
"rewards/margins": 17.06700325012207, |
|
"rewards/rejected": -25.802648544311523, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.3532182103610675, |
|
"grad_norm": 57.277868825940814, |
|
"learning_rate": 4.087638171951401e-07, |
|
"logits/chosen": 6479.9736328125, |
|
"logits/rejected": 4679.1142578125, |
|
"logps/chosen": -297.46343994140625, |
|
"logps/rejected": -259.014404296875, |
|
"loss": 0.5167, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -10.406542778015137, |
|
"rewards/margins": 22.608623504638672, |
|
"rewards/rejected": -33.015159606933594, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.36106750392464676, |
|
"grad_norm": 65.31589437332453, |
|
"learning_rate": 4.034108149278543e-07, |
|
"logits/chosen": 6699.00390625, |
|
"logits/rejected": 5186.462890625, |
|
"logps/chosen": -343.70062255859375, |
|
"logps/rejected": -298.53692626953125, |
|
"loss": 0.4645, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -10.98952865600586, |
|
"rewards/margins": 22.895919799804688, |
|
"rewards/rejected": -33.88544845581055, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.36891679748822603, |
|
"grad_norm": 96.27449799872079, |
|
"learning_rate": 3.979425312480629e-07, |
|
"logits/chosen": 5711.05224609375, |
|
"logits/rejected": 5017.3291015625, |
|
"logps/chosen": -306.458251953125, |
|
"logps/rejected": -319.9109802246094, |
|
"loss": 0.5455, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -18.672470092773438, |
|
"rewards/margins": 23.139759063720703, |
|
"rewards/rejected": -41.81222915649414, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.37676609105180536, |
|
"grad_norm": 55.55795766142816, |
|
"learning_rate": 3.923630753280357e-07, |
|
"logits/chosen": 6194.65673828125, |
|
"logits/rejected": 5330.5810546875, |
|
"logps/chosen": -294.33331298828125, |
|
"logps/rejected": -285.3929138183594, |
|
"loss": 0.5026, |
|
"rewards/accuracies": 0.7500000596046448, |
|
"rewards/chosen": -14.526718139648438, |
|
"rewards/margins": 24.07291030883789, |
|
"rewards/rejected": -38.59962844848633, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.38461538461538464, |
|
"grad_norm": 78.10688192233438, |
|
"learning_rate": 3.866766398810424e-07, |
|
"logits/chosen": 5823.94140625, |
|
"logits/rejected": 5543.9892578125, |
|
"logps/chosen": -248.40328979492188, |
|
"logps/rejected": -305.0126037597656, |
|
"loss": 0.4751, |
|
"rewards/accuracies": 0.7916666269302368, |
|
"rewards/chosen": -7.462246894836426, |
|
"rewards/margins": 24.950969696044922, |
|
"rewards/rejected": -32.41321563720703, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.3924646781789639, |
|
"grad_norm": 70.24124257704634, |
|
"learning_rate": 3.8088749801071496e-07, |
|
"logits/chosen": 6291.517578125, |
|
"logits/rejected": 4903.115234375, |
|
"logps/chosen": -314.2850646972656, |
|
"logps/rejected": -301.9085388183594, |
|
"loss": 0.5189, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -15.795580863952637, |
|
"rewards/margins": 21.68023109436035, |
|
"rewards/rejected": -37.47581100463867, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.4003139717425432, |
|
"grad_norm": 53.69297743558368, |
|
"learning_rate": 3.75e-07, |
|
"logits/chosen": 5080.75390625, |
|
"logits/rejected": 4507.572265625, |
|
"logps/chosen": -273.0891418457031, |
|
"logps/rejected": -269.4468688964844, |
|
"loss": 0.4849, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -16.22928237915039, |
|
"rewards/margins": 24.09990692138672, |
|
"rewards/rejected": -40.329185485839844, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 0.40816326530612246, |
|
"grad_norm": 80.31768469560811, |
|
"learning_rate": 3.6901857004211443e-07, |
|
"logits/chosen": 5401.982421875, |
|
"logits/rejected": 4994.92626953125, |
|
"logps/chosen": -286.5435791015625, |
|
"logps/rejected": -307.6182556152344, |
|
"loss": 0.5504, |
|
"rewards/accuracies": 0.6916666626930237, |
|
"rewards/chosen": -20.370731353759766, |
|
"rewards/margins": 18.12123680114746, |
|
"rewards/rejected": -38.491966247558594, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.41601255886970173, |
|
"grad_norm": 58.78911634916461, |
|
"learning_rate": 3.6294770291596076e-07, |
|
"logits/chosen": 6102.56005859375, |
|
"logits/rejected": 5029.2578125, |
|
"logps/chosen": -299.22210693359375, |
|
"logps/rejected": -295.10443115234375, |
|
"loss": 0.482, |
|
"rewards/accuracies": 0.7166666984558105, |
|
"rewards/chosen": -14.929966926574707, |
|
"rewards/margins": 16.103656768798828, |
|
"rewards/rejected": -31.03362464904785, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 0.423861852433281, |
|
"grad_norm": 80.65951431247957, |
|
"learning_rate": 3.5679196060850034e-07, |
|
"logits/chosen": 5810.88525390625, |
|
"logits/rejected": 5194.83935546875, |
|
"logps/chosen": -288.0130920410156, |
|
"logps/rejected": -287.81170654296875, |
|
"loss": 0.5021, |
|
"rewards/accuracies": 0.7333332896232605, |
|
"rewards/chosen": -10.629480361938477, |
|
"rewards/margins": 21.78488540649414, |
|
"rewards/rejected": -32.414371490478516, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.4317111459968603, |
|
"grad_norm": 75.60934568083842, |
|
"learning_rate": 3.505559688866229e-07, |
|
"logits/chosen": 5616.5888671875, |
|
"logits/rejected": 5239.4462890625, |
|
"logps/chosen": -293.79791259765625, |
|
"logps/rejected": -327.1747131347656, |
|
"loss": 0.5091, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -12.461262702941895, |
|
"rewards/margins": 23.530132293701172, |
|
"rewards/rejected": -35.99140167236328, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.43956043956043955, |
|
"grad_norm": 84.86627370577622, |
|
"learning_rate": 3.4424441382108826e-07, |
|
"logits/chosen": 5615.79833984375, |
|
"logits/rejected": 5273.56591796875, |
|
"logps/chosen": -290.41864013671875, |
|
"logps/rejected": -298.9422607421875, |
|
"loss": 0.5446, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -14.291006088256836, |
|
"rewards/margins": 17.873065948486328, |
|
"rewards/rejected": -32.1640739440918, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.4474097331240188, |
|
"grad_norm": 76.55175648368798, |
|
"learning_rate": 3.378620382651523e-07, |
|
"logits/chosen": 5950.259765625, |
|
"logits/rejected": 5518.1162109375, |
|
"logps/chosen": -326.0985412597656, |
|
"logps/rejected": -325.46246337890625, |
|
"loss": 0.4799, |
|
"rewards/accuracies": 0.7833333015441895, |
|
"rewards/chosen": -11.533608436584473, |
|
"rewards/margins": 21.987844467163086, |
|
"rewards/rejected": -33.521453857421875, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 0.4552590266875981, |
|
"grad_norm": 81.1438191032448, |
|
"learning_rate": 3.314136382905234e-07, |
|
"logits/chosen": 5932.8876953125, |
|
"logits/rejected": 5372.55078125, |
|
"logps/chosen": -293.031494140625, |
|
"logps/rejected": -319.82513427734375, |
|
"loss": 0.495, |
|
"rewards/accuracies": 0.8333333134651184, |
|
"rewards/chosen": -5.784700393676758, |
|
"rewards/margins": 29.190006256103516, |
|
"rewards/rejected": -34.974708557128906, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.4631083202511774, |
|
"grad_norm": 62.68029788470684, |
|
"learning_rate": 3.249040595833274e-07, |
|
"logits/chosen": 6448.5634765625, |
|
"logits/rejected": 5459.8603515625, |
|
"logps/chosen": -316.6751708984375, |
|
"logps/rejected": -283.12286376953125, |
|
"loss": 0.4824, |
|
"rewards/accuracies": 0.8083333969116211, |
|
"rewards/chosen": -10.52171516418457, |
|
"rewards/margins": 26.76095199584961, |
|
"rewards/rejected": -37.28267288208008, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 0.47095761381475665, |
|
"grad_norm": 85.34315127293867, |
|
"learning_rate": 3.1833819380279023e-07, |
|
"logits/chosen": 6080.92626953125, |
|
"logits/rejected": 5186.0732421875, |
|
"logps/chosen": -270.80426025390625, |
|
"logps/rejected": -296.1933898925781, |
|
"loss": 0.4672, |
|
"rewards/accuracies": 0.8333333730697632, |
|
"rewards/chosen": -18.313228607177734, |
|
"rewards/margins": 22.91641616821289, |
|
"rewards/rejected": -41.22964096069336, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.478806907378336, |
|
"grad_norm": 58.08462591355949, |
|
"learning_rate": 3.11720974905373e-07, |
|
"logits/chosen": 5836.931640625, |
|
"logits/rejected": 5111.87548828125, |
|
"logps/chosen": -290.5270080566406, |
|
"logps/rejected": -293.91949462890625, |
|
"loss": 0.461, |
|
"rewards/accuracies": 0.824999988079071, |
|
"rewards/chosen": -17.68552017211914, |
|
"rewards/margins": 25.742305755615234, |
|
"rewards/rejected": -43.427825927734375, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 0.48665620094191525, |
|
"grad_norm": 59.24314457292362, |
|
"learning_rate": 3.0505737543712275e-07, |
|
"logits/chosen": 4960.1611328125, |
|
"logits/rejected": 4097.82275390625, |
|
"logps/chosen": -264.2720031738281, |
|
"logps/rejected": -249.32626342773438, |
|
"loss": 0.4882, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -16.4460506439209, |
|
"rewards/margins": 19.30292510986328, |
|
"rewards/rejected": -35.74897766113281, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.4945054945054945, |
|
"grad_norm": 78.04378850224133, |
|
"learning_rate": 2.9835240279702513e-07, |
|
"logits/chosen": 6457.73974609375, |
|
"logits/rejected": 5500.73681640625, |
|
"logps/chosen": -325.96942138671875, |
|
"logps/rejected": -301.5501403808594, |
|
"loss": 0.4744, |
|
"rewards/accuracies": 0.7833333611488342, |
|
"rewards/chosen": -9.257902145385742, |
|
"rewards/margins": 26.257373809814453, |
|
"rewards/rejected": -35.51527404785156, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 0.5023547880690737, |
|
"grad_norm": 72.15746448236564, |
|
"learning_rate": 2.9161109547416667e-07, |
|
"logits/chosen": 6160.0927734375, |
|
"logits/rejected": 5292.87158203125, |
|
"logps/chosen": -296.0101318359375, |
|
"logps/rejected": -312.3030700683594, |
|
"loss": 0.5101, |
|
"rewards/accuracies": 0.675000011920929, |
|
"rewards/chosen": -17.3197078704834, |
|
"rewards/margins": 17.142587661743164, |
|
"rewards/rejected": -34.46229553222656, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.5102040816326531, |
|
"grad_norm": 67.34813898160822, |
|
"learning_rate": 2.848385192615339e-07, |
|
"logits/chosen": 5268.1103515625, |
|
"logits/rejected": 4316.62451171875, |
|
"logps/chosen": -275.38983154296875, |
|
"logps/rejected": -265.4394836425781, |
|
"loss": 0.4756, |
|
"rewards/accuracies": 0.7750000357627869, |
|
"rewards/chosen": -17.851306915283203, |
|
"rewards/margins": 20.486316680908203, |
|
"rewards/rejected": -38.337623596191406, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.5180533751962323, |
|
"grad_norm": 73.13924713953126, |
|
"learning_rate": 2.780397634492949e-07, |
|
"logits/chosen": 5867.3603515625, |
|
"logits/rejected": 4723.8388671875, |
|
"logps/chosen": -306.2716979980469, |
|
"logps/rejected": -297.45367431640625, |
|
"loss": 0.5284, |
|
"rewards/accuracies": 0.7916666269302368, |
|
"rewards/chosen": -15.953409194946289, |
|
"rewards/margins": 29.22943115234375, |
|
"rewards/rejected": -45.18284606933594, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.5259026687598116, |
|
"grad_norm": 92.53126905748795, |
|
"learning_rate": 2.71219937000424e-07, |
|
"logits/chosen": 5880.07958984375, |
|
"logits/rejected": 4863.40576171875, |
|
"logps/chosen": -291.5413513183594, |
|
"logps/rejected": -287.3309326171875, |
|
"loss": 0.5249, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -15.330841064453125, |
|
"rewards/margins": 20.9111385345459, |
|
"rewards/rejected": -36.24197769165039, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 0.533751962323391, |
|
"grad_norm": 67.62676868019088, |
|
"learning_rate": 2.6438416471154273e-07, |
|
"logits/chosen": 5770.3818359375, |
|
"logits/rejected": 4856.85791015625, |
|
"logps/chosen": -291.76312255859375, |
|
"logps/rejected": -270.06256103515625, |
|
"loss": 0.4897, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -13.935272216796875, |
|
"rewards/margins": 21.12171173095703, |
|
"rewards/rejected": -35.056983947753906, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 0.5416012558869702, |
|
"grad_norm": 73.05663859356332, |
|
"learning_rate": 2.5753758336186326e-07, |
|
"logits/chosen": 5670.96142578125, |
|
"logits/rejected": 5242.6416015625, |
|
"logps/chosen": -284.61090087890625, |
|
"logps/rejected": -310.2216796875, |
|
"loss": 0.4463, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -9.5217924118042, |
|
"rewards/margins": 23.241046905517578, |
|
"rewards/rejected": -32.76283645629883, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 0.5494505494505495, |
|
"grad_norm": 54.85355662529876, |
|
"learning_rate": 2.5068533785312666e-07, |
|
"logits/chosen": 5440.59716796875, |
|
"logits/rejected": 5240.3203125, |
|
"logps/chosen": -265.55499267578125, |
|
"logps/rejected": -282.92803955078125, |
|
"loss": 0.5399, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -15.438737869262695, |
|
"rewards/margins": 22.037721633911133, |
|
"rewards/rejected": -37.47645950317383, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.5572998430141287, |
|
"grad_norm": 60.42827942241951, |
|
"learning_rate": 2.4383257734343794e-07, |
|
"logits/chosen": 5401.58203125, |
|
"logits/rejected": 5404.7314453125, |
|
"logps/chosen": -260.6871643066406, |
|
"logps/rejected": -296.1017761230469, |
|
"loss": 0.4756, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -11.44128704071045, |
|
"rewards/margins": 20.571271896362305, |
|
"rewards/rejected": -32.0125617980957, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 0.565149136577708, |
|
"grad_norm": 67.98797911861922, |
|
"learning_rate": 2.3698445137790258e-07, |
|
"logits/chosen": 5796.3447265625, |
|
"logits/rejected": 5025.68603515625, |
|
"logps/chosen": -291.19708251953125, |
|
"logps/rejected": -288.0618591308594, |
|
"loss": 0.4765, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -6.432301998138428, |
|
"rewards/margins": 23.511104583740234, |
|
"rewards/rejected": -29.943408966064453, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 0.5729984301412873, |
|
"grad_norm": 61.33927353442037, |
|
"learning_rate": 2.3014610601897157e-07, |
|
"logits/chosen": 6256.431640625, |
|
"logits/rejected": 4830.943359375, |
|
"logps/chosen": -312.23529052734375, |
|
"logps/rejected": -280.59759521484375, |
|
"loss": 0.4773, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -17.176259994506836, |
|
"rewards/margins": 23.136552810668945, |
|
"rewards/rejected": -40.31281280517578, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 0.5808477237048666, |
|
"grad_norm": 58.17800831625638, |
|
"learning_rate": 2.2332267997940513e-07, |
|
"logits/chosen": 5208.26708984375, |
|
"logits/rejected": 4448.9052734375, |
|
"logps/chosen": -262.63885498046875, |
|
"logps/rejected": -255.3593292236328, |
|
"loss": 0.466, |
|
"rewards/accuracies": 0.7916666865348816, |
|
"rewards/chosen": -14.630803108215332, |
|
"rewards/margins": 23.1756591796875, |
|
"rewards/rejected": -37.806461334228516, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 0.5886970172684458, |
|
"grad_norm": 110.94118329677751, |
|
"learning_rate": 2.1651930076075723e-07, |
|
"logits/chosen": 5666.19775390625, |
|
"logits/rejected": 5149.2490234375, |
|
"logps/chosen": -257.9913635253906, |
|
"logps/rejected": -259.7269592285156, |
|
"loss": 0.497, |
|
"rewards/accuracies": 0.7083333730697632, |
|
"rewards/chosen": -17.729137420654297, |
|
"rewards/margins": 19.4478759765625, |
|
"rewards/rejected": -37.17700958251953, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.5965463108320251, |
|
"grad_norm": 76.60199494528777, |
|
"learning_rate": 2.0974108080028692e-07, |
|
"logits/chosen": 5950.09814453125, |
|
"logits/rejected": 4712.52197265625, |
|
"logps/chosen": -279.6956787109375, |
|
"logps/rejected": -278.8898010253906, |
|
"loss": 0.4727, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -16.335081100463867, |
|
"rewards/margins": 24.634830474853516, |
|
"rewards/rejected": -40.969913482666016, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 0.6043956043956044, |
|
"grad_norm": 80.46613010942175, |
|
"learning_rate": 2.0299311362918773e-07, |
|
"logits/chosen": 6163.73193359375, |
|
"logits/rejected": 5307.6240234375, |
|
"logps/chosen": -311.0682373046875, |
|
"logps/rejected": -323.7105407714844, |
|
"loss": 0.5202, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -16.42329216003418, |
|
"rewards/margins": 22.489013671875, |
|
"rewards/rejected": -38.91231155395508, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 0.6122448979591837, |
|
"grad_norm": 60.074991801115765, |
|
"learning_rate": 1.962804700450265e-07, |
|
"logits/chosen": 5973.45068359375, |
|
"logits/rejected": 5675.47412109375, |
|
"logps/chosen": -304.04522705078125, |
|
"logps/rejected": -360.00689697265625, |
|
"loss": 0.495, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -20.91264533996582, |
|
"rewards/margins": 24.53109359741211, |
|
"rewards/rejected": -45.4437370300293, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 0.6200941915227629, |
|
"grad_norm": 121.27262746222853, |
|
"learning_rate": 1.8960819430126334e-07, |
|
"logits/chosen": 5599.5244140625, |
|
"logits/rejected": 4970.3056640625, |
|
"logps/chosen": -278.2356262207031, |
|
"logps/rejected": -305.4681396484375, |
|
"loss": 0.4691, |
|
"rewards/accuracies": 0.8166667222976685, |
|
"rewards/chosen": -23.784244537353516, |
|
"rewards/margins": 26.33816909790039, |
|
"rewards/rejected": -50.122413635253906, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 0.6279434850863422, |
|
"grad_norm": 82.44548535694025, |
|
"learning_rate": 1.8298130031671972e-07, |
|
"logits/chosen": 5606.4052734375, |
|
"logits/rejected": 4922.90283203125, |
|
"logps/chosen": -291.6396179199219, |
|
"logps/rejected": -304.04962158203125, |
|
"loss": 0.5104, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -17.903173446655273, |
|
"rewards/margins": 19.6292667388916, |
|
"rewards/rejected": -37.532440185546875, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.6357927786499215, |
|
"grad_norm": 70.1420150131459, |
|
"learning_rate": 1.7640476790784075e-07, |
|
"logits/chosen": 5104.7275390625, |
|
"logits/rejected": 4670.8662109375, |
|
"logps/chosen": -292.78948974609375, |
|
"logps/rejected": -312.7052917480469, |
|
"loss": 0.5138, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -18.959609985351562, |
|
"rewards/margins": 19.970327377319336, |
|
"rewards/rejected": -38.92993927001953, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 0.6436420722135008, |
|
"grad_norm": 89.22388286182063, |
|
"learning_rate": 1.6988353904658492e-07, |
|
"logits/chosen": 5619.841796875, |
|
"logits/rejected": 4398.35595703125, |
|
"logps/chosen": -298.38201904296875, |
|
"logps/rejected": -252.72518920898438, |
|
"loss": 0.4789, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -15.309239387512207, |
|
"rewards/margins": 21.590890884399414, |
|
"rewards/rejected": -36.90012741088867, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 0.6514913657770801, |
|
"grad_norm": 78.38923638779514, |
|
"learning_rate": 1.634225141467513e-07, |
|
"logits/chosen": 5558.37255859375, |
|
"logits/rejected": 4999.9423828125, |
|
"logps/chosen": -281.6241455078125, |
|
"logps/rejected": -291.132568359375, |
|
"loss": 0.4816, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -18.16968536376953, |
|
"rewards/margins": 25.177610397338867, |
|
"rewards/rejected": -43.34729766845703, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 0.6593406593406593, |
|
"grad_norm": 60.020636851213084, |
|
"learning_rate": 1.570265483815364e-07, |
|
"logits/chosen": 6103.8017578125, |
|
"logits/rejected": 5022.09130859375, |
|
"logps/chosen": -300.73687744140625, |
|
"logps/rejected": -309.49188232421875, |
|
"loss": 0.5073, |
|
"rewards/accuracies": 0.7500000596046448, |
|
"rewards/chosen": -15.203628540039062, |
|
"rewards/margins": 20.09307289123535, |
|
"rewards/rejected": -35.29670333862305, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 0.6671899529042387, |
|
"grad_norm": 58.32492360881753, |
|
"learning_rate": 1.5070044803508691e-07, |
|
"logits/chosen": 5640.41357421875, |
|
"logits/rejected": 5135.97119140625, |
|
"logps/chosen": -290.51226806640625, |
|
"logps/rejected": -285.50213623046875, |
|
"loss": 0.4609, |
|
"rewards/accuracies": 0.7666666507720947, |
|
"rewards/chosen": -17.97010040283203, |
|
"rewards/margins": 21.574066162109375, |
|
"rewards/rejected": -39.544166564941406, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 0.6750392464678179, |
|
"grad_norm": 79.07046685006041, |
|
"learning_rate": 1.444489668907914e-07, |
|
"logits/chosen": 6106.01513671875, |
|
"logits/rejected": 5220.31689453125, |
|
"logps/chosen": -316.23028564453125, |
|
"logps/rejected": -288.64398193359375, |
|
"loss": 0.5233, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -13.361892700195312, |
|
"rewards/margins": 20.976970672607422, |
|
"rewards/rejected": -34.33885955810547, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 0.6828885400313972, |
|
"grad_norm": 47.127373222510464, |
|
"learning_rate": 1.3827680265902232e-07, |
|
"logits/chosen": 6052.7939453125, |
|
"logits/rejected": 5035.3828125, |
|
"logps/chosen": -300.17950439453125, |
|
"logps/rejected": -286.375244140625, |
|
"loss": 0.4843, |
|
"rewards/accuracies": 0.7833333015441895, |
|
"rewards/chosen": -10.767036437988281, |
|
"rewards/margins": 21.399105072021484, |
|
"rewards/rejected": -32.16614532470703, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 0.6907378335949764, |
|
"grad_norm": 74.09900524632, |
|
"learning_rate": 1.3218859344701632e-07, |
|
"logits/chosen": 5362.17041015625, |
|
"logits/rejected": 5107.20361328125, |
|
"logps/chosen": -265.02392578125, |
|
"logps/rejected": -314.38690185546875, |
|
"loss": 0.4774, |
|
"rewards/accuracies": 0.7666666507720947, |
|
"rewards/chosen": -6.2712297439575195, |
|
"rewards/margins": 19.80803680419922, |
|
"rewards/rejected": -26.079265594482422, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 0.6985871271585558, |
|
"grad_norm": 66.6052674675123, |
|
"learning_rate": 1.2618891427354172e-07, |
|
"logits/chosen": 6284.64111328125, |
|
"logits/rejected": 5166.4462890625, |
|
"logps/chosen": -317.2557067871094, |
|
"logps/rejected": -291.3585205078125, |
|
"loss": 0.4871, |
|
"rewards/accuracies": 0.8166667222976685, |
|
"rewards/chosen": -4.727200984954834, |
|
"rewards/margins": 24.60157012939453, |
|
"rewards/rejected": -29.328771591186523, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 0.706436420722135, |
|
"grad_norm": 68.28337798829429, |
|
"learning_rate": 1.202822736309758e-07, |
|
"logits/chosen": 5346.580078125, |
|
"logits/rejected": 4938.1474609375, |
|
"logps/chosen": -267.27410888671875, |
|
"logps/rejected": -296.4985046386719, |
|
"loss": 0.4859, |
|
"rewards/accuracies": 0.7666667103767395, |
|
"rewards/chosen": -9.232672691345215, |
|
"rewards/margins": 22.844303131103516, |
|
"rewards/rejected": -32.07697296142578, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.7142857142857143, |
|
"grad_norm": 63.48835238056808, |
|
"learning_rate": 1.1447311009737299e-07, |
|
"logits/chosen": 5272.83837890625, |
|
"logits/rejected": 5023.6845703125, |
|
"logps/chosen": -267.2757263183594, |
|
"logps/rejected": -292.3034362792969, |
|
"loss": 0.5038, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -9.708333015441895, |
|
"rewards/margins": 24.619693756103516, |
|
"rewards/rejected": -34.328025817871094, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 0.7221350078492935, |
|
"grad_norm": 60.97912439150742, |
|
"learning_rate": 1.0876578900107053e-07, |
|
"logits/chosen": 5796.75146484375, |
|
"logits/rejected": 4881.2841796875, |
|
"logps/chosen": -297.802001953125, |
|
"logps/rejected": -278.0216369628906, |
|
"loss": 0.5153, |
|
"rewards/accuracies": 0.8166667222976685, |
|
"rewards/chosen": -11.080281257629395, |
|
"rewards/margins": 21.987628936767578, |
|
"rewards/rejected": -33.067909240722656, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 0.7299843014128728, |
|
"grad_norm": 60.28907574898199, |
|
"learning_rate": 1.0316459914033793e-07, |
|
"logits/chosen": 5755.77197265625, |
|
"logits/rejected": 4348.8896484375, |
|
"logps/chosen": -303.0055236816406, |
|
"logps/rejected": -268.995849609375, |
|
"loss": 0.474, |
|
"rewards/accuracies": 0.7666667699813843, |
|
"rewards/chosen": -15.051956176757812, |
|
"rewards/margins": 20.99700355529785, |
|
"rewards/rejected": -36.04895782470703, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 0.7378335949764521, |
|
"grad_norm": 63.03798083693648, |
|
"learning_rate": 9.767374956053584e-08, |
|
"logits/chosen": 5592.4560546875, |
|
"logits/rejected": 4937.5419921875, |
|
"logps/chosen": -278.56890869140625, |
|
"logps/rejected": -285.54351806640625, |
|
"loss": 0.4986, |
|
"rewards/accuracies": 0.783333420753479, |
|
"rewards/chosen": -13.702900886535645, |
|
"rewards/margins": 24.679929733276367, |
|
"rewards/rejected": -38.38282775878906, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 0.7456828885400314, |
|
"grad_norm": 61.55603916303844, |
|
"learning_rate": 9.229736639120561e-08, |
|
"logits/chosen": 5777.404296875, |
|
"logits/rejected": 5323.2900390625, |
|
"logps/chosen": -278.0633850097656, |
|
"logps/rejected": -295.2113037109375, |
|
"loss": 0.5241, |
|
"rewards/accuracies": 0.7166666388511658, |
|
"rewards/chosen": -14.50324821472168, |
|
"rewards/margins": 17.914897918701172, |
|
"rewards/rejected": -32.41815185546875, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 0.7535321821036107, |
|
"grad_norm": 84.55760580604866, |
|
"learning_rate": 8.70394897454659e-08, |
|
"logits/chosen": 5628.0771484375, |
|
"logits/rejected": 5055.1318359375, |
|
"logps/chosen": -275.655517578125, |
|
"logps/rejected": -281.5873718261719, |
|
"loss": 0.4733, |
|
"rewards/accuracies": 0.7833333611488342, |
|
"rewards/chosen": -11.641939163208008, |
|
"rewards/margins": 23.622745513916016, |
|
"rewards/rejected": -35.26468276977539, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 0.7613814756671899, |
|
"grad_norm": 77.38466285245907, |
|
"learning_rate": 8.19040706840472e-08, |
|
"logits/chosen": 5724.43994140625, |
|
"logits/rejected": 4841.40234375, |
|
"logps/chosen": -301.0986328125, |
|
"logps/rejected": -283.05474853515625, |
|
"loss": 0.4868, |
|
"rewards/accuracies": 0.7666666507720947, |
|
"rewards/chosen": -11.967236518859863, |
|
"rewards/margins": 24.7530574798584, |
|
"rewards/rejected": -36.720298767089844, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 0.7692307692307693, |
|
"grad_norm": 76.26753722537488, |
|
"learning_rate": 7.689496824624525e-08, |
|
"logits/chosen": 5441.85009765625, |
|
"logits/rejected": 4387.2412109375, |
|
"logps/chosen": -287.865966796875, |
|
"logps/rejected": -277.3480224609375, |
|
"loss": 0.4696, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -11.223980903625488, |
|
"rewards/margins": 27.436904907226562, |
|
"rewards/rejected": -38.660888671875, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 0.7770800627943485, |
|
"grad_norm": 70.69138991463052, |
|
"learning_rate": 7.201594655002458e-08, |
|
"logits/chosen": 5780.7158203125, |
|
"logits/rejected": 4848.8974609375, |
|
"logps/chosen": -282.12384033203125, |
|
"logps/rejected": -271.3438720703125, |
|
"loss": 0.4823, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -14.01646900177002, |
|
"rewards/margins": 21.042787551879883, |
|
"rewards/rejected": -35.05925369262695, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 0.7849293563579278, |
|
"grad_norm": 67.08952299091601, |
|
"learning_rate": 6.727067196345099e-08, |
|
"logits/chosen": 5443.533203125, |
|
"logits/rejected": 4615.63134765625, |
|
"logps/chosen": -269.97344970703125, |
|
"logps/rejected": -256.36163330078125, |
|
"loss": 0.4742, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -9.651910781860352, |
|
"rewards/margins": 21.033966064453125, |
|
"rewards/rejected": -30.68587875366211, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.792778649921507, |
|
"grad_norm": 68.03255688040626, |
|
"learning_rate": 6.26627103495786e-08, |
|
"logits/chosen": 5637.0078125, |
|
"logits/rejected": 4722.42919921875, |
|
"logps/chosen": -267.38934326171875, |
|
"logps/rejected": -263.6007995605469, |
|
"loss": 0.4984, |
|
"rewards/accuracies": 0.6916666626930237, |
|
"rewards/chosen": -7.973989009857178, |
|
"rewards/margins": 20.111682891845703, |
|
"rewards/rejected": -28.085674285888672, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 0.8006279434850864, |
|
"grad_norm": 50.612164014066174, |
|
"learning_rate": 5.8195524386862374e-08, |
|
"logits/chosen": 5714.32763671875, |
|
"logits/rejected": 5093.7119140625, |
|
"logps/chosen": -302.93975830078125, |
|
"logps/rejected": -304.6034851074219, |
|
"loss": 0.4762, |
|
"rewards/accuracies": 0.783333420753479, |
|
"rewards/chosen": -5.043163299560547, |
|
"rewards/margins": 25.959774017333984, |
|
"rewards/rejected": -31.002941131591797, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 0.8084772370486656, |
|
"grad_norm": 85.87448292761121, |
|
"learning_rate": 5.38724709671092e-08, |
|
"logits/chosen": 6109.154296875, |
|
"logits/rejected": 5799.1044921875, |
|
"logps/chosen": -286.541015625, |
|
"logps/rejected": -309.1804504394531, |
|
"loss": 0.4762, |
|
"rewards/accuracies": 0.7916666269302368, |
|
"rewards/chosen": -8.740842819213867, |
|
"rewards/margins": 25.329326629638672, |
|
"rewards/rejected": -34.070167541503906, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 0.8163265306122449, |
|
"grad_norm": 66.40755472143236, |
|
"learning_rate": 4.969679867292276e-08, |
|
"logits/chosen": 5434.1982421875, |
|
"logits/rejected": 4960.38525390625, |
|
"logps/chosen": -274.160888671875, |
|
"logps/rejected": -286.50177001953125, |
|
"loss": 0.475, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -8.917675018310547, |
|
"rewards/margins": 24.010303497314453, |
|
"rewards/rejected": -32.927978515625, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 0.8241758241758241, |
|
"grad_norm": 75.62013708498031, |
|
"learning_rate": 4.5671645336537416e-08, |
|
"logits/chosen": 5481.43359375, |
|
"logits/rejected": 4999.4345703125, |
|
"logps/chosen": -289.73046875, |
|
"logps/rejected": -297.23944091796875, |
|
"loss": 0.4654, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -8.197317123413086, |
|
"rewards/margins": 23.31808090209961, |
|
"rewards/rejected": -31.515399932861328, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 0.8320251177394035, |
|
"grad_norm": 103.29471977365169, |
|
"learning_rate": 4.180003568187776e-08, |
|
"logits/chosen": 6785.52099609375, |
|
"logits/rejected": 5364.03271484375, |
|
"logps/chosen": -318.3176574707031, |
|
"logps/rejected": -291.8903503417969, |
|
"loss": 0.5068, |
|
"rewards/accuracies": 0.675000011920929, |
|
"rewards/chosen": -10.395001411437988, |
|
"rewards/margins": 16.960371017456055, |
|
"rewards/rejected": -27.355377197265625, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 0.8398744113029827, |
|
"grad_norm": 115.05057020857156, |
|
"learning_rate": 3.8084879051612144e-08, |
|
"logits/chosen": 5640.74072265625, |
|
"logits/rejected": 5185.1337890625, |
|
"logps/chosen": -280.8153991699219, |
|
"logps/rejected": -267.6299743652344, |
|
"loss": 0.5102, |
|
"rewards/accuracies": 0.7583333849906921, |
|
"rewards/chosen": -7.923087120056152, |
|
"rewards/margins": 22.471633911132812, |
|
"rewards/rejected": -30.39472007751465, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 0.847723704866562, |
|
"grad_norm": 51.487360894657684, |
|
"learning_rate": 3.452896722091128e-08, |
|
"logits/chosen": 6193.6533203125, |
|
"logits/rejected": 4817.9892578125, |
|
"logps/chosen": -319.95849609375, |
|
"logps/rejected": -274.63543701171875, |
|
"loss": 0.4487, |
|
"rewards/accuracies": 0.8083333969116211, |
|
"rewards/chosen": -7.470693111419678, |
|
"rewards/margins": 25.923675537109375, |
|
"rewards/rejected": -33.394371032714844, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 0.8555729984301413, |
|
"grad_norm": 59.46047724312089, |
|
"learning_rate": 3.11349722995527e-08, |
|
"logits/chosen": 6239.1337890625, |
|
"logits/rejected": 4728.5390625, |
|
"logps/chosen": -285.4170227050781, |
|
"logps/rejected": -280.19549560546875, |
|
"loss": 0.4794, |
|
"rewards/accuracies": 0.7166667580604553, |
|
"rewards/chosen": -11.619339942932129, |
|
"rewards/margins": 18.907350540161133, |
|
"rewards/rejected": -30.526691436767578, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 0.8634222919937206, |
|
"grad_norm": 67.790017961291, |
|
"learning_rate": 2.7905444723949762e-08, |
|
"logits/chosen": 6030.23193359375, |
|
"logits/rejected": 5021.29248046875, |
|
"logps/chosen": -295.6999816894531, |
|
"logps/rejected": -271.0659484863281, |
|
"loss": 0.5037, |
|
"rewards/accuracies": 0.7916666865348816, |
|
"rewards/chosen": -12.477197647094727, |
|
"rewards/margins": 26.47970199584961, |
|
"rewards/rejected": -38.95690155029297, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 0.8712715855572999, |
|
"grad_norm": 76.11643315595336, |
|
"learning_rate": 2.484281134061142e-08, |
|
"logits/chosen": 6397.3251953125, |
|
"logits/rejected": 5203.5615234375, |
|
"logps/chosen": -322.0311584472656, |
|
"logps/rejected": -299.73577880859375, |
|
"loss": 0.4742, |
|
"rewards/accuracies": 0.7916666269302368, |
|
"rewards/chosen": -12.75678825378418, |
|
"rewards/margins": 24.258211135864258, |
|
"rewards/rejected": -37.01499938964844, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 0.8791208791208791, |
|
"grad_norm": 66.07828474570364, |
|
"learning_rate": 2.194937358247506e-08, |
|
"logits/chosen": 6264.66796875, |
|
"logits/rejected": 5134.44189453125, |
|
"logps/chosen": -303.7208557128906, |
|
"logps/rejected": -294.8185729980469, |
|
"loss": 0.4697, |
|
"rewards/accuracies": 0.7166666984558105, |
|
"rewards/chosen": -12.382735252380371, |
|
"rewards/margins": 22.952373504638672, |
|
"rewards/rejected": -35.335105895996094, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 0.8869701726844584, |
|
"grad_norm": 70.44227234088717, |
|
"learning_rate": 1.9227305739481612e-08, |
|
"logits/chosen": 5715.337890625, |
|
"logits/rejected": 4504.00537109375, |
|
"logps/chosen": -285.2997741699219, |
|
"logps/rejected": -257.0673522949219, |
|
"loss": 0.4646, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -6.729764461517334, |
|
"rewards/margins": 25.735387802124023, |
|
"rewards/rejected": -32.46515655517578, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 0.8948194662480377, |
|
"grad_norm": 61.208533994127144, |
|
"learning_rate": 1.6678653324693787e-08, |
|
"logits/chosen": 6276.97900390625, |
|
"logits/rejected": 5115.7861328125, |
|
"logps/chosen": -315.98541259765625, |
|
"logps/rejected": -297.05108642578125, |
|
"loss": 0.4635, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -10.545991897583008, |
|
"rewards/margins": 24.89821434020996, |
|
"rewards/rejected": -35.44420623779297, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 0.902668759811617, |
|
"grad_norm": 65.39132230299009, |
|
"learning_rate": 1.4305331537183384e-08, |
|
"logits/chosen": 5591.3408203125, |
|
"logits/rejected": 5121.63916015625, |
|
"logps/chosen": -279.34210205078125, |
|
"logps/rejected": -297.5921936035156, |
|
"loss": 0.47, |
|
"rewards/accuracies": 0.7000000476837158, |
|
"rewards/chosen": -12.146747589111328, |
|
"rewards/margins": 21.974903106689453, |
|
"rewards/rejected": -34.12165069580078, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 0.9105180533751962, |
|
"grad_norm": 75.9548272739494, |
|
"learning_rate": 1.2109123822844653e-08, |
|
"logits/chosen": 5726.4619140625, |
|
"logits/rejected": 4543.42919921875, |
|
"logps/chosen": -283.1075134277344, |
|
"logps/rejected": -266.6970520019531, |
|
"loss": 0.4804, |
|
"rewards/accuracies": 0.7500000596046448, |
|
"rewards/chosen": -12.42963981628418, |
|
"rewards/margins": 20.248872756958008, |
|
"rewards/rejected": -32.67850875854492, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 0.9183673469387755, |
|
"grad_norm": 81.7313814355171, |
|
"learning_rate": 1.0091680534213387e-08, |
|
"logits/chosen": 6237.8369140625, |
|
"logits/rejected": 6018.0205078125, |
|
"logps/chosen": -303.80438232421875, |
|
"logps/rejected": -340.64825439453125, |
|
"loss": 0.5009, |
|
"rewards/accuracies": 0.7166666984558105, |
|
"rewards/chosen": -11.677328109741211, |
|
"rewards/margins": 25.162853240966797, |
|
"rewards/rejected": -36.840179443359375, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 0.9262166405023547, |
|
"grad_norm": 68.30654325467928, |
|
"learning_rate": 8.254517690300944e-09, |
|
"logits/chosen": 5522.61376953125, |
|
"logits/rejected": 4998.84326171875, |
|
"logps/chosen": -286.980224609375, |
|
"logps/rejected": -291.48883056640625, |
|
"loss": 0.4715, |
|
"rewards/accuracies": 0.7833333611488342, |
|
"rewards/chosen": -8.028700828552246, |
|
"rewards/margins": 24.647029876708984, |
|
"rewards/rejected": -32.67572784423828, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 0.9340659340659341, |
|
"grad_norm": 51.93666448252585, |
|
"learning_rate": 6.599015837372907e-09, |
|
"logits/chosen": 6022.2080078125, |
|
"logits/rejected": 5236.82177734375, |
|
"logps/chosen": -306.70306396484375, |
|
"logps/rejected": -300.5201110839844, |
|
"loss": 0.4905, |
|
"rewards/accuracies": 0.7166666984558105, |
|
"rewards/chosen": -14.384405136108398, |
|
"rewards/margins": 22.299158096313477, |
|
"rewards/rejected": -36.683563232421875, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 0.9419152276295133, |
|
"grad_norm": 129.14533113743295, |
|
"learning_rate": 5.126419011529992e-09, |
|
"logits/chosen": 6208.3291015625, |
|
"logits/rejected": 5302.0283203125, |
|
"logps/chosen": -313.00469970703125, |
|
"logps/rejected": -295.9210205078125, |
|
"loss": 0.4549, |
|
"rewards/accuracies": 0.841666579246521, |
|
"rewards/chosen": -8.256613731384277, |
|
"rewards/margins": 29.51548194885254, |
|
"rewards/rejected": -37.7720947265625, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 0.9497645211930926, |
|
"grad_norm": 75.35776224261268, |
|
"learning_rate": 3.837833803870177e-09, |
|
"logits/chosen": 5844.8037109375, |
|
"logits/rejected": 5141.6005859375, |
|
"logps/chosen": -290.4355773925781, |
|
"logps/rejected": -290.68804931640625, |
|
"loss": 0.4652, |
|
"rewards/accuracies": 0.783333420753479, |
|
"rewards/chosen": -10.632657051086426, |
|
"rewards/margins": 27.73859214782715, |
|
"rewards/rejected": -38.371253967285156, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 0.957613814756672, |
|
"grad_norm": 120.6788482461967, |
|
"learning_rate": 2.734228528934679e-09, |
|
"logits/chosen": 7238.70458984375, |
|
"logits/rejected": 5344.28125, |
|
"logps/chosen": -358.9611511230469, |
|
"logps/rejected": -323.251953125, |
|
"loss": 0.4772, |
|
"rewards/accuracies": 0.7416666746139526, |
|
"rewards/chosen": -12.545631408691406, |
|
"rewards/margins": 25.407411575317383, |
|
"rewards/rejected": -37.953041076660156, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 0.9654631083202512, |
|
"grad_norm": 67.03962393569711, |
|
"learning_rate": 1.8164324970625645e-09, |
|
"logits/chosen": 6408.5908203125, |
|
"logits/rejected": 5049.8037109375, |
|
"logps/chosen": -316.5514831542969, |
|
"logps/rejected": -297.5099182128906, |
|
"loss": 0.5263, |
|
"rewards/accuracies": 0.7333333492279053, |
|
"rewards/chosen": -8.196043014526367, |
|
"rewards/margins": 24.2534122467041, |
|
"rewards/rejected": -32.44945526123047, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 0.9733124018838305, |
|
"grad_norm": 66.59250985044233, |
|
"learning_rate": 1.0851353912008642e-09, |
|
"logits/chosen": 5524.4716796875, |
|
"logits/rejected": 5108.125, |
|
"logps/chosen": -285.8993835449219, |
|
"logps/rejected": -311.83306884765625, |
|
"loss": 0.4822, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -12.945849418640137, |
|
"rewards/margins": 22.67105484008789, |
|
"rewards/rejected": -35.61690139770508, |
|
"step": 1240 |
|
}, |
|
{ |
|
"epoch": 0.9811616954474097, |
|
"grad_norm": 63.8522226531417, |
|
"learning_rate": 5.408867486384471e-10, |
|
"logits/chosen": 5626.33837890625, |
|
"logits/rejected": 4768.8837890625, |
|
"logps/chosen": -280.3611755371094, |
|
"logps/rejected": -256.4848327636719, |
|
"loss": 0.4825, |
|
"rewards/accuracies": 0.7833333611488342, |
|
"rewards/chosen": -9.922576904296875, |
|
"rewards/margins": 23.209577560424805, |
|
"rewards/rejected": -33.13215255737305, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 0.989010989010989, |
|
"grad_norm": 66.09847251390043, |
|
"learning_rate": 1.840955480532924e-10, |
|
"logits/chosen": 5381.37744140625, |
|
"logits/rejected": 5083.3779296875, |
|
"logps/chosen": -277.90594482421875, |
|
"logps/rejected": -292.15655517578125, |
|
"loss": 0.4461, |
|
"rewards/accuracies": 0.7583333253860474, |
|
"rewards/chosen": -11.253137588500977, |
|
"rewards/margins": 23.19699478149414, |
|
"rewards/rejected": -34.45013427734375, |
|
"step": 1260 |
|
}, |
|
{ |
|
"epoch": 0.9968602825745683, |
|
"grad_norm": 81.13121346879733, |
|
"learning_rate": 1.502990218302247e-11, |
|
"logits/chosen": 5605.18603515625, |
|
"logits/rejected": 4559.984375, |
|
"logps/chosen": -273.18988037109375, |
|
"logps/rejected": -262.9356994628906, |
|
"loss": 0.4819, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": -11.716361045837402, |
|
"rewards/margins": 23.061031341552734, |
|
"rewards/rejected": -34.77738952636719, |
|
"step": 1270 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"step": 1274, |
|
"total_flos": 0.0, |
|
"train_loss": 0.515247391381855, |
|
"train_runtime": 13095.6882, |
|
"train_samples_per_second": 4.668, |
|
"train_steps_per_second": 0.097 |
|
} |
|
], |
|
"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 |
|
} |
|
|