Invalid JSON: Unexpected token 'N', ..."/chosen": NaN,
"... is not valid JSON
| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 50, | |
| "global_step": 3884, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0012873326467559218, | |
| "grad_norm": 199.0, | |
| "learning_rate": 4.99356333676622e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.1812500953674316, | |
| "logps/chosen": -255.39999389648438, | |
| "logps/rejected": -215.60000610351562, | |
| "loss": 0.6883, | |
| "rewards/accuracies": 0.22499999403953552, | |
| "rewards/chosen": -0.002044677734375, | |
| "rewards/margins": 0.01108398474752903, | |
| "rewards/rejected": -0.013134765438735485, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0025746652935118436, | |
| "grad_norm": 189.0, | |
| "learning_rate": 4.987126673532441e-07, | |
| "logits/chosen": -2.8062500953674316, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -274.20001220703125, | |
| "logps/rejected": -200.60000610351562, | |
| "loss": 0.6695, | |
| "rewards/accuracies": 0.23749999701976776, | |
| "rewards/chosen": 0.09026489406824112, | |
| "rewards/margins": 0.0977783203125, | |
| "rewards/rejected": -0.0075096129439771175, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0038619979402677654, | |
| "grad_norm": 201.0, | |
| "learning_rate": 4.980690010298661e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.2281250953674316, | |
| "logps/chosen": -298.20001220703125, | |
| "logps/rejected": -281.79998779296875, | |
| "loss": 0.6836, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.010467529296875, | |
| "rewards/margins": 0.015307617373764515, | |
| "rewards/rejected": -0.004855346865952015, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.005149330587023687, | |
| "grad_norm": 202.0, | |
| "learning_rate": 4.974253347064881e-07, | |
| "logits/chosen": -3.2093749046325684, | |
| "logits/rejected": -3.2125000953674316, | |
| "logps/chosen": -322.79998779296875, | |
| "logps/rejected": -262.6000061035156, | |
| "loss": 0.6883, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": 0.01413574256002903, | |
| "rewards/margins": 0.011029052548110485, | |
| "rewards/rejected": 0.0031387328635901213, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.006436663233779609, | |
| "grad_norm": 206.0, | |
| "learning_rate": 4.967816683831102e-07, | |
| "logits/chosen": -3.215625047683716, | |
| "logits/rejected": -3.21875, | |
| "logps/chosen": -303.3999938964844, | |
| "logps/rejected": -259.6000061035156, | |
| "loss": 0.6813, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.01656494103372097, | |
| "rewards/margins": 0.02910156175494194, | |
| "rewards/rejected": -0.01250610314309597, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.007723995880535531, | |
| "grad_norm": 184.0, | |
| "learning_rate": 4.961380020597322e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.299999952316284, | |
| "logps/chosen": -271.3999938964844, | |
| "logps/rejected": -239.39999389648438, | |
| "loss": 0.6891, | |
| "rewards/accuracies": 0.2874999940395355, | |
| "rewards/chosen": 0.014312744140625, | |
| "rewards/margins": 0.008190917782485485, | |
| "rewards/rejected": 0.0061096190474927425, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.009011328527291453, | |
| "grad_norm": 147.0, | |
| "learning_rate": 4.954943357363543e-07, | |
| "logits/chosen": -3.0718750953674316, | |
| "logits/rejected": -3.1187500953674316, | |
| "logps/chosen": -230.1999969482422, | |
| "logps/rejected": -213.60000610351562, | |
| "loss": 0.6852, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.03256835788488388, | |
| "rewards/margins": 0.02081298828125, | |
| "rewards/rejected": 0.01170196570456028, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.010298661174047374, | |
| "grad_norm": 161.0, | |
| "learning_rate": 4.948506694129763e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -292.6000061035156, | |
| "logps/rejected": -255.0, | |
| "loss": 0.6813, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": 0.03703613206744194, | |
| "rewards/margins": 0.02081146277487278, | |
| "rewards/rejected": 0.01628417894244194, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.011585993820803296, | |
| "grad_norm": 251.0, | |
| "learning_rate": 4.942070030895984e-07, | |
| "logits/chosen": -3.293750047683716, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -319.20001220703125, | |
| "logps/rejected": -286.3999938964844, | |
| "loss": 0.6906, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.02326660230755806, | |
| "rewards/margins": 0.0087890625, | |
| "rewards/rejected": 0.01454315148293972, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.012873326467559218, | |
| "grad_norm": 217.0, | |
| "learning_rate": 4.935633367662204e-07, | |
| "logits/chosen": -3.0687499046325684, | |
| "logits/rejected": -3.137500047683716, | |
| "logps/chosen": -275.0, | |
| "logps/rejected": -237.8000030517578, | |
| "loss": 0.6656, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.06899414211511612, | |
| "rewards/margins": 0.06474609673023224, | |
| "rewards/rejected": 0.0042175292037427425, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01416065911431514, | |
| "grad_norm": 231.0, | |
| "learning_rate": 4.929196704428423e-07, | |
| "logits/chosen": -3.0875000953674316, | |
| "logits/rejected": -2.8187499046325684, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -223.60000610351562, | |
| "loss": 0.6867, | |
| "rewards/accuracies": 0.2750000059604645, | |
| "rewards/chosen": 0.03863525390625, | |
| "rewards/margins": 0.01720886304974556, | |
| "rewards/rejected": 0.02142944373190403, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.015447991761071062, | |
| "grad_norm": 156.0, | |
| "learning_rate": 4.922760041194645e-07, | |
| "logits/chosen": -3.1312499046325684, | |
| "logits/rejected": -3.106250047683716, | |
| "logps/chosen": -206.8000030517578, | |
| "logps/rejected": -197.0, | |
| "loss": 0.6977, | |
| "rewards/accuracies": 0.25, | |
| "rewards/chosen": 0.02888183668255806, | |
| "rewards/margins": -0.01297912560403347, | |
| "rewards/rejected": 0.04189453274011612, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.016735324407826983, | |
| "grad_norm": 240.0, | |
| "learning_rate": 4.916323377960865e-07, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.203125, | |
| "logps/chosen": -285.3999938964844, | |
| "logps/rejected": -249.8000030517578, | |
| "loss": 0.6742, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.05961913987994194, | |
| "rewards/margins": 0.0439453125, | |
| "rewards/rejected": 0.015620422549545765, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.018022657054582905, | |
| "grad_norm": 211.0, | |
| "learning_rate": 4.909886714727085e-07, | |
| "logits/chosen": -2.987499952316284, | |
| "logits/rejected": -2.9390625953674316, | |
| "logps/chosen": -315.0, | |
| "logps/rejected": -268.70001220703125, | |
| "loss": 0.6805, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 0.05659179762005806, | |
| "rewards/margins": 0.03374023362994194, | |
| "rewards/rejected": 0.02286682091653347, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.019309989701338827, | |
| "grad_norm": 159.0, | |
| "learning_rate": 4.903450051493306e-07, | |
| "logits/chosen": -3.0218749046325684, | |
| "logits/rejected": -3.206249952316284, | |
| "logps/chosen": -320.20001220703125, | |
| "logps/rejected": -233.10000610351562, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.15390625596046448, | |
| "rewards/margins": 0.12485351413488388, | |
| "rewards/rejected": 0.02894287183880806, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02059732234809475, | |
| "grad_norm": 266.0, | |
| "learning_rate": 4.897013388259526e-07, | |
| "logits/chosen": -3.1343750953674316, | |
| "logits/rejected": -3.096874952316284, | |
| "logps/chosen": -378.3999938964844, | |
| "logps/rejected": -347.79998779296875, | |
| "loss": 0.6617, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.11794433742761612, | |
| "rewards/margins": 0.06697998195886612, | |
| "rewards/rejected": 0.05112304538488388, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.02188465499485067, | |
| "grad_norm": 220.0, | |
| "learning_rate": 4.890576725025746e-07, | |
| "logits/chosen": -3.3375000953674316, | |
| "logits/rejected": -3.325000047683716, | |
| "logps/chosen": -291.0, | |
| "logps/rejected": -302.79998779296875, | |
| "loss": 0.6766, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.0849609375, | |
| "rewards/margins": 0.04403076320886612, | |
| "rewards/rejected": 0.04079589992761612, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.023171987641606592, | |
| "grad_norm": 231.0, | |
| "learning_rate": 4.884140061791967e-07, | |
| "logits/chosen": -3.262500047683716, | |
| "logits/rejected": -3.171875, | |
| "logps/chosen": -290.6000061035156, | |
| "logps/rejected": -253.0, | |
| "loss": 0.6813, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.06650390475988388, | |
| "rewards/margins": 0.03142089769244194, | |
| "rewards/rejected": 0.03498535230755806, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.024459320288362514, | |
| "grad_norm": 220.0, | |
| "learning_rate": 4.877703398558187e-07, | |
| "logits/chosen": -3.1500000953674316, | |
| "logits/rejected": -3.1937499046325684, | |
| "logps/chosen": -304.3999938964844, | |
| "logps/rejected": -240.39999389648438, | |
| "loss": 0.6703, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.06552734225988388, | |
| "rewards/margins": 0.04541015625, | |
| "rewards/rejected": 0.0201416015625, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.025746652935118436, | |
| "grad_norm": 201.0, | |
| "learning_rate": 4.871266735324407e-07, | |
| "logits/chosen": -3.2874999046325684, | |
| "logits/rejected": -3.2593750953674316, | |
| "logps/chosen": -330.20001220703125, | |
| "logps/rejected": -278.79998779296875, | |
| "loss": 0.6789, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.07391357421875, | |
| "rewards/margins": 0.05015258863568306, | |
| "rewards/rejected": 0.02381591871380806, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.027033985581874358, | |
| "grad_norm": 260.0, | |
| "learning_rate": 4.864830072090629e-07, | |
| "logits/chosen": -3.1031250953674316, | |
| "logits/rejected": -3.078125, | |
| "logps/chosen": -301.6000061035156, | |
| "logps/rejected": -261.6000061035156, | |
| "loss": 0.6828, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.10205078125, | |
| "rewards/margins": 0.03461914137005806, | |
| "rewards/rejected": 0.06721191108226776, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.02832131822863028, | |
| "grad_norm": 230.0, | |
| "learning_rate": 4.858393408856848e-07, | |
| "logits/chosen": -3.109375, | |
| "logits/rejected": -3.265625, | |
| "logps/chosen": -294.6000061035156, | |
| "logps/rejected": -269.0, | |
| "loss": 0.6711, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 0.09199218451976776, | |
| "rewards/margins": 0.054443359375, | |
| "rewards/rejected": 0.03750000149011612, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0296086508753862, | |
| "grad_norm": 249.0, | |
| "learning_rate": 4.851956745623069e-07, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.1812500953674316, | |
| "logps/chosen": -312.20001220703125, | |
| "logps/rejected": -278.20001220703125, | |
| "loss": 0.6711, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.12714843451976776, | |
| "rewards/margins": 0.06899414211511612, | |
| "rewards/rejected": 0.05832519382238388, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.030895983522142123, | |
| "grad_norm": 207.0, | |
| "learning_rate": 4.845520082389289e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -294.3999938964844, | |
| "logps/rejected": -248.8000030517578, | |
| "loss": 0.6867, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 0.06540222465991974, | |
| "rewards/margins": 0.01856689527630806, | |
| "rewards/rejected": 0.04683838039636612, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.032183316168898045, | |
| "grad_norm": 448.0, | |
| "learning_rate": 4.839083419155509e-07, | |
| "logits/chosen": -3.2125000953674316, | |
| "logits/rejected": -3.028125047683716, | |
| "logps/chosen": -309.3999938964844, | |
| "logps/rejected": -240.8000030517578, | |
| "loss": 0.6805, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.14213867485523224, | |
| "rewards/margins": 0.03446044772863388, | |
| "rewards/rejected": 0.10739745944738388, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.03347064881565397, | |
| "grad_norm": 198.0, | |
| "learning_rate": 4.83264675592173e-07, | |
| "logits/chosen": -3.2406249046325684, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -264.79998779296875, | |
| "logps/rejected": -245.8000030517578, | |
| "loss": 0.6789, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.07900390774011612, | |
| "rewards/margins": 0.051544189453125, | |
| "rewards/rejected": 0.02736816368997097, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.03475798146240989, | |
| "grad_norm": 232.0, | |
| "learning_rate": 4.82621009268795e-07, | |
| "logits/chosen": -3.075000047683716, | |
| "logits/rejected": -3.028125047683716, | |
| "logps/chosen": -293.3999938964844, | |
| "logps/rejected": -212.1999969482422, | |
| "loss": 0.6805, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.09550781548023224, | |
| "rewards/margins": 0.03429412841796875, | |
| "rewards/rejected": 0.06123046949505806, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.03604531410916581, | |
| "grad_norm": 247.0, | |
| "learning_rate": 4.819773429454171e-07, | |
| "logits/chosen": -3.015625, | |
| "logits/rejected": -3.075000047683716, | |
| "logps/chosen": -290.0, | |
| "logps/rejected": -212.39999389648438, | |
| "loss": 0.6664, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.09230957180261612, | |
| "rewards/margins": 0.06367187201976776, | |
| "rewards/rejected": 0.028591156005859375, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.03733264675592173, | |
| "grad_norm": 136.0, | |
| "learning_rate": 4.813336766220391e-07, | |
| "logits/chosen": -3.2093749046325684, | |
| "logits/rejected": -3.293750047683716, | |
| "logps/chosen": -268.3999938964844, | |
| "logps/rejected": -201.8000030517578, | |
| "loss": 0.6727, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.08615722507238388, | |
| "rewards/margins": 0.04490356519818306, | |
| "rewards/rejected": 0.04128418117761612, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.038619979402677654, | |
| "grad_norm": 197.0, | |
| "learning_rate": 4.806900102986612e-07, | |
| "logits/chosen": -3.090625047683716, | |
| "logits/rejected": -3.203125, | |
| "logps/chosen": -289.3999938964844, | |
| "logps/rejected": -240.1999969482422, | |
| "loss": 0.6656, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.11083984375, | |
| "rewards/margins": 0.07034911960363388, | |
| "rewards/rejected": 0.04030761867761612, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.039907312049433576, | |
| "grad_norm": 171.0, | |
| "learning_rate": 4.800463439752832e-07, | |
| "logits/chosen": -3.0718750953674316, | |
| "logits/rejected": -3.2249999046325684, | |
| "logps/chosen": -165.1999969482422, | |
| "logps/rejected": -154.8000030517578, | |
| "loss": 0.6867, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.060546875, | |
| "rewards/margins": 0.01654052734375, | |
| "rewards/rejected": 0.04400634765625, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.0411946446961895, | |
| "grad_norm": 180.0, | |
| "learning_rate": 4.794026776519052e-07, | |
| "logits/chosen": -2.96875, | |
| "logits/rejected": -3.1781249046325684, | |
| "logps/chosen": -192.39999389648438, | |
| "logps/rejected": -168.14999389648438, | |
| "loss": 0.6797, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.07846679538488388, | |
| "rewards/margins": 0.04019775241613388, | |
| "rewards/rejected": 0.03828125074505806, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.04248197734294542, | |
| "grad_norm": 284.0, | |
| "learning_rate": 4.787590113285273e-07, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.1968750953674316, | |
| "logps/chosen": -251.1999969482422, | |
| "logps/rejected": -254.8000030517578, | |
| "loss": 0.6469, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.11611328274011612, | |
| "rewards/margins": 0.10654296725988388, | |
| "rewards/rejected": 0.00972900353372097, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.04376930998970134, | |
| "grad_norm": 340.0, | |
| "learning_rate": 4.781153450051493e-07, | |
| "logits/chosen": -3.081249952316284, | |
| "logits/rejected": -3.128124952316284, | |
| "logps/chosen": -277.20001220703125, | |
| "logps/rejected": -265.79998779296875, | |
| "loss": 0.6922, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.10166015475988388, | |
| "rewards/margins": 0.02373046800494194, | |
| "rewards/rejected": 0.0780029296875, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.04505664263645726, | |
| "grad_norm": 207.0, | |
| "learning_rate": 4.774716786817714e-07, | |
| "logits/chosen": -2.987499952316284, | |
| "logits/rejected": -3.075000047683716, | |
| "logps/chosen": -278.3999938964844, | |
| "logps/rejected": -268.79998779296875, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.15341797471046448, | |
| "rewards/margins": 0.13554687798023224, | |
| "rewards/rejected": 0.017845917493104935, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.046343975283213185, | |
| "grad_norm": 177.0, | |
| "learning_rate": 4.7682801235839336e-07, | |
| "logits/chosen": -3.0843749046325684, | |
| "logits/rejected": -3.0531249046325684, | |
| "logps/chosen": -242.8000030517578, | |
| "logps/rejected": -238.8000030517578, | |
| "loss": 0.6703, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 0.1038818359375, | |
| "rewards/margins": 0.04204101487994194, | |
| "rewards/rejected": 0.06177978590130806, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.047631307929969106, | |
| "grad_norm": 218.0, | |
| "learning_rate": 4.7618434603501545e-07, | |
| "logits/chosen": -3.200000047683716, | |
| "logits/rejected": -3.153125047683716, | |
| "logps/chosen": -303.79998779296875, | |
| "logps/rejected": -271.6000061035156, | |
| "loss": 0.6687, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.11247558891773224, | |
| "rewards/margins": 0.07473144680261612, | |
| "rewards/rejected": 0.03764648362994194, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.04891864057672503, | |
| "grad_norm": 195.0, | |
| "learning_rate": 4.755406797116375e-07, | |
| "logits/chosen": -3.3218750953674316, | |
| "logits/rejected": -3.0218749046325684, | |
| "logps/chosen": -256.3999938964844, | |
| "logps/rejected": -291.0, | |
| "loss": 0.6562, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.09619140625, | |
| "rewards/margins": 0.08134765923023224, | |
| "rewards/rejected": 0.0149078369140625, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.05020597322348095, | |
| "grad_norm": 206.0, | |
| "learning_rate": 4.748970133882595e-07, | |
| "logits/chosen": -3.2718749046325684, | |
| "logits/rejected": -3.2593750953674316, | |
| "logps/chosen": -255.8000030517578, | |
| "logps/rejected": -222.60000610351562, | |
| "loss": 0.6766, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.1097412109375, | |
| "rewards/margins": 0.04956970363855362, | |
| "rewards/rejected": 0.06021728366613388, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.05149330587023687, | |
| "grad_norm": 162.0, | |
| "learning_rate": 4.742533470648816e-07, | |
| "logits/chosen": -2.9156250953674316, | |
| "logits/rejected": -3.0875000953674316, | |
| "logps/chosen": -217.0, | |
| "logps/rejected": -215.0, | |
| "loss": 0.6641, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.13422851264476776, | |
| "rewards/margins": 0.07473144680261612, | |
| "rewards/rejected": 0.05949706956744194, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.052780638516992794, | |
| "grad_norm": 199.0, | |
| "learning_rate": 4.7360968074150357e-07, | |
| "logits/chosen": -3.3218750953674316, | |
| "logits/rejected": -3.3687500953674316, | |
| "logps/chosen": -273.20001220703125, | |
| "logps/rejected": -272.20001220703125, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.1318359375, | |
| "rewards/margins": 0.13054199516773224, | |
| "rewards/rejected": 0.0013916015159338713, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.054067971163748715, | |
| "grad_norm": 264.0, | |
| "learning_rate": 4.729660144181256e-07, | |
| "logits/chosen": -3.065624952316284, | |
| "logits/rejected": -3.1656250953674316, | |
| "logps/chosen": -314.0, | |
| "logps/rejected": -276.20001220703125, | |
| "loss": 0.6852, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 0.10917969048023224, | |
| "rewards/margins": 0.0269775390625, | |
| "rewards/rejected": 0.08232422173023224, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.05535530381050464, | |
| "grad_norm": 1000.0, | |
| "learning_rate": 4.7232234809474765e-07, | |
| "logits/chosen": -3.1343750953674316, | |
| "logits/rejected": -2.9156250953674316, | |
| "logps/chosen": -266.0, | |
| "logps/rejected": -241.1999969482422, | |
| "loss": 0.6703, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.11232910305261612, | |
| "rewards/margins": 0.05177002027630806, | |
| "rewards/rejected": 0.06040038913488388, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.05664263645726056, | |
| "grad_norm": 258.0, | |
| "learning_rate": 4.716786817713697e-07, | |
| "logits/chosen": -3.2125000953674316, | |
| "logits/rejected": -3.2718749046325684, | |
| "logps/chosen": -314.1000061035156, | |
| "logps/rejected": -245.3000030517578, | |
| "loss": 0.6484, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.10859374701976776, | |
| "rewards/margins": 0.11284179985523224, | |
| "rewards/rejected": -0.004223632626235485, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.05792996910401648, | |
| "grad_norm": 166.0, | |
| "learning_rate": 4.7103501544799174e-07, | |
| "logits/chosen": -3.206249952316284, | |
| "logits/rejected": -3.1968750953674316, | |
| "logps/chosen": -303.6000061035156, | |
| "logps/rejected": -278.20001220703125, | |
| "loss": 0.6727, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.11298827826976776, | |
| "rewards/margins": 0.05344238132238388, | |
| "rewards/rejected": 0.05970459058880806, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.0592173017507724, | |
| "grad_norm": 237.0, | |
| "learning_rate": 4.703913491246138e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.137500047683716, | |
| "logps/chosen": -311.0, | |
| "logps/rejected": -268.6000061035156, | |
| "loss": 0.668, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.12631836533546448, | |
| "rewards/margins": 0.06418456882238388, | |
| "rewards/rejected": 0.06206054612994194, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.060504634397528324, | |
| "grad_norm": 300.0, | |
| "learning_rate": 4.697476828012358e-07, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.078125, | |
| "logps/chosen": -330.0, | |
| "logps/rejected": -293.0, | |
| "loss": 0.6672, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.13554687798023224, | |
| "rewards/margins": 0.07490234076976776, | |
| "rewards/rejected": 0.06058349460363388, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.061791967044284246, | |
| "grad_norm": 222.0, | |
| "learning_rate": 4.6910401647785787e-07, | |
| "logits/chosen": -3.065624952316284, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -239.39999389648438, | |
| "logps/rejected": -217.8000030517578, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.12773437798023224, | |
| "rewards/margins": 0.08029785007238388, | |
| "rewards/rejected": 0.04746093600988388, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.06307929969104016, | |
| "grad_norm": 199.0, | |
| "learning_rate": 4.6846035015447986e-07, | |
| "logits/chosen": -3.065624952316284, | |
| "logits/rejected": -3.046875, | |
| "logps/chosen": -228.1999969482422, | |
| "logps/rejected": -224.1999969482422, | |
| "loss": 0.6617, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 0.11665038764476776, | |
| "rewards/margins": 0.08237304538488388, | |
| "rewards/rejected": 0.03426513820886612, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.06436663233779609, | |
| "grad_norm": 168.0, | |
| "learning_rate": 4.6781668383110195e-07, | |
| "logits/chosen": -3.2593750953674316, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -256.5, | |
| "logps/rejected": -290.79998779296875, | |
| "loss": 0.65, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.08144531399011612, | |
| "rewards/margins": 0.09887085109949112, | |
| "rewards/rejected": -0.01766357384622097, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.065653964984552, | |
| "grad_norm": 233.0, | |
| "learning_rate": 4.67173017507724e-07, | |
| "logits/chosen": -3.046875, | |
| "logits/rejected": -3.090625047683716, | |
| "logps/chosen": -295.0, | |
| "logps/rejected": -277.0, | |
| "loss": 0.6656, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.06113281100988388, | |
| "rewards/margins": 0.0748291015625, | |
| "rewards/rejected": -0.01352539099752903, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.06694129763130793, | |
| "grad_norm": 250.0, | |
| "learning_rate": 4.66529351184346e-07, | |
| "logits/chosen": -3.215625047683716, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -313.3999938964844, | |
| "logps/rejected": -288.20001220703125, | |
| "loss": 0.65, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.10113525390625, | |
| "rewards/margins": 0.10419921576976776, | |
| "rewards/rejected": -0.003173828125, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.06822863027806385, | |
| "grad_norm": 274.0, | |
| "learning_rate": 4.658856848609681e-07, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.293750047683716, | |
| "logps/chosen": -301.79998779296875, | |
| "logps/rejected": -232.60000610351562, | |
| "loss": 0.6367, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.08717040717601776, | |
| "rewards/margins": 0.13535156846046448, | |
| "rewards/rejected": -0.04812011867761612, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.06951596292481978, | |
| "grad_norm": 245.0, | |
| "learning_rate": 4.652420185375901e-07, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.231250047683716, | |
| "logps/chosen": -358.0, | |
| "logps/rejected": -302.3999938964844, | |
| "loss": 0.6422, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.12468262016773224, | |
| "rewards/margins": 0.13007202744483948, | |
| "rewards/rejected": -0.0052246092818677425, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.07080329557157569, | |
| "grad_norm": 168.0, | |
| "learning_rate": 4.645983522142121e-07, | |
| "logits/chosen": -3.2593750953674316, | |
| "logits/rejected": -3.2718749046325684, | |
| "logps/chosen": -277.6000061035156, | |
| "logps/rejected": -211.6999969482422, | |
| "loss": 0.6523, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.11538086086511612, | |
| "rewards/margins": 0.12126465141773224, | |
| "rewards/rejected": -0.005908203311264515, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.07209062821833162, | |
| "grad_norm": 254.0, | |
| "learning_rate": 4.639546858908342e-07, | |
| "logits/chosen": -2.987499952316284, | |
| "logits/rejected": -3.1156249046325684, | |
| "logps/chosen": -349.20001220703125, | |
| "logps/rejected": -261.6000061035156, | |
| "loss": 0.6219, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.19648437201976776, | |
| "rewards/margins": 0.20097656548023224, | |
| "rewards/rejected": -0.004687500186264515, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.07337796086508754, | |
| "grad_norm": 264.0, | |
| "learning_rate": 4.633110195674562e-07, | |
| "logits/chosen": -3.0250000953674316, | |
| "logits/rejected": -3.012500047683716, | |
| "logps/chosen": -285.79998779296875, | |
| "logps/rejected": -251.0, | |
| "loss": 0.6523, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.14013671875, | |
| "rewards/margins": 0.11826171725988388, | |
| "rewards/rejected": 0.02204589918255806, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.07466529351184346, | |
| "grad_norm": 209.0, | |
| "learning_rate": 4.6266735324407824e-07, | |
| "logits/chosen": -3.2125000953674316, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -263.79998779296875, | |
| "logps/rejected": -258.20001220703125, | |
| "loss": 0.6805, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.072509765625, | |
| "rewards/margins": 0.05144042894244194, | |
| "rewards/rejected": 0.02106017991900444, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.07595262615859938, | |
| "grad_norm": 231.0, | |
| "learning_rate": 4.620236869207003e-07, | |
| "logits/chosen": -3.1624999046325684, | |
| "logits/rejected": -3.1500000953674316, | |
| "logps/chosen": -332.20001220703125, | |
| "logps/rejected": -308.3999938964844, | |
| "loss": 0.6359, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.15419921278953552, | |
| "rewards/margins": 0.13994140923023224, | |
| "rewards/rejected": 0.014514160342514515, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.07723995880535531, | |
| "grad_norm": 184.0, | |
| "learning_rate": 4.613800205973223e-07, | |
| "logits/chosen": -3.1312499046325684, | |
| "logits/rejected": -3.081249952316284, | |
| "logps/chosen": -268.20001220703125, | |
| "logps/rejected": -237.89999389648438, | |
| "loss": 0.6672, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.094696044921875, | |
| "rewards/margins": 0.07314453274011612, | |
| "rewards/rejected": 0.02178955078125, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07852729145211122, | |
| "grad_norm": 232.0, | |
| "learning_rate": 4.6073635427394437e-07, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -313.79998779296875, | |
| "logps/rejected": -268.6000061035156, | |
| "loss": 0.6617, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.0489501953125, | |
| "rewards/margins": 0.09577026218175888, | |
| "rewards/rejected": -0.046875, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.07981462409886715, | |
| "grad_norm": 266.0, | |
| "learning_rate": 4.6009268795056636e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -302.0, | |
| "logps/rejected": -255.8000030517578, | |
| "loss": 0.6578, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.13081054389476776, | |
| "rewards/margins": 0.09918212890625, | |
| "rewards/rejected": 0.031494140625, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.08110195674562307, | |
| "grad_norm": 195.0, | |
| "learning_rate": 4.5944902162718845e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.1968750953674316, | |
| "logps/chosen": -243.39999389648438, | |
| "logps/rejected": -268.20001220703125, | |
| "loss": 0.6594, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.12099609524011612, | |
| "rewards/margins": 0.09287109225988388, | |
| "rewards/rejected": 0.027968596667051315, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.082389289392379, | |
| "grad_norm": 198.0, | |
| "learning_rate": 4.588053553038105e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.1031250953674316, | |
| "logps/chosen": -238.39999389648438, | |
| "logps/rejected": -197.2937469482422, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 0.12089844048023224, | |
| "rewards/margins": 0.0970611572265625, | |
| "rewards/rejected": 0.02369079552590847, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.08367662203913491, | |
| "grad_norm": 235.0, | |
| "learning_rate": 4.581616889804325e-07, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.15625, | |
| "logps/chosen": -286.0, | |
| "logps/rejected": -260.79998779296875, | |
| "loss": 0.675, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.12059326469898224, | |
| "rewards/margins": 0.06058349460363388, | |
| "rewards/rejected": 0.0599365234375, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.08496395468589084, | |
| "grad_norm": 220.0, | |
| "learning_rate": 4.575180226570546e-07, | |
| "logits/chosen": -3.265625, | |
| "logits/rejected": -3.1656250953674316, | |
| "logps/chosen": -261.3999938964844, | |
| "logps/rejected": -247.0, | |
| "loss": 0.6531, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.05754394456744194, | |
| "rewards/margins": 0.10883788764476776, | |
| "rewards/rejected": -0.05126953125, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.08625128733264675, | |
| "grad_norm": 242.0, | |
| "learning_rate": 4.568743563336766e-07, | |
| "logits/chosen": -3.09375, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -248.8000030517578, | |
| "logps/rejected": -188.6999969482422, | |
| "loss": 0.7148, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.09836425632238388, | |
| "rewards/margins": 0.02453002892434597, | |
| "rewards/rejected": 0.07387695461511612, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.08753861997940268, | |
| "grad_norm": 130.0, | |
| "learning_rate": 4.562306900102986e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.21875, | |
| "logps/chosen": -206.8000030517578, | |
| "logps/rejected": -196.8000030517578, | |
| "loss": 0.6836, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.06015624850988388, | |
| "rewards/margins": 0.03666992112994194, | |
| "rewards/rejected": 0.0233154296875, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.0888259526261586, | |
| "grad_norm": 284.0, | |
| "learning_rate": 4.555870236869207e-07, | |
| "logits/chosen": -3.328125, | |
| "logits/rejected": -2.895312547683716, | |
| "logps/chosen": -138.35000610351562, | |
| "logps/rejected": -167.8000030517578, | |
| "loss": 0.6555, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.05758056789636612, | |
| "rewards/margins": 0.08784179389476776, | |
| "rewards/rejected": -0.0302734375, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.09011328527291453, | |
| "grad_norm": 207.0, | |
| "learning_rate": 4.549433573635427e-07, | |
| "logits/chosen": -3.1156249046325684, | |
| "logits/rejected": -3.143749952316284, | |
| "logps/chosen": -300.3999938964844, | |
| "logps/rejected": -235.1999969482422, | |
| "loss": 0.6367, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.17333984375, | |
| "rewards/margins": 0.15162353217601776, | |
| "rewards/rejected": 0.02155761793255806, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.09140061791967044, | |
| "grad_norm": 196.0, | |
| "learning_rate": 4.5429969104016474e-07, | |
| "logits/chosen": -3.065624952316284, | |
| "logits/rejected": -3.0374999046325684, | |
| "logps/chosen": -264.6000061035156, | |
| "logps/rejected": -214.60000610351562, | |
| "loss": 0.6742, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.09343262016773224, | |
| "rewards/margins": 0.06411132961511612, | |
| "rewards/rejected": 0.02947998046875, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.09268795056642637, | |
| "grad_norm": 217.0, | |
| "learning_rate": 4.5365602471678684e-07, | |
| "logits/chosen": -3.2562499046325684, | |
| "logits/rejected": -3.143749952316284, | |
| "logps/chosen": -302.3999938964844, | |
| "logps/rejected": -266.79998779296875, | |
| "loss": 0.6687, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.12880858778953552, | |
| "rewards/margins": 0.08376464992761612, | |
| "rewards/rejected": 0.04526367038488388, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.09397528321318228, | |
| "grad_norm": 236.0, | |
| "learning_rate": 4.5301235839340883e-07, | |
| "logits/chosen": -3.121875047683716, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -329.6000061035156, | |
| "logps/rejected": -243.39999389648438, | |
| "loss": 0.6508, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.11865234375, | |
| "rewards/margins": 0.11799316108226776, | |
| "rewards/rejected": 0.0005126952892169356, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.09526261585993821, | |
| "grad_norm": 244.0, | |
| "learning_rate": 4.5236869207003087e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -275.79998779296875, | |
| "logps/rejected": -245.0, | |
| "loss": 0.6648, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.08018799126148224, | |
| "rewards/margins": 0.08574219048023224, | |
| "rewards/rejected": -0.005480957217514515, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.09654994850669413, | |
| "grad_norm": 236.0, | |
| "learning_rate": 4.517250257466529e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.2125000953674316, | |
| "logps/chosen": -328.20001220703125, | |
| "logps/rejected": -286.79998779296875, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.2138671875, | |
| "rewards/margins": 0.22158202528953552, | |
| "rewards/rejected": -0.0068359375, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.09783728115345006, | |
| "grad_norm": 213.0, | |
| "learning_rate": 4.5108135942327496e-07, | |
| "logits/chosen": -3.2249999046325684, | |
| "logits/rejected": -3.2593750953674316, | |
| "logps/chosen": -277.79998779296875, | |
| "logps/rejected": -210.6999969482422, | |
| "loss": 0.632, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.14072266221046448, | |
| "rewards/margins": 0.16083984076976776, | |
| "rewards/rejected": -0.02003173902630806, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.09912461380020597, | |
| "grad_norm": 194.0, | |
| "learning_rate": 4.50437693099897e-07, | |
| "logits/chosen": -3.1031250953674316, | |
| "logits/rejected": -3.1937499046325684, | |
| "logps/chosen": -298.20001220703125, | |
| "logps/rejected": -214.58749389648438, | |
| "loss": 0.6438, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.13935546576976776, | |
| "rewards/margins": 0.1220703125, | |
| "rewards/rejected": 0.01725158654153347, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.1004119464469619, | |
| "grad_norm": 223.0, | |
| "learning_rate": 4.49794026776519e-07, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -3.21875, | |
| "logps/chosen": -323.3999938964844, | |
| "logps/rejected": -264.3999938964844, | |
| "loss": 0.6391, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.15126952528953552, | |
| "rewards/margins": 0.11635742336511612, | |
| "rewards/rejected": 0.03485107421875, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.10169927909371781, | |
| "grad_norm": 130.0, | |
| "learning_rate": 4.491503604531411e-07, | |
| "logits/chosen": -3.034374952316284, | |
| "logits/rejected": -2.981250047683716, | |
| "logps/chosen": -265.0, | |
| "logps/rejected": -183.8000030517578, | |
| "loss": 0.6547, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.09814453125, | |
| "rewards/margins": 0.11323241889476776, | |
| "rewards/rejected": -0.01495361328125, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.10298661174047374, | |
| "grad_norm": 258.0, | |
| "learning_rate": 4.4850669412976313e-07, | |
| "logits/chosen": -2.984375, | |
| "logits/rejected": -3.0218749046325684, | |
| "logps/chosen": -251.39999389648438, | |
| "logps/rejected": -237.8000030517578, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.12263183295726776, | |
| "rewards/margins": 0.11127929389476776, | |
| "rewards/rejected": 0.01132812537252903, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.10427394438722966, | |
| "grad_norm": 238.0, | |
| "learning_rate": 4.478630278063851e-07, | |
| "logits/chosen": -3.1656250953674316, | |
| "logits/rejected": -3.2249999046325684, | |
| "logps/chosen": -266.3999938964844, | |
| "logps/rejected": -250.39999389648438, | |
| "loss": 0.6531, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.15058593451976776, | |
| "rewards/margins": 0.114990234375, | |
| "rewards/rejected": 0.03544921800494194, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.10556127703398559, | |
| "grad_norm": 202.0, | |
| "learning_rate": 4.472193614830072e-07, | |
| "logits/chosen": -3.1624999046325684, | |
| "logits/rejected": -3.284374952316284, | |
| "logps/chosen": -282.3999938964844, | |
| "logps/rejected": -236.0, | |
| "loss": 0.6297, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.12856444716453552, | |
| "rewards/margins": 0.15444335341453552, | |
| "rewards/rejected": -0.02585449256002903, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.1068486096807415, | |
| "grad_norm": 144.0, | |
| "learning_rate": 4.4657569515962926e-07, | |
| "logits/chosen": -2.9375, | |
| "logits/rejected": -3.0625, | |
| "logps/chosen": -228.39999389648438, | |
| "logps/rejected": -204.8000030517578, | |
| "loss": 0.6562, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.16166992485523224, | |
| "rewards/margins": 0.10384521633386612, | |
| "rewards/rejected": 0.05778808519244194, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.10813594232749743, | |
| "grad_norm": 219.0, | |
| "learning_rate": 4.4593202883625124e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.3031249046325684, | |
| "logps/chosen": -287.3999938964844, | |
| "logps/rejected": -257.0, | |
| "loss": 0.6477, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.12734374403953552, | |
| "rewards/margins": 0.13564452528953552, | |
| "rewards/rejected": -0.008105468936264515, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.10942327497425335, | |
| "grad_norm": 222.0, | |
| "learning_rate": 4.4528836251287334e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -265.6000061035156, | |
| "logps/rejected": -256.0, | |
| "loss": 0.6469, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.15068359673023224, | |
| "rewards/margins": 0.12324218451976776, | |
| "rewards/rejected": 0.02756347693502903, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.11071060762100927, | |
| "grad_norm": 148.0, | |
| "learning_rate": 4.4464469618949533e-07, | |
| "logits/chosen": -3.2125000953674316, | |
| "logits/rejected": -3.15625, | |
| "logps/chosen": -240.60000610351562, | |
| "logps/rejected": -245.60000610351562, | |
| "loss": 0.6562, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.11338500678539276, | |
| "rewards/margins": 0.10663451999425888, | |
| "rewards/rejected": 0.006579590030014515, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.11199794026776519, | |
| "grad_norm": 242.0, | |
| "learning_rate": 4.4400102986611737e-07, | |
| "logits/chosen": -3.2437500953674316, | |
| "logits/rejected": -3.237499952316284, | |
| "logps/chosen": -330.3999938964844, | |
| "logps/rejected": -300.79998779296875, | |
| "loss": 0.6648, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.13999024033546448, | |
| "rewards/margins": 0.0843505859375, | |
| "rewards/rejected": 0.05549926683306694, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.11328527291452112, | |
| "grad_norm": 248.0, | |
| "learning_rate": 4.4335736354273947e-07, | |
| "logits/chosen": -3.090625047683716, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -314.20001220703125, | |
| "logps/rejected": -277.3999938964844, | |
| "loss": 0.6633, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.17368164658546448, | |
| "rewards/margins": 0.10908202826976776, | |
| "rewards/rejected": 0.06459961086511612, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.11457260556127703, | |
| "grad_norm": 184.0, | |
| "learning_rate": 4.4271369721936146e-07, | |
| "logits/chosen": -3.106250047683716, | |
| "logits/rejected": -3.1031250953674316, | |
| "logps/chosen": -257.0, | |
| "logps/rejected": -234.60000610351562, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.10633544623851776, | |
| "rewards/margins": 0.1494140625, | |
| "rewards/rejected": -0.04318847507238388, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.11585993820803296, | |
| "grad_norm": 216.0, | |
| "learning_rate": 4.420700308959835e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.059375047683716, | |
| "logps/chosen": -292.79998779296875, | |
| "logps/rejected": -303.79998779296875, | |
| "loss": 0.6594, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.10972289741039276, | |
| "rewards/margins": 0.10012207180261612, | |
| "rewards/rejected": 0.00937500037252903, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.11714727085478888, | |
| "grad_norm": 217.0, | |
| "learning_rate": 4.414263645726055e-07, | |
| "logits/chosen": -3.28125, | |
| "logits/rejected": -3.315624952316284, | |
| "logps/chosen": -283.3999938964844, | |
| "logps/rejected": -266.6000061035156, | |
| "loss": 0.6289, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.04505004733800888, | |
| "rewards/margins": 0.17610473930835724, | |
| "rewards/rejected": -0.13100585341453552, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.1184346035015448, | |
| "grad_norm": 235.0, | |
| "learning_rate": 4.407826982492276e-07, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -254.39999389648438, | |
| "logps/rejected": -244.0, | |
| "loss": 0.6656, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.05616455152630806, | |
| "rewards/margins": 0.09648437798023224, | |
| "rewards/rejected": -0.04036254808306694, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.11972193614830072, | |
| "grad_norm": 230.0, | |
| "learning_rate": 4.4013903192584963e-07, | |
| "logits/chosen": -3.1156249046325684, | |
| "logits/rejected": -3.034374952316284, | |
| "logps/chosen": -234.8000030517578, | |
| "logps/rejected": -255.25, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.19570311903953552, | |
| "rewards/margins": 0.09677734225988388, | |
| "rewards/rejected": 0.09892578423023224, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.12100926879505665, | |
| "grad_norm": 266.0, | |
| "learning_rate": 4.394953656024716e-07, | |
| "logits/chosen": -3.1812500953674316, | |
| "logits/rejected": -3.1187500953674316, | |
| "logps/chosen": -300.6000061035156, | |
| "logps/rejected": -246.0, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.18437500298023224, | |
| "rewards/margins": 0.14165039360523224, | |
| "rewards/rejected": 0.04294433444738388, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.12229660144181256, | |
| "grad_norm": 174.0, | |
| "learning_rate": 4.388516992790937e-07, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.206249952316284, | |
| "logps/chosen": -257.20001220703125, | |
| "logps/rejected": -276.6000061035156, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.052734375, | |
| "rewards/margins": 0.160888671875, | |
| "rewards/rejected": -0.10834960639476776, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.12358393408856849, | |
| "grad_norm": 117.0, | |
| "learning_rate": 4.3820803295571576e-07, | |
| "logits/chosen": -3.028125047683716, | |
| "logits/rejected": -3.09375, | |
| "logps/chosen": -299.3999938964844, | |
| "logps/rejected": -247.875, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.21976622939109802, | |
| "rewards/margins": 0.18496093153953552, | |
| "rewards/rejected": 0.03480224683880806, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.12487126673532441, | |
| "grad_norm": 222.0, | |
| "learning_rate": 4.3756436663233775e-07, | |
| "logits/chosen": NaN, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -268.6000061035156, | |
| "logps/rejected": -255.74063110351562, | |
| "loss": 0.6594, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.11562500149011612, | |
| "rewards/margins": 0.10996093600988388, | |
| "rewards/rejected": 0.00555419921875, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.12615859938208032, | |
| "grad_norm": 236.0, | |
| "learning_rate": 4.3692070030895984e-07, | |
| "logits/chosen": -3.1968750953674316, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -239.1999969482422, | |
| "logps/rejected": -301.0, | |
| "loss": 0.6773, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.01584472693502903, | |
| "rewards/margins": 0.07509765774011612, | |
| "rewards/rejected": -0.05917968600988388, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.12744593202883625, | |
| "grad_norm": 230.0, | |
| "learning_rate": 4.3627703398558183e-07, | |
| "logits/chosen": -3.003124952316284, | |
| "logits/rejected": -2.9749999046325684, | |
| "logps/chosen": -252.0, | |
| "logps/rejected": -293.79998779296875, | |
| "loss": 0.6453, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.19011230766773224, | |
| "rewards/margins": 0.12675781548023224, | |
| "rewards/rejected": 0.0634765625, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.12873326467559218, | |
| "grad_norm": 186.0, | |
| "learning_rate": 4.356333676622039e-07, | |
| "logits/chosen": -3.090625047683716, | |
| "logits/rejected": -2.875, | |
| "logps/chosen": -249.6999969482422, | |
| "logps/rejected": -220.1999969482422, | |
| "loss": 0.6219, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.15830688178539276, | |
| "rewards/margins": 0.18085937201976776, | |
| "rewards/rejected": -0.02264709398150444, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1300205973223481, | |
| "grad_norm": 190.0, | |
| "learning_rate": 4.3498970133882597e-07, | |
| "logits/chosen": NaN, | |
| "logits/rejected": -3.043750047683716, | |
| "logps/chosen": -222.39999389648438, | |
| "logps/rejected": -179.83438110351562, | |
| "loss": 0.6375, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.12255859375, | |
| "rewards/margins": 0.14835968613624573, | |
| "rewards/rejected": -0.02584991417825222, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.131307929969104, | |
| "grad_norm": 171.0, | |
| "learning_rate": 4.3434603501544796e-07, | |
| "logits/chosen": -3.315624952316284, | |
| "logits/rejected": -3.3531250953674316, | |
| "logps/chosen": -284.3999938964844, | |
| "logps/rejected": -250.8000030517578, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.09062500298023224, | |
| "rewards/margins": 0.16804198920726776, | |
| "rewards/rejected": -0.07780762016773224, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.13259526261585994, | |
| "grad_norm": 126.5, | |
| "learning_rate": 4.3370236869207e-07, | |
| "logits/chosen": -3.0250000953674316, | |
| "logits/rejected": -3.125, | |
| "logps/chosen": -233.60000610351562, | |
| "logps/rejected": -213.8000030517578, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.14226074516773224, | |
| "rewards/margins": 0.10944823920726776, | |
| "rewards/rejected": 0.03278808668255806, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.13388259526261587, | |
| "grad_norm": 178.0, | |
| "learning_rate": 4.330587023686921e-07, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -3.1656250953674316, | |
| "logps/chosen": -270.20001220703125, | |
| "logps/rejected": -283.20001220703125, | |
| "loss": 0.6266, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.11562500149011612, | |
| "rewards/margins": 0.17392578721046448, | |
| "rewards/rejected": -0.05846252292394638, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.1351699279093718, | |
| "grad_norm": 215.0, | |
| "learning_rate": 4.324150360453141e-07, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.284374952316284, | |
| "logps/chosen": -307.20001220703125, | |
| "logps/rejected": -220.1999969482422, | |
| "loss": 0.632, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.22519531846046448, | |
| "rewards/margins": 0.15097656846046448, | |
| "rewards/rejected": 0.074371337890625, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.1364572605561277, | |
| "grad_norm": 160.0, | |
| "learning_rate": 4.3177136972193613e-07, | |
| "logits/chosen": -3.2093749046325684, | |
| "logits/rejected": -3.231250047683716, | |
| "logps/chosen": -271.79998779296875, | |
| "logps/rejected": -250.0, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.05512695387005806, | |
| "rewards/margins": 0.15380859375, | |
| "rewards/rejected": -0.09833984076976776, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.13774459320288363, | |
| "grad_norm": 227.0, | |
| "learning_rate": 4.311277033985581e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.121875047683716, | |
| "logps/chosen": -252.1999969482422, | |
| "logps/rejected": -277.79998779296875, | |
| "loss": 0.6875, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.12172851711511612, | |
| "rewards/margins": 0.052734375, | |
| "rewards/rejected": 0.06928100436925888, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.13903192584963955, | |
| "grad_norm": 245.0, | |
| "learning_rate": 4.304840370751802e-07, | |
| "logits/chosen": -3.0999999046325684, | |
| "logits/rejected": -2.7796874046325684, | |
| "logps/chosen": -268.6000061035156, | |
| "logps/rejected": -280.0, | |
| "loss": 0.6922, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.16083984076976776, | |
| "rewards/margins": 0.06669922173023224, | |
| "rewards/rejected": 0.09462890774011612, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.14031925849639545, | |
| "grad_norm": 208.0, | |
| "learning_rate": 4.2984037075180226e-07, | |
| "logits/chosen": -3.065624952316284, | |
| "logits/rejected": -3.143749952316284, | |
| "logps/chosen": -252.8000030517578, | |
| "logps/rejected": -230.60000610351562, | |
| "loss": 0.6078, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.13334961235523224, | |
| "rewards/margins": 0.21835938096046448, | |
| "rewards/rejected": -0.08524779975414276, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.14160659114315138, | |
| "grad_norm": 189.0, | |
| "learning_rate": 4.2919670442842425e-07, | |
| "logits/chosen": -3.296875, | |
| "logits/rejected": -3.3843750953674316, | |
| "logps/chosen": -306.6000061035156, | |
| "logps/rejected": -241.39999389648438, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.12783202528953552, | |
| "rewards/margins": 0.16796875, | |
| "rewards/rejected": -0.04018554836511612, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.1428939237899073, | |
| "grad_norm": 165.0, | |
| "learning_rate": 4.2855303810504634e-07, | |
| "logits/chosen": -3.140625, | |
| "logits/rejected": -3.153125047683716, | |
| "logps/chosen": -236.39999389648438, | |
| "logps/rejected": -216.39999389648438, | |
| "loss": 0.6523, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.14702148735523224, | |
| "rewards/margins": 0.10488281399011612, | |
| "rewards/rejected": 0.04229736328125, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.14418125643666324, | |
| "grad_norm": 219.0, | |
| "learning_rate": 4.2790937178166833e-07, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.1500000953674316, | |
| "logps/chosen": -243.1999969482422, | |
| "logps/rejected": -215.1999969482422, | |
| "loss": 0.643, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.13973388075828552, | |
| "rewards/margins": 0.13666991889476776, | |
| "rewards/rejected": 0.0029052733443677425, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.14546858908341914, | |
| "grad_norm": 227.0, | |
| "learning_rate": 4.272657054582904e-07, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -2.9749999046325684, | |
| "logps/chosen": -275.6000061035156, | |
| "logps/rejected": -243.6999969482422, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.13615722954273224, | |
| "rewards/margins": 0.208984375, | |
| "rewards/rejected": -0.07333984225988388, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.14675592173017507, | |
| "grad_norm": 184.0, | |
| "learning_rate": 4.2662203913491247e-07, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -2.8499999046325684, | |
| "logps/chosen": -263.3999938964844, | |
| "logps/rejected": -236.1999969482422, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.11201171576976776, | |
| "rewards/margins": 0.1708984375, | |
| "rewards/rejected": -0.05898437649011612, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.148043254376931, | |
| "grad_norm": 214.0, | |
| "learning_rate": 4.2597837281153446e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.1937499046325684, | |
| "logps/chosen": -226.1999969482422, | |
| "logps/rejected": -289.6000061035156, | |
| "loss": 0.6547, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.10014648735523224, | |
| "rewards/margins": 0.14655761420726776, | |
| "rewards/rejected": -0.04685058444738388, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.14933058702368693, | |
| "grad_norm": 193.0, | |
| "learning_rate": 4.253347064881565e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -2.9593749046325684, | |
| "logps/chosen": -270.79998779296875, | |
| "logps/rejected": -262.0, | |
| "loss": 0.6234, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.12539061903953552, | |
| "rewards/margins": 0.18266601860523224, | |
| "rewards/rejected": -0.05742187425494194, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.15061791967044283, | |
| "grad_norm": 202.0, | |
| "learning_rate": 4.246910401647786e-07, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -3.121875047683716, | |
| "logps/chosen": -298.3999938964844, | |
| "logps/rejected": -263.79998779296875, | |
| "loss": 0.6211, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.19150391221046448, | |
| "rewards/margins": 0.17910155653953552, | |
| "rewards/rejected": 0.012744140811264515, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.15190525231719876, | |
| "grad_norm": 189.0, | |
| "learning_rate": 4.240473738414006e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.09375, | |
| "logps/chosen": -266.20001220703125, | |
| "logps/rejected": -220.39999389648438, | |
| "loss": 0.6859, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.18232421576976776, | |
| "rewards/margins": 0.06546630710363388, | |
| "rewards/rejected": 0.1170654296875, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.1531925849639547, | |
| "grad_norm": 197.0, | |
| "learning_rate": 4.2340370751802263e-07, | |
| "logits/chosen": -3.1031250953674316, | |
| "logits/rejected": -3.246875047683716, | |
| "logps/chosen": -323.20001220703125, | |
| "logps/rejected": -257.20001220703125, | |
| "loss": 0.5813, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.15024414658546448, | |
| "rewards/margins": 0.29150390625, | |
| "rewards/rejected": -0.14128418266773224, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.15447991761071062, | |
| "grad_norm": 169.0, | |
| "learning_rate": 4.227600411946447e-07, | |
| "logits/chosen": -3.246875047683716, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -279.79998779296875, | |
| "logps/rejected": -205.10000610351562, | |
| "loss": 0.6406, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.08596191555261612, | |
| "rewards/margins": 0.15449218451976776, | |
| "rewards/rejected": -0.068572998046875, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.15576725025746652, | |
| "grad_norm": 251.0, | |
| "learning_rate": 4.221163748712667e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.253124952316284, | |
| "logps/chosen": -353.6000061035156, | |
| "logps/rejected": -286.0, | |
| "loss": 0.6492, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.15336914360523224, | |
| "rewards/margins": 0.14345702528953552, | |
| "rewards/rejected": 0.01035156287252903, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.15705458290422245, | |
| "grad_norm": 176.0, | |
| "learning_rate": 4.2147270854788876e-07, | |
| "logits/chosen": -3.2718749046325684, | |
| "logits/rejected": -3.174999952316284, | |
| "logps/chosen": -233.3000030517578, | |
| "logps/rejected": -212.10000610351562, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.087158203125, | |
| "rewards/margins": 0.16020508110523224, | |
| "rewards/rejected": -0.07280273735523224, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.15834191555097837, | |
| "grad_norm": 205.0, | |
| "learning_rate": 4.208290422245108e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.2249999046325684, | |
| "logps/chosen": -275.8999938964844, | |
| "logps/rejected": -256.79998779296875, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.10605468600988388, | |
| "rewards/margins": 0.10166015475988388, | |
| "rewards/rejected": 0.004638671875, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.1596292481977343, | |
| "grad_norm": 226.0, | |
| "learning_rate": 4.2018537590113285e-07, | |
| "logits/chosen": -3.2406249046325684, | |
| "logits/rejected": -3.246875047683716, | |
| "logps/chosen": -252.39999389648438, | |
| "logps/rejected": -307.6000061035156, | |
| "loss": 0.668, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.03529052808880806, | |
| "rewards/margins": 0.10805664211511612, | |
| "rewards/rejected": -0.07302246242761612, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.1609165808444902, | |
| "grad_norm": 172.0, | |
| "learning_rate": 4.195417095777549e-07, | |
| "logits/chosen": -3.1156249046325684, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -242.8000030517578, | |
| "logps/rejected": -213.1999969482422, | |
| "loss": 0.6023, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.15589599311351776, | |
| "rewards/margins": 0.23017577826976776, | |
| "rewards/rejected": -0.07406006008386612, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.16220391349124613, | |
| "grad_norm": 212.0, | |
| "learning_rate": 4.188980432543769e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.049999952316284, | |
| "logps/chosen": -242.3000030517578, | |
| "logps/rejected": -196.60000610351562, | |
| "loss": 0.6398, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.06118164211511612, | |
| "rewards/margins": 0.1611328125, | |
| "rewards/rejected": -0.09953002631664276, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.16349124613800206, | |
| "grad_norm": 242.0, | |
| "learning_rate": 4.18254376930999e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.121875047683716, | |
| "logps/chosen": -299.3999938964844, | |
| "logps/rejected": -277.79998779296875, | |
| "loss": 0.6164, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.12120361626148224, | |
| "rewards/margins": 0.20722655951976776, | |
| "rewards/rejected": -0.08637695014476776, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.164778578784758, | |
| "grad_norm": 219.0, | |
| "learning_rate": 4.1761071060762096e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.171875, | |
| "logps/chosen": -272.0, | |
| "logps/rejected": -258.20001220703125, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.11954345554113388, | |
| "rewards/margins": 0.16679687798023224, | |
| "rewards/rejected": -0.04699096828699112, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.1660659114315139, | |
| "grad_norm": 398.0, | |
| "learning_rate": 4.16967044284243e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.21875, | |
| "logps/chosen": -300.3999938964844, | |
| "logps/rejected": -268.0, | |
| "loss": 0.657, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.14911499619483948, | |
| "rewards/margins": 0.12636718153953552, | |
| "rewards/rejected": 0.0225830078125, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.16735324407826982, | |
| "grad_norm": 196.0, | |
| "learning_rate": 4.163233779608651e-07, | |
| "logits/chosen": -3.03125, | |
| "logits/rejected": -3.0218749046325684, | |
| "logps/chosen": -270.6000061035156, | |
| "logps/rejected": -232.0, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.23095703125, | |
| "rewards/margins": 0.17384032905101776, | |
| "rewards/rejected": 0.057403564453125, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.16864057672502575, | |
| "grad_norm": 172.0, | |
| "learning_rate": 4.156797116374871e-07, | |
| "logits/chosen": -2.903125047683716, | |
| "logits/rejected": -2.840625047683716, | |
| "logps/chosen": -170.75, | |
| "logps/rejected": -184.3000030517578, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 0.10405273735523224, | |
| "rewards/margins": 0.09428711235523224, | |
| "rewards/rejected": 0.010021972469985485, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.16992790937178168, | |
| "grad_norm": 207.0, | |
| "learning_rate": 4.1503604531410913e-07, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -343.20001220703125, | |
| "logps/rejected": -275.6000061035156, | |
| "loss": 0.5867, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.23427733778953552, | |
| "rewards/margins": 0.31171876192092896, | |
| "rewards/rejected": -0.07719726860523224, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.17121524201853758, | |
| "grad_norm": 191.0, | |
| "learning_rate": 4.1439237899073123e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -270.6000061035156, | |
| "logps/rejected": -298.6000061035156, | |
| "loss": 0.6469, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.08298339694738388, | |
| "rewards/margins": 0.14306640625, | |
| "rewards/rejected": -0.06015624850988388, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.1725025746652935, | |
| "grad_norm": 187.0, | |
| "learning_rate": 4.137487126673532e-07, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -2.9781250953674316, | |
| "logps/chosen": -223.1999969482422, | |
| "logps/rejected": -207.1999969482422, | |
| "loss": 0.6203, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.151123046875, | |
| "rewards/margins": 0.18344727158546448, | |
| "rewards/rejected": -0.03291015699505806, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.17378990731204944, | |
| "grad_norm": 298.0, | |
| "learning_rate": 4.1310504634397526e-07, | |
| "logits/chosen": -3.0875000953674316, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -348.0, | |
| "logps/rejected": -248.0, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.203125, | |
| "rewards/margins": 0.17519530653953552, | |
| "rewards/rejected": 0.0285491943359375, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.17507723995880536, | |
| "grad_norm": 203.0, | |
| "learning_rate": 4.124613800205973e-07, | |
| "logits/chosen": -3.1968750953674316, | |
| "logits/rejected": -3.09375, | |
| "logps/chosen": -326.6000061035156, | |
| "logps/rejected": -340.0, | |
| "loss": 0.6195, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.16599121689796448, | |
| "rewards/margins": 0.21269531548023224, | |
| "rewards/rejected": -0.046630859375, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.17636457260556127, | |
| "grad_norm": 224.0, | |
| "learning_rate": 4.1181771369721935e-07, | |
| "logits/chosen": -3.1500000953674316, | |
| "logits/rejected": -3.1812500953674316, | |
| "logps/chosen": -328.0, | |
| "logps/rejected": -251.60000610351562, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.24326172471046448, | |
| "rewards/margins": 0.18779297173023224, | |
| "rewards/rejected": 0.05568847805261612, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.1776519052523172, | |
| "grad_norm": 222.0, | |
| "learning_rate": 4.111740473738414e-07, | |
| "logits/chosen": -3.2562499046325684, | |
| "logits/rejected": -3.1781249046325684, | |
| "logps/chosen": -306.79998779296875, | |
| "logps/rejected": -326.3999938964844, | |
| "loss": 0.6789, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.11583252251148224, | |
| "rewards/margins": 0.09737548977136612, | |
| "rewards/rejected": 0.01840820349752903, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.17893923789907312, | |
| "grad_norm": 290.0, | |
| "learning_rate": 4.1053038105046343e-07, | |
| "logits/chosen": -3.231250047683716, | |
| "logits/rejected": -3.296875, | |
| "logps/chosen": -340.3999938964844, | |
| "logps/rejected": -301.20001220703125, | |
| "loss": 0.6133, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.17045898735523224, | |
| "rewards/margins": 0.25146484375, | |
| "rewards/rejected": -0.08159179985523224, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.18022657054582905, | |
| "grad_norm": 165.0, | |
| "learning_rate": 4.098867147270855e-07, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -3.2906250953674316, | |
| "logps/chosen": -289.0, | |
| "logps/rejected": -241.60000610351562, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.184814453125, | |
| "rewards/margins": 0.19721679389476776, | |
| "rewards/rejected": -0.012500000186264515, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.18151390319258495, | |
| "grad_norm": 198.0, | |
| "learning_rate": 4.0924304840370747e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.28125, | |
| "logps/chosen": -302.79998779296875, | |
| "logps/rejected": -291.6000061035156, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.10765381157398224, | |
| "rewards/margins": 0.23212890326976776, | |
| "rewards/rejected": -0.12421874701976776, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.18280123583934088, | |
| "grad_norm": 241.0, | |
| "learning_rate": 4.085993820803295e-07, | |
| "logits/chosen": -3.2437500953674316, | |
| "logits/rejected": -3.21875, | |
| "logps/chosen": -325.0, | |
| "logps/rejected": -301.3999938964844, | |
| "loss": 0.6078, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.23613281548023224, | |
| "rewards/margins": 0.22871093451976776, | |
| "rewards/rejected": 0.0077880858443677425, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.1840885684860968, | |
| "grad_norm": 198.0, | |
| "learning_rate": 4.079557157569516e-07, | |
| "logits/chosen": -2.871875047683716, | |
| "logits/rejected": -3.1187500953674316, | |
| "logps/chosen": -237.02499389648438, | |
| "logps/rejected": -221.10000610351562, | |
| "loss": 0.6695, | |
| "rewards/accuracies": 0.3375000059604645, | |
| "rewards/chosen": 0.15878906846046448, | |
| "rewards/margins": 0.10507812350988388, | |
| "rewards/rejected": 0.05382690578699112, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.18537590113285274, | |
| "grad_norm": 196.0, | |
| "learning_rate": 4.073120494335736e-07, | |
| "logits/chosen": -3.3531250953674316, | |
| "logits/rejected": -3.3343749046325684, | |
| "logps/chosen": -289.20001220703125, | |
| "logps/rejected": -235.0, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.12890625, | |
| "rewards/margins": 0.19033202528953552, | |
| "rewards/rejected": -0.06142578274011612, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.18666323377960864, | |
| "grad_norm": 181.0, | |
| "learning_rate": 4.0666838311019564e-07, | |
| "logits/chosen": -3.0062499046325684, | |
| "logits/rejected": -2.971874952316284, | |
| "logps/chosen": -288.6000061035156, | |
| "logps/rejected": -233.60000610351562, | |
| "loss": 0.6141, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.15761718153953552, | |
| "rewards/margins": 0.225830078125, | |
| "rewards/rejected": -0.06826172024011612, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.18795056642636457, | |
| "grad_norm": 314.0, | |
| "learning_rate": 4.0602471678681773e-07, | |
| "logits/chosen": -3.106250047683716, | |
| "logits/rejected": -3.015625, | |
| "logps/chosen": -275.6000061035156, | |
| "logps/rejected": -276.0, | |
| "loss": 0.7109, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": -0.01621093787252903, | |
| "rewards/margins": 0.052734375, | |
| "rewards/rejected": -0.06875000149011612, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.1892378990731205, | |
| "grad_norm": 217.0, | |
| "learning_rate": 4.053810504634397e-07, | |
| "logits/chosen": -3.0999999046325684, | |
| "logits/rejected": -3.0999999046325684, | |
| "logps/chosen": -321.20001220703125, | |
| "logps/rejected": -321.0, | |
| "loss": 0.632, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.20034179091453552, | |
| "rewards/margins": 0.17988280951976776, | |
| "rewards/rejected": 0.02063903771340847, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.19052523171987643, | |
| "grad_norm": 239.0, | |
| "learning_rate": 4.0473738414006176e-07, | |
| "logits/chosen": -3.109375, | |
| "logits/rejected": -3.1343750953674316, | |
| "logps/chosen": -238.0, | |
| "logps/rejected": -232.60000610351562, | |
| "loss": 0.6422, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.14543457329273224, | |
| "rewards/margins": 0.14316406846046448, | |
| "rewards/rejected": 0.0024902343284338713, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.19181256436663233, | |
| "grad_norm": 172.0, | |
| "learning_rate": 4.040937178166838e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.299999952316284, | |
| "logps/chosen": -314.6000061035156, | |
| "logps/rejected": -236.6999969482422, | |
| "loss": 0.5945, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.22003173828125, | |
| "rewards/margins": 0.26972657442092896, | |
| "rewards/rejected": -0.04985351487994194, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.19309989701338826, | |
| "grad_norm": 256.0, | |
| "learning_rate": 4.0345005149330585e-07, | |
| "logits/chosen": -3.003124952316284, | |
| "logits/rejected": -3.143749952316284, | |
| "logps/chosen": -283.6000061035156, | |
| "logps/rejected": -249.8000030517578, | |
| "loss": 0.675, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.12601928412914276, | |
| "rewards/margins": 0.08976440131664276, | |
| "rewards/rejected": 0.0361328125, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.19438722966014418, | |
| "grad_norm": 181.0, | |
| "learning_rate": 4.028063851699279e-07, | |
| "logits/chosen": -3.081249952316284, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -314.20001220703125, | |
| "logps/rejected": -282.0, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.22446289658546448, | |
| "rewards/margins": 0.22519531846046448, | |
| "rewards/rejected": -0.0007080078357830644, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.1956745623069001, | |
| "grad_norm": 225.0, | |
| "learning_rate": 4.0216271884654994e-07, | |
| "logits/chosen": -3.1500000953674316, | |
| "logits/rejected": -3.25, | |
| "logps/chosen": -265.20001220703125, | |
| "logps/rejected": -269.79998779296875, | |
| "loss": 0.6609, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.06442870944738388, | |
| "rewards/margins": 0.12302245944738388, | |
| "rewards/rejected": -0.05839843675494194, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.196961894953656, | |
| "grad_norm": 202.0, | |
| "learning_rate": 4.01519052523172e-07, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -247.0, | |
| "logps/rejected": -193.39999389648438, | |
| "loss": 0.6375, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.16249999403953552, | |
| "rewards/margins": 0.15114745497703552, | |
| "rewards/rejected": 0.01123046875, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.19824922760041194, | |
| "grad_norm": 209.0, | |
| "learning_rate": 4.00875386199794e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.268749952316284, | |
| "logps/chosen": -280.79998779296875, | |
| "logps/rejected": -238.0, | |
| "loss": 0.6195, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.19140625, | |
| "rewards/margins": 0.22636719048023224, | |
| "rewards/rejected": -0.03493041917681694, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.19953656024716787, | |
| "grad_norm": 169.0, | |
| "learning_rate": 4.0023171987641606e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.293750047683716, | |
| "logps/chosen": -278.20001220703125, | |
| "logps/rejected": -253.60000610351562, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.12055053561925888, | |
| "rewards/margins": 0.20175781846046448, | |
| "rewards/rejected": -0.08103027194738388, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.2008238928939238, | |
| "grad_norm": 188.0, | |
| "learning_rate": 3.995880535530381e-07, | |
| "logits/chosen": -3.0218749046325684, | |
| "logits/rejected": -3.340625047683716, | |
| "logps/chosen": -260.5, | |
| "logps/rejected": -247.6999969482422, | |
| "loss": 0.6195, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.18279418349266052, | |
| "rewards/margins": 0.18925781548023224, | |
| "rewards/rejected": -0.0063720704056322575, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.2021112255406797, | |
| "grad_norm": 190.0, | |
| "learning_rate": 3.989443872296601e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.2593750953674316, | |
| "logps/chosen": -243.8000030517578, | |
| "logps/rejected": -214.1999969482422, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.09213867038488388, | |
| "rewards/margins": 0.19478149712085724, | |
| "rewards/rejected": -0.10322265326976776, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.20339855818743563, | |
| "grad_norm": 171.0, | |
| "learning_rate": 3.9830072090628214e-07, | |
| "logits/chosen": -2.934375047683716, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -277.3999938964844, | |
| "logps/rejected": -225.1999969482422, | |
| "loss": 0.625, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.09014892578125, | |
| "rewards/margins": 0.21425780653953552, | |
| "rewards/rejected": -0.124267578125, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.20468589083419156, | |
| "grad_norm": 213.0, | |
| "learning_rate": 3.9765705458290423e-07, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.25, | |
| "logps/chosen": -290.6000061035156, | |
| "logps/rejected": -256.79998779296875, | |
| "loss": 0.6375, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.24746093153953552, | |
| "rewards/margins": 0.16259765625, | |
| "rewards/rejected": 0.08481445163488388, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.2059732234809475, | |
| "grad_norm": 175.0, | |
| "learning_rate": 3.970133882595262e-07, | |
| "logits/chosen": -3.2562499046325684, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -296.79998779296875, | |
| "logps/rejected": -274.6000061035156, | |
| "loss": 0.6359, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.06730346381664276, | |
| "rewards/margins": 0.16831055283546448, | |
| "rewards/rejected": -0.10146484524011612, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.2072605561277034, | |
| "grad_norm": 221.0, | |
| "learning_rate": 3.9636972193614827e-07, | |
| "logits/chosen": -3.2593750953674316, | |
| "logits/rejected": -3.262500047683716, | |
| "logps/chosen": -305.6000061035156, | |
| "logps/rejected": -262.3999938964844, | |
| "loss": 0.6422, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.19125977158546448, | |
| "rewards/margins": 0.17949219048023224, | |
| "rewards/rejected": 0.011387634091079235, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.20854788877445932, | |
| "grad_norm": 228.0, | |
| "learning_rate": 3.9572605561277036e-07, | |
| "logits/chosen": -3.200000047683716, | |
| "logits/rejected": -3.0718750953674316, | |
| "logps/chosen": -277.79998779296875, | |
| "logps/rejected": -270.6000061035156, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.1689453125, | |
| "rewards/margins": 0.18173828721046448, | |
| "rewards/rejected": -0.0126953125, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.20983522142121525, | |
| "grad_norm": 185.0, | |
| "learning_rate": 3.9508238928939235e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.003124952316284, | |
| "logps/chosen": -254.1999969482422, | |
| "logps/rejected": -239.60000610351562, | |
| "loss": 0.6191, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.19316406548023224, | |
| "rewards/margins": 0.19609375298023224, | |
| "rewards/rejected": -0.0029785155784338713, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.21112255406797117, | |
| "grad_norm": 183.0, | |
| "learning_rate": 3.944387229660144e-07, | |
| "logits/chosen": -3.2562499046325684, | |
| "logits/rejected": -3.2874999046325684, | |
| "logps/chosen": -237.39999389648438, | |
| "logps/rejected": -181.5, | |
| "loss": 0.602, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.17156982421875, | |
| "rewards/margins": 0.2676635682582855, | |
| "rewards/rejected": -0.09595946967601776, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.21240988671472708, | |
| "grad_norm": 270.0, | |
| "learning_rate": 3.9379505664263644e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.331249952316284, | |
| "logps/chosen": -333.20001220703125, | |
| "logps/rejected": -295.6000061035156, | |
| "loss": 0.6523, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.15825195610523224, | |
| "rewards/margins": 0.14663085341453552, | |
| "rewards/rejected": 0.01181640662252903, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.213697219361483, | |
| "grad_norm": 194.0, | |
| "learning_rate": 3.931513903192585e-07, | |
| "logits/chosen": -3.2125000953674316, | |
| "logits/rejected": -3.137500047683716, | |
| "logps/chosen": -297.6000061035156, | |
| "logps/rejected": -259.0, | |
| "loss": 0.6125, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.24428710341453552, | |
| "rewards/margins": 0.21621093153953552, | |
| "rewards/rejected": 0.02789306640625, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.21498455200823893, | |
| "grad_norm": 176.0, | |
| "learning_rate": 3.925077239958805e-07, | |
| "logits/chosen": -3.2093749046325684, | |
| "logits/rejected": -3.293750047683716, | |
| "logps/chosen": -268.0, | |
| "logps/rejected": -235.60000610351562, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.1953125, | |
| "rewards/margins": 0.23079833388328552, | |
| "rewards/rejected": -0.03544921800494194, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.21627188465499486, | |
| "grad_norm": 192.0, | |
| "learning_rate": 3.9186405767250257e-07, | |
| "logits/chosen": -3.03125, | |
| "logits/rejected": -3.003124952316284, | |
| "logps/chosen": -290.3999938964844, | |
| "logps/rejected": -214.10000610351562, | |
| "loss": 0.6578, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.10993652045726776, | |
| "rewards/margins": 0.12214355170726776, | |
| "rewards/rejected": -0.01230468787252903, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.21755921730175076, | |
| "grad_norm": 198.0, | |
| "learning_rate": 3.912203913491246e-07, | |
| "logits/chosen": -3.268749952316284, | |
| "logits/rejected": -3.2874999046325684, | |
| "logps/chosen": -251.1999969482422, | |
| "logps/rejected": -231.8000030517578, | |
| "loss": 0.6086, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.02813720703125, | |
| "rewards/margins": 0.22578124701976776, | |
| "rewards/rejected": -0.25312501192092896, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.2188465499485067, | |
| "grad_norm": 163.0, | |
| "learning_rate": 3.905767250257466e-07, | |
| "logits/chosen": -3.012500047683716, | |
| "logits/rejected": -3.168750047683716, | |
| "logps/chosen": -257.79998779296875, | |
| "logps/rejected": -228.0, | |
| "loss": 0.6266, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.2197265625, | |
| "rewards/margins": 0.17919921875, | |
| "rewards/rejected": 0.04022369533777237, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.22013388259526262, | |
| "grad_norm": 237.0, | |
| "learning_rate": 3.899330587023687e-07, | |
| "logits/chosen": -3.081249952316284, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -293.0, | |
| "logps/rejected": -241.1999969482422, | |
| "loss": 0.6078, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.30029296875, | |
| "rewards/margins": 0.23749999701976776, | |
| "rewards/rejected": 0.06264648586511612, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.22142121524201855, | |
| "grad_norm": 185.0, | |
| "learning_rate": 3.8928939237899074e-07, | |
| "logits/chosen": -3.140625, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -315.3999938964844, | |
| "logps/rejected": -248.60000610351562, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.20004883408546448, | |
| "rewards/margins": 0.21406249701976776, | |
| "rewards/rejected": -0.0142059326171875, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.22270854788877445, | |
| "grad_norm": 218.0, | |
| "learning_rate": 3.886457260556127e-07, | |
| "logits/chosen": -3.096874952316284, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -234.39999389648438, | |
| "logps/rejected": -194.8000030517578, | |
| "loss": 0.6203, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.16054686903953552, | |
| "rewards/margins": 0.19511719048023224, | |
| "rewards/rejected": -0.03474121168255806, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.22399588053553038, | |
| "grad_norm": 232.0, | |
| "learning_rate": 3.8800205973223477e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.128124952316284, | |
| "logps/chosen": -259.0, | |
| "logps/rejected": -314.6000061035156, | |
| "loss": 0.6625, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.13081054389476776, | |
| "rewards/margins": 0.1162109375, | |
| "rewards/rejected": 0.014599609188735485, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.2252832131822863, | |
| "grad_norm": 191.0, | |
| "learning_rate": 3.8735839340885686e-07, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.237499952316284, | |
| "logps/chosen": -267.6000061035156, | |
| "logps/rejected": -274.0, | |
| "loss": 0.5742, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.11304321140050888, | |
| "rewards/margins": 0.3338867127895355, | |
| "rewards/rejected": -0.22050781548023224, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.22657054582904224, | |
| "grad_norm": 181.0, | |
| "learning_rate": 3.8671472708547885e-07, | |
| "logits/chosen": -3.278125047683716, | |
| "logits/rejected": -3.2406249046325684, | |
| "logps/chosen": -263.8999938964844, | |
| "logps/rejected": -197.1999969482422, | |
| "loss": 0.6406, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.18015137314796448, | |
| "rewards/margins": 0.16845703125, | |
| "rewards/rejected": 0.011767578311264515, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.22785787847579814, | |
| "grad_norm": 240.0, | |
| "learning_rate": 3.860710607621009e-07, | |
| "logits/chosen": -3.078125, | |
| "logits/rejected": -3.0718750953674316, | |
| "logps/chosen": -317.6000061035156, | |
| "logps/rejected": -290.79998779296875, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.15424804389476776, | |
| "rewards/margins": 0.1982421875, | |
| "rewards/rejected": -0.04426269605755806, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.22914521112255407, | |
| "grad_norm": 193.0, | |
| "learning_rate": 3.8542739443872294e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.231250047683716, | |
| "logps/chosen": -290.20001220703125, | |
| "logps/rejected": -237.6999969482422, | |
| "loss": 0.618, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.093963623046875, | |
| "rewards/margins": 0.21689453721046448, | |
| "rewards/rejected": -0.122802734375, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.23043254376931, | |
| "grad_norm": 216.0, | |
| "learning_rate": 3.84783728115345e-07, | |
| "logits/chosen": -3.075000047683716, | |
| "logits/rejected": -3.0843749046325684, | |
| "logps/chosen": -236.39999389648438, | |
| "logps/rejected": -223.39999389648438, | |
| "loss": 0.6336, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.09038696438074112, | |
| "rewards/margins": 0.1766357421875, | |
| "rewards/rejected": -0.08623047173023224, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.23171987641606592, | |
| "grad_norm": 174.0, | |
| "learning_rate": 3.84140061791967e-07, | |
| "logits/chosen": -3.2406249046325684, | |
| "logits/rejected": -3.1968750953674316, | |
| "logps/chosen": -308.79998779296875, | |
| "logps/rejected": -234.39999389648438, | |
| "loss": 0.5992, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.12282714992761612, | |
| "rewards/margins": 0.271484375, | |
| "rewards/rejected": -0.14833983778953552, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.23300720906282182, | |
| "grad_norm": 142.0, | |
| "learning_rate": 3.8349639546858907e-07, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.3062500953674316, | |
| "logps/chosen": -295.79998779296875, | |
| "logps/rejected": -219.6999969482422, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.02926025353372097, | |
| "rewards/margins": 0.19501952826976776, | |
| "rewards/rejected": -0.16595458984375, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.23429454170957775, | |
| "grad_norm": 312.0, | |
| "learning_rate": 3.828527291452111e-07, | |
| "logits/chosen": -3.075000047683716, | |
| "logits/rejected": -3.143749952316284, | |
| "logps/chosen": -309.3999938964844, | |
| "logps/rejected": -304.3999938964844, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.29960936307907104, | |
| "rewards/margins": 0.21738281846046448, | |
| "rewards/rejected": 0.0821533203125, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.23558187435633368, | |
| "grad_norm": 177.0, | |
| "learning_rate": 3.8220906282183315e-07, | |
| "logits/chosen": -2.96875, | |
| "logits/rejected": -2.890625, | |
| "logps/chosen": -217.10000610351562, | |
| "logps/rejected": -261.79998779296875, | |
| "loss": 0.6273, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.13188476860523224, | |
| "rewards/margins": 0.15737304091453552, | |
| "rewards/rejected": -0.025634765625, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.2368692070030896, | |
| "grad_norm": 222.0, | |
| "learning_rate": 3.815653964984552e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.246875047683716, | |
| "logps/chosen": -200.39999389648438, | |
| "logps/rejected": -157.3000030517578, | |
| "loss": 0.6891, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 0.03823242336511612, | |
| "rewards/margins": 0.04287109524011612, | |
| "rewards/rejected": -0.0047851563431322575, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.2381565396498455, | |
| "grad_norm": 198.0, | |
| "learning_rate": 3.8092173017507724e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -248.1999969482422, | |
| "logps/rejected": -239.60000610351562, | |
| "loss": 0.6586, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.179443359375, | |
| "rewards/margins": 0.1416015625, | |
| "rewards/rejected": 0.03759765625, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.23944387229660144, | |
| "grad_norm": 204.0, | |
| "learning_rate": 3.8027806385169923e-07, | |
| "logits/chosen": -3.253124952316284, | |
| "logits/rejected": -3.3531250953674316, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -226.3000030517578, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.07831420749425888, | |
| "rewards/margins": 0.18691405653953552, | |
| "rewards/rejected": -0.10865478217601776, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.24073120494335737, | |
| "grad_norm": 292.0, | |
| "learning_rate": 3.796343975283213e-07, | |
| "logits/chosen": -3.140625, | |
| "logits/rejected": -3.1656250953674316, | |
| "logps/chosen": -252.0, | |
| "logps/rejected": -246.5, | |
| "loss": 0.6469, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.11220703274011612, | |
| "rewards/margins": 0.16547851264476776, | |
| "rewards/rejected": -0.05308837816119194, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.2420185375901133, | |
| "grad_norm": 202.0, | |
| "learning_rate": 3.7899073120494337e-07, | |
| "logits/chosen": -2.9781250953674316, | |
| "logits/rejected": -3.0062499046325684, | |
| "logps/chosen": -234.1999969482422, | |
| "logps/rejected": -233.8000030517578, | |
| "loss": 0.6438, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.16777344048023224, | |
| "rewards/margins": 0.160400390625, | |
| "rewards/rejected": 0.0077880858443677425, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.2433058702368692, | |
| "grad_norm": 168.0, | |
| "learning_rate": 3.7834706488156536e-07, | |
| "logits/chosen": -3.109375, | |
| "logits/rejected": -3.143749952316284, | |
| "logps/chosen": -265.20001220703125, | |
| "logps/rejected": -219.10000610351562, | |
| "loss": 0.6391, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.2583984434604645, | |
| "rewards/margins": 0.16342774033546448, | |
| "rewards/rejected": 0.09492187201976776, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.24459320288362513, | |
| "grad_norm": 243.0, | |
| "learning_rate": 3.777033985581874e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.0843749046325684, | |
| "logps/chosen": -231.0, | |
| "logps/rejected": -232.60000610351562, | |
| "loss": 0.6609, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.07631836086511612, | |
| "rewards/margins": 0.09954833984375, | |
| "rewards/rejected": -0.02333984337747097, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.24588053553038106, | |
| "grad_norm": 198.0, | |
| "learning_rate": 3.770597322348095e-07, | |
| "logits/chosen": -3.109375, | |
| "logits/rejected": -3.0687499046325684, | |
| "logps/chosen": -258.20001220703125, | |
| "logps/rejected": -241.60000610351562, | |
| "loss": 0.657, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.04843749850988388, | |
| "rewards/margins": 0.14091797173023224, | |
| "rewards/rejected": -0.09218750149011612, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.24716786817713698, | |
| "grad_norm": 264.0, | |
| "learning_rate": 3.764160659114315e-07, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.1624999046325684, | |
| "logps/chosen": -356.0, | |
| "logps/rejected": -296.20001220703125, | |
| "loss": 0.5797, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.21113280951976776, | |
| "rewards/margins": 0.3363281190395355, | |
| "rewards/rejected": -0.12607422471046448, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.24845520082389289, | |
| "grad_norm": 162.0, | |
| "learning_rate": 3.7577239958805353e-07, | |
| "logits/chosen": -3.03125, | |
| "logits/rejected": -3.081249952316284, | |
| "logps/chosen": -250.39999389648438, | |
| "logps/rejected": -202.60000610351562, | |
| "loss": 0.6086, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.20551757514476776, | |
| "rewards/margins": 0.25019532442092896, | |
| "rewards/rejected": -0.04458007961511612, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.24974253347064881, | |
| "grad_norm": 163.0, | |
| "learning_rate": 3.7512873326467557e-07, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.375, | |
| "logps/chosen": -326.3999938964844, | |
| "logps/rejected": -286.79998779296875, | |
| "loss": 0.6156, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.16152343153953552, | |
| "rewards/margins": 0.265625, | |
| "rewards/rejected": -0.10434570163488388, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.25102986611740474, | |
| "grad_norm": 179.0, | |
| "learning_rate": 3.744850669412976e-07, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.268749952316284, | |
| "logps/chosen": -317.20001220703125, | |
| "logps/rejected": -242.8000030517578, | |
| "loss": 0.6078, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.22177734971046448, | |
| "rewards/margins": 0.23251953721046448, | |
| "rewards/rejected": -0.01083984412252903, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.25231719876416064, | |
| "grad_norm": 256.0, | |
| "learning_rate": 3.7384140061791965e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.284374952316284, | |
| "logps/chosen": -307.3999938964844, | |
| "logps/rejected": -255.39999389648438, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.22988280653953552, | |
| "rewards/margins": 0.2294921875, | |
| "rewards/rejected": 0.0005371093866415322, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.2536045314109166, | |
| "grad_norm": 270.0, | |
| "learning_rate": 3.731977342945417e-07, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.2750000953674316, | |
| "logps/chosen": -308.0, | |
| "logps/rejected": -279.6000061035156, | |
| "loss": 0.6703, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.05672607570886612, | |
| "rewards/margins": 0.12509766221046448, | |
| "rewards/rejected": -0.06842041015625, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.2548918640576725, | |
| "grad_norm": 188.0, | |
| "learning_rate": 3.7255406797116374e-07, | |
| "logits/chosen": -3.200000047683716, | |
| "logits/rejected": -3.075000047683716, | |
| "logps/chosen": -308.20001220703125, | |
| "logps/rejected": -271.20001220703125, | |
| "loss": 0.5648, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.17021484673023224, | |
| "rewards/margins": 0.3773437440395355, | |
| "rewards/rejected": -0.20644530653953552, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.2561791967044284, | |
| "grad_norm": 213.0, | |
| "learning_rate": 3.7191040164778573e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.265625, | |
| "logps/chosen": -314.20001220703125, | |
| "logps/rejected": -252.60000610351562, | |
| "loss": 0.6281, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.06357421725988388, | |
| "rewards/margins": 0.1787109375, | |
| "rewards/rejected": -0.11472167819738388, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.25746652935118436, | |
| "grad_norm": 171.0, | |
| "learning_rate": 3.712667353244078e-07, | |
| "logits/chosen": -3.2406249046325684, | |
| "logits/rejected": -3.309375047683716, | |
| "logps/chosen": -260.6000061035156, | |
| "logps/rejected": -252.8000030517578, | |
| "loss": 0.6266, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.12783202528953552, | |
| "rewards/margins": 0.21494141221046448, | |
| "rewards/rejected": -0.0870361328125, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.25875386199794026, | |
| "grad_norm": 133.0, | |
| "learning_rate": 3.7062306900102987e-07, | |
| "logits/chosen": -3.2125000953674316, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -278.0, | |
| "logps/rejected": -234.3000030517578, | |
| "loss": 0.6727, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.11420898139476776, | |
| "rewards/margins": 0.1014404296875, | |
| "rewards/rejected": 0.012707519344985485, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.2600411946446962, | |
| "grad_norm": 198.0, | |
| "learning_rate": 3.6997940267765186e-07, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.0875000953674316, | |
| "logps/chosen": -343.20001220703125, | |
| "logps/rejected": -271.79998779296875, | |
| "loss": 0.5727, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.3257812559604645, | |
| "rewards/margins": 0.34765625, | |
| "rewards/rejected": -0.02120361290872097, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.2613285272914521, | |
| "grad_norm": 193.0, | |
| "learning_rate": 3.6933573635427395e-07, | |
| "logits/chosen": -3.0875000953674316, | |
| "logits/rejected": -3.106250047683716, | |
| "logps/chosen": -330.3999938964844, | |
| "logps/rejected": -262.20001220703125, | |
| "loss": 0.607, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.24892577528953552, | |
| "rewards/margins": 0.22353515028953552, | |
| "rewards/rejected": 0.02529296837747097, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.262615859938208, | |
| "grad_norm": 318.0, | |
| "learning_rate": 3.68692070030896e-07, | |
| "logits/chosen": -3.1812500953674316, | |
| "logits/rejected": -3.1781249046325684, | |
| "logps/chosen": -297.20001220703125, | |
| "logps/rejected": -300.6000061035156, | |
| "loss": 0.6344, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.16109618544578552, | |
| "rewards/margins": 0.22011718153953552, | |
| "rewards/rejected": -0.0595703125, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.263903192584964, | |
| "grad_norm": 187.0, | |
| "learning_rate": 3.68048403707518e-07, | |
| "logits/chosen": -3.078125, | |
| "logits/rejected": -3.0843749046325684, | |
| "logps/chosen": -249.1999969482422, | |
| "logps/rejected": -192.60000610351562, | |
| "loss": 0.6266, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.15595702826976776, | |
| "rewards/margins": 0.17451171576976776, | |
| "rewards/rejected": -0.01845703087747097, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.2651905252317199, | |
| "grad_norm": 216.0, | |
| "learning_rate": 3.6740473738414003e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.309375047683716, | |
| "logps/chosen": -284.1000061035156, | |
| "logps/rejected": -241.60000610351562, | |
| "loss": 0.618, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.22578124701976776, | |
| "rewards/margins": 0.20839843153953552, | |
| "rewards/rejected": 0.01716308668255806, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.2664778578784758, | |
| "grad_norm": 178.0, | |
| "learning_rate": 3.6676107106076207e-07, | |
| "logits/chosen": -3.096874952316284, | |
| "logits/rejected": -2.9937500953674316, | |
| "logps/chosen": -216.0, | |
| "logps/rejected": -207.8000030517578, | |
| "loss": 0.6484, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.16337890923023224, | |
| "rewards/margins": 0.13730469346046448, | |
| "rewards/rejected": 0.02592163160443306, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.26776519052523173, | |
| "grad_norm": 312.0, | |
| "learning_rate": 3.661174047373841e-07, | |
| "logits/chosen": -3.268749952316284, | |
| "logits/rejected": -3.21875, | |
| "logps/chosen": -285.6000061035156, | |
| "logps/rejected": -271.3999938964844, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.19863280653953552, | |
| "rewards/margins": 0.24919433891773224, | |
| "rewards/rejected": -0.05024414137005806, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.26905252317198763, | |
| "grad_norm": 314.0, | |
| "learning_rate": 3.6547373841400616e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.109375, | |
| "logps/chosen": -344.20001220703125, | |
| "logps/rejected": -270.20001220703125, | |
| "loss": 0.6367, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.19362792372703552, | |
| "rewards/margins": 0.17333984375, | |
| "rewards/rejected": 0.02009277418255806, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.2703398558187436, | |
| "grad_norm": 187.0, | |
| "learning_rate": 3.648300720906282e-07, | |
| "logits/chosen": -3.206249952316284, | |
| "logits/rejected": -3.168750047683716, | |
| "logps/chosen": -243.1999969482422, | |
| "logps/rejected": -225.1999969482422, | |
| "loss": 0.6531, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.09736327826976776, | |
| "rewards/margins": 0.12753906846046448, | |
| "rewards/rejected": -0.030517578125, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.2716271884654995, | |
| "grad_norm": 240.0, | |
| "learning_rate": 3.6418640576725024e-07, | |
| "logits/chosen": -3.2093749046325684, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -263.0, | |
| "logps/rejected": -284.3999938964844, | |
| "loss": 0.6625, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.18710938096046448, | |
| "rewards/margins": 0.12294922024011612, | |
| "rewards/rejected": 0.06440429389476776, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.2729145211122554, | |
| "grad_norm": 190.0, | |
| "learning_rate": 3.635427394438723e-07, | |
| "logits/chosen": -3.3187499046325684, | |
| "logits/rejected": -3.284374952316284, | |
| "logps/chosen": -240.8000030517578, | |
| "logps/rejected": -231.0, | |
| "loss": 0.65, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.13701172173023224, | |
| "rewards/margins": 0.15380859375, | |
| "rewards/rejected": -0.01640624925494194, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.27420185375901135, | |
| "grad_norm": 241.0, | |
| "learning_rate": 3.6289907312049433e-07, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.1156249046325684, | |
| "logps/chosen": -298.0, | |
| "logps/rejected": -281.20001220703125, | |
| "loss": 0.6492, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.15937499701976776, | |
| "rewards/margins": 0.15151366591453552, | |
| "rewards/rejected": 0.0077880858443677425, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.27548918640576725, | |
| "grad_norm": 214.0, | |
| "learning_rate": 3.6225540679711637e-07, | |
| "logits/chosen": -3.231250047683716, | |
| "logits/rejected": -3.03125, | |
| "logps/chosen": -197.3000030517578, | |
| "logps/rejected": -184.5, | |
| "loss": 0.6594, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.05141601711511612, | |
| "rewards/margins": 0.12026367336511612, | |
| "rewards/rejected": -0.0689697265625, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.27677651905252315, | |
| "grad_norm": 260.0, | |
| "learning_rate": 3.6161174047373836e-07, | |
| "logits/chosen": -3.278125047683716, | |
| "logits/rejected": -3.2093749046325684, | |
| "logps/chosen": -331.20001220703125, | |
| "logps/rejected": -293.79998779296875, | |
| "loss": 0.6555, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.06044922024011612, | |
| "rewards/margins": 0.13120117783546448, | |
| "rewards/rejected": -0.07094726711511612, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.2780638516992791, | |
| "grad_norm": 196.0, | |
| "learning_rate": 3.6096807415036046e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.2281250953674316, | |
| "logps/chosen": -257.6000061035156, | |
| "logps/rejected": -235.1999969482422, | |
| "loss": 0.6023, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.16972656548023224, | |
| "rewards/margins": 0.2525390684604645, | |
| "rewards/rejected": -0.08284912258386612, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.279351184346035, | |
| "grad_norm": 193.0, | |
| "learning_rate": 3.603244078269825e-07, | |
| "logits/chosen": -3.265625, | |
| "logits/rejected": -3.299999952316284, | |
| "logps/chosen": -364.0, | |
| "logps/rejected": -258.0, | |
| "loss": 0.5914, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.23099975287914276, | |
| "rewards/margins": 0.28789061307907104, | |
| "rewards/rejected": -0.05683593824505806, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.2806385169927909, | |
| "grad_norm": 195.0, | |
| "learning_rate": 3.596807415036045e-07, | |
| "logits/chosen": -3.0062499046325684, | |
| "logits/rejected": -3.046875, | |
| "logps/chosen": -239.89999389648438, | |
| "logps/rejected": -215.1999969482422, | |
| "loss": 0.643, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.20517578721046448, | |
| "rewards/margins": 0.14671631157398224, | |
| "rewards/rejected": 0.05817871168255806, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.28192584963954687, | |
| "grad_norm": 540.0, | |
| "learning_rate": 3.590370751802266e-07, | |
| "logits/chosen": -3.09375, | |
| "logits/rejected": -3.1031250953674316, | |
| "logps/chosen": -259.20001220703125, | |
| "logps/rejected": -230.3000030517578, | |
| "loss": 0.6969, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.090484619140625, | |
| "rewards/margins": 0.0458984375, | |
| "rewards/rejected": 0.04462890699505806, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.28321318228630277, | |
| "grad_norm": 199.0, | |
| "learning_rate": 3.583934088568486e-07, | |
| "logits/chosen": -3.1812500953674316, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -311.3999938964844, | |
| "logps/rejected": -260.20001220703125, | |
| "loss": 0.6102, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.184326171875, | |
| "rewards/margins": 0.23525390028953552, | |
| "rewards/rejected": -0.05076904222369194, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.2845005149330587, | |
| "grad_norm": 193.0, | |
| "learning_rate": 3.577497425334706e-07, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -3.09375, | |
| "logps/chosen": -248.60000610351562, | |
| "logps/rejected": -228.0, | |
| "loss": 0.6516, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.12456054985523224, | |
| "rewards/margins": 0.13632813096046448, | |
| "rewards/rejected": -0.01198730431497097, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.2857878475798146, | |
| "grad_norm": 240.0, | |
| "learning_rate": 3.571060762100927e-07, | |
| "logits/chosen": -3.2249999046325684, | |
| "logits/rejected": -3.3187499046325684, | |
| "logps/chosen": -279.0, | |
| "logps/rejected": -208.39999389648438, | |
| "loss": 0.6555, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.065185546875, | |
| "rewards/margins": 0.13657227158546448, | |
| "rewards/rejected": -0.071746826171875, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.2870751802265705, | |
| "grad_norm": 704.0, | |
| "learning_rate": 3.564624098867147e-07, | |
| "logits/chosen": -2.8968749046325684, | |
| "logits/rejected": -2.934375047683716, | |
| "logps/chosen": -246.8000030517578, | |
| "logps/rejected": -231.39999389648438, | |
| "loss": 0.6422, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.10878906399011612, | |
| "rewards/margins": 0.14351806044578552, | |
| "rewards/rejected": -0.03432617336511612, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.2883625128733265, | |
| "grad_norm": 203.0, | |
| "learning_rate": 3.5581874356333674e-07, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -239.60000610351562, | |
| "logps/rejected": -231.8000030517578, | |
| "loss": 0.5992, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.16972656548023224, | |
| "rewards/margins": 0.2578125, | |
| "rewards/rejected": -0.08847351372241974, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.2896498455200824, | |
| "grad_norm": 219.0, | |
| "learning_rate": 3.551750772399588e-07, | |
| "logits/chosen": -3.112499952316284, | |
| "logits/rejected": -3.171875, | |
| "logps/chosen": -279.79998779296875, | |
| "logps/rejected": -235.10000610351562, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.21406249701976776, | |
| "rewards/margins": 0.2740234434604645, | |
| "rewards/rejected": -0.06010742112994194, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.2909371781668383, | |
| "grad_norm": 219.0, | |
| "learning_rate": 3.5453141091658083e-07, | |
| "logits/chosen": -3.140625, | |
| "logits/rejected": -3.293750047683716, | |
| "logps/chosen": -259.0, | |
| "logps/rejected": -207.5500030517578, | |
| "loss": 0.6742, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.00341796875, | |
| "rewards/margins": 0.10895995795726776, | |
| "rewards/rejected": -0.10537109524011612, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.29222451081359424, | |
| "grad_norm": 186.0, | |
| "learning_rate": 3.5388774459320287e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.0687499046325684, | |
| "logps/chosen": -263.6000061035156, | |
| "logps/rejected": -275.20001220703125, | |
| "loss": 0.6117, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.22011718153953552, | |
| "rewards/margins": 0.23652343451976776, | |
| "rewards/rejected": -0.01643066480755806, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.29351184346035014, | |
| "grad_norm": 198.0, | |
| "learning_rate": 3.5324407826982486e-07, | |
| "logits/chosen": -3.0843749046325684, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -267.3999938964844, | |
| "logps/rejected": -211.8000030517578, | |
| "loss": 0.6094, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.198974609375, | |
| "rewards/margins": 0.2707763612270355, | |
| "rewards/rejected": -0.07192382961511612, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.2947991761071061, | |
| "grad_norm": 280.0, | |
| "learning_rate": 3.5260041194644696e-07, | |
| "logits/chosen": -3.1968750953674316, | |
| "logits/rejected": -3.128124952316284, | |
| "logps/chosen": -325.3999938964844, | |
| "logps/rejected": -296.6000061035156, | |
| "loss": 0.6227, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.216796875, | |
| "rewards/margins": 0.23458251357078552, | |
| "rewards/rejected": -0.01796874962747097, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.296086508753862, | |
| "grad_norm": 184.0, | |
| "learning_rate": 3.51956745623069e-07, | |
| "logits/chosen": -2.9906249046325684, | |
| "logits/rejected": -2.971874952316284, | |
| "logps/chosen": -365.3999938964844, | |
| "logps/rejected": -308.79998779296875, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.13627929985523224, | |
| "rewards/margins": 0.2923828065395355, | |
| "rewards/rejected": -0.15678711235523224, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.2973738414006179, | |
| "grad_norm": 192.0, | |
| "learning_rate": 3.51313079299691e-07, | |
| "logits/chosen": -3.2750000953674316, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -293.79998779296875, | |
| "logps/rejected": -257.6000061035156, | |
| "loss": 0.6117, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.06489257514476776, | |
| "rewards/margins": 0.2431640625, | |
| "rewards/rejected": -0.17817382514476776, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.29866117404737386, | |
| "grad_norm": 176.0, | |
| "learning_rate": 3.506694129763131e-07, | |
| "logits/chosen": -3.1500000953674316, | |
| "logits/rejected": -3.168750047683716, | |
| "logps/chosen": -296.3999938964844, | |
| "logps/rejected": -234.10000610351562, | |
| "loss": 0.598, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.08073730766773224, | |
| "rewards/margins": 0.2925781309604645, | |
| "rewards/rejected": -0.21230468153953552, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.29994850669412976, | |
| "grad_norm": 193.0, | |
| "learning_rate": 3.5002574665293513e-07, | |
| "logits/chosen": -3.1187500953674316, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -333.6000061035156, | |
| "logps/rejected": -250.8000030517578, | |
| "loss": 0.6531, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.18466797471046448, | |
| "rewards/margins": 0.17265625298023224, | |
| "rewards/rejected": 0.012188720516860485, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.30123583934088566, | |
| "grad_norm": 284.0, | |
| "learning_rate": 3.493820803295571e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.237499952316284, | |
| "logps/chosen": -333.6000061035156, | |
| "logps/rejected": -330.79998779296875, | |
| "loss": 0.6703, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.21591797471046448, | |
| "rewards/margins": 0.14858397841453552, | |
| "rewards/rejected": 0.06718750298023224, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.3025231719876416, | |
| "grad_norm": 193.0, | |
| "learning_rate": 3.487384140061792e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -290.79998779296875, | |
| "logps/rejected": -269.3999938964844, | |
| "loss": 0.6586, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.17792968451976776, | |
| "rewards/margins": 0.1546630859375, | |
| "rewards/rejected": 0.02363281324505806, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.3038105046343975, | |
| "grad_norm": 258.0, | |
| "learning_rate": 3.480947476828012e-07, | |
| "logits/chosen": -3.231250047683716, | |
| "logits/rejected": -3.3656249046325684, | |
| "logps/chosen": -306.0, | |
| "logps/rejected": -224.8000030517578, | |
| "loss": 0.6289, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.07225342094898224, | |
| "rewards/margins": 0.22258301079273224, | |
| "rewards/rejected": -0.15048828721046448, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.30509783728115347, | |
| "grad_norm": 207.0, | |
| "learning_rate": 3.4745108135942325e-07, | |
| "logits/chosen": -3.078125, | |
| "logits/rejected": -3.1656250953674316, | |
| "logps/chosen": -348.3999938964844, | |
| "logps/rejected": -262.20001220703125, | |
| "loss": 0.607, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.2523437440395355, | |
| "rewards/margins": 0.24042968451976776, | |
| "rewards/rejected": 0.011914062313735485, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.3063851699279094, | |
| "grad_norm": 266.0, | |
| "learning_rate": 3.4680741503604534e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -320.0, | |
| "logps/rejected": -305.6000061035156, | |
| "loss": 0.6227, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.14052733778953552, | |
| "rewards/margins": 0.220703125, | |
| "rewards/rejected": -0.079833984375, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.3076725025746653, | |
| "grad_norm": 198.0, | |
| "learning_rate": 3.4616374871266733e-07, | |
| "logits/chosen": -3.0625, | |
| "logits/rejected": -2.768749952316284, | |
| "logps/chosen": -241.8000030517578, | |
| "logps/rejected": -222.3000030517578, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.19296875596046448, | |
| "rewards/margins": 0.18662719428539276, | |
| "rewards/rejected": 0.0062255859375, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.30895983522142123, | |
| "grad_norm": 226.0, | |
| "learning_rate": 3.455200823892894e-07, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.081249952316284, | |
| "logps/chosen": -310.6000061035156, | |
| "logps/rejected": -239.1999969482422, | |
| "loss": 0.6176, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.18603515625, | |
| "rewards/margins": 0.25947266817092896, | |
| "rewards/rejected": -0.073486328125, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.31024716786817713, | |
| "grad_norm": 233.0, | |
| "learning_rate": 3.448764160659114e-07, | |
| "logits/chosen": -3.3375000953674316, | |
| "logits/rejected": -3.3343749046325684, | |
| "logps/chosen": -305.6000061035156, | |
| "logps/rejected": -299.20001220703125, | |
| "loss": 0.6164, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.12186889350414276, | |
| "rewards/margins": 0.22812500596046448, | |
| "rewards/rejected": -0.10634765774011612, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.31153450051493303, | |
| "grad_norm": 136.0, | |
| "learning_rate": 3.4423274974253346e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.3343749046325684, | |
| "logps/chosen": -338.79998779296875, | |
| "logps/rejected": -268.0, | |
| "loss": 0.5797, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.3148437440395355, | |
| "rewards/margins": 0.34394532442092896, | |
| "rewards/rejected": -0.02897949144244194, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.312821833161689, | |
| "grad_norm": 200.0, | |
| "learning_rate": 3.435890834191555e-07, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.25, | |
| "logps/chosen": -282.20001220703125, | |
| "logps/rejected": -234.39999389648438, | |
| "loss": 0.6156, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.11684570461511612, | |
| "rewards/margins": 0.22539062798023224, | |
| "rewards/rejected": -0.10791015625, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.3141091658084449, | |
| "grad_norm": 217.0, | |
| "learning_rate": 3.429454170957775e-07, | |
| "logits/chosen": -3.284374952316284, | |
| "logits/rejected": -3.34375, | |
| "logps/chosen": -310.6000061035156, | |
| "logps/rejected": -289.79998779296875, | |
| "loss": 0.5828, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.0625, | |
| "rewards/margins": 0.31640625, | |
| "rewards/rejected": -0.2544921934604645, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.31539649845520085, | |
| "grad_norm": 219.0, | |
| "learning_rate": 3.423017507723996e-07, | |
| "logits/chosen": -3.1312499046325684, | |
| "logits/rejected": -3.296875, | |
| "logps/chosen": -258.79998779296875, | |
| "logps/rejected": -259.6000061035156, | |
| "loss": 0.593, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.18674317002296448, | |
| "rewards/margins": 0.29082030057907104, | |
| "rewards/rejected": -0.10444335639476776, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.31668383110195675, | |
| "grad_norm": 197.0, | |
| "learning_rate": 3.4165808444902163e-07, | |
| "logits/chosen": -3.1156249046325684, | |
| "logits/rejected": -3.1343750953674316, | |
| "logps/chosen": -301.79998779296875, | |
| "logps/rejected": -273.0, | |
| "loss": 0.5797, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.20810547471046448, | |
| "rewards/margins": 0.3384765684604645, | |
| "rewards/rejected": -0.12978515028953552, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.31797116374871265, | |
| "grad_norm": 148.0, | |
| "learning_rate": 3.410144181256436e-07, | |
| "logits/chosen": -2.9312500953674316, | |
| "logits/rejected": -2.875, | |
| "logps/chosen": -293.3999938964844, | |
| "logps/rejected": -230.8000030517578, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.0911865234375, | |
| "rewards/margins": 0.14248046278953552, | |
| "rewards/rejected": -0.05090942233800888, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.3192584963954686, | |
| "grad_norm": 160.0, | |
| "learning_rate": 3.403707518022657e-07, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.3187499046325684, | |
| "logps/chosen": -267.79998779296875, | |
| "logps/rejected": -229.89999389648438, | |
| "loss": 0.6469, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.0364990234375, | |
| "rewards/margins": 0.17847900092601776, | |
| "rewards/rejected": -0.14199218153953552, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.3205458290422245, | |
| "grad_norm": 144.0, | |
| "learning_rate": 3.3972708547888776e-07, | |
| "logits/chosen": -3.1968750953674316, | |
| "logits/rejected": -3.2281250953674316, | |
| "logps/chosen": -257.6000061035156, | |
| "logps/rejected": -205.5500030517578, | |
| "loss": 0.6133, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.13740234076976776, | |
| "rewards/margins": 0.22358398139476776, | |
| "rewards/rejected": -0.08634643256664276, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.3218331616889804, | |
| "grad_norm": 264.0, | |
| "learning_rate": 3.3908341915550975e-07, | |
| "logits/chosen": -3.265625, | |
| "logits/rejected": -3.1781249046325684, | |
| "logps/chosen": -285.6000061035156, | |
| "logps/rejected": -272.0, | |
| "loss": 0.6578, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.143310546875, | |
| "rewards/margins": 0.16262206435203552, | |
| "rewards/rejected": -0.01953125, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.32312049433573636, | |
| "grad_norm": 250.0, | |
| "learning_rate": 3.3843975283213184e-07, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -3.0875000953674316, | |
| "logps/chosen": -304.79998779296875, | |
| "logps/rejected": -337.6000061035156, | |
| "loss": 0.6016, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.20830078423023224, | |
| "rewards/margins": 0.32832032442092896, | |
| "rewards/rejected": -0.11923827975988388, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.32440782698249226, | |
| "grad_norm": 200.0, | |
| "learning_rate": 3.3779608650875383e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.284374952316284, | |
| "logps/chosen": -264.0, | |
| "logps/rejected": -208.1999969482422, | |
| "loss": 0.6438, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.158447265625, | |
| "rewards/margins": 0.19901123642921448, | |
| "rewards/rejected": -0.04057617112994194, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.3256951596292482, | |
| "grad_norm": 171.0, | |
| "learning_rate": 3.371524201853759e-07, | |
| "logits/chosen": -3.2593750953674316, | |
| "logits/rejected": -3.293750047683716, | |
| "logps/chosen": -222.39999389648438, | |
| "logps/rejected": -190.0, | |
| "loss": 0.6016, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.17348632216453552, | |
| "rewards/margins": 0.2562499940395355, | |
| "rewards/rejected": -0.08303222805261612, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.3269824922760041, | |
| "grad_norm": 226.0, | |
| "learning_rate": 3.3650875386199797e-07, | |
| "logits/chosen": -3.278125047683716, | |
| "logits/rejected": -3.265625, | |
| "logps/chosen": -271.0, | |
| "logps/rejected": -245.1999969482422, | |
| "loss": 0.6, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.163818359375, | |
| "rewards/margins": 0.25743407011032104, | |
| "rewards/rejected": -0.0931396484375, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.32826982492276, | |
| "grad_norm": 214.0, | |
| "learning_rate": 3.3586508753861996e-07, | |
| "logits/chosen": -3.253124952316284, | |
| "logits/rejected": -3.2406249046325684, | |
| "logps/chosen": -339.3999938964844, | |
| "logps/rejected": -287.20001220703125, | |
| "loss": 0.6586, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.20332030951976776, | |
| "rewards/margins": 0.14089354872703552, | |
| "rewards/rejected": 0.06223144382238388, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.329557157569516, | |
| "grad_norm": 190.0, | |
| "learning_rate": 3.35221421215242e-07, | |
| "logits/chosen": -3.237499952316284, | |
| "logits/rejected": -3.262500047683716, | |
| "logps/chosen": -323.79998779296875, | |
| "logps/rejected": -311.79998779296875, | |
| "loss": 0.643, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.14921875298023224, | |
| "rewards/margins": 0.20458984375, | |
| "rewards/rejected": -0.05551757663488388, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.3308444902162719, | |
| "grad_norm": 204.0, | |
| "learning_rate": 3.34577754891864e-07, | |
| "logits/chosen": -2.8671875, | |
| "logits/rejected": -3.296875, | |
| "logps/chosen": -293.79998779296875, | |
| "logps/rejected": -197.4499969482422, | |
| "loss": 0.5855, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.23574218153953552, | |
| "rewards/margins": 0.30810546875, | |
| "rewards/rejected": -0.072662353515625, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.3321318228630278, | |
| "grad_norm": 190.0, | |
| "learning_rate": 3.339340885684861e-07, | |
| "logits/chosen": -3.1343750953674316, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -273.0, | |
| "logps/rejected": -258.6000061035156, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.12663574516773224, | |
| "rewards/margins": 0.2867187559604645, | |
| "rewards/rejected": -0.159759521484375, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.33341915550978374, | |
| "grad_norm": 163.0, | |
| "learning_rate": 3.3329042224510813e-07, | |
| "logits/chosen": -3.309375047683716, | |
| "logits/rejected": -3.3343749046325684, | |
| "logps/chosen": -295.3999938964844, | |
| "logps/rejected": -291.0, | |
| "loss": 0.6125, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.09621582180261612, | |
| "rewards/margins": 0.2633056640625, | |
| "rewards/rejected": -0.16728515923023224, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.33470648815653964, | |
| "grad_norm": 211.0, | |
| "learning_rate": 3.326467559217301e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.1656250953674316, | |
| "logps/chosen": -356.6000061035156, | |
| "logps/rejected": -326.3999938964844, | |
| "loss": 0.5914, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.2616210877895355, | |
| "rewards/margins": 0.3111328184604645, | |
| "rewards/rejected": -0.04887695237994194, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.3359938208032956, | |
| "grad_norm": 171.0, | |
| "learning_rate": 3.320030895983522e-07, | |
| "logits/chosen": -3.106250047683716, | |
| "logits/rejected": -3.2718749046325684, | |
| "logps/chosen": -248.1999969482422, | |
| "logps/rejected": -216.8000030517578, | |
| "loss": 0.632, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.08686523139476776, | |
| "rewards/margins": 0.19169922173023224, | |
| "rewards/rejected": -0.10455322265625, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.3372811534500515, | |
| "grad_norm": 147.0, | |
| "learning_rate": 3.3135942327497426e-07, | |
| "logits/chosen": -2.9593749046325684, | |
| "logits/rejected": -2.7874999046325684, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -256.70001220703125, | |
| "loss": 0.6477, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.11861572414636612, | |
| "rewards/margins": 0.16412353515625, | |
| "rewards/rejected": -0.04515380784869194, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.3385684860968074, | |
| "grad_norm": 181.0, | |
| "learning_rate": 3.3071575695159625e-07, | |
| "logits/chosen": -3.0, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -296.0, | |
| "logps/rejected": -241.60000610351562, | |
| "loss": 0.5645, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.14291992783546448, | |
| "rewards/margins": 0.36634522676467896, | |
| "rewards/rejected": -0.2230224609375, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.33985581874356335, | |
| "grad_norm": 182.0, | |
| "learning_rate": 3.3007209062821835e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.112499952316284, | |
| "logps/chosen": -263.3999938964844, | |
| "logps/rejected": -320.79998779296875, | |
| "loss": 0.6453, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.16687622666358948, | |
| "rewards/margins": 0.17246094346046448, | |
| "rewards/rejected": -0.005859375, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.34114315139031925, | |
| "grad_norm": 231.0, | |
| "learning_rate": 3.2942842430484033e-07, | |
| "logits/chosen": -3.1312499046325684, | |
| "logits/rejected": -3.2249999046325684, | |
| "logps/chosen": -319.6000061035156, | |
| "logps/rejected": -229.8000030517578, | |
| "loss": 0.582, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.23081055283546448, | |
| "rewards/margins": 0.30292969942092896, | |
| "rewards/rejected": -0.07221679389476776, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.34243048403707516, | |
| "grad_norm": 206.0, | |
| "learning_rate": 3.287847579814624e-07, | |
| "logits/chosen": -2.7890625, | |
| "logits/rejected": -3.1812500953674316, | |
| "logps/chosen": -274.70001220703125, | |
| "logps/rejected": -245.39999389648438, | |
| "loss": 0.5668, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.2953124940395355, | |
| "rewards/margins": 0.3467773497104645, | |
| "rewards/rejected": -0.05166015774011612, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.3437178166838311, | |
| "grad_norm": 200.0, | |
| "learning_rate": 3.2814109165808447e-07, | |
| "logits/chosen": -3.121875047683716, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -270.3999938964844, | |
| "logps/rejected": -261.3999938964844, | |
| "loss": 0.6438, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.11909179389476776, | |
| "rewards/margins": 0.20371094346046448, | |
| "rewards/rejected": -0.08505859225988388, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.345005149330587, | |
| "grad_norm": 250.0, | |
| "learning_rate": 3.2749742533470646e-07, | |
| "logits/chosen": -3.1968750953674316, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -270.20001220703125, | |
| "logps/rejected": -245.1999969482422, | |
| "loss": 0.6297, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.2701171934604645, | |
| "rewards/margins": 0.1953125, | |
| "rewards/rejected": 0.07443847507238388, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.34629248197734297, | |
| "grad_norm": 255.0, | |
| "learning_rate": 3.268537590113285e-07, | |
| "logits/chosen": -3.1156249046325684, | |
| "logits/rejected": -3.015625, | |
| "logps/chosen": -230.0, | |
| "logps/rejected": -231.8000030517578, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.13314208388328552, | |
| "rewards/margins": 0.13330078125, | |
| "rewards/rejected": -0.0003662109375, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 0.34757981462409887, | |
| "grad_norm": 248.0, | |
| "learning_rate": 3.262100926879506e-07, | |
| "logits/chosen": -2.9781250953674316, | |
| "logits/rejected": -2.9000000953674316, | |
| "logps/chosen": -299.6000061035156, | |
| "logps/rejected": -328.79998779296875, | |
| "loss": 0.6281, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.16464844346046448, | |
| "rewards/margins": 0.38671875, | |
| "rewards/rejected": -0.22099609673023224, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.34886714727085477, | |
| "grad_norm": 177.0, | |
| "learning_rate": 3.255664263645726e-07, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.25, | |
| "logps/chosen": -266.6000061035156, | |
| "logps/rejected": -253.8000030517578, | |
| "loss": 0.5953, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.07573242485523224, | |
| "rewards/margins": 0.29121094942092896, | |
| "rewards/rejected": -0.21552734076976776, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 0.35015447991761073, | |
| "grad_norm": 207.0, | |
| "learning_rate": 3.2492276004119463e-07, | |
| "logits/chosen": -3.2093749046325684, | |
| "logits/rejected": -3.203125, | |
| "logps/chosen": -258.79998779296875, | |
| "logps/rejected": -232.10000610351562, | |
| "loss": 0.6242, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.08806152641773224, | |
| "rewards/margins": 0.20458984375, | |
| "rewards/rejected": -0.11586914211511612, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.35144181256436663, | |
| "grad_norm": 206.0, | |
| "learning_rate": 3.242790937178166e-07, | |
| "logits/chosen": -3.28125, | |
| "logits/rejected": -3.3062500953674316, | |
| "logps/chosen": -285.79998779296875, | |
| "logps/rejected": -310.3999938964844, | |
| "loss": 0.682, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.14287109673023224, | |
| "rewards/margins": 0.12539061903953552, | |
| "rewards/rejected": 0.01786499097943306, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 0.35272914521112253, | |
| "grad_norm": 199.0, | |
| "learning_rate": 3.236354273944387e-07, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.09375, | |
| "logps/chosen": -241.8000030517578, | |
| "logps/rejected": -203.39999389648438, | |
| "loss": 0.6375, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.06069336086511612, | |
| "rewards/margins": 0.15908202528953552, | |
| "rewards/rejected": -0.09858398139476776, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.3540164778578785, | |
| "grad_norm": 904.0, | |
| "learning_rate": 3.2299176107106076e-07, | |
| "logits/chosen": -3.1187500953674316, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -306.20001220703125, | |
| "logps/rejected": -345.20001220703125, | |
| "loss": 0.7344, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.06350097805261612, | |
| "rewards/margins": 0.0406494140625, | |
| "rewards/rejected": 0.02266235277056694, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.3553038105046344, | |
| "grad_norm": 201.0, | |
| "learning_rate": 3.2234809474768275e-07, | |
| "logits/chosen": -3.296875, | |
| "logits/rejected": -3.315624952316284, | |
| "logps/chosen": -330.6000061035156, | |
| "logps/rejected": -269.3999938964844, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.12905272841453552, | |
| "rewards/margins": 0.20888671278953552, | |
| "rewards/rejected": -0.07963867485523224, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.35659114315139034, | |
| "grad_norm": 163.0, | |
| "learning_rate": 3.2170442842430485e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.1624999046325684, | |
| "logps/chosen": -229.0, | |
| "logps/rejected": -234.1999969482422, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.11601562798023224, | |
| "rewards/margins": 0.13569335639476776, | |
| "rewards/rejected": -0.019775390625, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 0.35787847579814624, | |
| "grad_norm": 209.0, | |
| "learning_rate": 3.2106076210092684e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.2281250953674316, | |
| "logps/chosen": -327.6000061035156, | |
| "logps/rejected": -281.79998779296875, | |
| "loss": 0.643, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.14169922471046448, | |
| "rewards/margins": 0.19189453125, | |
| "rewards/rejected": -0.04982910305261612, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.35916580844490215, | |
| "grad_norm": 206.0, | |
| "learning_rate": 3.204170957775489e-07, | |
| "logits/chosen": -3.003124952316284, | |
| "logits/rejected": -2.864062547683716, | |
| "logps/chosen": -221.39999389648438, | |
| "logps/rejected": -214.39999389648438, | |
| "loss": 0.6344, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.09531249850988388, | |
| "rewards/margins": 0.20234374701976776, | |
| "rewards/rejected": -0.107177734375, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 0.3604531410916581, | |
| "grad_norm": 217.0, | |
| "learning_rate": 3.19773429454171e-07, | |
| "logits/chosen": -3.237499952316284, | |
| "logits/rejected": -3.262500047683716, | |
| "logps/chosen": -332.0, | |
| "logps/rejected": -266.20001220703125, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.16628417372703552, | |
| "rewards/margins": 0.25, | |
| "rewards/rejected": -0.08369140326976776, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.361740473738414, | |
| "grad_norm": 248.0, | |
| "learning_rate": 3.1912976313079296e-07, | |
| "logits/chosen": -3.1500000953674316, | |
| "logits/rejected": -3.168750047683716, | |
| "logps/chosen": -290.20001220703125, | |
| "logps/rejected": -249.39999389648438, | |
| "loss": 0.6461, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.2943359315395355, | |
| "rewards/margins": 0.17158202826976776, | |
| "rewards/rejected": 0.12265624850988388, | |
| "step": 1405 | |
| }, | |
| { | |
| "epoch": 0.3630278063851699, | |
| "grad_norm": 229.0, | |
| "learning_rate": 3.18486096807415e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -2.9937500953674316, | |
| "logps/chosen": -276.3999938964844, | |
| "logps/rejected": -299.3999938964844, | |
| "loss": 0.6242, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.08261718600988388, | |
| "rewards/margins": 0.21806640923023224, | |
| "rewards/rejected": -0.135498046875, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.36431513903192586, | |
| "grad_norm": 204.0, | |
| "learning_rate": 3.178424304840371e-07, | |
| "logits/chosen": -3.0999999046325684, | |
| "logits/rejected": -3.090625047683716, | |
| "logps/chosen": -271.79998779296875, | |
| "logps/rejected": -245.1999969482422, | |
| "loss": 0.6133, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.15288695693016052, | |
| "rewards/margins": 0.23857422173023224, | |
| "rewards/rejected": -0.08603515475988388, | |
| "step": 1415 | |
| }, | |
| { | |
| "epoch": 0.36560247167868176, | |
| "grad_norm": 212.0, | |
| "learning_rate": 3.171987641606591e-07, | |
| "logits/chosen": -3.0218749046325684, | |
| "logits/rejected": -3.125, | |
| "logps/chosen": -266.8999938964844, | |
| "logps/rejected": -242.5, | |
| "loss": 0.632, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.08964844048023224, | |
| "rewards/margins": 0.19931641221046448, | |
| "rewards/rejected": -0.10957030951976776, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.3668898043254377, | |
| "grad_norm": 246.0, | |
| "learning_rate": 3.1655509783728114e-07, | |
| "logits/chosen": -3.2437500953674316, | |
| "logits/rejected": -3.28125, | |
| "logps/chosen": -312.3999938964844, | |
| "logps/rejected": -313.0, | |
| "loss": 0.6055, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.19541016221046448, | |
| "rewards/margins": 0.2750000059604645, | |
| "rewards/rejected": -0.079833984375, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.3681771369721936, | |
| "grad_norm": 158.0, | |
| "learning_rate": 3.159114315139032e-07, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -3.2281250953674316, | |
| "logps/chosen": -249.39999389648438, | |
| "logps/rejected": -180.39999389648438, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.11330566555261612, | |
| "rewards/margins": 0.20781250298023224, | |
| "rewards/rejected": -0.09456787258386612, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.3694644696189495, | |
| "grad_norm": 219.0, | |
| "learning_rate": 3.152677651905252e-07, | |
| "logits/chosen": -2.9437499046325684, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -226.60000610351562, | |
| "logps/rejected": -222.60000610351562, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.07968749850988388, | |
| "rewards/margins": 0.23671874403953552, | |
| "rewards/rejected": -0.15681152045726776, | |
| "step": 1435 | |
| }, | |
| { | |
| "epoch": 0.3707518022657055, | |
| "grad_norm": 219.0, | |
| "learning_rate": 3.1462409886714726e-07, | |
| "logits/chosen": -3.0843749046325684, | |
| "logits/rejected": -3.059375047683716, | |
| "logps/chosen": -269.1000061035156, | |
| "logps/rejected": -227.8000030517578, | |
| "loss": 0.6398, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.08621826022863388, | |
| "rewards/margins": 0.18442383408546448, | |
| "rewards/rejected": -0.09829101711511612, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.3720391349124614, | |
| "grad_norm": 282.0, | |
| "learning_rate": 3.1398043254376925e-07, | |
| "logits/chosen": -3.153125047683716, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -306.79998779296875, | |
| "logps/rejected": -265.6000061035156, | |
| "loss": 0.6141, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.1748046875, | |
| "rewards/margins": 0.2745117247104645, | |
| "rewards/rejected": -0.099609375, | |
| "step": 1445 | |
| }, | |
| { | |
| "epoch": 0.3733264675592173, | |
| "grad_norm": 236.0, | |
| "learning_rate": 3.1333676622039135e-07, | |
| "logits/chosen": -3.206249952316284, | |
| "logits/rejected": -3.231250047683716, | |
| "logps/chosen": -325.20001220703125, | |
| "logps/rejected": -258.20001220703125, | |
| "loss": 0.6344, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.24716797471046448, | |
| "rewards/margins": 0.20786133408546448, | |
| "rewards/rejected": 0.03964843600988388, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.37461380020597324, | |
| "grad_norm": 222.0, | |
| "learning_rate": 3.126930998970134e-07, | |
| "logits/chosen": -3.3125, | |
| "logits/rejected": -3.2281250953674316, | |
| "logps/chosen": -271.20001220703125, | |
| "logps/rejected": -250.60000610351562, | |
| "loss": 0.6086, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.02475585974752903, | |
| "rewards/margins": 0.28154295682907104, | |
| "rewards/rejected": -0.25664061307907104, | |
| "step": 1455 | |
| }, | |
| { | |
| "epoch": 0.37590113285272914, | |
| "grad_norm": 180.0, | |
| "learning_rate": 3.120494335736354e-07, | |
| "logits/chosen": -3.059375047683716, | |
| "logits/rejected": -3.1937499046325684, | |
| "logps/chosen": -312.3999938964844, | |
| "logps/rejected": -274.0, | |
| "loss": 0.6266, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.206695556640625, | |
| "rewards/margins": 0.21328124403953552, | |
| "rewards/rejected": -0.006787109188735485, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.3771884654994851, | |
| "grad_norm": 174.0, | |
| "learning_rate": 3.114057672502575e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.2125000953674316, | |
| "logps/chosen": -304.20001220703125, | |
| "logps/rejected": -266.6000061035156, | |
| "loss": 0.5941, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.19863280653953552, | |
| "rewards/margins": 0.28193360567092896, | |
| "rewards/rejected": -0.0836181640625, | |
| "step": 1465 | |
| }, | |
| { | |
| "epoch": 0.378475798146241, | |
| "grad_norm": 189.0, | |
| "learning_rate": 3.1076210092687947e-07, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -3.15625, | |
| "logps/chosen": -296.20001220703125, | |
| "logps/rejected": -233.0, | |
| "loss": 0.6133, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.189208984375, | |
| "rewards/margins": 0.23779296875, | |
| "rewards/rejected": -0.04912719875574112, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.3797631307929969, | |
| "grad_norm": 231.0, | |
| "learning_rate": 3.101184346035015e-07, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.268749952316284, | |
| "logps/chosen": -308.79998779296875, | |
| "logps/rejected": -296.5, | |
| "loss": 0.6609, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.08837890625, | |
| "rewards/margins": 0.13093261420726776, | |
| "rewards/rejected": -0.04282836988568306, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.38105046343975285, | |
| "grad_norm": 113.0, | |
| "learning_rate": 3.094747682801236e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -2.635937452316284, | |
| "logps/chosen": -197.9250030517578, | |
| "logps/rejected": -241.8000030517578, | |
| "loss": 0.6562, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": 0.09381103515625, | |
| "rewards/margins": 0.15015868842601776, | |
| "rewards/rejected": -0.05609435960650444, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.38233779608650875, | |
| "grad_norm": 182.0, | |
| "learning_rate": 3.088311019567456e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -301.20001220703125, | |
| "logps/rejected": -251.1999969482422, | |
| "loss": 0.6156, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.20888671278953552, | |
| "rewards/margins": 0.21640625596046448, | |
| "rewards/rejected": -0.00791015662252903, | |
| "step": 1485 | |
| }, | |
| { | |
| "epoch": 0.38362512873326465, | |
| "grad_norm": 168.0, | |
| "learning_rate": 3.0818743563336764e-07, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -308.79998779296875, | |
| "logps/rejected": -248.39999389648438, | |
| "loss": 0.5719, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.23994140326976776, | |
| "rewards/margins": 0.34375, | |
| "rewards/rejected": -0.10410156100988388, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.3849124613800206, | |
| "grad_norm": 184.0, | |
| "learning_rate": 3.0754376930998973e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.2093749046325684, | |
| "logps/chosen": -278.0, | |
| "logps/rejected": -241.10000610351562, | |
| "loss": 0.6133, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.2603515684604645, | |
| "rewards/margins": 0.2770446836948395, | |
| "rewards/rejected": -0.01699218712747097, | |
| "step": 1495 | |
| }, | |
| { | |
| "epoch": 0.3861997940267765, | |
| "grad_norm": 183.0, | |
| "learning_rate": 3.069001029866117e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.2281250953674316, | |
| "logps/chosen": -275.3999938964844, | |
| "logps/rejected": -270.20001220703125, | |
| "loss": 0.6031, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.20688477158546448, | |
| "rewards/margins": 0.2582031190395355, | |
| "rewards/rejected": -0.05107421800494194, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.38748712667353247, | |
| "grad_norm": 224.0, | |
| "learning_rate": 3.0625643666323377e-07, | |
| "logits/chosen": -3.2406249046325684, | |
| "logits/rejected": -3.2906250953674316, | |
| "logps/chosen": -312.3999938964844, | |
| "logps/rejected": -301.3999938964844, | |
| "loss": 0.6281, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.22246094048023224, | |
| "rewards/margins": 0.23886719346046448, | |
| "rewards/rejected": -0.01621093787252903, | |
| "step": 1505 | |
| }, | |
| { | |
| "epoch": 0.38877445932028837, | |
| "grad_norm": 212.0, | |
| "learning_rate": 3.056127703398558e-07, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -3.049999952316284, | |
| "logps/chosen": -299.20001220703125, | |
| "logps/rejected": -233.1999969482422, | |
| "loss": 0.618, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.18349608778953552, | |
| "rewards/margins": 0.22832031548023224, | |
| "rewards/rejected": -0.04473266750574112, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.39006179196704427, | |
| "grad_norm": 139.0, | |
| "learning_rate": 3.0496910401647785e-07, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.09375, | |
| "logps/chosen": -296.0, | |
| "logps/rejected": -287.3999938964844, | |
| "loss": 0.6102, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.17763671278953552, | |
| "rewards/margins": 0.23884277045726776, | |
| "rewards/rejected": -0.06157226487994194, | |
| "step": 1515 | |
| }, | |
| { | |
| "epoch": 0.3913491246138002, | |
| "grad_norm": 204.0, | |
| "learning_rate": 3.043254376930999e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -304.79998779296875, | |
| "logps/rejected": -311.6000061035156, | |
| "loss": 0.625, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.11054687201976776, | |
| "rewards/margins": 0.27656251192092896, | |
| "rewards/rejected": -0.16623535752296448, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.3926364572605561, | |
| "grad_norm": 189.0, | |
| "learning_rate": 3.036817713697219e-07, | |
| "logits/chosen": -3.065624952316284, | |
| "logits/rejected": -3.121875047683716, | |
| "logps/chosen": -245.39999389648438, | |
| "logps/rejected": -207.60000610351562, | |
| "loss": 0.6195, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.20927734673023224, | |
| "rewards/margins": 0.20187988877296448, | |
| "rewards/rejected": 0.0075439452193677425, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.393923789907312, | |
| "grad_norm": 224.0, | |
| "learning_rate": 3.03038105046344e-07, | |
| "logits/chosen": -3.0999999046325684, | |
| "logits/rejected": -3.1781249046325684, | |
| "logps/chosen": -256.79998779296875, | |
| "logps/rejected": -259.6499938964844, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.15146484971046448, | |
| "rewards/margins": 0.2601562440395355, | |
| "rewards/rejected": -0.10888671875, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.395211122554068, | |
| "grad_norm": 148.0, | |
| "learning_rate": 3.0239443872296597e-07, | |
| "logits/chosen": -3.153125047683716, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -256.79998779296875, | |
| "logps/rejected": -238.1999969482422, | |
| "loss": 0.6453, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.08495483547449112, | |
| "rewards/margins": 0.177734375, | |
| "rewards/rejected": -0.0928955078125, | |
| "step": 1535 | |
| }, | |
| { | |
| "epoch": 0.3964984552008239, | |
| "grad_norm": 282.0, | |
| "learning_rate": 3.01750772399588e-07, | |
| "logits/chosen": -3.065624952316284, | |
| "logits/rejected": -3.1343750953674316, | |
| "logps/chosen": -304.79998779296875, | |
| "logps/rejected": -321.79998779296875, | |
| "loss": 0.6875, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.09736327826976776, | |
| "rewards/margins": 0.13884277641773224, | |
| "rewards/rejected": -0.04184570163488388, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.39778578784757984, | |
| "grad_norm": 172.0, | |
| "learning_rate": 3.011071060762101e-07, | |
| "logits/chosen": -2.9781250953674316, | |
| "logits/rejected": -3.043750047683716, | |
| "logps/chosen": -262.6000061035156, | |
| "logps/rejected": -290.79998779296875, | |
| "loss": 0.6039, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.158203125, | |
| "rewards/margins": 0.2740234434604645, | |
| "rewards/rejected": -0.115966796875, | |
| "step": 1545 | |
| }, | |
| { | |
| "epoch": 0.39907312049433574, | |
| "grad_norm": 255.0, | |
| "learning_rate": 3.004634397528321e-07, | |
| "logits/chosen": -3.143749952316284, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -263.6000061035156, | |
| "logps/rejected": -204.60000610351562, | |
| "loss": 0.6234, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.12197265774011612, | |
| "rewards/margins": 0.23247070610523224, | |
| "rewards/rejected": -0.11093749850988388, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.40036045314109164, | |
| "grad_norm": 187.0, | |
| "learning_rate": 2.9981977342945414e-07, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.2406249046325684, | |
| "logps/chosen": -289.3999938964844, | |
| "logps/rejected": -277.0, | |
| "loss": 0.6203, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.19389648735523224, | |
| "rewards/margins": 0.22929687798023224, | |
| "rewards/rejected": -0.03549804538488388, | |
| "step": 1555 | |
| }, | |
| { | |
| "epoch": 0.4016477857878476, | |
| "grad_norm": 236.0, | |
| "learning_rate": 2.9917610710607623e-07, | |
| "logits/chosen": -3.1624999046325684, | |
| "logits/rejected": -3.15625, | |
| "logps/chosen": -340.3999938964844, | |
| "logps/rejected": -279.20001220703125, | |
| "loss": 0.5852, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.2955078184604645, | |
| "rewards/margins": 0.33203125, | |
| "rewards/rejected": -0.037109375, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.4029351184346035, | |
| "grad_norm": 212.0, | |
| "learning_rate": 2.985324407826982e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.1343750953674316, | |
| "logps/chosen": -279.0, | |
| "logps/rejected": -237.1999969482422, | |
| "loss": 0.6398, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.17666015028953552, | |
| "rewards/margins": 0.17475585639476776, | |
| "rewards/rejected": 0.0019592284224927425, | |
| "step": 1565 | |
| }, | |
| { | |
| "epoch": 0.4042224510813594, | |
| "grad_norm": 262.0, | |
| "learning_rate": 2.9788877445932027e-07, | |
| "logits/chosen": -3.34375, | |
| "logits/rejected": -3.331249952316284, | |
| "logps/chosen": -296.0, | |
| "logps/rejected": -272.0, | |
| "loss": 0.6719, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.014453125186264515, | |
| "rewards/margins": 0.13984374701976776, | |
| "rewards/rejected": -0.1251220703125, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.40550978372811536, | |
| "grad_norm": 213.0, | |
| "learning_rate": 2.972451081359423e-07, | |
| "logits/chosen": -3.1500000953674316, | |
| "logits/rejected": -3.174999952316284, | |
| "logps/chosen": -266.6000061035156, | |
| "logps/rejected": -247.39999389648438, | |
| "loss": 0.6133, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.03063354454934597, | |
| "rewards/margins": 0.2533203065395355, | |
| "rewards/rejected": -0.28339844942092896, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.40679711637487126, | |
| "grad_norm": 176.0, | |
| "learning_rate": 2.9660144181256435e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.325000047683716, | |
| "logps/chosen": -228.8000030517578, | |
| "logps/rejected": -162.60000610351562, | |
| "loss": 0.6391, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.18046875298023224, | |
| "rewards/margins": 0.16850586235523224, | |
| "rewards/rejected": 0.01180419884622097, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.4080844490216272, | |
| "grad_norm": 221.0, | |
| "learning_rate": 2.959577754891864e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -289.6000061035156, | |
| "logps/rejected": -245.1999969482422, | |
| "loss": 0.6617, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.20478515326976776, | |
| "rewards/margins": 0.15556640923023224, | |
| "rewards/rejected": 0.04921875149011612, | |
| "step": 1585 | |
| }, | |
| { | |
| "epoch": 0.4093717816683831, | |
| "grad_norm": 223.0, | |
| "learning_rate": 2.9531410916580844e-07, | |
| "logits/chosen": -3.2874999046325684, | |
| "logits/rejected": -3.1500000953674316, | |
| "logps/chosen": -270.20001220703125, | |
| "logps/rejected": -248.8000030517578, | |
| "loss": 0.5789, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.16352538764476776, | |
| "rewards/margins": 0.310546875, | |
| "rewards/rejected": -0.1468505859375, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.410659114315139, | |
| "grad_norm": 274.0, | |
| "learning_rate": 2.946704428424305e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.028125047683716, | |
| "logps/chosen": -294.3999938964844, | |
| "logps/rejected": -245.60000610351562, | |
| "loss": 0.6055, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.18427734076976776, | |
| "rewards/margins": 0.2530273497104645, | |
| "rewards/rejected": -0.0692138671875, | |
| "step": 1595 | |
| }, | |
| { | |
| "epoch": 0.411946446961895, | |
| "grad_norm": 185.0, | |
| "learning_rate": 2.940267765190525e-07, | |
| "logits/chosen": -3.049999952316284, | |
| "logits/rejected": -3.128124952316284, | |
| "logps/chosen": -261.0, | |
| "logps/rejected": -219.39999389648438, | |
| "loss": 0.6078, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.20844726264476776, | |
| "rewards/margins": 0.22207030653953552, | |
| "rewards/rejected": -0.013427734375, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.4132337796086509, | |
| "grad_norm": 178.0, | |
| "learning_rate": 2.933831101956745e-07, | |
| "logits/chosen": -3.09375, | |
| "logits/rejected": -3.0718750953674316, | |
| "logps/chosen": -266.79998779296875, | |
| "logps/rejected": -215.1999969482422, | |
| "loss": 0.5855, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.23085936903953552, | |
| "rewards/margins": 0.2978515625, | |
| "rewards/rejected": -0.06662597507238388, | |
| "step": 1605 | |
| }, | |
| { | |
| "epoch": 0.4145211122554068, | |
| "grad_norm": 193.0, | |
| "learning_rate": 2.927394438722966e-07, | |
| "logits/chosen": -3.237499952316284, | |
| "logits/rejected": -3.2593750953674316, | |
| "logps/chosen": -245.0, | |
| "logps/rejected": -231.8000030517578, | |
| "loss": 0.6227, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.15673828125, | |
| "rewards/margins": 0.237335205078125, | |
| "rewards/rejected": -0.080810546875, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.41580844490216273, | |
| "grad_norm": 197.0, | |
| "learning_rate": 2.920957775489186e-07, | |
| "logits/chosen": -3.075000047683716, | |
| "logits/rejected": -3.0875000953674316, | |
| "logps/chosen": -301.6000061035156, | |
| "logps/rejected": -281.20001220703125, | |
| "loss": 0.6617, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.15205077826976776, | |
| "rewards/margins": 0.13051757216453552, | |
| "rewards/rejected": 0.02145996131002903, | |
| "step": 1615 | |
| }, | |
| { | |
| "epoch": 0.41709577754891863, | |
| "grad_norm": 228.0, | |
| "learning_rate": 2.9145211122554064e-07, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.096874952316284, | |
| "logps/chosen": -297.8999938964844, | |
| "logps/rejected": -242.39999389648438, | |
| "loss": 0.6422, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.19160155951976776, | |
| "rewards/margins": 0.16884765028953552, | |
| "rewards/rejected": 0.02295226976275444, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.41838311019567453, | |
| "grad_norm": 194.0, | |
| "learning_rate": 2.9080844490216274e-07, | |
| "logits/chosen": -3.2406249046325684, | |
| "logits/rejected": -3.153125047683716, | |
| "logps/chosen": -238.0, | |
| "logps/rejected": -216.60000610351562, | |
| "loss": 0.6703, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.08037109673023224, | |
| "rewards/margins": 0.10625000298023224, | |
| "rewards/rejected": -0.02565917931497097, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.4196704428424305, | |
| "grad_norm": 183.0, | |
| "learning_rate": 2.9016477857878473e-07, | |
| "logits/chosen": -3.2874999046325684, | |
| "logits/rejected": -3.309375047683716, | |
| "logps/chosen": -305.6000061035156, | |
| "logps/rejected": -253.39999389648438, | |
| "loss": 0.6328, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.21123047173023224, | |
| "rewards/margins": 0.21079102158546448, | |
| "rewards/rejected": 0.0001464843808207661, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.4209577754891864, | |
| "grad_norm": 252.0, | |
| "learning_rate": 2.8952111225540677e-07, | |
| "logits/chosen": -2.879687547683716, | |
| "logits/rejected": -2.9000000953674316, | |
| "logps/chosen": -269.0, | |
| "logps/rejected": -246.89999389648438, | |
| "loss": 0.6477, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.09267578274011612, | |
| "rewards/margins": 0.18876953423023224, | |
| "rewards/rejected": -0.09602050483226776, | |
| "step": 1635 | |
| }, | |
| { | |
| "epoch": 0.42224510813594235, | |
| "grad_norm": 183.0, | |
| "learning_rate": 2.8887744593202886e-07, | |
| "logits/chosen": -3.299999952316284, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -308.79998779296875, | |
| "logps/rejected": -269.79998779296875, | |
| "loss": 0.584, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.34687501192092896, | |
| "rewards/margins": 0.3173828125, | |
| "rewards/rejected": 0.029052734375, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.42353244078269825, | |
| "grad_norm": 320.0, | |
| "learning_rate": 2.8823377960865085e-07, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -286.0, | |
| "logps/rejected": -259.79998779296875, | |
| "loss": 0.6438, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.201171875, | |
| "rewards/margins": 0.17412109673023224, | |
| "rewards/rejected": 0.02714843675494194, | |
| "step": 1645 | |
| }, | |
| { | |
| "epoch": 0.42481977342945415, | |
| "grad_norm": 199.0, | |
| "learning_rate": 2.875901132852729e-07, | |
| "logits/chosen": -3.106250047683716, | |
| "logits/rejected": -3.25, | |
| "logps/chosen": -256.79998779296875, | |
| "logps/rejected": -245.1999969482422, | |
| "loss": 0.582, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.18779297173023224, | |
| "rewards/margins": 0.322265625, | |
| "rewards/rejected": -0.13395996391773224, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.4261071060762101, | |
| "grad_norm": 180.0, | |
| "learning_rate": 2.8694644696189494e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.2125000953674316, | |
| "logps/chosen": -247.1999969482422, | |
| "logps/rejected": -251.0, | |
| "loss": 0.6406, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.1630859375, | |
| "rewards/margins": 0.17138671875, | |
| "rewards/rejected": -0.00797119177877903, | |
| "step": 1655 | |
| }, | |
| { | |
| "epoch": 0.427394438722966, | |
| "grad_norm": 234.0, | |
| "learning_rate": 2.86302780638517e-07, | |
| "logits/chosen": -3.1343750953674316, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -370.79998779296875, | |
| "logps/rejected": -335.6000061035156, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.1650390625, | |
| "rewards/margins": 0.23837891221046448, | |
| "rewards/rejected": -0.07275390625, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.4286817713697219, | |
| "grad_norm": 176.0, | |
| "learning_rate": 2.85659114315139e-07, | |
| "logits/chosen": -3.0093750953674316, | |
| "logits/rejected": -3.075000047683716, | |
| "logps/chosen": -273.6000061035156, | |
| "logps/rejected": -331.3999938964844, | |
| "loss": 0.6336, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.18115234375, | |
| "rewards/margins": 0.19731445610523224, | |
| "rewards/rejected": -0.01633605919778347, | |
| "step": 1665 | |
| }, | |
| { | |
| "epoch": 0.42996910401647787, | |
| "grad_norm": 207.0, | |
| "learning_rate": 2.8501544799176107e-07, | |
| "logits/chosen": -3.096874952316284, | |
| "logits/rejected": -3.012500047683716, | |
| "logps/chosen": -302.3999938964844, | |
| "logps/rejected": -265.79998779296875, | |
| "loss": 0.6188, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.2586914002895355, | |
| "rewards/margins": 0.22509765625, | |
| "rewards/rejected": 0.03300781175494194, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.43125643666323377, | |
| "grad_norm": 604.0, | |
| "learning_rate": 2.843717816683831e-07, | |
| "logits/chosen": -3.0250000953674316, | |
| "logits/rejected": -3.0562500953674316, | |
| "logps/chosen": -221.39999389648438, | |
| "logps/rejected": -178.1999969482422, | |
| "loss": 0.6273, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.13044433295726776, | |
| "rewards/margins": 0.20537109673023224, | |
| "rewards/rejected": -0.07467041164636612, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.4325437693099897, | |
| "grad_norm": 206.0, | |
| "learning_rate": 2.837281153450051e-07, | |
| "logits/chosen": -3.3218750953674316, | |
| "logits/rejected": -3.3843750953674316, | |
| "logps/chosen": -281.20001220703125, | |
| "logps/rejected": -200.5, | |
| "loss": 0.6414, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.10593261569738388, | |
| "rewards/margins": 0.17817382514476776, | |
| "rewards/rejected": -0.07177734375, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.4338311019567456, | |
| "grad_norm": 262.0, | |
| "learning_rate": 2.830844490216272e-07, | |
| "logits/chosen": -3.143749952316284, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -332.79998779296875, | |
| "logps/rejected": -277.0, | |
| "loss": 0.6336, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.23095703125, | |
| "rewards/margins": 0.19321289658546448, | |
| "rewards/rejected": 0.03779296949505806, | |
| "step": 1685 | |
| }, | |
| { | |
| "epoch": 0.4351184346035015, | |
| "grad_norm": 232.0, | |
| "learning_rate": 2.8244078269824924e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.2125000953674316, | |
| "logps/chosen": -311.3999938964844, | |
| "logps/rejected": -287.20001220703125, | |
| "loss": 0.6289, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.12458495795726776, | |
| "rewards/margins": 0.23125000298023224, | |
| "rewards/rejected": -0.106689453125, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.4364057672502575, | |
| "grad_norm": 213.0, | |
| "learning_rate": 2.8179711637487123e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.153125047683716, | |
| "logps/chosen": -270.0, | |
| "logps/rejected": -255.0, | |
| "loss": 0.6125, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.06340332329273224, | |
| "rewards/margins": 0.26958006620407104, | |
| "rewards/rejected": -0.20673827826976776, | |
| "step": 1695 | |
| }, | |
| { | |
| "epoch": 0.4376930998970134, | |
| "grad_norm": 149.0, | |
| "learning_rate": 2.8115345005149327e-07, | |
| "logits/chosen": -3.012500047683716, | |
| "logits/rejected": -2.9046874046325684, | |
| "logps/chosen": -282.29998779296875, | |
| "logps/rejected": -254.6999969482422, | |
| "loss": 0.6398, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.25810545682907104, | |
| "rewards/margins": 0.17900390923023224, | |
| "rewards/rejected": 0.07868652045726776, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.4389804325437693, | |
| "grad_norm": 219.0, | |
| "learning_rate": 2.8050978372811537e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -2.9749999046325684, | |
| "logps/chosen": -274.79998779296875, | |
| "logps/rejected": -266.3999938964844, | |
| "loss": 0.6406, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.19199219346046448, | |
| "rewards/margins": 0.22148437798023224, | |
| "rewards/rejected": -0.02949218824505806, | |
| "step": 1705 | |
| }, | |
| { | |
| "epoch": 0.44026776519052524, | |
| "grad_norm": 234.0, | |
| "learning_rate": 2.7986611740473736e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.1343750953674316, | |
| "logps/chosen": -328.79998779296875, | |
| "logps/rejected": -318.3999938964844, | |
| "loss": 0.625, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.13188476860523224, | |
| "rewards/margins": 0.24580077826976776, | |
| "rewards/rejected": -0.11406250298023224, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.44155509783728114, | |
| "grad_norm": 180.0, | |
| "learning_rate": 2.792224510813594e-07, | |
| "logits/chosen": -3.393749952316284, | |
| "logits/rejected": -2.7007813453674316, | |
| "logps/chosen": -237.4375, | |
| "logps/rejected": -261.20001220703125, | |
| "loss": 0.6195, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.10927734524011612, | |
| "rewards/margins": 0.27558594942092896, | |
| "rewards/rejected": -0.16616210341453552, | |
| "step": 1715 | |
| }, | |
| { | |
| "epoch": 0.4428424304840371, | |
| "grad_norm": 214.0, | |
| "learning_rate": 2.7857878475798144e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -332.79998779296875, | |
| "logps/rejected": -282.0, | |
| "loss": 0.6508, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.22739258408546448, | |
| "rewards/margins": 0.15092773735523224, | |
| "rewards/rejected": 0.07656250149011612, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.444129763130793, | |
| "grad_norm": 149.0, | |
| "learning_rate": 2.779351184346035e-07, | |
| "logits/chosen": -3.112499952316284, | |
| "logits/rejected": -3.109375, | |
| "logps/chosen": -260.0, | |
| "logps/rejected": -223.10000610351562, | |
| "loss": 0.5992, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.19814452528953552, | |
| "rewards/margins": 0.2529296875, | |
| "rewards/rejected": -0.0543212890625, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.4454170957775489, | |
| "grad_norm": 180.0, | |
| "learning_rate": 2.7729145211122553e-07, | |
| "logits/chosen": -3.2718749046325684, | |
| "logits/rejected": -3.265625, | |
| "logps/chosen": -315.0, | |
| "logps/rejected": -291.20001220703125, | |
| "loss": 0.6289, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.20634765923023224, | |
| "rewards/margins": 0.23808594048023224, | |
| "rewards/rejected": -0.03139648586511612, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.44670442842430486, | |
| "grad_norm": 180.0, | |
| "learning_rate": 2.7664778578784757e-07, | |
| "logits/chosen": -2.9781250953674316, | |
| "logits/rejected": -2.8453125953674316, | |
| "logps/chosen": -249.39999389648438, | |
| "logps/rejected": -203.5, | |
| "loss": 0.6406, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.24238280951976776, | |
| "rewards/margins": 0.187255859375, | |
| "rewards/rejected": 0.05506591871380806, | |
| "step": 1735 | |
| }, | |
| { | |
| "epoch": 0.44799176107106076, | |
| "grad_norm": 221.0, | |
| "learning_rate": 2.760041194644696e-07, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -252.1999969482422, | |
| "logps/rejected": -202.8000030517578, | |
| "loss": 0.6016, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.10740967094898224, | |
| "rewards/margins": 0.29804688692092896, | |
| "rewards/rejected": -0.19096679985523224, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.44927909371781666, | |
| "grad_norm": 232.0, | |
| "learning_rate": 2.7536045314109166e-07, | |
| "logits/chosen": -3.1624999046325684, | |
| "logits/rejected": -3.2750000953674316, | |
| "logps/chosen": -218.6999969482422, | |
| "logps/rejected": -177.0, | |
| "loss": 0.6211, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.22895507514476776, | |
| "rewards/margins": 0.20820312201976776, | |
| "rewards/rejected": 0.02060546912252903, | |
| "step": 1745 | |
| }, | |
| { | |
| "epoch": 0.4505664263645726, | |
| "grad_norm": 235.0, | |
| "learning_rate": 2.747167868177137e-07, | |
| "logits/chosen": -3.2718749046325684, | |
| "logits/rejected": -3.2593750953674316, | |
| "logps/chosen": -309.20001220703125, | |
| "logps/rejected": -291.20001220703125, | |
| "loss": 0.6492, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.07080078125, | |
| "rewards/margins": 0.18525390326976776, | |
| "rewards/rejected": -0.11406250298023224, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.4518537590113285, | |
| "grad_norm": 150.0, | |
| "learning_rate": 2.7407312049433574e-07, | |
| "logits/chosen": -3.043750047683716, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -332.3999938964844, | |
| "logps/rejected": -241.5, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.30058592557907104, | |
| "rewards/margins": 0.2603515684604645, | |
| "rewards/rejected": 0.03996581956744194, | |
| "step": 1755 | |
| }, | |
| { | |
| "epoch": 0.45314109165808447, | |
| "grad_norm": 181.0, | |
| "learning_rate": 2.7342945417095773e-07, | |
| "logits/chosen": -3.090625047683716, | |
| "logits/rejected": -3.0374999046325684, | |
| "logps/chosen": -250.0, | |
| "logps/rejected": -238.10000610351562, | |
| "loss": 0.6438, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.10273437201976776, | |
| "rewards/margins": 0.16862793266773224, | |
| "rewards/rejected": -0.06611327826976776, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.4544284243048404, | |
| "grad_norm": 262.0, | |
| "learning_rate": 2.727857878475798e-07, | |
| "logits/chosen": -3.2093749046325684, | |
| "logits/rejected": -3.1968750953674316, | |
| "logps/chosen": -291.79998779296875, | |
| "logps/rejected": -327.79998779296875, | |
| "loss": 0.6992, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.09501953423023224, | |
| "rewards/margins": 0.08768310397863388, | |
| "rewards/rejected": 0.0074707032181322575, | |
| "step": 1765 | |
| }, | |
| { | |
| "epoch": 0.4557157569515963, | |
| "grad_norm": 166.0, | |
| "learning_rate": 2.7214212152420187e-07, | |
| "logits/chosen": -2.9375, | |
| "logits/rejected": -3.0093750953674316, | |
| "logps/chosen": -289.3999938964844, | |
| "logps/rejected": -259.3999938964844, | |
| "loss": 0.6121, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.24169921875, | |
| "rewards/margins": 0.258544921875, | |
| "rewards/rejected": -0.01669921912252903, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.45700308959835223, | |
| "grad_norm": 224.0, | |
| "learning_rate": 2.7149845520082386e-07, | |
| "logits/chosen": -3.059375047683716, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -338.79998779296875, | |
| "logps/rejected": -246.5, | |
| "loss": 0.6242, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.18941040337085724, | |
| "rewards/margins": 0.24492187798023224, | |
| "rewards/rejected": -0.0552978515625, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.45829042224510813, | |
| "grad_norm": 237.0, | |
| "learning_rate": 2.708547888774459e-07, | |
| "logits/chosen": -3.206249952316284, | |
| "logits/rejected": -3.25, | |
| "logps/chosen": -293.0, | |
| "logps/rejected": -273.3999938964844, | |
| "loss": 0.6359, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.1737060546875, | |
| "rewards/margins": 0.22207030653953552, | |
| "rewards/rejected": -0.048583984375, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.45957775489186403, | |
| "grad_norm": 172.0, | |
| "learning_rate": 2.70211122554068e-07, | |
| "logits/chosen": -3.3031249046325684, | |
| "logits/rejected": -3.309375047683716, | |
| "logps/chosen": -275.3999938964844, | |
| "logps/rejected": -221.8000030517578, | |
| "loss": 0.593, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.21213379502296448, | |
| "rewards/margins": 0.2945312559604645, | |
| "rewards/rejected": -0.08222655951976776, | |
| "step": 1785 | |
| }, | |
| { | |
| "epoch": 0.46086508753862, | |
| "grad_norm": 141.0, | |
| "learning_rate": 2.6956745623069e-07, | |
| "logits/chosen": -3.0062499046325684, | |
| "logits/rejected": -3.112499952316284, | |
| "logps/chosen": -246.6999969482422, | |
| "logps/rejected": -239.10000610351562, | |
| "loss": 0.8523, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.15545654296875, | |
| "rewards/margins": -0.01125488243997097, | |
| "rewards/rejected": 0.16716308891773224, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.4621524201853759, | |
| "grad_norm": 168.0, | |
| "learning_rate": 2.6892378990731203e-07, | |
| "logits/chosen": -3.1968750953674316, | |
| "logits/rejected": -3.21875, | |
| "logps/chosen": -288.6000061035156, | |
| "logps/rejected": -254.8000030517578, | |
| "loss": 0.6391, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.18300780653953552, | |
| "rewards/margins": 0.21665039658546448, | |
| "rewards/rejected": -0.03423462063074112, | |
| "step": 1795 | |
| }, | |
| { | |
| "epoch": 0.46343975283213185, | |
| "grad_norm": 198.0, | |
| "learning_rate": 2.6828012358393407e-07, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -257.20001220703125, | |
| "logps/rejected": -275.6000061035156, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.21480712294578552, | |
| "rewards/margins": 0.26972657442092896, | |
| "rewards/rejected": -0.05454101413488388, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.46472708547888775, | |
| "grad_norm": 159.0, | |
| "learning_rate": 2.676364572605561e-07, | |
| "logits/chosen": -3.012500047683716, | |
| "logits/rejected": -3.0999999046325684, | |
| "logps/chosen": -184.39999389648438, | |
| "logps/rejected": -180.0500030517578, | |
| "loss": 0.657, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 0.16542968153953552, | |
| "rewards/margins": 0.12382812798023224, | |
| "rewards/rejected": 0.041015625, | |
| "step": 1805 | |
| }, | |
| { | |
| "epoch": 0.46601441812564365, | |
| "grad_norm": 242.0, | |
| "learning_rate": 2.6699279093717816e-07, | |
| "logits/chosen": -2.799999952316284, | |
| "logits/rejected": -2.671875, | |
| "logps/chosen": -226.10000610351562, | |
| "logps/rejected": -234.89999389648438, | |
| "loss": 0.6375, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.10688476264476776, | |
| "rewards/margins": 0.16912230849266052, | |
| "rewards/rejected": -0.06193847581744194, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.4673017507723996, | |
| "grad_norm": 233.0, | |
| "learning_rate": 2.663491246138002e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.174999952316284, | |
| "logps/chosen": -241.8000030517578, | |
| "logps/rejected": -251.60000610351562, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.18002930283546448, | |
| "rewards/margins": 0.23554687201976776, | |
| "rewards/rejected": -0.05510253831744194, | |
| "step": 1815 | |
| }, | |
| { | |
| "epoch": 0.4685890834191555, | |
| "grad_norm": 180.0, | |
| "learning_rate": 2.6570545829042224e-07, | |
| "logits/chosen": -3.1031250953674316, | |
| "logits/rejected": -2.987499952316284, | |
| "logps/chosen": -247.1999969482422, | |
| "logps/rejected": -253.1999969482422, | |
| "loss": 0.584, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.20942382514476776, | |
| "rewards/margins": 0.31135255098342896, | |
| "rewards/rejected": -0.10251464694738388, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.4698764160659114, | |
| "grad_norm": 192.0, | |
| "learning_rate": 2.6506179196704423e-07, | |
| "logits/chosen": -3.153125047683716, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -319.20001220703125, | |
| "logps/rejected": -285.6000061035156, | |
| "loss": 0.5687, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.21826171875, | |
| "rewards/margins": 0.38398438692092896, | |
| "rewards/rejected": -0.165283203125, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.47116374871266736, | |
| "grad_norm": 189.0, | |
| "learning_rate": 2.6441812564366633e-07, | |
| "logits/chosen": -3.278125047683716, | |
| "logits/rejected": -3.3687500953674316, | |
| "logps/chosen": -216.10000610351562, | |
| "logps/rejected": -206.1999969482422, | |
| "loss": 0.6297, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.03642578050494194, | |
| "rewards/margins": 0.20051880180835724, | |
| "rewards/rejected": -0.16423340141773224, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.47245108135942326, | |
| "grad_norm": 189.0, | |
| "learning_rate": 2.6377445932028837e-07, | |
| "logits/chosen": -3.0374999046325684, | |
| "logits/rejected": -2.987499952316284, | |
| "logps/chosen": -278.20001220703125, | |
| "logps/rejected": -250.1999969482422, | |
| "loss": 0.6188, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.18671874701976776, | |
| "rewards/margins": 0.21689453721046448, | |
| "rewards/rejected": -0.03024902381002903, | |
| "step": 1835 | |
| }, | |
| { | |
| "epoch": 0.4737384140061792, | |
| "grad_norm": 183.0, | |
| "learning_rate": 2.6313079299691036e-07, | |
| "logits/chosen": -3.075000047683716, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -302.3999938964844, | |
| "logps/rejected": -222.1999969482422, | |
| "loss": 0.5742, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.3318847715854645, | |
| "rewards/margins": 0.3651367127895355, | |
| "rewards/rejected": -0.03415527194738388, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.4750257466529351, | |
| "grad_norm": 192.0, | |
| "learning_rate": 2.6248712667353246e-07, | |
| "logits/chosen": -3.2562499046325684, | |
| "logits/rejected": -3.3531250953674316, | |
| "logps/chosen": -318.0, | |
| "logps/rejected": -267.6000061035156, | |
| "loss": 0.5813, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.20419922471046448, | |
| "rewards/margins": 0.3343749940395355, | |
| "rewards/rejected": -0.13017578423023224, | |
| "step": 1845 | |
| }, | |
| { | |
| "epoch": 0.476313079299691, | |
| "grad_norm": 197.0, | |
| "learning_rate": 2.618434603501545e-07, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.21875, | |
| "logps/chosen": -315.79998779296875, | |
| "logps/rejected": -273.6000061035156, | |
| "loss": 0.5789, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.3955078125, | |
| "rewards/margins": 0.3462890684604645, | |
| "rewards/rejected": 0.04929199069738388, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.477600411946447, | |
| "grad_norm": 221.0, | |
| "learning_rate": 2.611997940267765e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.3187499046325684, | |
| "logps/chosen": -234.3000030517578, | |
| "logps/rejected": -188.89999389648438, | |
| "loss": 0.6523, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.0576171875, | |
| "rewards/margins": 0.15253905951976776, | |
| "rewards/rejected": -0.09494628757238388, | |
| "step": 1855 | |
| }, | |
| { | |
| "epoch": 0.4788877445932029, | |
| "grad_norm": 182.0, | |
| "learning_rate": 2.6055612770339853e-07, | |
| "logits/chosen": -2.9781250953674316, | |
| "logits/rejected": -2.9781250953674316, | |
| "logps/chosen": -179.0, | |
| "logps/rejected": -200.75, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.19633789360523224, | |
| "rewards/margins": 0.19267578423023224, | |
| "rewards/rejected": 0.0036254883743822575, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.4801750772399588, | |
| "grad_norm": 262.0, | |
| "learning_rate": 2.599124613800206e-07, | |
| "logits/chosen": -3.237499952316284, | |
| "logits/rejected": -3.2593750953674316, | |
| "logps/chosen": -280.79998779296875, | |
| "logps/rejected": -252.60000610351562, | |
| "loss": 0.6937, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -0.05029296875, | |
| "rewards/margins": 0.07781982421875, | |
| "rewards/rejected": -0.12812499701976776, | |
| "step": 1865 | |
| }, | |
| { | |
| "epoch": 0.48146240988671474, | |
| "grad_norm": 170.0, | |
| "learning_rate": 2.592687950566426e-07, | |
| "logits/chosen": -3.2593750953674316, | |
| "logits/rejected": -3.34375, | |
| "logps/chosen": -307.79998779296875, | |
| "logps/rejected": -276.79998779296875, | |
| "loss": 0.5355, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.20371094346046448, | |
| "rewards/margins": 0.4320312440395355, | |
| "rewards/rejected": -0.22829589247703552, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.48274974253347064, | |
| "grad_norm": 284.0, | |
| "learning_rate": 2.5862512873326466e-07, | |
| "logits/chosen": -3.112499952316284, | |
| "logits/rejected": -3.2562499046325684, | |
| "logps/chosen": -355.0, | |
| "logps/rejected": -276.3999938964844, | |
| "loss": 0.6336, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.33281248807907104, | |
| "rewards/margins": 0.21098633110523224, | |
| "rewards/rejected": 0.12187500298023224, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.4840370751802266, | |
| "grad_norm": 195.0, | |
| "learning_rate": 2.579814624098867e-07, | |
| "logits/chosen": -3.106250047683716, | |
| "logits/rejected": -3.03125, | |
| "logps/chosen": -171.3000030517578, | |
| "logps/rejected": -151.1999969482422, | |
| "loss": 0.6578, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.003875732421875, | |
| "rewards/margins": 0.12470702826976776, | |
| "rewards/rejected": -0.12070312350988388, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.4853244078269825, | |
| "grad_norm": 211.0, | |
| "learning_rate": 2.5733779608650874e-07, | |
| "logits/chosen": -3.0406250953674316, | |
| "logits/rejected": -3.043750047683716, | |
| "logps/chosen": -331.3999938964844, | |
| "logps/rejected": -259.79998779296875, | |
| "loss": 0.5875, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.33671873807907104, | |
| "rewards/margins": 0.3287109434604645, | |
| "rewards/rejected": 0.0078125, | |
| "step": 1885 | |
| }, | |
| { | |
| "epoch": 0.4866117404737384, | |
| "grad_norm": 260.0, | |
| "learning_rate": 2.566941297631308e-07, | |
| "logits/chosen": -3.2562499046325684, | |
| "logits/rejected": -3.265625, | |
| "logps/chosen": -261.79998779296875, | |
| "logps/rejected": -246.0, | |
| "loss": 0.6109, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.06113281100988388, | |
| "rewards/margins": 0.27812498807907104, | |
| "rewards/rejected": -0.21699218451976776, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.48789907312049435, | |
| "grad_norm": 186.0, | |
| "learning_rate": 2.5605046343975283e-07, | |
| "logits/chosen": -2.715625047683716, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -163.0, | |
| "logps/rejected": -164.39999389648438, | |
| "loss": 0.6703, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.04927978664636612, | |
| "rewards/margins": 0.08072509616613388, | |
| "rewards/rejected": -0.03144531324505806, | |
| "step": 1895 | |
| }, | |
| { | |
| "epoch": 0.48918640576725025, | |
| "grad_norm": 128.0, | |
| "learning_rate": 2.5540679711637487e-07, | |
| "logits/chosen": -3.265625, | |
| "logits/rejected": -3.284374952316284, | |
| "logps/chosen": -267.79998779296875, | |
| "logps/rejected": -221.0, | |
| "loss": 0.6375, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.16044922173023224, | |
| "rewards/margins": 0.19255371391773224, | |
| "rewards/rejected": -0.03203124925494194, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.49047373841400616, | |
| "grad_norm": 251.0, | |
| "learning_rate": 2.5476313079299686e-07, | |
| "logits/chosen": -3.046875, | |
| "logits/rejected": -3.0250000953674316, | |
| "logps/chosen": -270.79998779296875, | |
| "logps/rejected": -246.0, | |
| "loss": 0.6508, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.15754394233226776, | |
| "rewards/margins": 0.15517577528953552, | |
| "rewards/rejected": 0.0026367187965661287, | |
| "step": 1905 | |
| }, | |
| { | |
| "epoch": 0.4917610710607621, | |
| "grad_norm": 180.0, | |
| "learning_rate": 2.5411946446961896e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -283.0, | |
| "logps/rejected": -282.0, | |
| "loss": 0.5598, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.1455078125, | |
| "rewards/margins": 0.3936523497104645, | |
| "rewards/rejected": -0.24858398735523224, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.493048403707518, | |
| "grad_norm": 235.0, | |
| "learning_rate": 2.53475798146241e-07, | |
| "logits/chosen": -3.0843749046325684, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -269.20001220703125, | |
| "logps/rejected": -273.20001220703125, | |
| "loss": 0.6219, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.07382812350988388, | |
| "rewards/margins": 0.23505859076976776, | |
| "rewards/rejected": -0.16079100966453552, | |
| "step": 1915 | |
| }, | |
| { | |
| "epoch": 0.49433573635427397, | |
| "grad_norm": 165.0, | |
| "learning_rate": 2.52832131822863e-07, | |
| "logits/chosen": -3.1812500953674316, | |
| "logits/rejected": -3.2093749046325684, | |
| "logps/chosen": -301.20001220703125, | |
| "logps/rejected": -250.0, | |
| "loss": 0.5797, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.23466797173023224, | |
| "rewards/margins": 0.33613282442092896, | |
| "rewards/rejected": -0.10170898586511612, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.49562306900102987, | |
| "grad_norm": 243.0, | |
| "learning_rate": 2.521884654994851e-07, | |
| "logits/chosen": -3.2249999046325684, | |
| "logits/rejected": -3.28125, | |
| "logps/chosen": -259.20001220703125, | |
| "logps/rejected": -213.1999969482422, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.06973876804113388, | |
| "rewards/margins": 0.16960449516773224, | |
| "rewards/rejected": -0.09931640326976776, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.49691040164778577, | |
| "grad_norm": 138.0, | |
| "learning_rate": 2.5154479917610713e-07, | |
| "logits/chosen": -3.0999999046325684, | |
| "logits/rejected": -3.215625047683716, | |
| "logps/chosen": -275.79998779296875, | |
| "logps/rejected": -228.8000030517578, | |
| "loss": 0.6438, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.10273437201976776, | |
| "rewards/margins": 0.19697265326976776, | |
| "rewards/rejected": -0.09445800632238388, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.4981977342945417, | |
| "grad_norm": 47.0, | |
| "learning_rate": 2.509011328527291e-07, | |
| "logits/chosen": NaN, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -226.8000030517578, | |
| "logps/rejected": -170.02499389648438, | |
| "loss": 0.6273, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.13081054389476776, | |
| "rewards/margins": 0.19884033501148224, | |
| "rewards/rejected": -0.06790771335363388, | |
| "step": 1935 | |
| }, | |
| { | |
| "epoch": 0.49948506694129763, | |
| "grad_norm": 207.0, | |
| "learning_rate": 2.5025746652935116e-07, | |
| "logits/chosen": -3.043750047683716, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -261.20001220703125, | |
| "logps/rejected": -236.89999389648438, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.26142579317092896, | |
| "rewards/margins": 0.2642578184604645, | |
| "rewards/rejected": -0.0025268555618822575, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.5007723995880535, | |
| "grad_norm": 150.0, | |
| "learning_rate": 2.496138002059732e-07, | |
| "logits/chosen": -3.1656250953674316, | |
| "logits/rejected": -3.078125, | |
| "logps/chosen": -235.39999389648438, | |
| "logps/rejected": -192.60000610351562, | |
| "loss": 0.6578, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.12978515028953552, | |
| "rewards/margins": 0.14687499403953552, | |
| "rewards/rejected": -0.01726074144244194, | |
| "step": 1945 | |
| }, | |
| { | |
| "epoch": 0.5020597322348095, | |
| "grad_norm": 900.0, | |
| "learning_rate": 2.4897013388259525e-07, | |
| "logits/chosen": -3.1187500953674316, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -348.79998779296875, | |
| "logps/rejected": -294.0, | |
| "loss": 0.6672, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.16757813096046448, | |
| "rewards/margins": 0.17722168564796448, | |
| "rewards/rejected": -0.0098876953125, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.5033470648815654, | |
| "grad_norm": 253.0, | |
| "learning_rate": 2.483264675592173e-07, | |
| "logits/chosen": -3.015625, | |
| "logits/rejected": -3.028125047683716, | |
| "logps/chosen": -255.8000030517578, | |
| "logps/rejected": -233.60000610351562, | |
| "loss": 0.6203, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.07962646335363388, | |
| "rewards/margins": 0.22207030653953552, | |
| "rewards/rejected": -0.14228515326976776, | |
| "step": 1955 | |
| }, | |
| { | |
| "epoch": 0.5046343975283213, | |
| "grad_norm": 245.0, | |
| "learning_rate": 2.4768280123583933e-07, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -291.20001220703125, | |
| "logps/rejected": -298.3999938964844, | |
| "loss": 0.5957, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.09934692084789276, | |
| "rewards/margins": 0.3407226502895355, | |
| "rewards/rejected": -0.24044188857078552, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.5059217301750772, | |
| "grad_norm": 300.0, | |
| "learning_rate": 2.470391349124614e-07, | |
| "logits/chosen": -3.0687499046325684, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -241.60000610351562, | |
| "logps/rejected": -215.0, | |
| "loss": 0.7, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.09638671576976776, | |
| "rewards/margins": 0.05019531399011612, | |
| "rewards/rejected": 0.04624023288488388, | |
| "step": 1965 | |
| }, | |
| { | |
| "epoch": 0.5072090628218332, | |
| "grad_norm": 247.0, | |
| "learning_rate": 2.463954685890834e-07, | |
| "logits/chosen": -3.049999952316284, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -336.6000061035156, | |
| "logps/rejected": -275.0, | |
| "loss": 0.5906, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.33723145723342896, | |
| "rewards/margins": 0.301675409078598, | |
| "rewards/rejected": 0.03554687649011612, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.508496395468589, | |
| "grad_norm": 222.0, | |
| "learning_rate": 2.4575180226570546e-07, | |
| "logits/chosen": -3.081249952316284, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -292.20001220703125, | |
| "logps/rejected": -275.3999938964844, | |
| "loss": 0.6883, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.15937499701976776, | |
| "rewards/margins": 0.14775390923023224, | |
| "rewards/rejected": 0.011669921688735485, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.509783728115345, | |
| "grad_norm": 218.0, | |
| "learning_rate": 2.451081359423275e-07, | |
| "logits/chosen": -3.231250047683716, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -278.3999938964844, | |
| "logps/rejected": -309.6000061035156, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.13261719048023224, | |
| "rewards/margins": 0.16766357421875, | |
| "rewards/rejected": -0.03515625, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.511071060762101, | |
| "grad_norm": 200.0, | |
| "learning_rate": 2.4446446961894955e-07, | |
| "logits/chosen": -3.215625047683716, | |
| "logits/rejected": -3.2874999046325684, | |
| "logps/chosen": -258.5, | |
| "logps/rejected": -269.0, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.13178710639476776, | |
| "rewards/margins": 0.20761719346046448, | |
| "rewards/rejected": -0.07619629055261612, | |
| "step": 1985 | |
| }, | |
| { | |
| "epoch": 0.5123583934088568, | |
| "grad_norm": 251.0, | |
| "learning_rate": 2.4382080329557153e-07, | |
| "logits/chosen": -3.0718750953674316, | |
| "logits/rejected": -2.9281249046325684, | |
| "logps/chosen": -301.79998779296875, | |
| "logps/rejected": -240.0, | |
| "loss": 0.5938, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.22773437201976776, | |
| "rewards/margins": 0.28291016817092896, | |
| "rewards/rejected": -0.05488281324505806, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.5136457260556128, | |
| "grad_norm": 235.0, | |
| "learning_rate": 2.4317713697219363e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.3125, | |
| "logps/chosen": -250.9499969482422, | |
| "logps/rejected": -267.20001220703125, | |
| "loss": 0.7086, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.05007324367761612, | |
| "rewards/margins": 0.05624999850988388, | |
| "rewards/rejected": -0.006298827938735485, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 0.5149330587023687, | |
| "grad_norm": 137.0, | |
| "learning_rate": 2.425334706488156e-07, | |
| "logits/chosen": -3.253124952316284, | |
| "logits/rejected": -3.026562452316284, | |
| "logps/chosen": -256.3999938964844, | |
| "logps/rejected": -219.5, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.29877930879592896, | |
| "rewards/margins": 0.24169921875, | |
| "rewards/rejected": 0.05776367336511612, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.5162203913491246, | |
| "grad_norm": 186.0, | |
| "learning_rate": 2.4188980432543766e-07, | |
| "logits/chosen": -3.2750000953674316, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -287.79998779296875, | |
| "logps/rejected": -259.79998779296875, | |
| "loss": 0.6094, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.21562500298023224, | |
| "rewards/margins": 0.25800782442092896, | |
| "rewards/rejected": -0.04311523586511612, | |
| "step": 2005 | |
| }, | |
| { | |
| "epoch": 0.5175077239958805, | |
| "grad_norm": 196.0, | |
| "learning_rate": 2.412461380020597e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.1968750953674316, | |
| "logps/chosen": -331.20001220703125, | |
| "logps/rejected": -271.0, | |
| "loss": 0.5813, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.24384765326976776, | |
| "rewards/margins": 0.3294921815395355, | |
| "rewards/rejected": -0.085540771484375, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.5187950566426365, | |
| "grad_norm": 266.0, | |
| "learning_rate": 2.4060247167868175e-07, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -2.9437499046325684, | |
| "logps/chosen": -235.1999969482422, | |
| "logps/rejected": -234.10000610351562, | |
| "loss": 0.6617, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.15516968071460724, | |
| "rewards/margins": 0.14814452826976776, | |
| "rewards/rejected": 0.006542968563735485, | |
| "step": 2015 | |
| }, | |
| { | |
| "epoch": 0.5200823892893924, | |
| "grad_norm": 290.0, | |
| "learning_rate": 2.399588053553038e-07, | |
| "logits/chosen": -2.9906249046325684, | |
| "logits/rejected": -2.832812547683716, | |
| "logps/chosen": -226.0, | |
| "logps/rejected": -203.89999389648438, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.25703126192092896, | |
| "rewards/margins": 0.23193359375, | |
| "rewards/rejected": 0.02456054650247097, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.5213697219361483, | |
| "grad_norm": 242.0, | |
| "learning_rate": 2.3931513903192583e-07, | |
| "logits/chosen": -3.2906250953674316, | |
| "logits/rejected": -3.340625047683716, | |
| "logps/chosen": -290.0, | |
| "logps/rejected": -245.60000610351562, | |
| "loss": 0.5891, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.15097351372241974, | |
| "rewards/margins": 0.3495727479457855, | |
| "rewards/rejected": -0.19843749701976776, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.5226570545829042, | |
| "grad_norm": 183.0, | |
| "learning_rate": 2.386714727085479e-07, | |
| "logits/chosen": -3.231250047683716, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -324.3999938964844, | |
| "logps/rejected": -281.20001220703125, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.2529296875, | |
| "rewards/margins": 0.2691406309604645, | |
| "rewards/rejected": -0.015576171688735485, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.5239443872296602, | |
| "grad_norm": 164.0, | |
| "learning_rate": 2.3802780638516992e-07, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -223.5, | |
| "logps/rejected": -184.3000030517578, | |
| "loss": 0.625, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.13789062201976776, | |
| "rewards/margins": 0.21318359673023224, | |
| "rewards/rejected": -0.07587890326976776, | |
| "step": 2035 | |
| }, | |
| { | |
| "epoch": 0.525231719876416, | |
| "grad_norm": 252.0, | |
| "learning_rate": 2.3738414006179194e-07, | |
| "logits/chosen": -3.106250047683716, | |
| "logits/rejected": -2.9625000953674316, | |
| "logps/chosen": -265.79998779296875, | |
| "logps/rejected": -270.20001220703125, | |
| "loss": 0.6461, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.1839599609375, | |
| "rewards/margins": 0.19843749701976776, | |
| "rewards/rejected": -0.01425781287252903, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.526519052523172, | |
| "grad_norm": 217.0, | |
| "learning_rate": 2.36740473738414e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -350.3999938964844, | |
| "logps/rejected": -276.3999938964844, | |
| "loss": 0.5883, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.24199219048023224, | |
| "rewards/margins": 0.3148437440395355, | |
| "rewards/rejected": -0.07353515923023224, | |
| "step": 2045 | |
| }, | |
| { | |
| "epoch": 0.527806385169928, | |
| "grad_norm": 163.0, | |
| "learning_rate": 2.3609680741503605e-07, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -2.8968749046325684, | |
| "logps/chosen": -255.3000030517578, | |
| "logps/rejected": -207.8000030517578, | |
| "loss": 0.607, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.18314819037914276, | |
| "rewards/margins": 0.2763671875, | |
| "rewards/rejected": -0.09285278618335724, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.5290937178166838, | |
| "grad_norm": 213.0, | |
| "learning_rate": 2.3545314109165806e-07, | |
| "logits/chosen": -3.1968750953674316, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -257.79998779296875, | |
| "logps/rejected": -219.60000610351562, | |
| "loss": 0.6867, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.04345703125, | |
| "rewards/margins": 0.13066406548023224, | |
| "rewards/rejected": -0.0869140625, | |
| "step": 2055 | |
| }, | |
| { | |
| "epoch": 0.5303810504634398, | |
| "grad_norm": 221.0, | |
| "learning_rate": 2.348094747682801e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -276.20001220703125, | |
| "logps/rejected": -223.8000030517578, | |
| "loss": 0.6016, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.15185546875, | |
| "rewards/margins": 0.2880859375, | |
| "rewards/rejected": -0.13643798232078552, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.5316683831101957, | |
| "grad_norm": 183.0, | |
| "learning_rate": 2.3416580844490218e-07, | |
| "logits/chosen": -3.1187500953674316, | |
| "logits/rejected": -3.1781249046325684, | |
| "logps/chosen": -290.3999938964844, | |
| "logps/rejected": -289.0, | |
| "loss": 0.6234, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.2718749940395355, | |
| "rewards/margins": 0.23867186903953552, | |
| "rewards/rejected": 0.0338134765625, | |
| "step": 2065 | |
| }, | |
| { | |
| "epoch": 0.5329557157569516, | |
| "grad_norm": 215.0, | |
| "learning_rate": 2.335221421215242e-07, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.3062500953674316, | |
| "logps/chosen": -252.60000610351562, | |
| "logps/rejected": -247.60000610351562, | |
| "loss": 0.6359, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.11689452826976776, | |
| "rewards/margins": 0.2119140625, | |
| "rewards/rejected": -0.09511718899011612, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.5342430484037075, | |
| "grad_norm": 210.0, | |
| "learning_rate": 2.3287847579814623e-07, | |
| "logits/chosen": -3.265625, | |
| "logits/rejected": -3.121875047683716, | |
| "logps/chosen": -214.8000030517578, | |
| "logps/rejected": -182.1999969482422, | |
| "loss": 0.6445, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.175537109375, | |
| "rewards/margins": 0.16020508110523224, | |
| "rewards/rejected": 0.01513671875, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.5355303810504635, | |
| "grad_norm": 249.0, | |
| "learning_rate": 2.3223480947476825e-07, | |
| "logits/chosen": -3.112499952316284, | |
| "logits/rejected": -3.0562500953674316, | |
| "logps/chosen": -334.3999938964844, | |
| "logps/rejected": -254.1999969482422, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.2630859315395355, | |
| "rewards/margins": 0.17460937798023224, | |
| "rewards/rejected": 0.08876953274011612, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.5368177136972193, | |
| "grad_norm": 282.0, | |
| "learning_rate": 2.315911431513903e-07, | |
| "logits/chosen": -3.153125047683716, | |
| "logits/rejected": -3.03125, | |
| "logps/chosen": -301.20001220703125, | |
| "logps/rejected": -263.3999938964844, | |
| "loss": 0.6625, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.19023437798023224, | |
| "rewards/margins": 0.13609619438648224, | |
| "rewards/rejected": 0.05424804612994194, | |
| "step": 2085 | |
| }, | |
| { | |
| "epoch": 0.5381050463439753, | |
| "grad_norm": 111.5, | |
| "learning_rate": 2.3094747682801236e-07, | |
| "logits/chosen": -3.034374952316284, | |
| "logits/rejected": -3.003124952316284, | |
| "logps/chosen": -243.0, | |
| "logps/rejected": -201.39999389648438, | |
| "loss": 0.6484, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.15345458686351776, | |
| "rewards/margins": 0.15859374403953552, | |
| "rewards/rejected": -0.0055480957962572575, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.5393923789907312, | |
| "grad_norm": 408.0, | |
| "learning_rate": 2.3030381050463438e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.143749952316284, | |
| "logps/chosen": -262.6000061035156, | |
| "logps/rejected": -270.6000061035156, | |
| "loss": 0.6484, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.18515625596046448, | |
| "rewards/margins": 0.18642577528953552, | |
| "rewards/rejected": -0.0014526366721838713, | |
| "step": 2095 | |
| }, | |
| { | |
| "epoch": 0.5406797116374872, | |
| "grad_norm": 239.0, | |
| "learning_rate": 2.2966014418125642e-07, | |
| "logits/chosen": -3.237499952316284, | |
| "logits/rejected": -3.246875047683716, | |
| "logps/chosen": -282.6000061035156, | |
| "logps/rejected": -248.0, | |
| "loss": 0.6406, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.16347655653953552, | |
| "rewards/margins": 0.18940429389476776, | |
| "rewards/rejected": -0.02578124962747097, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.541967044284243, | |
| "grad_norm": 220.0, | |
| "learning_rate": 2.2901647785787846e-07, | |
| "logits/chosen": -3.262500047683716, | |
| "logits/rejected": -3.2906250953674316, | |
| "logps/chosen": -321.20001220703125, | |
| "logps/rejected": -300.20001220703125, | |
| "loss": 0.6195, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.19921875, | |
| "rewards/margins": 0.25224608182907104, | |
| "rewards/rejected": -0.05312500149011612, | |
| "step": 2105 | |
| }, | |
| { | |
| "epoch": 0.543254376930999, | |
| "grad_norm": 266.0, | |
| "learning_rate": 2.283728115345005e-07, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -3.2093749046325684, | |
| "logps/chosen": -289.6000061035156, | |
| "logps/rejected": -217.3000030517578, | |
| "loss": 0.643, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.22792968153953552, | |
| "rewards/margins": 0.18303222954273224, | |
| "rewards/rejected": 0.04534912109375, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.5445417095777549, | |
| "grad_norm": 241.0, | |
| "learning_rate": 2.2772914521112255e-07, | |
| "logits/chosen": -3.153125047683716, | |
| "logits/rejected": -3.109375, | |
| "logps/chosen": -306.0, | |
| "logps/rejected": -268.6000061035156, | |
| "loss": 0.5844, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.27314454317092896, | |
| "rewards/margins": 0.31914061307907104, | |
| "rewards/rejected": -0.04618530347943306, | |
| "step": 2115 | |
| }, | |
| { | |
| "epoch": 0.5458290422245108, | |
| "grad_norm": 125.0, | |
| "learning_rate": 2.2708547888774457e-07, | |
| "logits/chosen": -3.2593750953674316, | |
| "logits/rejected": -3.299999952316284, | |
| "logps/chosen": -278.20001220703125, | |
| "logps/rejected": -242.0, | |
| "loss": 0.618, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.11508788913488388, | |
| "rewards/margins": 0.22187499701976776, | |
| "rewards/rejected": -0.1068115234375, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.5471163748712667, | |
| "grad_norm": 244.0, | |
| "learning_rate": 2.264418125643666e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.168750047683716, | |
| "logps/chosen": -314.6000061035156, | |
| "logps/rejected": -302.5, | |
| "loss": 0.5844, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.23789063096046448, | |
| "rewards/margins": 0.32861328125, | |
| "rewards/rejected": -0.0904541015625, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.5484037075180227, | |
| "grad_norm": 223.0, | |
| "learning_rate": 2.2579814624098868e-07, | |
| "logits/chosen": -2.971874952316284, | |
| "logits/rejected": -2.965625047683716, | |
| "logps/chosen": -309.6000061035156, | |
| "logps/rejected": -222.8000030517578, | |
| "loss": 0.5828, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.239990234375, | |
| "rewards/margins": 0.32695311307907104, | |
| "rewards/rejected": -0.08745117485523224, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.5496910401647785, | |
| "grad_norm": 203.0, | |
| "learning_rate": 2.251544799176107e-07, | |
| "logits/chosen": -3.1812500953674316, | |
| "logits/rejected": -3.206249952316284, | |
| "logps/chosen": -309.6000061035156, | |
| "logps/rejected": -273.79998779296875, | |
| "loss": 0.6398, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.19921875, | |
| "rewards/margins": 0.23674316704273224, | |
| "rewards/rejected": -0.036865234375, | |
| "step": 2135 | |
| }, | |
| { | |
| "epoch": 0.5509783728115345, | |
| "grad_norm": 190.0, | |
| "learning_rate": 2.2451081359423274e-07, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.206249952316284, | |
| "logps/chosen": -228.8000030517578, | |
| "logps/rejected": -245.89999389648438, | |
| "loss": 0.5836, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.1448974609375, | |
| "rewards/margins": 0.3125, | |
| "rewards/rejected": -0.16767577826976776, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.5522657054582905, | |
| "grad_norm": 196.0, | |
| "learning_rate": 2.2386714727085478e-07, | |
| "logits/chosen": -3.0625, | |
| "logits/rejected": -3.137500047683716, | |
| "logps/chosen": -332.79998779296875, | |
| "logps/rejected": -250.0, | |
| "loss": 0.5922, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.33642578125, | |
| "rewards/margins": 0.31181639432907104, | |
| "rewards/rejected": 0.02421874925494194, | |
| "step": 2145 | |
| }, | |
| { | |
| "epoch": 0.5535530381050463, | |
| "grad_norm": 250.0, | |
| "learning_rate": 2.2322348094747682e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.2406249046325684, | |
| "logps/chosen": -275.3999938964844, | |
| "logps/rejected": -275.79998779296875, | |
| "loss": 0.575, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.2503906190395355, | |
| "rewards/margins": 0.34492188692092896, | |
| "rewards/rejected": -0.09409179538488388, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.5548403707518023, | |
| "grad_norm": 178.0, | |
| "learning_rate": 2.2257981462409886e-07, | |
| "logits/chosen": -3.200000047683716, | |
| "logits/rejected": -3.128124952316284, | |
| "logps/chosen": -297.20001220703125, | |
| "logps/rejected": -272.3999938964844, | |
| "loss": 0.582, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.18327637016773224, | |
| "rewards/margins": 0.31425780057907104, | |
| "rewards/rejected": -0.13046875596046448, | |
| "step": 2155 | |
| }, | |
| { | |
| "epoch": 0.5561277033985582, | |
| "grad_norm": 294.0, | |
| "learning_rate": 2.219361483007209e-07, | |
| "logits/chosen": -3.0562500953674316, | |
| "logits/rejected": -3.0625, | |
| "logps/chosen": -306.3999938964844, | |
| "logps/rejected": -290.20001220703125, | |
| "loss": 0.6203, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.27460938692092896, | |
| "rewards/margins": 0.2589355409145355, | |
| "rewards/rejected": 0.01606445387005806, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.5574150360453141, | |
| "grad_norm": 248.0, | |
| "learning_rate": 2.2129248197734292e-07, | |
| "logits/chosen": -3.159374952316284, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -320.79998779296875, | |
| "logps/rejected": -247.8000030517578, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.20292969048023224, | |
| "rewards/margins": 0.2021484375, | |
| "rewards/rejected": 0.0007812500116415322, | |
| "step": 2165 | |
| }, | |
| { | |
| "epoch": 0.55870236869207, | |
| "grad_norm": 292.0, | |
| "learning_rate": 2.20648815653965e-07, | |
| "logits/chosen": -3.159374952316284, | |
| "logits/rejected": -3.215625047683716, | |
| "logps/chosen": -251.1999969482422, | |
| "logps/rejected": -218.60000610351562, | |
| "loss": 0.6273, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.15583495795726776, | |
| "rewards/margins": 0.21074219048023224, | |
| "rewards/rejected": -0.05479736253619194, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.559989701338826, | |
| "grad_norm": 222.0, | |
| "learning_rate": 2.20005149330587e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -265.6000061035156, | |
| "logps/rejected": -224.0, | |
| "loss": 0.5711, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.17280273139476776, | |
| "rewards/margins": 0.3550781309604645, | |
| "rewards/rejected": -0.18265685439109802, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.5612770339855818, | |
| "grad_norm": 241.0, | |
| "learning_rate": 2.1936148300720905e-07, | |
| "logits/chosen": -3.2406249046325684, | |
| "logits/rejected": -3.315624952316284, | |
| "logps/chosen": -268.6000061035156, | |
| "logps/rejected": -251.0, | |
| "loss": 0.5609, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.19423827528953552, | |
| "rewards/margins": 0.3753906190395355, | |
| "rewards/rejected": -0.18095703423023224, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.5625643666323378, | |
| "grad_norm": 213.0, | |
| "learning_rate": 2.187178166838311e-07, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.2593750953674316, | |
| "logps/chosen": -242.8000030517578, | |
| "logps/rejected": -245.39999389648438, | |
| "loss": 0.6008, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.20649413764476776, | |
| "rewards/margins": 0.30205076932907104, | |
| "rewards/rejected": -0.09604492038488388, | |
| "step": 2185 | |
| }, | |
| { | |
| "epoch": 0.5638516992790937, | |
| "grad_norm": 264.0, | |
| "learning_rate": 2.1807415036045314e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -295.79998779296875, | |
| "logps/rejected": -271.79998779296875, | |
| "loss": 0.5875, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.26469725370407104, | |
| "rewards/margins": 0.30791014432907104, | |
| "rewards/rejected": -0.04282226413488388, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.5651390319258497, | |
| "grad_norm": 223.0, | |
| "learning_rate": 2.1743048403707518e-07, | |
| "logits/chosen": -3.1624999046325684, | |
| "logits/rejected": NaN, | |
| "logps/chosen": -250.89999389648438, | |
| "logps/rejected": -227.0, | |
| "loss": 0.6211, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.18212890625, | |
| "rewards/margins": 0.2294921875, | |
| "rewards/rejected": -0.04721679538488388, | |
| "step": 2195 | |
| }, | |
| { | |
| "epoch": 0.5664263645726055, | |
| "grad_norm": 262.0, | |
| "learning_rate": 2.1678681771369722e-07, | |
| "logits/chosen": -3.2093749046325684, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -280.6000061035156, | |
| "logps/rejected": -252.1999969482422, | |
| "loss": 0.6156, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.10905762016773224, | |
| "rewards/margins": 0.27617186307907104, | |
| "rewards/rejected": -0.16706542670726776, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.5677136972193615, | |
| "grad_norm": 179.0, | |
| "learning_rate": 2.1614315139031924e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.1343750953674316, | |
| "logps/chosen": -289.79998779296875, | |
| "logps/rejected": -282.0, | |
| "loss": 0.6461, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.3091796934604645, | |
| "rewards/margins": 0.18393555283546448, | |
| "rewards/rejected": 0.12529297173023224, | |
| "step": 2205 | |
| }, | |
| { | |
| "epoch": 0.5690010298661174, | |
| "grad_norm": 270.0, | |
| "learning_rate": 2.154994850669413e-07, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -3.128124952316284, | |
| "logps/chosen": -274.0, | |
| "logps/rejected": -236.8000030517578, | |
| "loss": 0.607, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.15400390326976776, | |
| "rewards/margins": 0.2728515565395355, | |
| "rewards/rejected": -0.11923827975988388, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.5702883625128733, | |
| "grad_norm": 228.0, | |
| "learning_rate": 2.1485581874356332e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.1624999046325684, | |
| "logps/chosen": -275.20001220703125, | |
| "logps/rejected": -267.0, | |
| "loss": 0.5863, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.29188233613967896, | |
| "rewards/margins": 0.35527342557907104, | |
| "rewards/rejected": -0.0630645751953125, | |
| "step": 2215 | |
| }, | |
| { | |
| "epoch": 0.5715756951596292, | |
| "grad_norm": 220.0, | |
| "learning_rate": 2.1421215242018537e-07, | |
| "logits/chosen": -3.106250047683716, | |
| "logits/rejected": -3.2125000953674316, | |
| "logps/chosen": -256.20001220703125, | |
| "logps/rejected": -244.39999389648438, | |
| "loss": 0.5547, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.18437500298023224, | |
| "rewards/margins": 0.39453125, | |
| "rewards/rejected": -0.21025390923023224, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.5728630278063852, | |
| "grad_norm": 186.0, | |
| "learning_rate": 2.135684860968074e-07, | |
| "logits/chosen": -3.2406249046325684, | |
| "logits/rejected": -3.0875000953674316, | |
| "logps/chosen": -237.8000030517578, | |
| "logps/rejected": -280.6000061035156, | |
| "loss": 0.5953, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.23701171576976776, | |
| "rewards/margins": 0.29326170682907104, | |
| "rewards/rejected": -0.05666504055261612, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.574150360453141, | |
| "grad_norm": 188.0, | |
| "learning_rate": 2.1292481977342942e-07, | |
| "logits/chosen": -3.121875047683716, | |
| "logits/rejected": -3.0843749046325684, | |
| "logps/chosen": -277.20001220703125, | |
| "logps/rejected": -249.8000030517578, | |
| "loss": 0.5867, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.23574218153953552, | |
| "rewards/margins": 0.30683594942092896, | |
| "rewards/rejected": -0.07119140774011612, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.575437693099897, | |
| "grad_norm": 268.0, | |
| "learning_rate": 2.122811534500515e-07, | |
| "logits/chosen": -3.049999952316284, | |
| "logits/rejected": -3.112499952316284, | |
| "logps/chosen": -300.3999938964844, | |
| "logps/rejected": -247.1999969482422, | |
| "loss": 0.6344, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.26152342557907104, | |
| "rewards/margins": 0.21196289360523224, | |
| "rewards/rejected": 0.04934082180261612, | |
| "step": 2235 | |
| }, | |
| { | |
| "epoch": 0.576725025746653, | |
| "grad_norm": 192.0, | |
| "learning_rate": 2.1163748712667354e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.0406250953674316, | |
| "logps/chosen": -223.89999389648438, | |
| "logps/rejected": -194.60000610351562, | |
| "loss": 0.6734, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.15932616591453552, | |
| "rewards/margins": 0.11433105170726776, | |
| "rewards/rejected": 0.04541015625, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.5780123583934088, | |
| "grad_norm": 197.0, | |
| "learning_rate": 2.1099382080329555e-07, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.25, | |
| "logps/chosen": -276.6000061035156, | |
| "logps/rejected": -227.39999389648438, | |
| "loss": 0.5961, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.2646484375, | |
| "rewards/margins": 0.3133789002895355, | |
| "rewards/rejected": -0.04853515699505806, | |
| "step": 2245 | |
| }, | |
| { | |
| "epoch": 0.5792996910401648, | |
| "grad_norm": 212.0, | |
| "learning_rate": 2.103501544799176e-07, | |
| "logits/chosen": -2.9281249046325684, | |
| "logits/rejected": -3.206249952316284, | |
| "logps/chosen": -258.3999938964844, | |
| "logps/rejected": -299.20001220703125, | |
| "loss": 0.6781, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.18105468153953552, | |
| "rewards/margins": 0.09600524604320526, | |
| "rewards/rejected": 0.08544921875, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.5805870236869207, | |
| "grad_norm": 170.0, | |
| "learning_rate": 2.0970648815653964e-07, | |
| "logits/chosen": -3.1656250953674316, | |
| "logits/rejected": -3.174999952316284, | |
| "logps/chosen": -352.6000061035156, | |
| "logps/rejected": -268.79998779296875, | |
| "loss": 0.5844, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.35566407442092896, | |
| "rewards/margins": 0.32929688692092896, | |
| "rewards/rejected": 0.02617187425494194, | |
| "step": 2255 | |
| }, | |
| { | |
| "epoch": 0.5818743563336766, | |
| "grad_norm": 260.0, | |
| "learning_rate": 2.0906282183316168e-07, | |
| "logits/chosen": -3.0687499046325684, | |
| "logits/rejected": -3.112499952316284, | |
| "logps/chosen": -255.39999389648438, | |
| "logps/rejected": -229.10000610351562, | |
| "loss": 0.6398, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.099609375, | |
| "rewards/margins": 0.17695312201976776, | |
| "rewards/rejected": -0.07749023288488388, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.5831616889804325, | |
| "grad_norm": 244.0, | |
| "learning_rate": 2.0841915550978372e-07, | |
| "logits/chosen": -2.934375047683716, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -283.20001220703125, | |
| "logps/rejected": -227.10000610351562, | |
| "loss": 0.6484, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.18437500298023224, | |
| "rewards/margins": 0.18364258110523224, | |
| "rewards/rejected": 0.000244140625, | |
| "step": 2265 | |
| }, | |
| { | |
| "epoch": 0.5844490216271885, | |
| "grad_norm": 199.0, | |
| "learning_rate": 2.0777548918640574e-07, | |
| "logits/chosen": -3.0406250953674316, | |
| "logits/rejected": -3.109375, | |
| "logps/chosen": -256.79998779296875, | |
| "logps/rejected": -239.1999969482422, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.08652343600988388, | |
| "rewards/margins": 0.22734375298023224, | |
| "rewards/rejected": -0.14111328125, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.5857363542739444, | |
| "grad_norm": 211.0, | |
| "learning_rate": 2.071318228630278e-07, | |
| "logits/chosen": -3.112499952316284, | |
| "logits/rejected": -3.1187500953674316, | |
| "logps/chosen": -299.79998779296875, | |
| "logps/rejected": -292.20001220703125, | |
| "loss": 0.6023, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.3091796934604645, | |
| "rewards/margins": 0.2818359434604645, | |
| "rewards/rejected": 0.02714843675494194, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.5870236869207003, | |
| "grad_norm": 214.0, | |
| "learning_rate": 2.0648815653964985e-07, | |
| "logits/chosen": -3.075000047683716, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -249.60000610351562, | |
| "logps/rejected": -222.1999969482422, | |
| "loss": 0.675, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.1318359375, | |
| "rewards/margins": 0.16914062201976776, | |
| "rewards/rejected": -0.3006347715854645, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.5883110195674562, | |
| "grad_norm": 214.0, | |
| "learning_rate": 2.0584449021627187e-07, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -291.0, | |
| "logps/rejected": -243.39999389648438, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.28681641817092896, | |
| "rewards/margins": 0.20273438096046448, | |
| "rewards/rejected": 0.08359374850988388, | |
| "step": 2285 | |
| }, | |
| { | |
| "epoch": 0.5895983522142122, | |
| "grad_norm": 220.0, | |
| "learning_rate": 2.052008238928939e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.1937499046325684, | |
| "logps/chosen": -288.79998779296875, | |
| "logps/rejected": -279.3999938964844, | |
| "loss": 0.6188, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.21010741591453552, | |
| "rewards/margins": 0.24492187798023224, | |
| "rewards/rejected": -0.03475341945886612, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.590885684860968, | |
| "grad_norm": 189.0, | |
| "learning_rate": 2.0455715756951595e-07, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -254.1999969482422, | |
| "logps/rejected": -254.0, | |
| "loss": 0.6445, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.16611328721046448, | |
| "rewards/margins": 0.17724609375, | |
| "rewards/rejected": -0.0113525390625, | |
| "step": 2295 | |
| }, | |
| { | |
| "epoch": 0.592173017507724, | |
| "grad_norm": 176.0, | |
| "learning_rate": 2.03913491246138e-07, | |
| "logits/chosen": -3.046875, | |
| "logits/rejected": -3.1187500953674316, | |
| "logps/chosen": -286.3999938964844, | |
| "logps/rejected": -255.39999389648438, | |
| "loss": 0.5687, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.35566407442092896, | |
| "rewards/margins": 0.353515625, | |
| "rewards/rejected": 0.0027099610306322575, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.59346035015448, | |
| "grad_norm": 247.0, | |
| "learning_rate": 2.0326982492276004e-07, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.1187500953674316, | |
| "logps/chosen": -314.3999938964844, | |
| "logps/rejected": -301.79998779296875, | |
| "loss": 0.6016, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.3011718690395355, | |
| "rewards/margins": 0.2562499940395355, | |
| "rewards/rejected": 0.04561157152056694, | |
| "step": 2305 | |
| }, | |
| { | |
| "epoch": 0.5947476828012358, | |
| "grad_norm": 201.0, | |
| "learning_rate": 2.0262615859938205e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.2593750953674316, | |
| "logps/chosen": -363.79998779296875, | |
| "logps/rejected": -336.3999938964844, | |
| "loss": 0.5813, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.2500976622104645, | |
| "rewards/margins": 0.3675781190395355, | |
| "rewards/rejected": -0.11767578125, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.5960350154479918, | |
| "grad_norm": 181.0, | |
| "learning_rate": 2.0198249227600412e-07, | |
| "logits/chosen": -3.0250000953674316, | |
| "logits/rejected": -3.0625, | |
| "logps/chosen": -265.79998779296875, | |
| "logps/rejected": -226.39999389648438, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.2574218809604645, | |
| "rewards/margins": 0.21552734076976776, | |
| "rewards/rejected": 0.04213867336511612, | |
| "step": 2315 | |
| }, | |
| { | |
| "epoch": 0.5973223480947477, | |
| "grad_norm": 180.0, | |
| "learning_rate": 2.0133882595262617e-07, | |
| "logits/chosen": -3.0999999046325684, | |
| "logits/rejected": -2.953125, | |
| "logps/chosen": -294.79998779296875, | |
| "logps/rejected": -269.6000061035156, | |
| "loss": 0.6461, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.06816406548023224, | |
| "rewards/margins": 0.18093261122703552, | |
| "rewards/rejected": -0.11289062350988388, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.5986096807415036, | |
| "grad_norm": 204.0, | |
| "learning_rate": 2.0069515962924818e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -223.1999969482422, | |
| "logps/rejected": -257.79998779296875, | |
| "loss": 0.6141, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.10140381008386612, | |
| "rewards/margins": 0.23793944716453552, | |
| "rewards/rejected": -0.13662108778953552, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.5998970133882595, | |
| "grad_norm": 170.0, | |
| "learning_rate": 2.0005149330587023e-07, | |
| "logits/chosen": -3.0718750953674316, | |
| "logits/rejected": -3.018749952316284, | |
| "logps/chosen": -180.5, | |
| "logps/rejected": -163.6999969482422, | |
| "loss": 0.6164, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.22880859673023224, | |
| "rewards/margins": 0.21855469048023224, | |
| "rewards/rejected": 0.009838867001235485, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.6011843460350155, | |
| "grad_norm": 298.0, | |
| "learning_rate": 1.9940782698249227e-07, | |
| "logits/chosen": -3.215625047683716, | |
| "logits/rejected": -3.137500047683716, | |
| "logps/chosen": -281.20001220703125, | |
| "logps/rejected": -302.6000061035156, | |
| "loss": 0.6262, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.22172851860523224, | |
| "rewards/margins": 0.23955078423023224, | |
| "rewards/rejected": -0.01821289025247097, | |
| "step": 2335 | |
| }, | |
| { | |
| "epoch": 0.6024716786817713, | |
| "grad_norm": 198.0, | |
| "learning_rate": 1.987641606591143e-07, | |
| "logits/chosen": -3.012500047683716, | |
| "logits/rejected": -2.971874952316284, | |
| "logps/chosen": -293.3999938964844, | |
| "logps/rejected": -311.79998779296875, | |
| "loss": 0.6094, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.30158692598342896, | |
| "rewards/margins": 0.33183592557907104, | |
| "rewards/rejected": -0.03037109412252903, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.6037590113285273, | |
| "grad_norm": 226.0, | |
| "learning_rate": 1.9812049433573635e-07, | |
| "logits/chosen": -3.1624999046325684, | |
| "logits/rejected": -3.1187500953674316, | |
| "logps/chosen": -269.0, | |
| "logps/rejected": -252.39999389648438, | |
| "loss": 0.6219, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.15341797471046448, | |
| "rewards/margins": 0.23007813096046448, | |
| "rewards/rejected": -0.076416015625, | |
| "step": 2345 | |
| }, | |
| { | |
| "epoch": 0.6050463439752832, | |
| "grad_norm": 170.0, | |
| "learning_rate": 1.9747682801235837e-07, | |
| "logits/chosen": -3.121875047683716, | |
| "logits/rejected": -2.8031249046325684, | |
| "logps/chosen": -229.60000610351562, | |
| "logps/rejected": -221.60000610351562, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.0733642578125, | |
| "rewards/margins": 0.23745116591453552, | |
| "rewards/rejected": -0.16416016221046448, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.6063336766220392, | |
| "grad_norm": 235.0, | |
| "learning_rate": 1.9683316168898044e-07, | |
| "logits/chosen": -3.0718750953674316, | |
| "logits/rejected": -3.1968750953674316, | |
| "logps/chosen": -279.3999938964844, | |
| "logps/rejected": -245.1999969482422, | |
| "loss": 0.5813, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.2594238221645355, | |
| "rewards/margins": 0.3062500059604645, | |
| "rewards/rejected": -0.04672851413488388, | |
| "step": 2355 | |
| }, | |
| { | |
| "epoch": 0.607621009268795, | |
| "grad_norm": 320.0, | |
| "learning_rate": 1.9618949536560248e-07, | |
| "logits/chosen": -3.2437500953674316, | |
| "logits/rejected": -3.246875047683716, | |
| "logps/chosen": -286.0, | |
| "logps/rejected": -287.6000061035156, | |
| "loss": 0.6328, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.02031249925494194, | |
| "rewards/margins": 0.216796875, | |
| "rewards/rejected": -0.19652099907398224, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.608908341915551, | |
| "grad_norm": 170.0, | |
| "learning_rate": 1.955458290422245e-07, | |
| "logits/chosen": -3.200000047683716, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -251.39999389648438, | |
| "logps/rejected": -220.0, | |
| "loss": 0.6164, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.19868163764476776, | |
| "rewards/margins": 0.23828125, | |
| "rewards/rejected": -0.0391845703125, | |
| "step": 2365 | |
| }, | |
| { | |
| "epoch": 0.6101956745623069, | |
| "grad_norm": 192.0, | |
| "learning_rate": 1.9490216271884654e-07, | |
| "logits/chosen": -3.012500047683716, | |
| "logits/rejected": -2.831249952316284, | |
| "logps/chosen": -258.79998779296875, | |
| "logps/rejected": -239.60000610351562, | |
| "loss": 0.5977, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.21865233778953552, | |
| "rewards/margins": 0.30644530057907104, | |
| "rewards/rejected": -0.08742065727710724, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.6114830072090628, | |
| "grad_norm": 194.0, | |
| "learning_rate": 1.9425849639546856e-07, | |
| "logits/chosen": -3.106250047683716, | |
| "logits/rejected": -3.109375, | |
| "logps/chosen": -221.8000030517578, | |
| "logps/rejected": -227.60000610351562, | |
| "loss": 0.6211, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.13427734375, | |
| "rewards/margins": 0.23911133408546448, | |
| "rewards/rejected": -0.10505370795726776, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.6127703398558187, | |
| "grad_norm": 233.0, | |
| "learning_rate": 1.9361483007209063e-07, | |
| "logits/chosen": -3.2750000953674316, | |
| "logits/rejected": -3.253124952316284, | |
| "logps/chosen": -286.6000061035156, | |
| "logps/rejected": -258.6000061035156, | |
| "loss": 0.5836, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.28730469942092896, | |
| "rewards/margins": 0.30048829317092896, | |
| "rewards/rejected": -0.01279296912252903, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.6140576725025747, | |
| "grad_norm": 152.0, | |
| "learning_rate": 1.9297116374871267e-07, | |
| "logits/chosen": -3.143749952316284, | |
| "logits/rejected": -2.924999952316284, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -188.75, | |
| "loss": 0.6328, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.14228515326976776, | |
| "rewards/margins": 0.18496093153953552, | |
| "rewards/rejected": -0.04262084886431694, | |
| "step": 2385 | |
| }, | |
| { | |
| "epoch": 0.6153450051493305, | |
| "grad_norm": 494.0, | |
| "learning_rate": 1.9232749742533468e-07, | |
| "logits/chosen": -3.0218749046325684, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -217.5, | |
| "logps/rejected": -214.3000030517578, | |
| "loss": 0.6617, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.16738280653953552, | |
| "rewards/margins": 0.13740234076976776, | |
| "rewards/rejected": 0.02995605394244194, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.6166323377960865, | |
| "grad_norm": 412.0, | |
| "learning_rate": 1.9168383110195673e-07, | |
| "logits/chosen": -3.121875047683716, | |
| "logits/rejected": -3.174999952316284, | |
| "logps/chosen": -339.20001220703125, | |
| "logps/rejected": -298.79998779296875, | |
| "loss": 0.6328, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.1923828125, | |
| "rewards/margins": 0.26249998807907104, | |
| "rewards/rejected": -0.070068359375, | |
| "step": 2395 | |
| }, | |
| { | |
| "epoch": 0.6179196704428425, | |
| "grad_norm": 584.0, | |
| "learning_rate": 1.910401647785788e-07, | |
| "logits/chosen": -3.159374952316284, | |
| "logits/rejected": -3.0687499046325684, | |
| "logps/chosen": -261.6000061035156, | |
| "logps/rejected": -279.6000061035156, | |
| "loss": 0.6359, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.06833495944738388, | |
| "rewards/margins": 0.2583984434604645, | |
| "rewards/rejected": -0.189697265625, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.6192070030895983, | |
| "grad_norm": 181.0, | |
| "learning_rate": 1.903964984552008e-07, | |
| "logits/chosen": -3.09375, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -259.20001220703125, | |
| "logps/rejected": -201.1999969482422, | |
| "loss": 0.6445, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.10537109524011612, | |
| "rewards/margins": 0.16997070610523224, | |
| "rewards/rejected": -0.06503906100988388, | |
| "step": 2405 | |
| }, | |
| { | |
| "epoch": 0.6204943357363543, | |
| "grad_norm": 302.0, | |
| "learning_rate": 1.8975283213182286e-07, | |
| "logits/chosen": -3.1656250953674316, | |
| "logits/rejected": -3.265625, | |
| "logps/chosen": -354.3999938964844, | |
| "logps/rejected": -282.6000061035156, | |
| "loss": 0.557, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.221923828125, | |
| "rewards/margins": 0.39140623807907104, | |
| "rewards/rejected": -0.1693115234375, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.6217816683831102, | |
| "grad_norm": 231.0, | |
| "learning_rate": 1.8910916580844487e-07, | |
| "logits/chosen": -3.065624952316284, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -288.79998779296875, | |
| "logps/rejected": -236.39999389648438, | |
| "loss": 0.6531, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.2689453065395355, | |
| "rewards/margins": 0.14755859971046448, | |
| "rewards/rejected": 0.12128905951976776, | |
| "step": 2415 | |
| }, | |
| { | |
| "epoch": 0.6230690010298661, | |
| "grad_norm": 203.0, | |
| "learning_rate": 1.8846549948506694e-07, | |
| "logits/chosen": -3.262500047683716, | |
| "logits/rejected": -3.2562499046325684, | |
| "logps/chosen": -272.0, | |
| "logps/rejected": -247.0, | |
| "loss": 0.6055, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.16733399033546448, | |
| "rewards/margins": 0.2699218690395355, | |
| "rewards/rejected": -0.1025390625, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.624356333676622, | |
| "grad_norm": 236.0, | |
| "learning_rate": 1.8782183316168898e-07, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.112499952316284, | |
| "logps/chosen": -301.79998779296875, | |
| "logps/rejected": -274.3999938964844, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.13203124701976776, | |
| "rewards/margins": 0.2669921815395355, | |
| "rewards/rejected": -0.1357421875, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.625643666323378, | |
| "grad_norm": 296.0, | |
| "learning_rate": 1.87178166838311e-07, | |
| "logits/chosen": -3.153125047683716, | |
| "logits/rejected": -3.2125000953674316, | |
| "logps/chosen": -298.3999938964844, | |
| "logps/rejected": -266.6000061035156, | |
| "loss": 0.65, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.20014648139476776, | |
| "rewards/margins": 0.16228027641773224, | |
| "rewards/rejected": 0.03798828274011612, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.6269309989701339, | |
| "grad_norm": 276.0, | |
| "learning_rate": 1.8653450051493304e-07, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.25, | |
| "logps/chosen": -303.3999938964844, | |
| "logps/rejected": -267.20001220703125, | |
| "loss": 0.6906, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.17695312201976776, | |
| "rewards/margins": 0.09438476711511612, | |
| "rewards/rejected": 0.08271484076976776, | |
| "step": 2435 | |
| }, | |
| { | |
| "epoch": 0.6282183316168898, | |
| "grad_norm": 232.0, | |
| "learning_rate": 1.858908341915551e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.2125000953674316, | |
| "logps/chosen": -277.3999938964844, | |
| "logps/rejected": -289.20001220703125, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.11538086086511612, | |
| "rewards/margins": 0.21562500298023224, | |
| "rewards/rejected": -0.10004882514476776, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.6295056642636457, | |
| "grad_norm": 185.0, | |
| "learning_rate": 1.8524716786817713e-07, | |
| "logits/chosen": -2.956249952316284, | |
| "logits/rejected": -3.03125, | |
| "logps/chosen": -220.1999969482422, | |
| "logps/rejected": -212.39999389648438, | |
| "loss": 0.6398, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.21103516221046448, | |
| "rewards/margins": 0.17292480170726776, | |
| "rewards/rejected": 0.03803711012005806, | |
| "step": 2445 | |
| }, | |
| { | |
| "epoch": 0.6307929969104017, | |
| "grad_norm": 204.0, | |
| "learning_rate": 1.8460350154479917e-07, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.143749952316284, | |
| "logps/chosen": -282.79998779296875, | |
| "logps/rejected": -249.1999969482422, | |
| "loss": 0.6008, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.20258788764476776, | |
| "rewards/margins": 0.26708984375, | |
| "rewards/rejected": -0.064697265625, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.6320803295571575, | |
| "grad_norm": 260.0, | |
| "learning_rate": 1.8395983522142119e-07, | |
| "logits/chosen": -3.2249999046325684, | |
| "logits/rejected": -3.2093749046325684, | |
| "logps/chosen": -309.6000061035156, | |
| "logps/rejected": -305.0, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.091796875, | |
| "rewards/margins": 0.20615234971046448, | |
| "rewards/rejected": -0.11445312201976776, | |
| "step": 2455 | |
| }, | |
| { | |
| "epoch": 0.6333676622039135, | |
| "grad_norm": 203.0, | |
| "learning_rate": 1.8331616889804326e-07, | |
| "logits/chosen": -3.200000047683716, | |
| "logits/rejected": -3.253124952316284, | |
| "logps/chosen": -306.79998779296875, | |
| "logps/rejected": -292.3999938964844, | |
| "loss": 0.5852, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.18314209580421448, | |
| "rewards/margins": 0.3568359315395355, | |
| "rewards/rejected": -0.1737060546875, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.6346549948506695, | |
| "grad_norm": 250.0, | |
| "learning_rate": 1.826725025746653e-07, | |
| "logits/chosen": -3.246875047683716, | |
| "logits/rejected": -3.206249952316284, | |
| "logps/chosen": -330.3999938964844, | |
| "logps/rejected": -320.20001220703125, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.191162109375, | |
| "rewards/margins": 0.19968871772289276, | |
| "rewards/rejected": -0.00830078125, | |
| "step": 2465 | |
| }, | |
| { | |
| "epoch": 0.6359423274974253, | |
| "grad_norm": 195.0, | |
| "learning_rate": 1.8202883625128731e-07, | |
| "logits/chosen": -3.1187500953674316, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -249.10000610351562, | |
| "logps/rejected": -241.8000030517578, | |
| "loss": 0.6348, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.22966918349266052, | |
| "rewards/margins": 0.18662109971046448, | |
| "rewards/rejected": 0.04374999925494194, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.6372296601441813, | |
| "grad_norm": 266.0, | |
| "learning_rate": 1.8138516992790936e-07, | |
| "logits/chosen": -3.112499952316284, | |
| "logits/rejected": -3.2406249046325684, | |
| "logps/chosen": -301.3999938964844, | |
| "logps/rejected": -262.79998779296875, | |
| "loss": 0.6047, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.12451171875, | |
| "rewards/margins": 0.2754882872104645, | |
| "rewards/rejected": -0.15073242783546448, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.6385169927909372, | |
| "grad_norm": 230.0, | |
| "learning_rate": 1.8074150360453143e-07, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.2406249046325684, | |
| "logps/chosen": -266.3999938964844, | |
| "logps/rejected": -282.20001220703125, | |
| "loss": 0.607, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.14589843153953552, | |
| "rewards/margins": 0.28886717557907104, | |
| "rewards/rejected": -0.142578125, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.6398043254376931, | |
| "grad_norm": 145.0, | |
| "learning_rate": 1.8009783728115344e-07, | |
| "logits/chosen": -3.284374952316284, | |
| "logits/rejected": -3.2249999046325684, | |
| "logps/chosen": -293.79998779296875, | |
| "logps/rejected": -266.6000061035156, | |
| "loss": 0.6203, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.04648437350988388, | |
| "rewards/margins": 0.2421875, | |
| "rewards/rejected": -0.19560547173023224, | |
| "step": 2485 | |
| }, | |
| { | |
| "epoch": 0.641091658084449, | |
| "grad_norm": 253.0, | |
| "learning_rate": 1.7945417095777549e-07, | |
| "logits/chosen": -3.2125000953674316, | |
| "logits/rejected": -3.1156249046325684, | |
| "logps/chosen": -227.8000030517578, | |
| "logps/rejected": -243.1999969482422, | |
| "loss": 0.5828, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.12412109225988388, | |
| "rewards/margins": 0.3359375, | |
| "rewards/rejected": -0.21201172471046448, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.642378990731205, | |
| "grad_norm": 214.0, | |
| "learning_rate": 1.788105046343975e-07, | |
| "logits/chosen": -3.078125, | |
| "logits/rejected": -3.034374952316284, | |
| "logps/chosen": -222.89999389648438, | |
| "logps/rejected": -198.60000610351562, | |
| "loss": 0.6469, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.10048828274011612, | |
| "rewards/margins": 0.17402343451976776, | |
| "rewards/rejected": -0.07329101860523224, | |
| "step": 2495 | |
| }, | |
| { | |
| "epoch": 0.6436663233779608, | |
| "grad_norm": 218.0, | |
| "learning_rate": 1.7816683831101954e-07, | |
| "logits/chosen": -3.3125, | |
| "logits/rejected": -3.296875, | |
| "logps/chosen": -315.79998779296875, | |
| "logps/rejected": -280.79998779296875, | |
| "loss": 0.643, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.1494140625, | |
| "rewards/margins": 0.21464844048023224, | |
| "rewards/rejected": -0.06499023735523224, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.6449536560247168, | |
| "grad_norm": 270.0, | |
| "learning_rate": 1.7752317198764161e-07, | |
| "logits/chosen": -3.143749952316284, | |
| "logits/rejected": -3.215625047683716, | |
| "logps/chosen": -274.3999938964844, | |
| "logps/rejected": -219.89999389648438, | |
| "loss": 0.6484, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.21113280951976776, | |
| "rewards/margins": 0.18190917372703552, | |
| "rewards/rejected": 0.02900390699505806, | |
| "step": 2505 | |
| }, | |
| { | |
| "epoch": 0.6462409886714727, | |
| "grad_norm": 218.0, | |
| "learning_rate": 1.7687950566426363e-07, | |
| "logits/chosen": -2.9156250953674316, | |
| "logits/rejected": -3.0062499046325684, | |
| "logps/chosen": -251.1999969482422, | |
| "logps/rejected": -221.85000610351562, | |
| "loss": 0.6484, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.2829833924770355, | |
| "rewards/margins": 0.17363281548023224, | |
| "rewards/rejected": 0.10885009914636612, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.6475283213182287, | |
| "grad_norm": 260.0, | |
| "learning_rate": 1.7623583934088567e-07, | |
| "logits/chosen": -3.096874952316284, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -338.79998779296875, | |
| "logps/rejected": -247.8000030517578, | |
| "loss": 0.6578, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.22734375298023224, | |
| "rewards/margins": 0.13430175185203552, | |
| "rewards/rejected": 0.093017578125, | |
| "step": 2515 | |
| }, | |
| { | |
| "epoch": 0.6488156539649845, | |
| "grad_norm": 238.0, | |
| "learning_rate": 1.7559217301750771e-07, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -319.79998779296875, | |
| "logps/rejected": -280.79998779296875, | |
| "loss": 0.607, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.25617676973342896, | |
| "rewards/margins": 0.2607421875, | |
| "rewards/rejected": -0.004345702938735485, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.6501029866117405, | |
| "grad_norm": 183.0, | |
| "learning_rate": 1.7494850669412976e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.262500047683716, | |
| "logps/chosen": -335.79998779296875, | |
| "logps/rejected": -289.3999938964844, | |
| "loss": 0.6031, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.23066405951976776, | |
| "rewards/margins": 0.28662109375, | |
| "rewards/rejected": -0.05527343600988388, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.6513903192584964, | |
| "grad_norm": 199.0, | |
| "learning_rate": 1.743048403707518e-07, | |
| "logits/chosen": -3.2125000953674316, | |
| "logits/rejected": -3.1656250953674316, | |
| "logps/chosen": -312.20001220703125, | |
| "logps/rejected": -280.0, | |
| "loss": 0.5516, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.20456543564796448, | |
| "rewards/margins": 0.4339843690395355, | |
| "rewards/rejected": -0.22929687798023224, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.6526776519052523, | |
| "grad_norm": 143.0, | |
| "learning_rate": 1.7366117404737382e-07, | |
| "logits/chosen": -3.046875, | |
| "logits/rejected": -3.203125, | |
| "logps/chosen": -305.6000061035156, | |
| "logps/rejected": -170.8000030517578, | |
| "loss": 0.5867, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.298828125, | |
| "rewards/margins": 0.30830079317092896, | |
| "rewards/rejected": -0.010009765625, | |
| "step": 2535 | |
| }, | |
| { | |
| "epoch": 0.6539649845520082, | |
| "grad_norm": 181.0, | |
| "learning_rate": 1.7301750772399586e-07, | |
| "logits/chosen": -3.3531250953674316, | |
| "logits/rejected": -3.3375000953674316, | |
| "logps/chosen": -305.3999938964844, | |
| "logps/rejected": -257.6000061035156, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.0038085938431322575, | |
| "rewards/margins": 0.21328124403953552, | |
| "rewards/rejected": -0.217529296875, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.6552523171987642, | |
| "grad_norm": 251.0, | |
| "learning_rate": 1.7237384140061793e-07, | |
| "logits/chosen": -3.215625047683716, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -218.8000030517578, | |
| "logps/rejected": -210.1999969482422, | |
| "loss": 0.6125, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.03925781324505806, | |
| "rewards/margins": 0.2427978515625, | |
| "rewards/rejected": -0.28173828125, | |
| "step": 2545 | |
| }, | |
| { | |
| "epoch": 0.65653964984552, | |
| "grad_norm": 203.0, | |
| "learning_rate": 1.7173017507723994e-07, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.325000047683716, | |
| "logps/chosen": -269.0, | |
| "logps/rejected": -239.39999389648438, | |
| "loss": 0.6023, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.13291016221046448, | |
| "rewards/margins": 0.3218750059604645, | |
| "rewards/rejected": -0.18854980170726776, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.657826982492276, | |
| "grad_norm": 492.0, | |
| "learning_rate": 1.71086508753862e-07, | |
| "logits/chosen": -3.075000047683716, | |
| "logits/rejected": -2.9593749046325684, | |
| "logps/chosen": -274.3999938964844, | |
| "logps/rejected": -224.8000030517578, | |
| "loss": 0.5766, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.14765624701976776, | |
| "rewards/margins": 0.3375000059604645, | |
| "rewards/rejected": -0.189849853515625, | |
| "step": 2555 | |
| }, | |
| { | |
| "epoch": 0.659114315139032, | |
| "grad_norm": 141.0, | |
| "learning_rate": 1.7044284243048403e-07, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -224.39999389648438, | |
| "logps/rejected": -226.5, | |
| "loss": 0.707, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.14760741591453552, | |
| "rewards/margins": 0.05874023586511612, | |
| "rewards/rejected": 0.08852539211511612, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.6604016477857878, | |
| "grad_norm": 225.0, | |
| "learning_rate": 1.6979917610710607e-07, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -217.10000610351562, | |
| "logps/rejected": -245.5, | |
| "loss": 0.6105, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.1728515625, | |
| "rewards/margins": 0.25312501192092896, | |
| "rewards/rejected": -0.08078613132238388, | |
| "step": 2565 | |
| }, | |
| { | |
| "epoch": 0.6616889804325438, | |
| "grad_norm": 208.0, | |
| "learning_rate": 1.6915550978372812e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -347.6000061035156, | |
| "logps/rejected": -263.79998779296875, | |
| "loss": 0.6141, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.30976563692092896, | |
| "rewards/margins": 0.2735351622104645, | |
| "rewards/rejected": 0.03591308742761612, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.6629763130792997, | |
| "grad_norm": 194.0, | |
| "learning_rate": 1.6851184346035013e-07, | |
| "logits/chosen": -3.2125000953674316, | |
| "logits/rejected": -3.2249999046325684, | |
| "logps/chosen": -317.3999938964844, | |
| "logps/rejected": -232.60000610351562, | |
| "loss": 0.5773, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.287353515625, | |
| "rewards/margins": 0.34062498807907104, | |
| "rewards/rejected": -0.05349121242761612, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.6642636457260556, | |
| "grad_norm": 256.0, | |
| "learning_rate": 1.6786817713697217e-07, | |
| "logits/chosen": -3.2249999046325684, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -340.3999938964844, | |
| "logps/rejected": -346.0, | |
| "loss": 0.6102, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.12221374362707138, | |
| "rewards/margins": 0.28691405057907104, | |
| "rewards/rejected": -0.16484375298023224, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.6655509783728115, | |
| "grad_norm": 175.0, | |
| "learning_rate": 1.6722451081359424e-07, | |
| "logits/chosen": -3.3218750953674316, | |
| "logits/rejected": -3.2750000953674316, | |
| "logps/chosen": -247.60000610351562, | |
| "logps/rejected": -263.6000061035156, | |
| "loss": 0.5922, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.11604003608226776, | |
| "rewards/margins": 0.2845703065395355, | |
| "rewards/rejected": -0.16790771484375, | |
| "step": 2585 | |
| }, | |
| { | |
| "epoch": 0.6668383110195675, | |
| "grad_norm": 233.0, | |
| "learning_rate": 1.6658084449021626e-07, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.1937499046325684, | |
| "logps/chosen": -237.60000610351562, | |
| "logps/rejected": -194.3000030517578, | |
| "loss": 0.5945, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.2515625059604645, | |
| "rewards/margins": 0.2845703065395355, | |
| "rewards/rejected": -0.03293456882238388, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.6681256436663234, | |
| "grad_norm": 243.0, | |
| "learning_rate": 1.659371781668383e-07, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.2718749046325684, | |
| "logps/chosen": -264.79998779296875, | |
| "logps/rejected": -253.3000030517578, | |
| "loss": 0.6066, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.21767577528953552, | |
| "rewards/margins": 0.3008789122104645, | |
| "rewards/rejected": -0.08293457329273224, | |
| "step": 2595 | |
| }, | |
| { | |
| "epoch": 0.6694129763130793, | |
| "grad_norm": 159.0, | |
| "learning_rate": 1.6529351184346034e-07, | |
| "logits/chosen": -3.237499952316284, | |
| "logits/rejected": -3.2874999046325684, | |
| "logps/chosen": -313.6000061035156, | |
| "logps/rejected": -257.3999938964844, | |
| "loss": 0.5488, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.24384765326976776, | |
| "rewards/margins": 0.4052734375, | |
| "rewards/rejected": -0.16083984076976776, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.6707003089598352, | |
| "grad_norm": 253.0, | |
| "learning_rate": 1.646498455200824e-07, | |
| "logits/chosen": -3.078125, | |
| "logits/rejected": -3.106250047683716, | |
| "logps/chosen": -252.39999389648438, | |
| "logps/rejected": -249.39999389648438, | |
| "loss": 0.5945, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.15615233778953552, | |
| "rewards/margins": 0.2783203125, | |
| "rewards/rejected": -0.12250976264476776, | |
| "step": 2605 | |
| }, | |
| { | |
| "epoch": 0.6719876416065912, | |
| "grad_norm": 286.0, | |
| "learning_rate": 1.6400617919670443e-07, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": NaN, | |
| "logps/chosen": -278.5, | |
| "logps/rejected": -210.39999389648438, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.17027588188648224, | |
| "rewards/margins": 0.19407348334789276, | |
| "rewards/rejected": -0.02363281324505806, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.673274974253347, | |
| "grad_norm": 164.0, | |
| "learning_rate": 1.6336251287332645e-07, | |
| "logits/chosen": -3.143749952316284, | |
| "logits/rejected": -3.2562499046325684, | |
| "logps/chosen": -315.0, | |
| "logps/rejected": -278.6000061035156, | |
| "loss": 0.6328, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.16367188096046448, | |
| "rewards/margins": 0.2591796815395355, | |
| "rewards/rejected": -0.09555663913488388, | |
| "step": 2615 | |
| }, | |
| { | |
| "epoch": 0.674562306900103, | |
| "grad_norm": 227.0, | |
| "learning_rate": 1.627188465499485e-07, | |
| "logits/chosen": -3.0843749046325684, | |
| "logits/rejected": -3.1500000953674316, | |
| "logps/chosen": -274.3999938964844, | |
| "logps/rejected": -293.3999938964844, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.20078125596046448, | |
| "rewards/margins": 0.24541015923023224, | |
| "rewards/rejected": -0.0445556640625, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.675849639546859, | |
| "grad_norm": 153.0, | |
| "learning_rate": 1.6207518022657056e-07, | |
| "logits/chosen": -3.081249952316284, | |
| "logits/rejected": -2.7054686546325684, | |
| "logps/chosen": -217.74374389648438, | |
| "logps/rejected": -170.0, | |
| "loss": 0.6398, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 0.22426757216453552, | |
| "rewards/margins": 0.19833984971046448, | |
| "rewards/rejected": 0.026611328125, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.6771369721936148, | |
| "grad_norm": 199.0, | |
| "learning_rate": 1.6143151390319257e-07, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.2593750953674316, | |
| "logps/chosen": -255.39999389648438, | |
| "logps/rejected": -241.8000030517578, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.01312866248190403, | |
| "rewards/margins": 0.24882812798023224, | |
| "rewards/rejected": -0.23544922471046448, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.6784243048403708, | |
| "grad_norm": 175.0, | |
| "learning_rate": 1.6078784757981462e-07, | |
| "logits/chosen": -3.0406250953674316, | |
| "logits/rejected": -3.059375047683716, | |
| "logps/chosen": -228.5, | |
| "logps/rejected": -178.1999969482422, | |
| "loss": 0.6242, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.19785156846046448, | |
| "rewards/margins": 0.22812500596046448, | |
| "rewards/rejected": -0.02994995191693306, | |
| "step": 2635 | |
| }, | |
| { | |
| "epoch": 0.6797116374871267, | |
| "grad_norm": 192.0, | |
| "learning_rate": 1.6014418125643666e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.28125, | |
| "logps/chosen": -279.20001220703125, | |
| "logps/rejected": -256.6000061035156, | |
| "loss": 0.5855, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.32428282499313354, | |
| "rewards/margins": 0.3451171815395355, | |
| "rewards/rejected": -0.02031249925494194, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.6809989701338826, | |
| "grad_norm": 158.0, | |
| "learning_rate": 1.5950051493305868e-07, | |
| "logits/chosen": -2.971874952316284, | |
| "logits/rejected": -3.018749952316284, | |
| "logps/chosen": -200.6999969482422, | |
| "logps/rejected": -172.60000610351562, | |
| "loss": 0.6211, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.21367187798023224, | |
| "rewards/margins": 0.20864257216453552, | |
| "rewards/rejected": 0.0049682618118822575, | |
| "step": 2645 | |
| }, | |
| { | |
| "epoch": 0.6822863027806385, | |
| "grad_norm": 225.0, | |
| "learning_rate": 1.5885684860968075e-07, | |
| "logits/chosen": -3.296875, | |
| "logits/rejected": -3.3031249046325684, | |
| "logps/chosen": -281.79998779296875, | |
| "logps/rejected": -249.39999389648438, | |
| "loss": 0.5992, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.09018554538488388, | |
| "rewards/margins": 0.28730469942092896, | |
| "rewards/rejected": -0.19697265326976776, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.6835736354273945, | |
| "grad_norm": 205.0, | |
| "learning_rate": 1.5821318228630276e-07, | |
| "logits/chosen": -3.265625, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -287.79998779296875, | |
| "logps/rejected": -282.20001220703125, | |
| "loss": 0.5852, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.15566405653953552, | |
| "rewards/margins": 0.34746092557907104, | |
| "rewards/rejected": -0.19218750298023224, | |
| "step": 2655 | |
| }, | |
| { | |
| "epoch": 0.6848609680741503, | |
| "grad_norm": 169.0, | |
| "learning_rate": 1.575695159629248e-07, | |
| "logits/chosen": -3.2750000953674316, | |
| "logits/rejected": -3.3031249046325684, | |
| "logps/chosen": -335.6000061035156, | |
| "logps/rejected": -245.1999969482422, | |
| "loss": 0.5977, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.05734863132238388, | |
| "rewards/margins": 0.3460937440395355, | |
| "rewards/rejected": -0.289306640625, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.6861483007209063, | |
| "grad_norm": 214.0, | |
| "learning_rate": 1.5692584963954685e-07, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -234.6999969482422, | |
| "logps/rejected": -227.1999969482422, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.17617186903953552, | |
| "rewards/margins": 0.23095703125, | |
| "rewards/rejected": -0.05429687350988388, | |
| "step": 2665 | |
| }, | |
| { | |
| "epoch": 0.6874356333676622, | |
| "grad_norm": 185.0, | |
| "learning_rate": 1.562821833161689e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.309375047683716, | |
| "logps/chosen": -269.20001220703125, | |
| "logps/rejected": -241.1999969482422, | |
| "loss": 0.6273, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.13525390625, | |
| "rewards/margins": 0.24384765326976776, | |
| "rewards/rejected": -0.10883788764476776, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.6887229660144182, | |
| "grad_norm": 186.0, | |
| "learning_rate": 1.5563851699279093e-07, | |
| "logits/chosen": -3.237499952316284, | |
| "logits/rejected": -3.1812500953674316, | |
| "logps/chosen": -253.0, | |
| "logps/rejected": -238.0, | |
| "loss": 0.6035, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.20380859076976776, | |
| "rewards/margins": 0.2653137147426605, | |
| "rewards/rejected": -0.06123046949505806, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.690010298661174, | |
| "grad_norm": 235.0, | |
| "learning_rate": 1.5499485066941297e-07, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.262500047683716, | |
| "logps/chosen": -315.79998779296875, | |
| "logps/rejected": -310.3999938964844, | |
| "loss": 0.5875, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.12932129204273224, | |
| "rewards/margins": 0.35205078125, | |
| "rewards/rejected": -0.22285155951976776, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.69129763130793, | |
| "grad_norm": 176.0, | |
| "learning_rate": 1.54351184346035e-07, | |
| "logits/chosen": -3.1343750953674316, | |
| "logits/rejected": -3.0562500953674316, | |
| "logps/chosen": -253.39999389648438, | |
| "logps/rejected": -256.20001220703125, | |
| "loss": 0.6117, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.17226561903953552, | |
| "rewards/margins": 0.24160155653953552, | |
| "rewards/rejected": -0.06953124701976776, | |
| "step": 2685 | |
| }, | |
| { | |
| "epoch": 0.6925849639546859, | |
| "grad_norm": 171.0, | |
| "learning_rate": 1.5370751802265706e-07, | |
| "logits/chosen": -3.1656250953674316, | |
| "logits/rejected": -3.1781249046325684, | |
| "logps/chosen": -301.20001220703125, | |
| "logps/rejected": -277.20001220703125, | |
| "loss": 0.5695, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.221435546875, | |
| "rewards/margins": 0.3511718809604645, | |
| "rewards/rejected": -0.12939453125, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.6938722966014418, | |
| "grad_norm": 220.0, | |
| "learning_rate": 1.5306385169927908e-07, | |
| "logits/chosen": -3.0999999046325684, | |
| "logits/rejected": -3.125, | |
| "logps/chosen": -304.0, | |
| "logps/rejected": -229.0, | |
| "loss": 0.6301, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.19176940619945526, | |
| "rewards/margins": 0.26225584745407104, | |
| "rewards/rejected": -0.07011719048023224, | |
| "step": 2695 | |
| }, | |
| { | |
| "epoch": 0.6951596292481977, | |
| "grad_norm": 272.0, | |
| "learning_rate": 1.5242018537590112e-07, | |
| "logits/chosen": -2.9906249046325684, | |
| "logits/rejected": -2.549999952316284, | |
| "logps/chosen": -255.39999389648438, | |
| "logps/rejected": -243.1999969482422, | |
| "loss": 0.6234, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.2697997987270355, | |
| "rewards/margins": 0.2080078125, | |
| "rewards/rejected": 0.06196289137005806, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.6964469618949537, | |
| "grad_norm": 183.0, | |
| "learning_rate": 1.5177651905252316e-07, | |
| "logits/chosen": -3.0875000953674316, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -297.0, | |
| "logps/rejected": -237.39999389648438, | |
| "loss": 0.6102, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.24013671278953552, | |
| "rewards/margins": 0.26679688692092896, | |
| "rewards/rejected": -0.0263671875, | |
| "step": 2705 | |
| }, | |
| { | |
| "epoch": 0.6977342945417095, | |
| "grad_norm": 201.0, | |
| "learning_rate": 1.511328527291452e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.2093749046325684, | |
| "logps/chosen": -298.3999938964844, | |
| "logps/rejected": -270.6000061035156, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.14985351264476776, | |
| "rewards/margins": 0.236572265625, | |
| "rewards/rejected": -0.08706054836511612, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.6990216271884655, | |
| "grad_norm": 205.0, | |
| "learning_rate": 1.5048918640576725e-07, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.106250047683716, | |
| "logps/chosen": -282.3999938964844, | |
| "logps/rejected": -284.79998779296875, | |
| "loss": 0.575, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.09931640326976776, | |
| "rewards/margins": 0.373046875, | |
| "rewards/rejected": -0.2734375, | |
| "step": 2715 | |
| }, | |
| { | |
| "epoch": 0.7003089598352215, | |
| "grad_norm": 254.0, | |
| "learning_rate": 1.498455200823893e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.2562499046325684, | |
| "logps/chosen": -330.79998779296875, | |
| "logps/rejected": -273.6000061035156, | |
| "loss": 0.6242, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.270498663187027, | |
| "rewards/margins": 0.2603515684604645, | |
| "rewards/rejected": 0.01022949256002903, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.7015962924819773, | |
| "grad_norm": 197.0, | |
| "learning_rate": 1.492018537590113e-07, | |
| "logits/chosen": -3.265625, | |
| "logits/rejected": -3.315624952316284, | |
| "logps/chosen": -301.79998779296875, | |
| "logps/rejected": -230.1999969482422, | |
| "loss": 0.5953, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.27373045682907104, | |
| "rewards/margins": 0.3296875059604645, | |
| "rewards/rejected": -0.05585937574505806, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.7028836251287333, | |
| "grad_norm": 222.0, | |
| "learning_rate": 1.4855818743563338e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.253124952316284, | |
| "logps/chosen": -291.20001220703125, | |
| "logps/rejected": -259.3999938964844, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.3197265565395355, | |
| "rewards/margins": 0.2008056640625, | |
| "rewards/rejected": 0.11953125149011612, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.7041709577754892, | |
| "grad_norm": 241.0, | |
| "learning_rate": 1.4791452111225542e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.215625047683716, | |
| "logps/chosen": -268.3999938964844, | |
| "logps/rejected": -258.0, | |
| "loss": 0.6094, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.181884765625, | |
| "rewards/margins": 0.3034912049770355, | |
| "rewards/rejected": -0.1224365234375, | |
| "step": 2735 | |
| }, | |
| { | |
| "epoch": 0.7054582904222451, | |
| "grad_norm": 232.0, | |
| "learning_rate": 1.4727085478887743e-07, | |
| "logits/chosen": -3.1624999046325684, | |
| "logits/rejected": -3.1812500953674316, | |
| "logps/chosen": -264.79998779296875, | |
| "logps/rejected": -230.1999969482422, | |
| "loss": 0.6203, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.168701171875, | |
| "rewards/margins": 0.2572265565395355, | |
| "rewards/rejected": -0.08847656100988388, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.706745623069001, | |
| "grad_norm": 173.0, | |
| "learning_rate": 1.4662718846549948e-07, | |
| "logits/chosen": -3.2750000953674316, | |
| "logits/rejected": -3.1937499046325684, | |
| "logps/chosen": -268.79998779296875, | |
| "logps/rejected": -259.6000061035156, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.13642577826976776, | |
| "rewards/margins": 0.21910400688648224, | |
| "rewards/rejected": -0.08266601711511612, | |
| "step": 2745 | |
| }, | |
| { | |
| "epoch": 0.708032955715757, | |
| "grad_norm": 210.0, | |
| "learning_rate": 1.4598352214212152e-07, | |
| "logits/chosen": -3.159374952316284, | |
| "logits/rejected": -2.953125, | |
| "logps/chosen": -293.20001220703125, | |
| "logps/rejected": -273.6000061035156, | |
| "loss": 0.6773, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.1478721648454666, | |
| "rewards/margins": 0.11484374850988388, | |
| "rewards/rejected": 0.03281249850988388, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.7093202883625128, | |
| "grad_norm": 179.0, | |
| "learning_rate": 1.4533985581874356e-07, | |
| "logits/chosen": -3.143749952316284, | |
| "logits/rejected": -3.253124952316284, | |
| "logps/chosen": -265.6000061035156, | |
| "logps/rejected": -233.0, | |
| "loss": 0.6156, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.28515625, | |
| "rewards/margins": 0.25468748807907104, | |
| "rewards/rejected": 0.03103027306497097, | |
| "step": 2755 | |
| }, | |
| { | |
| "epoch": 0.7106076210092688, | |
| "grad_norm": 192.0, | |
| "learning_rate": 1.446961894953656e-07, | |
| "logits/chosen": -3.106250047683716, | |
| "logits/rejected": -3.174999952316284, | |
| "logps/chosen": -234.8000030517578, | |
| "logps/rejected": -235.8000030517578, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.09719238430261612, | |
| "rewards/margins": 0.21435546875, | |
| "rewards/rejected": -0.11704101413488388, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.7118949536560247, | |
| "grad_norm": 197.0, | |
| "learning_rate": 1.4405252317198762e-07, | |
| "logits/chosen": -3.081249952316284, | |
| "logits/rejected": -2.832812547683716, | |
| "logps/chosen": -202.5500030517578, | |
| "logps/rejected": -210.3000030517578, | |
| "loss": 0.6406, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.14389649033546448, | |
| "rewards/margins": 0.18117065727710724, | |
| "rewards/rejected": -0.03696288913488388, | |
| "step": 2765 | |
| }, | |
| { | |
| "epoch": 0.7131822863027807, | |
| "grad_norm": 296.0, | |
| "learning_rate": 1.434088568486097e-07, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.1624999046325684, | |
| "logps/chosen": -304.3999938964844, | |
| "logps/rejected": -283.79998779296875, | |
| "loss": 0.6883, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.16162109375, | |
| "rewards/margins": 0.104736328125, | |
| "rewards/rejected": 0.05673827975988388, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.7144696189495365, | |
| "grad_norm": 210.0, | |
| "learning_rate": 1.4276519052523173e-07, | |
| "logits/chosen": -3.0562500953674316, | |
| "logits/rejected": -3.293750047683716, | |
| "logps/chosen": -215.1999969482422, | |
| "logps/rejected": -189.89999389648438, | |
| "loss": 0.7156, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.05120849609375, | |
| "rewards/margins": 0.01025390625, | |
| "rewards/rejected": 0.040771484375, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.7157569515962925, | |
| "grad_norm": 187.0, | |
| "learning_rate": 1.4212152420185375e-07, | |
| "logits/chosen": -3.215625047683716, | |
| "logits/rejected": -3.21875, | |
| "logps/chosen": -306.79998779296875, | |
| "logps/rejected": -256.0, | |
| "loss": 0.5715, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.21845702826976776, | |
| "rewards/margins": 0.3544921875, | |
| "rewards/rejected": -0.13618163764476776, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.7170442842430484, | |
| "grad_norm": 235.0, | |
| "learning_rate": 1.414778578784758e-07, | |
| "logits/chosen": -3.059375047683716, | |
| "logits/rejected": -2.875, | |
| "logps/chosen": -220.0, | |
| "logps/rejected": -250.0, | |
| "loss": 0.6406, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.15087890625, | |
| "rewards/margins": 0.17545166611671448, | |
| "rewards/rejected": -0.02436523512005806, | |
| "step": 2785 | |
| }, | |
| { | |
| "epoch": 0.7183316168898043, | |
| "grad_norm": 195.0, | |
| "learning_rate": 1.408341915550978e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.1937499046325684, | |
| "logps/chosen": -304.20001220703125, | |
| "logps/rejected": -281.0, | |
| "loss": 0.5676, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.25642091035842896, | |
| "rewards/margins": 0.3583984375, | |
| "rewards/rejected": -0.10175780951976776, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.7196189495365602, | |
| "grad_norm": 254.0, | |
| "learning_rate": 1.4019052523171988e-07, | |
| "logits/chosen": -3.143749952316284, | |
| "logits/rejected": -3.1156249046325684, | |
| "logps/chosen": -279.0, | |
| "logps/rejected": -288.6000061035156, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.10868529975414276, | |
| "rewards/margins": 0.30683594942092896, | |
| "rewards/rejected": -0.19814452528953552, | |
| "step": 2795 | |
| }, | |
| { | |
| "epoch": 0.7209062821833162, | |
| "grad_norm": 150.0, | |
| "learning_rate": 1.3954685890834192e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.1968750953674316, | |
| "logps/chosen": -205.1999969482422, | |
| "logps/rejected": -223.60000610351562, | |
| "loss": 0.6156, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.12163086235523224, | |
| "rewards/margins": 0.21054688096046448, | |
| "rewards/rejected": -0.0889892578125, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.722193614830072, | |
| "grad_norm": 213.0, | |
| "learning_rate": 1.3890319258496394e-07, | |
| "logits/chosen": -3.1312499046325684, | |
| "logits/rejected": -3.1187500953674316, | |
| "logps/chosen": -263.20001220703125, | |
| "logps/rejected": -255.10000610351562, | |
| "loss": 0.6883, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.17109374701976776, | |
| "rewards/margins": 0.07288818061351776, | |
| "rewards/rejected": 0.09833984076976776, | |
| "step": 2805 | |
| }, | |
| { | |
| "epoch": 0.723480947476828, | |
| "grad_norm": 167.0, | |
| "learning_rate": 1.3825952626158598e-07, | |
| "logits/chosen": -3.296875, | |
| "logits/rejected": -3.2562499046325684, | |
| "logps/chosen": -252.0, | |
| "logps/rejected": -222.89999389648438, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.17724609375, | |
| "rewards/margins": 0.24296875298023224, | |
| "rewards/rejected": -0.06564941257238388, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.724768280123584, | |
| "grad_norm": 165.0, | |
| "learning_rate": 1.3761585993820805e-07, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.1937499046325684, | |
| "logps/chosen": -279.0, | |
| "logps/rejected": -250.39999389648438, | |
| "loss": 0.6406, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.2177734375, | |
| "rewards/margins": 0.1986083984375, | |
| "rewards/rejected": 0.01894531212747097, | |
| "step": 2815 | |
| }, | |
| { | |
| "epoch": 0.7260556127703398, | |
| "grad_norm": 181.0, | |
| "learning_rate": 1.3697219361483006e-07, | |
| "logits/chosen": -3.0406250953674316, | |
| "logits/rejected": -3.003124952316284, | |
| "logps/chosen": -260.20001220703125, | |
| "logps/rejected": -234.8000030517578, | |
| "loss": 0.6133, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.08548583835363388, | |
| "rewards/margins": 0.26386719942092896, | |
| "rewards/rejected": -0.17869262397289276, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.7273429454170958, | |
| "grad_norm": 168.0, | |
| "learning_rate": 1.363285272914521e-07, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.2874999046325684, | |
| "logps/chosen": -293.20001220703125, | |
| "logps/rejected": -257.3999938964844, | |
| "loss": 0.5977, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.03195800632238388, | |
| "rewards/margins": 0.30732423067092896, | |
| "rewards/rejected": -0.27485352754592896, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.7286302780638517, | |
| "grad_norm": 199.0, | |
| "learning_rate": 1.3568486096807412e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.075000047683716, | |
| "logps/chosen": -291.0, | |
| "logps/rejected": -227.10000610351562, | |
| "loss": 0.566, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.2621093690395355, | |
| "rewards/margins": 0.3773437440395355, | |
| "rewards/rejected": -0.11538086086511612, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.7299176107106076, | |
| "grad_norm": 276.0, | |
| "learning_rate": 1.350411946446962e-07, | |
| "logits/chosen": -3.143749952316284, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -290.20001220703125, | |
| "logps/rejected": -272.0, | |
| "loss": 0.5801, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.18046875298023224, | |
| "rewards/margins": 0.3517089784145355, | |
| "rewards/rejected": -0.17050781846046448, | |
| "step": 2835 | |
| }, | |
| { | |
| "epoch": 0.7312049433573635, | |
| "grad_norm": 205.0, | |
| "learning_rate": 1.3439752832131823e-07, | |
| "logits/chosen": -3.1656250953674316, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -367.20001220703125, | |
| "logps/rejected": -296.0, | |
| "loss": 0.5621, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.29216307401657104, | |
| "rewards/margins": 0.3802734315395355, | |
| "rewards/rejected": -0.08845214545726776, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.7324922760041195, | |
| "grad_norm": 218.0, | |
| "learning_rate": 1.3375386199794025e-07, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -3.328125, | |
| "logps/chosen": -254.0, | |
| "logps/rejected": -205.0, | |
| "loss": 0.6199, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.13925781846046448, | |
| "rewards/margins": 0.28608399629592896, | |
| "rewards/rejected": -0.14638671278953552, | |
| "step": 2845 | |
| }, | |
| { | |
| "epoch": 0.7337796086508754, | |
| "grad_norm": 187.0, | |
| "learning_rate": 1.331101956745623e-07, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -3.1031250953674316, | |
| "logps/chosen": -256.3999938964844, | |
| "logps/rejected": -240.1999969482422, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.13737793266773224, | |
| "rewards/margins": 0.18437500298023224, | |
| "rewards/rejected": -0.04707031324505806, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.7350669412976313, | |
| "grad_norm": 197.0, | |
| "learning_rate": 1.3246652935118436e-07, | |
| "logits/chosen": -3.237499952316284, | |
| "logits/rejected": -3.296875, | |
| "logps/chosen": -281.20001220703125, | |
| "logps/rejected": -239.39999389648438, | |
| "loss": 0.6328, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.10371093451976776, | |
| "rewards/margins": 0.23598632216453552, | |
| "rewards/rejected": -0.13215331733226776, | |
| "step": 2855 | |
| }, | |
| { | |
| "epoch": 0.7363542739443872, | |
| "grad_norm": 115.5, | |
| "learning_rate": 1.3182286302780638e-07, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -236.6999969482422, | |
| "logps/rejected": -205.39999389648438, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.06386718899011612, | |
| "rewards/margins": 0.20868530869483948, | |
| "rewards/rejected": -0.1446533203125, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.7376416065911432, | |
| "grad_norm": 262.0, | |
| "learning_rate": 1.3117919670442842e-07, | |
| "logits/chosen": -3.0562500953674316, | |
| "logits/rejected": -3.125, | |
| "logps/chosen": -213.0, | |
| "logps/rejected": -211.60000610351562, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.05859375, | |
| "rewards/margins": 0.06704101711511612, | |
| "rewards/rejected": -0.008496093563735485, | |
| "step": 2865 | |
| }, | |
| { | |
| "epoch": 0.738928939237899, | |
| "grad_norm": 221.0, | |
| "learning_rate": 1.3053553038105044e-07, | |
| "logits/chosen": -3.171875, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -292.6000061035156, | |
| "logps/rejected": -267.20001220703125, | |
| "loss": 0.6328, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.173583984375, | |
| "rewards/margins": 0.17744140326976776, | |
| "rewards/rejected": -0.004199218936264515, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.740216271884655, | |
| "grad_norm": 216.0, | |
| "learning_rate": 1.298918640576725e-07, | |
| "logits/chosen": -3.28125, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -349.79998779296875, | |
| "logps/rejected": -297.20001220703125, | |
| "loss": 0.5777, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.184326171875, | |
| "rewards/margins": 0.4085937440395355, | |
| "rewards/rejected": -0.22529296576976776, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.741503604531411, | |
| "grad_norm": 236.0, | |
| "learning_rate": 1.2924819773429455e-07, | |
| "logits/chosen": -3.0218749046325684, | |
| "logits/rejected": -2.96875, | |
| "logps/chosen": -233.0, | |
| "logps/rejected": -185.60000610351562, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.11445312201976776, | |
| "rewards/margins": 0.19432373344898224, | |
| "rewards/rejected": -0.08022461086511612, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.7427909371781668, | |
| "grad_norm": 201.0, | |
| "learning_rate": 1.2860453141091657e-07, | |
| "logits/chosen": -3.059375047683716, | |
| "logits/rejected": -3.0843749046325684, | |
| "logps/chosen": -233.1999969482422, | |
| "logps/rejected": -228.60000610351562, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.17636719346046448, | |
| "rewards/margins": 0.14597168564796448, | |
| "rewards/rejected": 0.03078613243997097, | |
| "step": 2885 | |
| }, | |
| { | |
| "epoch": 0.7440782698249228, | |
| "grad_norm": 220.0, | |
| "learning_rate": 1.279608650875386e-07, | |
| "logits/chosen": -3.215625047683716, | |
| "logits/rejected": -3.203125, | |
| "logps/chosen": -269.3999938964844, | |
| "logps/rejected": -268.79998779296875, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.13935546576976776, | |
| "rewards/margins": 0.20600585639476776, | |
| "rewards/rejected": -0.06650390475988388, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.7453656024716787, | |
| "grad_norm": 209.0, | |
| "learning_rate": 1.2731719876416068e-07, | |
| "logits/chosen": -3.0687499046325684, | |
| "logits/rejected": -3.1624999046325684, | |
| "logps/chosen": -359.0, | |
| "logps/rejected": -255.0, | |
| "loss": 0.7203, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.1328125, | |
| "rewards/margins": 0.13798828423023224, | |
| "rewards/rejected": -0.005566406063735485, | |
| "step": 2895 | |
| }, | |
| { | |
| "epoch": 0.7466529351184346, | |
| "grad_norm": 242.0, | |
| "learning_rate": 1.266735324407827e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.2281250953674316, | |
| "logps/chosen": -295.79998779296875, | |
| "logps/rejected": -350.79998779296875, | |
| "loss": 0.6266, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.06953124701976776, | |
| "rewards/margins": 0.20888671278953552, | |
| "rewards/rejected": -0.13896484673023224, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.7479402677651905, | |
| "grad_norm": 226.0, | |
| "learning_rate": 1.2602986611740474e-07, | |
| "logits/chosen": -3.0843749046325684, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -279.6000061035156, | |
| "logps/rejected": -217.60000610351562, | |
| "loss": 0.6078, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.15292969346046448, | |
| "rewards/margins": 0.2955078184604645, | |
| "rewards/rejected": -0.1429443359375, | |
| "step": 2905 | |
| }, | |
| { | |
| "epoch": 0.7492276004119465, | |
| "grad_norm": 207.0, | |
| "learning_rate": 1.2538619979402675e-07, | |
| "logits/chosen": -2.971874952316284, | |
| "logits/rejected": -2.8125, | |
| "logps/chosen": -219.10000610351562, | |
| "logps/rejected": -178.89999389648438, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.2874999940395355, | |
| "rewards/margins": 0.24589844048023224, | |
| "rewards/rejected": 0.04213256761431694, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.7505149330587023, | |
| "grad_norm": 167.0, | |
| "learning_rate": 1.247425334706488e-07, | |
| "logits/chosen": -3.2406249046325684, | |
| "logits/rejected": -3.1156249046325684, | |
| "logps/chosen": -233.60000610351562, | |
| "logps/rejected": -232.0, | |
| "loss": 0.6242, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.12187500298023224, | |
| "rewards/margins": 0.22260741889476776, | |
| "rewards/rejected": -0.10065917670726776, | |
| "step": 2915 | |
| }, | |
| { | |
| "epoch": 0.7518022657054583, | |
| "grad_norm": 236.0, | |
| "learning_rate": 1.2409886714727084e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.2093749046325684, | |
| "logps/chosen": -321.79998779296875, | |
| "logps/rejected": -274.79998779296875, | |
| "loss": 0.5941, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.33713072538375854, | |
| "rewards/margins": 0.3564209043979645, | |
| "rewards/rejected": -0.01933593675494194, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.7530895983522142, | |
| "grad_norm": 233.0, | |
| "learning_rate": 1.2345520082389288e-07, | |
| "logits/chosen": -3.1656250953674316, | |
| "logits/rejected": -3.137500047683716, | |
| "logps/chosen": -289.3999938964844, | |
| "logps/rejected": -256.0, | |
| "loss": 0.6188, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.18232421576976776, | |
| "rewards/margins": 0.24062499403953552, | |
| "rewards/rejected": -0.057861328125, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.7543769309989702, | |
| "grad_norm": 205.0, | |
| "learning_rate": 1.2281153450051492e-07, | |
| "logits/chosen": -3.0531249046325684, | |
| "logits/rejected": -3.0406250953674316, | |
| "logps/chosen": -274.0, | |
| "logps/rejected": -249.8000030517578, | |
| "loss": 0.6531, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 0.191914364695549, | |
| "rewards/margins": 0.14501953125, | |
| "rewards/rejected": 0.04692382737994194, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.755664263645726, | |
| "grad_norm": 193.0, | |
| "learning_rate": 1.2216786817713697e-07, | |
| "logits/chosen": -3.231250047683716, | |
| "logits/rejected": -3.231250047683716, | |
| "logps/chosen": -263.3999938964844, | |
| "logps/rejected": -220.6999969482422, | |
| "loss": 0.582, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.16728515923023224, | |
| "rewards/margins": 0.2789062559604645, | |
| "rewards/rejected": -0.11127624660730362, | |
| "step": 2935 | |
| }, | |
| { | |
| "epoch": 0.756951596292482, | |
| "grad_norm": 208.0, | |
| "learning_rate": 1.21524201853759e-07, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.2125000953674316, | |
| "logps/chosen": -317.0, | |
| "logps/rejected": -267.70001220703125, | |
| "loss": 0.6227, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.29707032442092896, | |
| "rewards/margins": 0.20937499403953552, | |
| "rewards/rejected": 0.08777771145105362, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.7582389289392379, | |
| "grad_norm": 247.0, | |
| "learning_rate": 1.2088053553038105e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.065624952316284, | |
| "logps/chosen": -272.79998779296875, | |
| "logps/rejected": -252.60000610351562, | |
| "loss": 0.668, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.18734130263328552, | |
| "rewards/margins": 0.14833983778953552, | |
| "rewards/rejected": 0.03929443284869194, | |
| "step": 2945 | |
| }, | |
| { | |
| "epoch": 0.7595262615859938, | |
| "grad_norm": 170.0, | |
| "learning_rate": 1.202368692070031e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -275.6000061035156, | |
| "logps/rejected": -247.1999969482422, | |
| "loss": 0.5695, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.1568603515625, | |
| "rewards/margins": 0.36054688692092896, | |
| "rewards/rejected": -0.20375975966453552, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.7608135942327497, | |
| "grad_norm": 221.0, | |
| "learning_rate": 1.195932028836251e-07, | |
| "logits/chosen": -3.096874952316284, | |
| "logits/rejected": -3.1968750953674316, | |
| "logps/chosen": -277.0, | |
| "logps/rejected": -238.8000030517578, | |
| "loss": 0.618, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.21791991591453552, | |
| "rewards/margins": 0.2508300840854645, | |
| "rewards/rejected": -0.03251953050494194, | |
| "step": 2955 | |
| }, | |
| { | |
| "epoch": 0.7621009268795057, | |
| "grad_norm": 236.0, | |
| "learning_rate": 1.1894953656024715e-07, | |
| "logits/chosen": -3.1624999046325684, | |
| "logits/rejected": -3.0999999046325684, | |
| "logps/chosen": -254.8000030517578, | |
| "logps/rejected": -271.0, | |
| "loss": 0.6453, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.06796874850988388, | |
| "rewards/margins": 0.22634276747703552, | |
| "rewards/rejected": -0.1583251953125, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.7633882595262615, | |
| "grad_norm": 158.0, | |
| "learning_rate": 1.1830587023686921e-07, | |
| "logits/chosen": -3.2750000953674316, | |
| "logits/rejected": -3.1812500953674316, | |
| "logps/chosen": -265.0, | |
| "logps/rejected": -218.3000030517578, | |
| "loss": 0.6125, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.28300780057907104, | |
| "rewards/margins": 0.22773437201976776, | |
| "rewards/rejected": 0.05532226711511612, | |
| "step": 2965 | |
| }, | |
| { | |
| "epoch": 0.7646755921730175, | |
| "grad_norm": 179.0, | |
| "learning_rate": 1.1766220391349124e-07, | |
| "logits/chosen": -2.831249952316284, | |
| "logits/rejected": -3.028125047683716, | |
| "logps/chosen": -305.20001220703125, | |
| "logps/rejected": -209.89999389648438, | |
| "loss": 0.5762, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.11469726264476776, | |
| "rewards/margins": 0.32744139432907104, | |
| "rewards/rejected": -0.21269531548023224, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.7659629248197735, | |
| "grad_norm": 272.0, | |
| "learning_rate": 1.1701853759011328e-07, | |
| "logits/chosen": -3.206249952316284, | |
| "logits/rejected": -3.206249952316284, | |
| "logps/chosen": -307.20001220703125, | |
| "logps/rejected": -265.6000061035156, | |
| "loss": 0.6438, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.14497070014476776, | |
| "rewards/margins": 0.19462890923023224, | |
| "rewards/rejected": -0.04965820163488388, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.7672502574665293, | |
| "grad_norm": 169.0, | |
| "learning_rate": 1.1637487126673531e-07, | |
| "logits/chosen": -3.0687499046325684, | |
| "logits/rejected": -3.1156249046325684, | |
| "logps/chosen": -289.6000061035156, | |
| "logps/rejected": -234.6999969482422, | |
| "loss": 0.5363, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.2718749940395355, | |
| "rewards/margins": 0.43769532442092896, | |
| "rewards/rejected": -0.16616210341453552, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.7685375901132853, | |
| "grad_norm": 151.0, | |
| "learning_rate": 1.1573120494335737e-07, | |
| "logits/chosen": -3.03125, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -248.0, | |
| "logps/rejected": -202.10000610351562, | |
| "loss": 0.6289, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.22539062798023224, | |
| "rewards/margins": 0.22050170600414276, | |
| "rewards/rejected": 0.0052246092818677425, | |
| "step": 2985 | |
| }, | |
| { | |
| "epoch": 0.7698249227600412, | |
| "grad_norm": 162.0, | |
| "learning_rate": 1.150875386199794e-07, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.143749952316284, | |
| "logps/chosen": -268.0, | |
| "logps/rejected": -185.6999969482422, | |
| "loss": 0.6219, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.31132811307907104, | |
| "rewards/margins": 0.20234374701976776, | |
| "rewards/rejected": 0.10885925590991974, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.7711122554067971, | |
| "grad_norm": 235.0, | |
| "learning_rate": 1.1444387229660144e-07, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.171875, | |
| "logps/chosen": -260.20001220703125, | |
| "logps/rejected": -239.1999969482422, | |
| "loss": 0.6094, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.23457030951976776, | |
| "rewards/margins": 0.21171875298023224, | |
| "rewards/rejected": 0.02299804612994194, | |
| "step": 2995 | |
| }, | |
| { | |
| "epoch": 0.772399588053553, | |
| "grad_norm": 260.0, | |
| "learning_rate": 1.1380020597322347e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -336.0, | |
| "logps/rejected": -281.20001220703125, | |
| "loss": 0.6555, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.31904298067092896, | |
| "rewards/margins": 0.17702636122703552, | |
| "rewards/rejected": 0.1416015625, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.773686920700309, | |
| "grad_norm": 216.0, | |
| "learning_rate": 1.1315653964984552e-07, | |
| "logits/chosen": -3.121875047683716, | |
| "logits/rejected": -3.137500047683716, | |
| "logps/chosen": -314.0, | |
| "logps/rejected": -249.1999969482422, | |
| "loss": 0.6086, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.23691406846046448, | |
| "rewards/margins": 0.28203123807907104, | |
| "rewards/rejected": -0.04509887844324112, | |
| "step": 3005 | |
| }, | |
| { | |
| "epoch": 0.7749742533470649, | |
| "grad_norm": 224.0, | |
| "learning_rate": 1.1251287332646755e-07, | |
| "logits/chosen": -2.901562452316284, | |
| "logits/rejected": -2.9000000953674316, | |
| "logps/chosen": -257.8999938964844, | |
| "logps/rejected": -232.60000610351562, | |
| "loss": 0.5727, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.16884765028953552, | |
| "rewards/margins": 0.34164732694625854, | |
| "rewards/rejected": -0.17307129502296448, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.7762615859938208, | |
| "grad_norm": 1168.0, | |
| "learning_rate": 1.118692070030896e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.0875000953674316, | |
| "logps/chosen": -282.79998779296875, | |
| "logps/rejected": -253.1999969482422, | |
| "loss": 0.6344, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.2674804627895355, | |
| "rewards/margins": 0.224853515625, | |
| "rewards/rejected": 0.04257812350988388, | |
| "step": 3015 | |
| }, | |
| { | |
| "epoch": 0.7775489186405767, | |
| "grad_norm": 134.0, | |
| "learning_rate": 1.1122554067971163e-07, | |
| "logits/chosen": -3.3187499046325684, | |
| "logits/rejected": -3.387500047683716, | |
| "logps/chosen": -283.3999938964844, | |
| "logps/rejected": -193.10000610351562, | |
| "loss": 0.6098, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.10302734375, | |
| "rewards/margins": 0.3109374940395355, | |
| "rewards/rejected": -0.20771484076976776, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.7788362512873327, | |
| "grad_norm": 300.0, | |
| "learning_rate": 1.1058187435633368e-07, | |
| "logits/chosen": -3.121875047683716, | |
| "logits/rejected": -3.237499952316284, | |
| "logps/chosen": -298.6000061035156, | |
| "logps/rejected": -241.39999389648438, | |
| "loss": 0.5805, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.304443359375, | |
| "rewards/margins": 0.32988280057907104, | |
| "rewards/rejected": -0.02508544921875, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.7801235839340885, | |
| "grad_norm": 212.0, | |
| "learning_rate": 1.0993820803295571e-07, | |
| "logits/chosen": -3.215625047683716, | |
| "logits/rejected": -3.090625047683716, | |
| "logps/chosen": -293.79998779296875, | |
| "logps/rejected": -266.20001220703125, | |
| "loss": 0.6078, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.17001953721046448, | |
| "rewards/margins": 0.2769531309604645, | |
| "rewards/rejected": -0.10675048828125, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.7814109165808445, | |
| "grad_norm": 254.0, | |
| "learning_rate": 1.0929454170957775e-07, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.203125, | |
| "logps/chosen": -258.5, | |
| "logps/rejected": -205.77499389648438, | |
| "loss": 0.5887, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.0811767578125, | |
| "rewards/margins": 0.33966064453125, | |
| "rewards/rejected": -0.2586303651332855, | |
| "step": 3035 | |
| }, | |
| { | |
| "epoch": 0.7826982492276005, | |
| "grad_norm": 172.0, | |
| "learning_rate": 1.0865087538619978e-07, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -3.143749952316284, | |
| "logps/chosen": -233.1999969482422, | |
| "logps/rejected": -263.6000061035156, | |
| "loss": 0.5945, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.08522949367761612, | |
| "rewards/margins": 0.3082031309604645, | |
| "rewards/rejected": -0.22207030653953552, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.7839855818743563, | |
| "grad_norm": 233.0, | |
| "learning_rate": 1.0800720906282184e-07, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.168750047683716, | |
| "logps/chosen": -261.5, | |
| "logps/rejected": -281.79998779296875, | |
| "loss": 0.6543, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.142333984375, | |
| "rewards/margins": 0.17509765923023224, | |
| "rewards/rejected": -0.03281249850988388, | |
| "step": 3045 | |
| }, | |
| { | |
| "epoch": 0.7852729145211123, | |
| "grad_norm": 5024.0, | |
| "learning_rate": 1.0736354273944387e-07, | |
| "logits/chosen": -3.121875047683716, | |
| "logits/rejected": -3.121875047683716, | |
| "logps/chosen": -214.1999969482422, | |
| "logps/rejected": -212.1999969482422, | |
| "loss": 0.8156, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.14052733778953552, | |
| "rewards/margins": -0.02724609337747097, | |
| "rewards/rejected": 0.16826172173023224, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.7865602471678682, | |
| "grad_norm": 173.0, | |
| "learning_rate": 1.067198764160659e-07, | |
| "logits/chosen": -3.206249952316284, | |
| "logits/rejected": -3.096874952316284, | |
| "logps/chosen": -299.20001220703125, | |
| "logps/rejected": -308.6000061035156, | |
| "loss": 0.6484, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.28862303495407104, | |
| "rewards/margins": 0.21522216498851776, | |
| "rewards/rejected": 0.073486328125, | |
| "step": 3055 | |
| }, | |
| { | |
| "epoch": 0.787847579814624, | |
| "grad_norm": 210.0, | |
| "learning_rate": 1.0607621009268794e-07, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.246875047683716, | |
| "logps/chosen": -300.79998779296875, | |
| "logps/rejected": -267.3999938964844, | |
| "loss": 0.5777, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.2520507872104645, | |
| "rewards/margins": 0.36894530057907104, | |
| "rewards/rejected": -0.1162109375, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.78913491246138, | |
| "grad_norm": 212.0, | |
| "learning_rate": 1.0543254376930998e-07, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.106250047683716, | |
| "logps/chosen": -254.1999969482422, | |
| "logps/rejected": -238.1999969482422, | |
| "loss": 0.6055, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.1461181640625, | |
| "rewards/margins": 0.25712889432907104, | |
| "rewards/rejected": -0.11127929389476776, | |
| "step": 3065 | |
| }, | |
| { | |
| "epoch": 0.790422245108136, | |
| "grad_norm": 249.0, | |
| "learning_rate": 1.0478887744593203e-07, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -3.0625, | |
| "logps/chosen": -274.20001220703125, | |
| "logps/rejected": -209.0, | |
| "loss": 0.6406, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.34589844942092896, | |
| "rewards/margins": 0.20351561903953552, | |
| "rewards/rejected": 0.14287109673023224, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.7917095777548918, | |
| "grad_norm": 227.0, | |
| "learning_rate": 1.0414521112255406e-07, | |
| "logits/chosen": -3.159374952316284, | |
| "logits/rejected": -3.046875, | |
| "logps/chosen": -255.60000610351562, | |
| "logps/rejected": -222.3000030517578, | |
| "loss": 0.627, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.17460937798023224, | |
| "rewards/margins": 0.22822265326976776, | |
| "rewards/rejected": -0.05366211012005806, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.7929969104016478, | |
| "grad_norm": 230.0, | |
| "learning_rate": 1.035015447991761e-07, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -3.0218749046325684, | |
| "logps/chosen": -232.60000610351562, | |
| "logps/rejected": -242.8000030517578, | |
| "loss": 0.6203, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.17025145888328552, | |
| "rewards/margins": 0.20881347358226776, | |
| "rewards/rejected": -0.03837890550494194, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.7942842430484037, | |
| "grad_norm": 186.0, | |
| "learning_rate": 1.0285787847579814e-07, | |
| "logits/chosen": -3.2093749046325684, | |
| "logits/rejected": -3.25, | |
| "logps/chosen": -294.20001220703125, | |
| "logps/rejected": -263.20001220703125, | |
| "loss": 0.6641, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.173828125, | |
| "rewards/margins": 0.1417800933122635, | |
| "rewards/rejected": 0.03212890774011612, | |
| "step": 3085 | |
| }, | |
| { | |
| "epoch": 0.7955715756951597, | |
| "grad_norm": 179.0, | |
| "learning_rate": 1.0221421215242018e-07, | |
| "logits/chosen": -3.2562499046325684, | |
| "logits/rejected": -3.2906250953674316, | |
| "logps/chosen": -342.79998779296875, | |
| "logps/rejected": -294.79998779296875, | |
| "loss": 0.5773, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.11787109076976776, | |
| "rewards/margins": 0.362548828125, | |
| "rewards/rejected": -0.24560546875, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.7968589083419155, | |
| "grad_norm": 194.0, | |
| "learning_rate": 1.0157054582904221e-07, | |
| "logits/chosen": -3.0718750953674316, | |
| "logits/rejected": -2.887500047683716, | |
| "logps/chosen": -164.25, | |
| "logps/rejected": -157.1750030517578, | |
| "loss": 0.6297, | |
| "rewards/accuracies": 0.38749998807907104, | |
| "rewards/chosen": 0.12177734076976776, | |
| "rewards/margins": 0.17550048232078552, | |
| "rewards/rejected": -0.05329589918255806, | |
| "step": 3095 | |
| }, | |
| { | |
| "epoch": 0.7981462409886715, | |
| "grad_norm": 207.0, | |
| "learning_rate": 1.0092687950566426e-07, | |
| "logits/chosen": -3.1812500953674316, | |
| "logits/rejected": -3.2562499046325684, | |
| "logps/chosen": -246.0, | |
| "logps/rejected": -212.89999389648438, | |
| "loss": 0.5672, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.06630859524011612, | |
| "rewards/margins": 0.3277343809604645, | |
| "rewards/rejected": -0.26093751192092896, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.7994335736354274, | |
| "grad_norm": 328.0, | |
| "learning_rate": 1.002832131822863e-07, | |
| "logits/chosen": -2.859375, | |
| "logits/rejected": -2.7718749046325684, | |
| "logps/chosen": -263.6000061035156, | |
| "logps/rejected": -242.0, | |
| "loss": 0.6617, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.18535156548023224, | |
| "rewards/margins": 0.12255859375, | |
| "rewards/rejected": 0.06303711235523224, | |
| "step": 3105 | |
| }, | |
| { | |
| "epoch": 0.8007209062821833, | |
| "grad_norm": 186.0, | |
| "learning_rate": 9.963954685890834e-08, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.215625047683716, | |
| "logps/chosen": -276.0, | |
| "logps/rejected": -267.79998779296875, | |
| "loss": 0.5898, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.20708008110523224, | |
| "rewards/margins": 0.3060546815395355, | |
| "rewards/rejected": -0.09881591796875, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.8020082389289392, | |
| "grad_norm": 160.0, | |
| "learning_rate": 9.899588053553037e-08, | |
| "logits/chosen": -2.9749999046325684, | |
| "logits/rejected": -2.9906249046325684, | |
| "logps/chosen": -235.1999969482422, | |
| "logps/rejected": -210.3000030517578, | |
| "loss": 0.6594, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.05439453199505806, | |
| "rewards/margins": 0.12880858778953552, | |
| "rewards/rejected": -0.07449951022863388, | |
| "step": 3115 | |
| }, | |
| { | |
| "epoch": 0.8032955715756952, | |
| "grad_norm": 197.0, | |
| "learning_rate": 9.835221421215241e-08, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.0562500953674316, | |
| "logps/chosen": -240.1999969482422, | |
| "logps/rejected": -247.0, | |
| "loss": 0.6344, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.13937988877296448, | |
| "rewards/margins": 0.18964843451976776, | |
| "rewards/rejected": -0.05014648288488388, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.804582904222451, | |
| "grad_norm": 213.0, | |
| "learning_rate": 9.770854788877446e-08, | |
| "logits/chosen": -3.153125047683716, | |
| "logits/rejected": -3.012500047683716, | |
| "logps/chosen": -283.79998779296875, | |
| "logps/rejected": -278.6000061035156, | |
| "loss": 0.5961, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.09360351413488388, | |
| "rewards/margins": 0.31884765625, | |
| "rewards/rejected": -0.22545166313648224, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.805870236869207, | |
| "grad_norm": 181.0, | |
| "learning_rate": 9.70648815653965e-08, | |
| "logits/chosen": -3.0, | |
| "logits/rejected": -3.065624952316284, | |
| "logps/chosen": -338.0, | |
| "logps/rejected": -290.6000061035156, | |
| "loss": 0.625, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.24199219048023224, | |
| "rewards/margins": 0.27265626192092896, | |
| "rewards/rejected": -0.03085937537252903, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.807157569515963, | |
| "grad_norm": 238.0, | |
| "learning_rate": 9.642121524201853e-08, | |
| "logits/chosen": -3.1812500953674316, | |
| "logits/rejected": -3.231250047683716, | |
| "logps/chosen": -306.79998779296875, | |
| "logps/rejected": -278.6000061035156, | |
| "loss": 0.6461, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.11765136569738388, | |
| "rewards/margins": 0.2158203125, | |
| "rewards/rejected": -0.09804687649011612, | |
| "step": 3135 | |
| }, | |
| { | |
| "epoch": 0.8084449021627188, | |
| "grad_norm": 284.0, | |
| "learning_rate": 9.577754891864057e-08, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.1968750953674316, | |
| "logps/chosen": -273.0, | |
| "logps/rejected": -252.1999969482422, | |
| "loss": 0.6781, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.02207031287252903, | |
| "rewards/margins": 0.12158203125, | |
| "rewards/rejected": -0.099609375, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.8097322348094748, | |
| "grad_norm": 220.0, | |
| "learning_rate": 9.513388259526261e-08, | |
| "logits/chosen": -3.0406250953674316, | |
| "logits/rejected": -3.0687499046325684, | |
| "logps/chosen": -231.10000610351562, | |
| "logps/rejected": -196.8000030517578, | |
| "loss": 0.6477, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.04638671875, | |
| "rewards/margins": 0.16733399033546448, | |
| "rewards/rejected": -0.12111816555261612, | |
| "step": 3145 | |
| }, | |
| { | |
| "epoch": 0.8110195674562307, | |
| "grad_norm": 204.0, | |
| "learning_rate": 9.449021627188466e-08, | |
| "logits/chosen": -3.065624952316284, | |
| "logits/rejected": -2.984375, | |
| "logps/chosen": -291.20001220703125, | |
| "logps/rejected": -213.60000610351562, | |
| "loss": 0.6141, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.16501465439796448, | |
| "rewards/margins": 0.263671875, | |
| "rewards/rejected": -0.09833984076976776, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.8123069001029866, | |
| "grad_norm": 184.0, | |
| "learning_rate": 9.384654994850669e-08, | |
| "logits/chosen": -3.1031250953674316, | |
| "logits/rejected": -3.171875, | |
| "logps/chosen": -236.39999389648438, | |
| "logps/rejected": -213.8000030517578, | |
| "loss": 0.5969, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.15568847954273224, | |
| "rewards/margins": 0.2978515625, | |
| "rewards/rejected": -0.14213867485523224, | |
| "step": 3155 | |
| }, | |
| { | |
| "epoch": 0.8135942327497425, | |
| "grad_norm": 224.0, | |
| "learning_rate": 9.320288362512873e-08, | |
| "logits/chosen": -3.3125, | |
| "logits/rejected": -3.268749952316284, | |
| "logps/chosen": -250.8000030517578, | |
| "logps/rejected": -276.79998779296875, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.21230468153953552, | |
| "rewards/margins": 0.2767578065395355, | |
| "rewards/rejected": -0.06413574516773224, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.8148815653964985, | |
| "grad_norm": 194.0, | |
| "learning_rate": 9.255921730175077e-08, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.2093749046325684, | |
| "logps/chosen": -256.3999938964844, | |
| "logps/rejected": -242.1999969482422, | |
| "loss": 0.6375, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.00913085974752903, | |
| "rewards/margins": 0.1767578125, | |
| "rewards/rejected": -0.16748046875, | |
| "step": 3165 | |
| }, | |
| { | |
| "epoch": 0.8161688980432544, | |
| "grad_norm": 206.0, | |
| "learning_rate": 9.191555097837281e-08, | |
| "logits/chosen": -3.265625, | |
| "logits/rejected": -3.262500047683716, | |
| "logps/chosen": -262.79998779296875, | |
| "logps/rejected": -196.8000030517578, | |
| "loss": 0.6398, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.15390625596046448, | |
| "rewards/margins": 0.21748046576976776, | |
| "rewards/rejected": -0.06396484375, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.8174562306900103, | |
| "grad_norm": 193.0, | |
| "learning_rate": 9.127188465499484e-08, | |
| "logits/chosen": -3.1343750953674316, | |
| "logits/rejected": -3.096874952316284, | |
| "logps/chosen": -327.3999938964844, | |
| "logps/rejected": -218.60000610351562, | |
| "loss": 0.5859, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.37109375, | |
| "rewards/margins": 0.33544921875, | |
| "rewards/rejected": 0.03547363355755806, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.8187435633367662, | |
| "grad_norm": 338.0, | |
| "learning_rate": 9.062821833161689e-08, | |
| "logits/chosen": -3.159374952316284, | |
| "logits/rejected": -3.1343750953674316, | |
| "logps/chosen": -290.6000061035156, | |
| "logps/rejected": -263.20001220703125, | |
| "loss": 0.657, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.24335937201976776, | |
| "rewards/margins": 0.18593749403953552, | |
| "rewards/rejected": 0.05717773362994194, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.8200308959835222, | |
| "grad_norm": 450.0, | |
| "learning_rate": 8.998455200823893e-08, | |
| "logits/chosen": -3.1156249046325684, | |
| "logits/rejected": -3.174999952316284, | |
| "logps/chosen": -273.20001220703125, | |
| "logps/rejected": -204.60000610351562, | |
| "loss": 0.5922, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.14890137314796448, | |
| "rewards/margins": 0.3138671815395355, | |
| "rewards/rejected": -0.16470947861671448, | |
| "step": 3185 | |
| }, | |
| { | |
| "epoch": 0.821318228630278, | |
| "grad_norm": 238.0, | |
| "learning_rate": 8.934088568486097e-08, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.315624952316284, | |
| "logps/chosen": -283.20001220703125, | |
| "logps/rejected": -231.60000610351562, | |
| "loss": 0.6953, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.1865234375, | |
| "rewards/margins": 0.09453125298023224, | |
| "rewards/rejected": 0.09201660007238388, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.822605561277034, | |
| "grad_norm": 211.0, | |
| "learning_rate": 8.8697219361483e-08, | |
| "logits/chosen": -3.1031250953674316, | |
| "logits/rejected": -3.2093749046325684, | |
| "logps/chosen": -281.6000061035156, | |
| "logps/rejected": -237.89999389648438, | |
| "loss": 0.6102, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.16245117783546448, | |
| "rewards/margins": 0.2657226622104645, | |
| "rewards/rejected": -0.10354004055261612, | |
| "step": 3195 | |
| }, | |
| { | |
| "epoch": 0.82389289392379, | |
| "grad_norm": 292.0, | |
| "learning_rate": 8.805355303810503e-08, | |
| "logits/chosen": -3.0562500953674316, | |
| "logits/rejected": -3.1187500953674316, | |
| "logps/chosen": -295.20001220703125, | |
| "logps/rejected": -280.1000061035156, | |
| "loss": 0.6109, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.2842773497104645, | |
| "rewards/margins": 0.291015625, | |
| "rewards/rejected": -0.0061279297806322575, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.8251802265705458, | |
| "grad_norm": 223.0, | |
| "learning_rate": 8.740988671472709e-08, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -246.1999969482422, | |
| "logps/rejected": -237.0, | |
| "loss": 0.6633, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.09824218600988388, | |
| "rewards/margins": 0.1552734375, | |
| "rewards/rejected": -0.05703125149011612, | |
| "step": 3205 | |
| }, | |
| { | |
| "epoch": 0.8264675592173018, | |
| "grad_norm": 227.0, | |
| "learning_rate": 8.676622039134912e-08, | |
| "logits/chosen": -3.293750047683716, | |
| "logits/rejected": -3.2874999046325684, | |
| "logps/chosen": -358.6000061035156, | |
| "logps/rejected": -301.6000061035156, | |
| "loss": 0.5914, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.2513671815395355, | |
| "rewards/margins": 0.3316894471645355, | |
| "rewards/rejected": -0.08022461086511612, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.8277548918640577, | |
| "grad_norm": 223.0, | |
| "learning_rate": 8.612255406797116e-08, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.096874952316284, | |
| "logps/chosen": -288.0, | |
| "logps/rejected": -239.89999389648438, | |
| "loss": 0.5695, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.23005370795726776, | |
| "rewards/margins": 0.3765625059604645, | |
| "rewards/rejected": -0.14638671278953552, | |
| "step": 3215 | |
| }, | |
| { | |
| "epoch": 0.8290422245108136, | |
| "grad_norm": 204.0, | |
| "learning_rate": 8.547888774459319e-08, | |
| "logits/chosen": -2.996875047683716, | |
| "logits/rejected": -2.9625000953674316, | |
| "logps/chosen": -234.0, | |
| "logps/rejected": -192.6999969482422, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.1552734375, | |
| "rewards/margins": 0.18242187798023224, | |
| "rewards/rejected": -0.02705078199505806, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.8303295571575695, | |
| "grad_norm": 235.0, | |
| "learning_rate": 8.483522142121524e-08, | |
| "logits/chosen": -3.231250047683716, | |
| "logits/rejected": -3.1937499046325684, | |
| "logps/chosen": -294.0, | |
| "logps/rejected": -255.60000610351562, | |
| "loss": 0.582, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.1722412109375, | |
| "rewards/margins": 0.3197265565395355, | |
| "rewards/rejected": -0.1474609375, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.8316168898043255, | |
| "grad_norm": 212.0, | |
| "learning_rate": 8.419155509783727e-08, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.2406249046325684, | |
| "logps/chosen": -299.6000061035156, | |
| "logps/rejected": -268.20001220703125, | |
| "loss": 0.5758, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.28229981660842896, | |
| "rewards/margins": 0.35273438692092896, | |
| "rewards/rejected": -0.07060547173023224, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 0.8329042224510813, | |
| "grad_norm": 178.0, | |
| "learning_rate": 8.354788877445932e-08, | |
| "logits/chosen": -3.237499952316284, | |
| "logits/rejected": -3.1156249046325684, | |
| "logps/chosen": -307.3999938964844, | |
| "logps/rejected": -265.0, | |
| "loss": 0.5977, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.13041992485523224, | |
| "rewards/margins": 0.2681640684604645, | |
| "rewards/rejected": -0.13789062201976776, | |
| "step": 3235 | |
| }, | |
| { | |
| "epoch": 0.8341915550978373, | |
| "grad_norm": 155.0, | |
| "learning_rate": 8.290422245108136e-08, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.2750000953674316, | |
| "logps/chosen": -252.60000610351562, | |
| "logps/rejected": -258.20001220703125, | |
| "loss": 0.6008, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.23749999701976776, | |
| "rewards/margins": 0.28093260526657104, | |
| "rewards/rejected": -0.04326171800494194, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.8354788877445932, | |
| "grad_norm": 190.0, | |
| "learning_rate": 8.22605561277034e-08, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -258.0, | |
| "logps/rejected": -256.6000061035156, | |
| "loss": 0.6133, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.14921875298023224, | |
| "rewards/margins": 0.23847655951976776, | |
| "rewards/rejected": -0.08962402492761612, | |
| "step": 3245 | |
| }, | |
| { | |
| "epoch": 0.8367662203913491, | |
| "grad_norm": 188.0, | |
| "learning_rate": 8.161688980432543e-08, | |
| "logits/chosen": -3.278125047683716, | |
| "logits/rejected": -3.3125, | |
| "logps/chosen": -295.79998779296875, | |
| "logps/rejected": -218.1999969482422, | |
| "loss": 0.5859, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.118896484375, | |
| "rewards/margins": 0.34521484375, | |
| "rewards/rejected": -0.2265625, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.838053553038105, | |
| "grad_norm": 237.0, | |
| "learning_rate": 8.097322348094747e-08, | |
| "logits/chosen": -3.268749952316284, | |
| "logits/rejected": -2.6273436546325684, | |
| "logps/chosen": -219.1750030517578, | |
| "logps/rejected": -217.8000030517578, | |
| "loss": 0.625, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.15632323920726776, | |
| "rewards/margins": 0.23422852158546448, | |
| "rewards/rejected": -0.07841797173023224, | |
| "step": 3255 | |
| }, | |
| { | |
| "epoch": 0.839340885684861, | |
| "grad_norm": 189.0, | |
| "learning_rate": 8.032955715756952e-08, | |
| "logits/chosen": -3.1968750953674316, | |
| "logits/rejected": -3.237499952316284, | |
| "logps/chosen": -281.3999938964844, | |
| "logps/rejected": -311.3999938964844, | |
| "loss": 0.6668, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.15664061903953552, | |
| "rewards/margins": 0.15244141221046448, | |
| "rewards/rejected": 0.0042968750931322575, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.8406282183316169, | |
| "grad_norm": 200.0, | |
| "learning_rate": 7.968589083419156e-08, | |
| "logits/chosen": -3.3031249046325684, | |
| "logits/rejected": -3.265625, | |
| "logps/chosen": -303.79998779296875, | |
| "logps/rejected": -263.79998779296875, | |
| "loss": 0.6, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.08432617038488388, | |
| "rewards/margins": 0.29707032442092896, | |
| "rewards/rejected": -0.21308593451976776, | |
| "step": 3265 | |
| }, | |
| { | |
| "epoch": 0.8419155509783728, | |
| "grad_norm": 180.0, | |
| "learning_rate": 7.904222451081359e-08, | |
| "logits/chosen": -3.0, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -312.6000061035156, | |
| "logps/rejected": -251.8000030517578, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.2560058534145355, | |
| "rewards/margins": 0.25615233182907104, | |
| "rewards/rejected": -0.0002929687616415322, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 0.8432028836251287, | |
| "grad_norm": 223.0, | |
| "learning_rate": 7.839855818743563e-08, | |
| "logits/chosen": -3.096874952316284, | |
| "logits/rejected": -3.1500000953674316, | |
| "logps/chosen": -282.20001220703125, | |
| "logps/rejected": -223.5, | |
| "loss": 0.5891, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.24453124403953552, | |
| "rewards/margins": 0.318359375, | |
| "rewards/rejected": -0.07368163764476776, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.8444902162718847, | |
| "grad_norm": 204.0, | |
| "learning_rate": 7.775489186405767e-08, | |
| "logits/chosen": -2.9781250953674316, | |
| "logits/rejected": -3.018749952316284, | |
| "logps/chosen": -274.8999938964844, | |
| "logps/rejected": -226.6999969482422, | |
| "loss": 0.6477, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.17036132514476776, | |
| "rewards/margins": 0.17939452826976776, | |
| "rewards/rejected": -0.00891723670065403, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.8457775489186405, | |
| "grad_norm": 205.0, | |
| "learning_rate": 7.711122554067972e-08, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.143749952316284, | |
| "logps/chosen": -258.8999938964844, | |
| "logps/rejected": -261.6000061035156, | |
| "loss": 0.6586, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.12392578274011612, | |
| "rewards/margins": 0.15998534858226776, | |
| "rewards/rejected": -0.03615722805261612, | |
| "step": 3285 | |
| }, | |
| { | |
| "epoch": 0.8470648815653965, | |
| "grad_norm": 243.0, | |
| "learning_rate": 7.646755921730175e-08, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.206249952316284, | |
| "logps/chosen": -303.70001220703125, | |
| "logps/rejected": -304.70001220703125, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.20252227783203125, | |
| "rewards/margins": 0.24544982612133026, | |
| "rewards/rejected": -0.04345703125, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 0.8483522142121525, | |
| "grad_norm": 180.0, | |
| "learning_rate": 7.582389289392379e-08, | |
| "logits/chosen": -3.2562499046325684, | |
| "logits/rejected": -3.309375047683716, | |
| "logps/chosen": -332.79998779296875, | |
| "logps/rejected": -325.6000061035156, | |
| "loss": 0.5332, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 0.2530761659145355, | |
| "rewards/margins": 0.48750001192092896, | |
| "rewards/rejected": -0.234619140625, | |
| "step": 3295 | |
| }, | |
| { | |
| "epoch": 0.8496395468589083, | |
| "grad_norm": 266.0, | |
| "learning_rate": 7.518022657054583e-08, | |
| "logits/chosen": -3.059375047683716, | |
| "logits/rejected": -2.9937500953674316, | |
| "logps/chosen": -287.3999938964844, | |
| "logps/rejected": -250.60000610351562, | |
| "loss": 0.6562, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.20493164658546448, | |
| "rewards/margins": 0.16724853217601776, | |
| "rewards/rejected": 0.03715820237994194, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.8509268795056643, | |
| "grad_norm": 168.0, | |
| "learning_rate": 7.453656024716787e-08, | |
| "logits/chosen": -3.296875, | |
| "logits/rejected": -3.2562499046325684, | |
| "logps/chosen": -310.79998779296875, | |
| "logps/rejected": -301.3999938964844, | |
| "loss": 0.6523, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.2889404296875, | |
| "rewards/margins": 0.15336914360523224, | |
| "rewards/rejected": 0.13486328721046448, | |
| "step": 3305 | |
| }, | |
| { | |
| "epoch": 0.8522142121524202, | |
| "grad_norm": 260.0, | |
| "learning_rate": 7.38928939237899e-08, | |
| "logits/chosen": -3.1500000953674316, | |
| "logits/rejected": -3.190624952316284, | |
| "logps/chosen": -338.20001220703125, | |
| "logps/rejected": -269.29998779296875, | |
| "loss": 0.5953, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.2984375059604645, | |
| "rewards/margins": 0.30683594942092896, | |
| "rewards/rejected": -0.00823974609375, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 0.8535015447991761, | |
| "grad_norm": 192.0, | |
| "learning_rate": 7.324922760041195e-08, | |
| "logits/chosen": -3.1656250953674316, | |
| "logits/rejected": -3.2125000953674316, | |
| "logps/chosen": -295.6000061035156, | |
| "logps/rejected": -302.79998779296875, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.3109374940395355, | |
| "rewards/margins": 0.27031248807907104, | |
| "rewards/rejected": 0.04132080078125, | |
| "step": 3315 | |
| }, | |
| { | |
| "epoch": 0.854788877445932, | |
| "grad_norm": 232.0, | |
| "learning_rate": 7.260556127703399e-08, | |
| "logits/chosen": -3.153125047683716, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -275.29998779296875, | |
| "logps/rejected": -241.39999389648438, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.08115234225988388, | |
| "rewards/margins": 0.2535156309604645, | |
| "rewards/rejected": -0.17241211235523224, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.856076210092688, | |
| "grad_norm": 218.0, | |
| "learning_rate": 7.196189495365603e-08, | |
| "logits/chosen": -3.206249952316284, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -296.3999938964844, | |
| "logps/rejected": -279.20001220703125, | |
| "loss": 0.618, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.05078125, | |
| "rewards/margins": 0.3037109375, | |
| "rewards/rejected": -0.2530273497104645, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.8573635427394438, | |
| "grad_norm": 208.0, | |
| "learning_rate": 7.131822863027806e-08, | |
| "logits/chosen": -3.2249999046325684, | |
| "logits/rejected": -3.231250047683716, | |
| "logps/chosen": -341.6000061035156, | |
| "logps/rejected": -290.0, | |
| "loss": 0.6562, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.12347412109375, | |
| "rewards/margins": 0.24443359673023224, | |
| "rewards/rejected": -0.12041015923023224, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 0.8586508753861998, | |
| "grad_norm": 180.0, | |
| "learning_rate": 7.067456230690009e-08, | |
| "logits/chosen": -3.1156249046325684, | |
| "logits/rejected": -3.174999952316284, | |
| "logps/chosen": -224.5, | |
| "logps/rejected": -202.39999389648438, | |
| "loss": 0.6086, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.04217529296875, | |
| "rewards/margins": 0.2533203065395355, | |
| "rewards/rejected": -0.21077576279640198, | |
| "step": 3335 | |
| }, | |
| { | |
| "epoch": 0.8599382080329557, | |
| "grad_norm": 276.0, | |
| "learning_rate": 7.003089598352215e-08, | |
| "logits/chosen": -3.106250047683716, | |
| "logits/rejected": -3.206249952316284, | |
| "logps/chosen": -263.20001220703125, | |
| "logps/rejected": -233.8000030517578, | |
| "loss": 0.6438, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.18300780653953552, | |
| "rewards/margins": 0.17587891221046448, | |
| "rewards/rejected": 0.0071777342818677425, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.8612255406797117, | |
| "grad_norm": 208.0, | |
| "learning_rate": 6.938722966014417e-08, | |
| "logits/chosen": -3.215625047683716, | |
| "logits/rejected": -3.299999952316284, | |
| "logps/chosen": -269.6000061035156, | |
| "logps/rejected": -221.0, | |
| "loss": 0.6094, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.110595703125, | |
| "rewards/margins": 0.2718749940395355, | |
| "rewards/rejected": -0.16103515028953552, | |
| "step": 3345 | |
| }, | |
| { | |
| "epoch": 0.8625128733264675, | |
| "grad_norm": 294.0, | |
| "learning_rate": 6.874356333676622e-08, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.090625047683716, | |
| "logps/chosen": -325.20001220703125, | |
| "logps/rejected": -282.0, | |
| "loss": 0.6273, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.19501952826976776, | |
| "rewards/margins": 0.2542968690395355, | |
| "rewards/rejected": -0.05959472805261612, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.8638002059732235, | |
| "grad_norm": 199.0, | |
| "learning_rate": 6.809989701338825e-08, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.168750047683716, | |
| "logps/chosen": -305.20001220703125, | |
| "logps/rejected": -289.79998779296875, | |
| "loss": 0.6227, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.10878906399011612, | |
| "rewards/margins": 0.26640623807907104, | |
| "rewards/rejected": -0.15761718153953552, | |
| "step": 3355 | |
| }, | |
| { | |
| "epoch": 0.8650875386199794, | |
| "grad_norm": 199.0, | |
| "learning_rate": 6.74562306900103e-08, | |
| "logits/chosen": NaN, | |
| "logits/rejected": -2.7593750953674316, | |
| "logps/chosen": -271.6000061035156, | |
| "logps/rejected": -234.5, | |
| "loss": 0.6027, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.21005859971046448, | |
| "rewards/margins": 0.25556641817092896, | |
| "rewards/rejected": -0.04523925855755806, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.8663748712667353, | |
| "grad_norm": 188.0, | |
| "learning_rate": 6.681256436663233e-08, | |
| "logits/chosen": -3.2125000953674316, | |
| "logits/rejected": -3.1031250953674316, | |
| "logps/chosen": -284.6000061035156, | |
| "logps/rejected": -250.60000610351562, | |
| "loss": 0.5336, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 0.2632812559604645, | |
| "rewards/margins": 0.45703125, | |
| "rewards/rejected": -0.19403076171875, | |
| "step": 3365 | |
| }, | |
| { | |
| "epoch": 0.8676622039134912, | |
| "grad_norm": 298.0, | |
| "learning_rate": 6.616889804325438e-08, | |
| "logits/chosen": -3.096874952316284, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -292.29998779296875, | |
| "logps/rejected": -208.25, | |
| "loss": 0.6297, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.24238280951976776, | |
| "rewards/margins": 0.22958984971046448, | |
| "rewards/rejected": 0.01253662072122097, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 0.8689495365602472, | |
| "grad_norm": 224.0, | |
| "learning_rate": 6.55252317198764e-08, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.0562500953674316, | |
| "logps/chosen": -274.20001220703125, | |
| "logps/rejected": -239.0, | |
| "loss": 0.6375, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.12563475966453552, | |
| "rewards/margins": 0.2025146484375, | |
| "rewards/rejected": -0.076416015625, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.870236869207003, | |
| "grad_norm": 219.0, | |
| "learning_rate": 6.488156539649846e-08, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.1781249046325684, | |
| "logps/chosen": -296.20001220703125, | |
| "logps/rejected": -222.1999969482422, | |
| "loss": 0.5953, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.27363282442092896, | |
| "rewards/margins": 0.3023437559604645, | |
| "rewards/rejected": -0.028564453125, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.871524201853759, | |
| "grad_norm": 158.0, | |
| "learning_rate": 6.423789907312049e-08, | |
| "logits/chosen": -3.237499952316284, | |
| "logits/rejected": -3.128124952316284, | |
| "logps/chosen": -270.3999938964844, | |
| "logps/rejected": -257.79998779296875, | |
| "loss": 0.6234, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.05593261867761612, | |
| "rewards/margins": 0.22177734971046448, | |
| "rewards/rejected": -0.16511841118335724, | |
| "step": 3385 | |
| }, | |
| { | |
| "epoch": 0.872811534500515, | |
| "grad_norm": 222.0, | |
| "learning_rate": 6.359423274974253e-08, | |
| "logits/chosen": -3.0562500953674316, | |
| "logits/rejected": -3.2125000953674316, | |
| "logps/chosen": -276.79998779296875, | |
| "logps/rejected": -230.39999389648438, | |
| "loss": 0.5809, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.10698242485523224, | |
| "rewards/margins": 0.32951658964157104, | |
| "rewards/rejected": -0.22214356064796448, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 0.8740988671472708, | |
| "grad_norm": 241.0, | |
| "learning_rate": 6.295056642636456e-08, | |
| "logits/chosen": -3.1312499046325684, | |
| "logits/rejected": -3.125, | |
| "logps/chosen": -374.0, | |
| "logps/rejected": -298.20001220703125, | |
| "loss": 0.6078, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.4013671875, | |
| "rewards/margins": 0.2583984434604645, | |
| "rewards/rejected": 0.14279937744140625, | |
| "step": 3395 | |
| }, | |
| { | |
| "epoch": 0.8753861997940268, | |
| "grad_norm": 195.0, | |
| "learning_rate": 6.23069001029866e-08, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -2.642578125, | |
| "logps/chosen": -262.20001220703125, | |
| "logps/rejected": -234.39999389648438, | |
| "loss": 0.707, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.06767578423023224, | |
| "rewards/margins": 0.10616455227136612, | |
| "rewards/rejected": -0.03852539137005806, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.8766735324407827, | |
| "grad_norm": 272.0, | |
| "learning_rate": 6.166323377960865e-08, | |
| "logits/chosen": -3.075000047683716, | |
| "logits/rejected": -3.109375, | |
| "logps/chosen": -351.79998779296875, | |
| "logps/rejected": -305.0, | |
| "loss": 0.5875, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.26606446504592896, | |
| "rewards/margins": 0.3348632752895355, | |
| "rewards/rejected": -0.06806640326976776, | |
| "step": 3405 | |
| }, | |
| { | |
| "epoch": 0.8779608650875386, | |
| "grad_norm": 178.0, | |
| "learning_rate": 6.101956745623069e-08, | |
| "logits/chosen": -3.253124952316284, | |
| "logits/rejected": -3.328125, | |
| "logps/chosen": -304.6000061035156, | |
| "logps/rejected": -257.79998779296875, | |
| "loss": 0.5988, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.0709228515625, | |
| "rewards/margins": 0.3402343690395355, | |
| "rewards/rejected": -0.2691406309604645, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 0.8792481977342945, | |
| "grad_norm": 208.0, | |
| "learning_rate": 6.037590113285273e-08, | |
| "logits/chosen": -3.1187500953674316, | |
| "logits/rejected": -2.8531250953674316, | |
| "logps/chosen": -204.5, | |
| "logps/rejected": -226.10000610351562, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.19028320908546448, | |
| "rewards/margins": 0.21250000596046448, | |
| "rewards/rejected": -0.02207031287252903, | |
| "step": 3415 | |
| }, | |
| { | |
| "epoch": 0.8805355303810505, | |
| "grad_norm": 344.0, | |
| "learning_rate": 5.973223480947476e-08, | |
| "logits/chosen": -2.8656249046325684, | |
| "logits/rejected": -2.598437547683716, | |
| "logps/chosen": -234.8000030517578, | |
| "logps/rejected": -224.8000030517578, | |
| "loss": 0.6453, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.20136718451976776, | |
| "rewards/margins": 0.12967529892921448, | |
| "rewards/rejected": 0.07163085788488388, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.8818228630278064, | |
| "grad_norm": 199.0, | |
| "learning_rate": 5.9088568486096805e-08, | |
| "logits/chosen": -3.0875000953674316, | |
| "logits/rejected": -3.0250000953674316, | |
| "logps/chosen": -325.20001220703125, | |
| "logps/rejected": -231.39999389648438, | |
| "loss": 0.607, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.24267578125, | |
| "rewards/margins": 0.25996094942092896, | |
| "rewards/rejected": -0.01762695237994194, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.8831101956745623, | |
| "grad_norm": 175.0, | |
| "learning_rate": 5.844490216271884e-08, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.121875047683716, | |
| "logps/chosen": -325.3999938964844, | |
| "logps/rejected": -291.0, | |
| "loss": 0.55, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.23781737685203552, | |
| "rewards/margins": 0.4085937440395355, | |
| "rewards/rejected": -0.17041015625, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 0.8843975283213182, | |
| "grad_norm": 229.0, | |
| "learning_rate": 5.7801235839340884e-08, | |
| "logits/chosen": -2.762500047683716, | |
| "logits/rejected": -2.9515624046325684, | |
| "logps/chosen": -217.6999969482422, | |
| "logps/rejected": -169.39999389648438, | |
| "loss": 0.6305, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.3077148497104645, | |
| "rewards/margins": 0.20998534560203552, | |
| "rewards/rejected": 0.09767456352710724, | |
| "step": 3435 | |
| }, | |
| { | |
| "epoch": 0.8856848609680742, | |
| "grad_norm": 230.0, | |
| "learning_rate": 5.715756951596292e-08, | |
| "logits/chosen": -3.1156249046325684, | |
| "logits/rejected": -3.1781249046325684, | |
| "logps/chosen": -249.0, | |
| "logps/rejected": -254.60000610351562, | |
| "loss": 0.6031, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.19316406548023224, | |
| "rewards/margins": 0.29150390625, | |
| "rewards/rejected": -0.09834595024585724, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.88697219361483, | |
| "grad_norm": 200.0, | |
| "learning_rate": 5.651390319258496e-08, | |
| "logits/chosen": -2.9906249046325684, | |
| "logits/rejected": -3.153125047683716, | |
| "logps/chosen": -250.1999969482422, | |
| "logps/rejected": -280.6000061035156, | |
| "loss": 0.5977, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.2623535096645355, | |
| "rewards/margins": 0.29931640625, | |
| "rewards/rejected": -0.03725586086511612, | |
| "step": 3445 | |
| }, | |
| { | |
| "epoch": 0.888259526261586, | |
| "grad_norm": 170.0, | |
| "learning_rate": 5.5870236869207e-08, | |
| "logits/chosen": -3.043750047683716, | |
| "logits/rejected": -3.2281250953674316, | |
| "logps/chosen": -265.3999938964844, | |
| "logps/rejected": -177.6999969482422, | |
| "loss": 0.5742, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.2548828125, | |
| "rewards/margins": 0.3193359375, | |
| "rewards/rejected": -0.06376953423023224, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.889546858908342, | |
| "grad_norm": 214.0, | |
| "learning_rate": 5.522657054582904e-08, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -268.79998779296875, | |
| "logps/rejected": -253.60000610351562, | |
| "loss": 0.6547, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.07691650092601776, | |
| "rewards/margins": 0.15390625596046448, | |
| "rewards/rejected": -0.07675781100988388, | |
| "step": 3455 | |
| }, | |
| { | |
| "epoch": 0.8908341915550978, | |
| "grad_norm": 227.0, | |
| "learning_rate": 5.458290422245108e-08, | |
| "logits/chosen": -3.1781249046325684, | |
| "logits/rejected": -3.215625047683716, | |
| "logps/chosen": -313.20001220703125, | |
| "logps/rejected": -257.3999938964844, | |
| "loss": 0.6227, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.13955077528953552, | |
| "rewards/margins": 0.23447266221046448, | |
| "rewards/rejected": -0.09486083686351776, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.8921215242018538, | |
| "grad_norm": 117.0, | |
| "learning_rate": 5.393923789907312e-08, | |
| "logits/chosen": -3.2406249046325684, | |
| "logits/rejected": -3.125, | |
| "logps/chosen": -202.0, | |
| "logps/rejected": -222.89999389648438, | |
| "loss": 0.6695, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": 0.12324218451976776, | |
| "rewards/margins": 0.14050903916358948, | |
| "rewards/rejected": -0.017333984375, | |
| "step": 3465 | |
| }, | |
| { | |
| "epoch": 0.8934088568486097, | |
| "grad_norm": 204.0, | |
| "learning_rate": 5.3295571575695156e-08, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.159374952316284, | |
| "logps/chosen": -244.1999969482422, | |
| "logps/rejected": -219.1999969482422, | |
| "loss": 0.5883, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.16069336235523224, | |
| "rewards/margins": 0.31336671113967896, | |
| "rewards/rejected": -0.15283203125, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 0.8946961894953656, | |
| "grad_norm": 302.0, | |
| "learning_rate": 5.26519052523172e-08, | |
| "logits/chosen": -3.168750047683716, | |
| "logits/rejected": -3.078125, | |
| "logps/chosen": -302.0, | |
| "logps/rejected": -281.6000061035156, | |
| "loss": 0.577, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.3382202088832855, | |
| "rewards/margins": 0.3521484434604645, | |
| "rewards/rejected": -0.014453125186264515, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.8959835221421215, | |
| "grad_norm": 251.0, | |
| "learning_rate": 5.2008238928939235e-08, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.0250000953674316, | |
| "logps/chosen": -296.79998779296875, | |
| "logps/rejected": -286.79998779296875, | |
| "loss": 0.6266, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.23872070014476776, | |
| "rewards/margins": 0.26806640625, | |
| "rewards/rejected": -0.02919921837747097, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.8972708547888775, | |
| "grad_norm": 194.0, | |
| "learning_rate": 5.136457260556128e-08, | |
| "logits/chosen": -2.996875047683716, | |
| "logits/rejected": -3.0562500953674316, | |
| "logps/chosen": -256.0, | |
| "logps/rejected": -240.60000610351562, | |
| "loss": 0.6047, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.15751953423023224, | |
| "rewards/margins": 0.31328123807907104, | |
| "rewards/rejected": -0.15534667670726776, | |
| "step": 3485 | |
| }, | |
| { | |
| "epoch": 0.8985581874356333, | |
| "grad_norm": 272.0, | |
| "learning_rate": 5.0720906282183313e-08, | |
| "logits/chosen": -3.0406250953674316, | |
| "logits/rejected": -3.0875000953674316, | |
| "logps/chosen": -284.6000061035156, | |
| "logps/rejected": -288.79998779296875, | |
| "loss": 0.5863, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.06549987941980362, | |
| "rewards/margins": 0.3589843809604645, | |
| "rewards/rejected": -0.29350584745407104, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 0.8998455200823893, | |
| "grad_norm": 223.0, | |
| "learning_rate": 5.0077239958805356e-08, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.253124952316284, | |
| "logps/chosen": -340.79998779296875, | |
| "logps/rejected": -257.6000061035156, | |
| "loss": 0.6016, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.345703125, | |
| "rewards/margins": 0.3196777403354645, | |
| "rewards/rejected": 0.02479248121380806, | |
| "step": 3495 | |
| }, | |
| { | |
| "epoch": 0.9011328527291452, | |
| "grad_norm": 251.0, | |
| "learning_rate": 4.943357363542739e-08, | |
| "logits/chosen": -3.203125, | |
| "logits/rejected": -3.1156249046325684, | |
| "logps/chosen": -288.70001220703125, | |
| "logps/rejected": -234.60000610351562, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.15566405653953552, | |
| "rewards/margins": 0.28558349609375, | |
| "rewards/rejected": -0.13051757216453552, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.9024201853759012, | |
| "grad_norm": 220.0, | |
| "learning_rate": 4.8789907312049435e-08, | |
| "logits/chosen": -3.2562499046325684, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -333.0, | |
| "logps/rejected": -298.0, | |
| "loss": 0.573, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.19880370795726776, | |
| "rewards/margins": 0.384765625, | |
| "rewards/rejected": -0.18662109971046448, | |
| "step": 3505 | |
| }, | |
| { | |
| "epoch": 0.903707518022657, | |
| "grad_norm": 203.0, | |
| "learning_rate": 4.814624098867147e-08, | |
| "logits/chosen": -3.043750047683716, | |
| "logits/rejected": -3.231250047683716, | |
| "logps/chosen": -259.5, | |
| "logps/rejected": -231.8000030517578, | |
| "loss": 0.6281, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.26025390625, | |
| "rewards/margins": 0.2529540956020355, | |
| "rewards/rejected": 0.007250976748764515, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 0.904994850669413, | |
| "grad_norm": 252.0, | |
| "learning_rate": 4.7502574665293514e-08, | |
| "logits/chosen": -3.234375, | |
| "logits/rejected": -3.3375000953674316, | |
| "logps/chosen": -348.0, | |
| "logps/rejected": -290.6000061035156, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.08777465671300888, | |
| "rewards/margins": 0.3062500059604645, | |
| "rewards/rejected": -0.21816405653953552, | |
| "step": 3515 | |
| }, | |
| { | |
| "epoch": 0.9062821833161689, | |
| "grad_norm": 306.0, | |
| "learning_rate": 4.685890834191555e-08, | |
| "logits/chosen": -3.2750000953674316, | |
| "logits/rejected": -3.237499952316284, | |
| "logps/chosen": -281.0, | |
| "logps/rejected": -262.79998779296875, | |
| "loss": 0.5789, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.1484375, | |
| "rewards/margins": 0.38612669706344604, | |
| "rewards/rejected": -0.23789063096046448, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.9075695159629248, | |
| "grad_norm": 236.0, | |
| "learning_rate": 4.621524201853759e-08, | |
| "logits/chosen": -3.1312499046325684, | |
| "logits/rejected": -3.265625, | |
| "logps/chosen": -282.5, | |
| "logps/rejected": -285.79998779296875, | |
| "loss": 0.6016, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.14414063096046448, | |
| "rewards/margins": 0.28386229276657104, | |
| "rewards/rejected": -0.138916015625, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.9088568486096807, | |
| "grad_norm": 185.0, | |
| "learning_rate": 4.557157569515963e-08, | |
| "logits/chosen": -3.184375047683716, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -308.6000061035156, | |
| "logps/rejected": -269.79998779296875, | |
| "loss": 0.6352, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.16005858778953552, | |
| "rewards/margins": 0.20595702528953552, | |
| "rewards/rejected": -0.04582519456744194, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 0.9101441812564367, | |
| "grad_norm": 212.0, | |
| "learning_rate": 4.492790937178167e-08, | |
| "logits/chosen": -3.3062500953674316, | |
| "logits/rejected": -3.3031249046325684, | |
| "logps/chosen": -225.39999389648438, | |
| "logps/rejected": -234.60000610351562, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -0.02724609337747097, | |
| "rewards/margins": 0.25312501192092896, | |
| "rewards/rejected": -0.2801269590854645, | |
| "step": 3535 | |
| }, | |
| { | |
| "epoch": 0.9114315139031925, | |
| "grad_norm": 152.0, | |
| "learning_rate": 4.42842430484037e-08, | |
| "logits/chosen": -3.078125, | |
| "logits/rejected": -3.090625047683716, | |
| "logps/chosen": -293.79998779296875, | |
| "logps/rejected": -256.0, | |
| "loss": 0.5945, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.21572265028953552, | |
| "rewards/margins": 0.30693358182907104, | |
| "rewards/rejected": -0.09058227390050888, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.9127188465499485, | |
| "grad_norm": 232.0, | |
| "learning_rate": 4.364057672502574e-08, | |
| "logits/chosen": -3.143749952316284, | |
| "logits/rejected": -3.2249999046325684, | |
| "logps/chosen": -331.6000061035156, | |
| "logps/rejected": -318.3999938964844, | |
| "loss": 0.6211, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.23469237983226776, | |
| "rewards/margins": 0.2720703184604645, | |
| "rewards/rejected": -0.03725586086511612, | |
| "step": 3545 | |
| }, | |
| { | |
| "epoch": 0.9140061791967045, | |
| "grad_norm": 204.0, | |
| "learning_rate": 4.299691040164778e-08, | |
| "logits/chosen": -3.153125047683716, | |
| "logits/rejected": -3.153125047683716, | |
| "logps/chosen": -305.79998779296875, | |
| "logps/rejected": -303.3999938964844, | |
| "loss": 0.65, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.18559570610523224, | |
| "rewards/margins": 0.16357421875, | |
| "rewards/rejected": 0.02163085900247097, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.9152935118434603, | |
| "grad_norm": 229.0, | |
| "learning_rate": 4.235324407826982e-08, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.246875047683716, | |
| "logps/chosen": -274.6000061035156, | |
| "logps/rejected": -271.79998779296875, | |
| "loss": 0.6086, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.163330078125, | |
| "rewards/margins": 0.2552734315395355, | |
| "rewards/rejected": -0.09162597358226776, | |
| "step": 3555 | |
| }, | |
| { | |
| "epoch": 0.9165808444902163, | |
| "grad_norm": 183.0, | |
| "learning_rate": 4.170957775489186e-08, | |
| "logits/chosen": -3.1624999046325684, | |
| "logits/rejected": -3.2406249046325684, | |
| "logps/chosen": -307.20001220703125, | |
| "logps/rejected": -247.10000610351562, | |
| "loss": 0.6336, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.06923828274011612, | |
| "rewards/margins": 0.23739013075828552, | |
| "rewards/rejected": -0.16816405951976776, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.9178681771369722, | |
| "grad_norm": 211.0, | |
| "learning_rate": 4.10659114315139e-08, | |
| "logits/chosen": -3.0999999046325684, | |
| "logits/rejected": -2.890625, | |
| "logps/chosen": -260.6000061035156, | |
| "logps/rejected": -264.1000061035156, | |
| "loss": 0.6516, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.22031250596046448, | |
| "rewards/margins": 0.1630859375, | |
| "rewards/rejected": 0.05756225436925888, | |
| "step": 3565 | |
| }, | |
| { | |
| "epoch": 0.9191555097837281, | |
| "grad_norm": 408.0, | |
| "learning_rate": 4.042224510813594e-08, | |
| "logits/chosen": -3.0062499046325684, | |
| "logits/rejected": -2.7750000953674316, | |
| "logps/chosen": -267.6000061035156, | |
| "logps/rejected": -237.3249969482422, | |
| "loss": 0.5992, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.20205077528953552, | |
| "rewards/margins": 0.2645263671875, | |
| "rewards/rejected": -0.062255859375, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 0.920442842430484, | |
| "grad_norm": 245.0, | |
| "learning_rate": 3.977857878475798e-08, | |
| "logits/chosen": -3.1500000953674316, | |
| "logits/rejected": -3.1624999046325684, | |
| "logps/chosen": -287.0, | |
| "logps/rejected": -205.8000030517578, | |
| "loss": 0.5961, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.08537597954273224, | |
| "rewards/margins": 0.29316407442092896, | |
| "rewards/rejected": -0.20732422173023224, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.92173017507724, | |
| "grad_norm": 245.0, | |
| "learning_rate": 3.9134912461380015e-08, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -3.21875, | |
| "logps/chosen": -267.0, | |
| "logps/rejected": -265.6000061035156, | |
| "loss": 0.6133, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.22695311903953552, | |
| "rewards/margins": 0.25529783964157104, | |
| "rewards/rejected": -0.02851562574505806, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.9230175077239959, | |
| "grad_norm": 248.0, | |
| "learning_rate": 3.849124613800206e-08, | |
| "logits/chosen": -3.206249952316284, | |
| "logits/rejected": -3.3125, | |
| "logps/chosen": -309.20001220703125, | |
| "logps/rejected": -227.6999969482422, | |
| "loss": 0.5891, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.24960938096046448, | |
| "rewards/margins": 0.31132811307907104, | |
| "rewards/rejected": -0.06132812425494194, | |
| "step": 3585 | |
| }, | |
| { | |
| "epoch": 0.9243048403707518, | |
| "grad_norm": 158.0, | |
| "learning_rate": 3.7847579814624094e-08, | |
| "logits/chosen": -3.1500000953674316, | |
| "logits/rejected": -2.901562452316284, | |
| "logps/chosen": -243.39999389648438, | |
| "logps/rejected": -213.89999389648438, | |
| "loss": 0.6094, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.24399414658546448, | |
| "rewards/margins": 0.255859375, | |
| "rewards/rejected": -0.0120849609375, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 0.9255921730175077, | |
| "grad_norm": 190.0, | |
| "learning_rate": 3.720391349124614e-08, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.0999999046325684, | |
| "logps/chosen": -250.8000030517578, | |
| "logps/rejected": -258.0, | |
| "loss": 0.6219, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.11809082329273224, | |
| "rewards/margins": 0.23867186903953552, | |
| "rewards/rejected": -0.12075195461511612, | |
| "step": 3595 | |
| }, | |
| { | |
| "epoch": 0.9268795056642637, | |
| "grad_norm": 183.0, | |
| "learning_rate": 3.656024716786817e-08, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -3.0999999046325684, | |
| "logps/chosen": -241.0, | |
| "logps/rejected": -257.3999938964844, | |
| "loss": 0.5773, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.19086913764476776, | |
| "rewards/margins": 0.4173828065395355, | |
| "rewards/rejected": -0.2261962890625, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.9281668383110195, | |
| "grad_norm": 199.0, | |
| "learning_rate": 3.5916580844490216e-08, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -2.926562547683716, | |
| "logps/chosen": -258.20001220703125, | |
| "logps/rejected": -251.39999389648438, | |
| "loss": 0.6094, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.12124023586511612, | |
| "rewards/margins": 0.2586914002895355, | |
| "rewards/rejected": -0.13750000298023224, | |
| "step": 3605 | |
| }, | |
| { | |
| "epoch": 0.9294541709577755, | |
| "grad_norm": 258.0, | |
| "learning_rate": 3.527291452111225e-08, | |
| "logits/chosen": -3.21875, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -288.20001220703125, | |
| "logps/rejected": -239.1999969482422, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.02497558668255806, | |
| "rewards/margins": 0.24150390923023224, | |
| "rewards/rejected": -0.21640625596046448, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 0.9307415036045315, | |
| "grad_norm": 226.0, | |
| "learning_rate": 3.4629248197734294e-08, | |
| "logits/chosen": -3.159374952316284, | |
| "logits/rejected": -3.203125, | |
| "logps/chosen": -343.20001220703125, | |
| "logps/rejected": -298.6000061035156, | |
| "loss": 0.607, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.2844482362270355, | |
| "rewards/margins": 0.3106445372104645, | |
| "rewards/rejected": -0.02644043043255806, | |
| "step": 3615 | |
| }, | |
| { | |
| "epoch": 0.9320288362512873, | |
| "grad_norm": 156.0, | |
| "learning_rate": 3.398558187435633e-08, | |
| "logits/chosen": -3.237499952316284, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -241.8000030517578, | |
| "logps/rejected": -198.6999969482422, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.16362304985523224, | |
| "rewards/margins": 0.2529296875, | |
| "rewards/rejected": -0.08906249701976776, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.9333161688980433, | |
| "grad_norm": 210.0, | |
| "learning_rate": 3.334191555097837e-08, | |
| "logits/chosen": -3.109375, | |
| "logits/rejected": -3.121875047683716, | |
| "logps/chosen": -272.6000061035156, | |
| "logps/rejected": -222.0, | |
| "loss": 0.5984, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.14384765923023224, | |
| "rewards/margins": 0.2730468809604645, | |
| "rewards/rejected": -0.12919922173023224, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.9346035015447992, | |
| "grad_norm": 173.0, | |
| "learning_rate": 3.269824922760041e-08, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.065624952316284, | |
| "logps/chosen": -289.6000061035156, | |
| "logps/rejected": -301.3999938964844, | |
| "loss": 0.6539, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.22919921576976776, | |
| "rewards/margins": 0.19843749701976776, | |
| "rewards/rejected": 0.03022461012005806, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 0.935890834191555, | |
| "grad_norm": 204.0, | |
| "learning_rate": 3.205458290422245e-08, | |
| "logits/chosen": -3.1343750953674316, | |
| "logits/rejected": -3.1500000953674316, | |
| "logps/chosen": -275.3999938964844, | |
| "logps/rejected": -245.60000610351562, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.12482909858226776, | |
| "rewards/margins": 0.20070800185203552, | |
| "rewards/rejected": -0.07583007961511612, | |
| "step": 3635 | |
| }, | |
| { | |
| "epoch": 0.937178166838311, | |
| "grad_norm": 111.0, | |
| "learning_rate": 3.141091658084449e-08, | |
| "logits/chosen": -3.112499952316284, | |
| "logits/rejected": -3.1812500953674316, | |
| "logps/chosen": -282.3999938964844, | |
| "logps/rejected": -241.1999969482422, | |
| "loss": 0.5992, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.31269532442092896, | |
| "rewards/margins": 0.30009764432907104, | |
| "rewards/rejected": 0.012939453125, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.938465499485067, | |
| "grad_norm": 209.0, | |
| "learning_rate": 3.076725025746653e-08, | |
| "logits/chosen": -3.2593750953674316, | |
| "logits/rejected": -3.2593750953674316, | |
| "logps/chosen": -356.79998779296875, | |
| "logps/rejected": -272.0, | |
| "loss": 0.6547, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.20356445014476776, | |
| "rewards/margins": 0.19970703125, | |
| "rewards/rejected": 0.0038085938431322575, | |
| "step": 3645 | |
| }, | |
| { | |
| "epoch": 0.9397528321318228, | |
| "grad_norm": 188.0, | |
| "learning_rate": 3.0123583934088567e-08, | |
| "logits/chosen": -2.801562547683716, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -316.3999938964844, | |
| "logps/rejected": -264.3999938964844, | |
| "loss": 0.6156, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.2706054747104645, | |
| "rewards/margins": 0.24980469048023224, | |
| "rewards/rejected": 0.02082519605755806, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.9410401647785788, | |
| "grad_norm": 187.0, | |
| "learning_rate": 2.9479917610710606e-08, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -306.3999938964844, | |
| "logps/rejected": -272.79998779296875, | |
| "loss": 0.6008, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.2607421875, | |
| "rewards/margins": 0.29853516817092896, | |
| "rewards/rejected": -0.03745117038488388, | |
| "step": 3655 | |
| }, | |
| { | |
| "epoch": 0.9423274974253347, | |
| "grad_norm": 274.0, | |
| "learning_rate": 2.8836251287332645e-08, | |
| "logits/chosen": -3.28125, | |
| "logits/rejected": -3.2562499046325684, | |
| "logps/chosen": -304.0, | |
| "logps/rejected": -256.6000061035156, | |
| "loss": 0.6383, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.21503905951976776, | |
| "rewards/margins": 0.21127930283546448, | |
| "rewards/rejected": 0.00408935546875, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.9436148300720907, | |
| "grad_norm": 190.0, | |
| "learning_rate": 2.8192584963954685e-08, | |
| "logits/chosen": -3.25, | |
| "logits/rejected": -3.331249952316284, | |
| "logps/chosen": -275.20001220703125, | |
| "logps/rejected": -254.60000610351562, | |
| "loss": 0.5961, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.22675780951976776, | |
| "rewards/margins": 0.31328123807907104, | |
| "rewards/rejected": -0.08681640774011612, | |
| "step": 3665 | |
| }, | |
| { | |
| "epoch": 0.9449021627188465, | |
| "grad_norm": 201.0, | |
| "learning_rate": 2.7548918640576724e-08, | |
| "logits/chosen": -3.0625, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -243.1999969482422, | |
| "logps/rejected": -188.0, | |
| "loss": 0.6172, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.1763153076171875, | |
| "rewards/margins": 0.2109375, | |
| "rewards/rejected": -0.03471069410443306, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 0.9461894953656025, | |
| "grad_norm": 212.0, | |
| "learning_rate": 2.6905252317198764e-08, | |
| "logits/chosen": -3.090625047683716, | |
| "logits/rejected": -3.1812500953674316, | |
| "logps/chosen": -230.60000610351562, | |
| "logps/rejected": -155.0, | |
| "loss": 0.6469, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.2138671875, | |
| "rewards/margins": 0.15771484375, | |
| "rewards/rejected": 0.05624999850988388, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.9474768280123584, | |
| "grad_norm": 200.0, | |
| "learning_rate": 2.6261585993820803e-08, | |
| "logits/chosen": -3.034374952316284, | |
| "logits/rejected": -2.9749999046325684, | |
| "logps/chosen": -235.6999969482422, | |
| "logps/rejected": -212.8000030517578, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.1455078125, | |
| "rewards/margins": 0.2255859375, | |
| "rewards/rejected": -0.08006592094898224, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.9487641606591143, | |
| "grad_norm": 330.0, | |
| "learning_rate": 2.5617919670442842e-08, | |
| "logits/chosen": -3.0374999046325684, | |
| "logits/rejected": -2.9781250953674316, | |
| "logps/chosen": -246.89999389648438, | |
| "logps/rejected": -256.79998779296875, | |
| "loss": 0.6254, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.21298828721046448, | |
| "rewards/margins": 0.2412109375, | |
| "rewards/rejected": -0.02812499925494194, | |
| "step": 3685 | |
| }, | |
| { | |
| "epoch": 0.9500514933058702, | |
| "grad_norm": 225.0, | |
| "learning_rate": 2.497425334706488e-08, | |
| "logits/chosen": -3.034374952316284, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -301.3999938964844, | |
| "logps/rejected": -232.8000030517578, | |
| "loss": 0.6125, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.19809570908546448, | |
| "rewards/margins": 0.2586425840854645, | |
| "rewards/rejected": -0.06069336086511612, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 0.9513388259526262, | |
| "grad_norm": 228.0, | |
| "learning_rate": 2.433058702368692e-08, | |
| "logits/chosen": -3.15625, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -293.6000061035156, | |
| "logps/rejected": -239.8000030517578, | |
| "loss": 0.6219, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.00048828125, | |
| "rewards/margins": 0.25859373807907104, | |
| "rewards/rejected": -0.2585037350654602, | |
| "step": 3695 | |
| }, | |
| { | |
| "epoch": 0.952626158599382, | |
| "grad_norm": 227.0, | |
| "learning_rate": 2.368692070030896e-08, | |
| "logits/chosen": -3.059375047683716, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -289.70001220703125, | |
| "logps/rejected": -250.0, | |
| "loss": 0.5922, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.3734374940395355, | |
| "rewards/margins": 0.35136717557907104, | |
| "rewards/rejected": 0.02231445349752903, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.953913491246138, | |
| "grad_norm": 216.0, | |
| "learning_rate": 2.3043254376931e-08, | |
| "logits/chosen": -3.1875, | |
| "logits/rejected": -3.237499952316284, | |
| "logps/chosen": -300.3999938964844, | |
| "logps/rejected": -213.60000610351562, | |
| "loss": 0.6617, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.15087890625, | |
| "rewards/margins": 0.16909179091453552, | |
| "rewards/rejected": -0.01850586012005806, | |
| "step": 3705 | |
| }, | |
| { | |
| "epoch": 0.955200823892894, | |
| "grad_norm": 194.0, | |
| "learning_rate": 2.239958805355304e-08, | |
| "logits/chosen": -3.018749952316284, | |
| "logits/rejected": -3.203125, | |
| "logps/chosen": -231.39999389648438, | |
| "logps/rejected": -190.6999969482422, | |
| "loss": 0.6602, | |
| "rewards/accuracies": 0.3499999940395355, | |
| "rewards/chosen": 0.11689452826976776, | |
| "rewards/margins": 0.1304931640625, | |
| "rewards/rejected": -0.0137939453125, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 0.9564881565396498, | |
| "grad_norm": 143.0, | |
| "learning_rate": 2.1755921730175075e-08, | |
| "logits/chosen": -3.0093750953674316, | |
| "logits/rejected": -3.1312499046325684, | |
| "logps/chosen": -240.5, | |
| "logps/rejected": -194.0, | |
| "loss": 0.65, | |
| "rewards/accuracies": 0.4749999940395355, | |
| "rewards/chosen": 0.14113768935203552, | |
| "rewards/margins": 0.14545898139476776, | |
| "rewards/rejected": -0.0045166015625, | |
| "step": 3715 | |
| }, | |
| { | |
| "epoch": 0.9577754891864058, | |
| "grad_norm": 246.0, | |
| "learning_rate": 2.1112255406797115e-08, | |
| "logits/chosen": -3.159374952316284, | |
| "logits/rejected": -3.234375, | |
| "logps/chosen": -287.3999938964844, | |
| "logps/rejected": -222.39999389648438, | |
| "loss": 0.6453, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.07766113430261612, | |
| "rewards/margins": 0.22709961235523224, | |
| "rewards/rejected": -0.14912109076976776, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.9590628218331617, | |
| "grad_norm": 262.0, | |
| "learning_rate": 2.0468589083419154e-08, | |
| "logits/chosen": -3.206249952316284, | |
| "logits/rejected": -3.003124952316284, | |
| "logps/chosen": -239.8000030517578, | |
| "logps/rejected": -220.89999389648438, | |
| "loss": 0.657, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.04658203199505806, | |
| "rewards/margins": 0.18295899033546448, | |
| "rewards/rejected": -0.22978515923023224, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.9603501544799176, | |
| "grad_norm": 204.0, | |
| "learning_rate": 1.9824922760041193e-08, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -3.1343750953674316, | |
| "logps/chosen": -295.6000061035156, | |
| "logps/rejected": -228.60000610351562, | |
| "loss": 0.6312, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.19609375298023224, | |
| "rewards/margins": 0.24118652939796448, | |
| "rewards/rejected": -0.044921875, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 0.9616374871266735, | |
| "grad_norm": 192.0, | |
| "learning_rate": 1.9181256436663233e-08, | |
| "logits/chosen": -3.221874952316284, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -277.0, | |
| "logps/rejected": -259.79998779296875, | |
| "loss": 0.5859, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.30292969942092896, | |
| "rewards/margins": 0.30976563692092896, | |
| "rewards/rejected": -0.0072998045943677425, | |
| "step": 3735 | |
| }, | |
| { | |
| "epoch": 0.9629248197734295, | |
| "grad_norm": 194.0, | |
| "learning_rate": 1.8537590113285272e-08, | |
| "logits/chosen": -3.0, | |
| "logits/rejected": -3.049999952316284, | |
| "logps/chosen": -199.5, | |
| "logps/rejected": -200.39999389648438, | |
| "loss": 0.6156, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.06118164211511612, | |
| "rewards/margins": 0.21455077826976776, | |
| "rewards/rejected": -0.15314941108226776, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.9642121524201854, | |
| "grad_norm": 199.0, | |
| "learning_rate": 1.789392378990731e-08, | |
| "logits/chosen": -3.096874952316284, | |
| "logits/rejected": -3.065624952316284, | |
| "logps/chosen": -267.79998779296875, | |
| "logps/rejected": -241.8000030517578, | |
| "loss": 0.5625, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.28271484375, | |
| "rewards/margins": 0.37910157442092896, | |
| "rewards/rejected": -0.09638671576976776, | |
| "step": 3745 | |
| }, | |
| { | |
| "epoch": 0.9654994850669413, | |
| "grad_norm": 208.0, | |
| "learning_rate": 1.725025746652935e-08, | |
| "logits/chosen": -2.971874952316284, | |
| "logits/rejected": -3.253124952316284, | |
| "logps/chosen": -189.60000610351562, | |
| "logps/rejected": -194.5, | |
| "loss": 0.6891, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": 0.15531006455421448, | |
| "rewards/margins": 0.07856445014476776, | |
| "rewards/rejected": 0.07664795219898224, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.9667868177136972, | |
| "grad_norm": 185.0, | |
| "learning_rate": 1.660659114315139e-08, | |
| "logits/chosen": -3.1468749046325684, | |
| "logits/rejected": -3.253124952316284, | |
| "logps/chosen": -255.0, | |
| "logps/rejected": -234.39999389648438, | |
| "loss": 0.6078, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.00991210900247097, | |
| "rewards/margins": 0.2867187559604645, | |
| "rewards/rejected": -0.2763671875, | |
| "step": 3755 | |
| }, | |
| { | |
| "epoch": 0.9680741503604532, | |
| "grad_norm": 213.0, | |
| "learning_rate": 1.596292481977343e-08, | |
| "logits/chosen": -3.2125000953674316, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -262.0, | |
| "logps/rejected": -230.0, | |
| "loss": 0.6035, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.0032470703590661287, | |
| "rewards/margins": 0.2923828065395355, | |
| "rewards/rejected": -0.2958007752895355, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.969361483007209, | |
| "grad_norm": 292.0, | |
| "learning_rate": 1.531925849639547e-08, | |
| "logits/chosen": -3.0687499046325684, | |
| "logits/rejected": -2.965625047683716, | |
| "logps/chosen": -257.3999938964844, | |
| "logps/rejected": -216.0, | |
| "loss": 0.6375, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.102569580078125, | |
| "rewards/margins": 0.2158203125, | |
| "rewards/rejected": -0.1134033203125, | |
| "step": 3765 | |
| }, | |
| { | |
| "epoch": 0.970648815653965, | |
| "grad_norm": 222.0, | |
| "learning_rate": 1.4675592173017507e-08, | |
| "logits/chosen": -3.090625047683716, | |
| "logits/rejected": -2.7874999046325684, | |
| "logps/chosen": -258.70001220703125, | |
| "logps/rejected": -255.39999389648438, | |
| "loss": 0.5938, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.18955078721046448, | |
| "rewards/margins": 0.30000001192092896, | |
| "rewards/rejected": -0.11054687201976776, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 0.971936148300721, | |
| "grad_norm": 276.0, | |
| "learning_rate": 1.4031925849639546e-08, | |
| "logits/chosen": -3.0999999046325684, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -302.3999938964844, | |
| "logps/rejected": -265.3999938964844, | |
| "loss": 0.5875, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.3929687440395355, | |
| "rewards/margins": 0.3042968809604645, | |
| "rewards/rejected": 0.08896484225988388, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.9732234809474768, | |
| "grad_norm": 177.0, | |
| "learning_rate": 1.3388259526261585e-08, | |
| "logits/chosen": -3.1187500953674316, | |
| "logits/rejected": -3.0625, | |
| "logps/chosen": -263.6000061035156, | |
| "logps/rejected": -254.0, | |
| "loss": 0.5758, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.24560546875, | |
| "rewards/margins": 0.3070312440395355, | |
| "rewards/rejected": -0.06125488132238388, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.9745108135942327, | |
| "grad_norm": 248.0, | |
| "learning_rate": 1.2744593202883625e-08, | |
| "logits/chosen": -3.0531249046325684, | |
| "logits/rejected": -3.1468749046325684, | |
| "logps/chosen": -290.20001220703125, | |
| "logps/rejected": -230.8000030517578, | |
| "loss": 0.6344, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.23293456435203552, | |
| "rewards/margins": 0.19414062798023224, | |
| "rewards/rejected": 0.03820495679974556, | |
| "step": 3785 | |
| }, | |
| { | |
| "epoch": 0.9757981462409887, | |
| "grad_norm": 204.0, | |
| "learning_rate": 1.2100926879505664e-08, | |
| "logits/chosen": -3.078125, | |
| "logits/rejected": -3.171875, | |
| "logps/chosen": -225.39999389648438, | |
| "logps/rejected": -225.4499969482422, | |
| "loss": 0.6203, | |
| "rewards/accuracies": 0.4625000059604645, | |
| "rewards/chosen": 0.184326171875, | |
| "rewards/margins": 0.24467773735523224, | |
| "rewards/rejected": -0.06074218824505806, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 0.9770854788877446, | |
| "grad_norm": 200.0, | |
| "learning_rate": 1.1457260556127703e-08, | |
| "logits/chosen": -3.128124952316284, | |
| "logits/rejected": -3.253124952316284, | |
| "logps/chosen": -242.1999969482422, | |
| "logps/rejected": -254.8000030517578, | |
| "loss": 0.6164, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.09205321967601776, | |
| "rewards/margins": 0.24550780653953552, | |
| "rewards/rejected": -0.15328064560890198, | |
| "step": 3795 | |
| }, | |
| { | |
| "epoch": 0.9783728115345005, | |
| "grad_norm": 268.0, | |
| "learning_rate": 1.0813594232749741e-08, | |
| "logits/chosen": -3.2093749046325684, | |
| "logits/rejected": -3.1875, | |
| "logps/chosen": -297.0, | |
| "logps/rejected": -257.79998779296875, | |
| "loss": 0.6148, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.09401855617761612, | |
| "rewards/margins": 0.27314454317092896, | |
| "rewards/rejected": -0.17918701469898224, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.9796601441812565, | |
| "grad_norm": 194.0, | |
| "learning_rate": 1.016992790937178e-08, | |
| "logits/chosen": -3.253124952316284, | |
| "logits/rejected": -3.221874952316284, | |
| "logps/chosen": -234.8000030517578, | |
| "logps/rejected": -246.1999969482422, | |
| "loss": 0.5922, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -0.04604492336511612, | |
| "rewards/margins": 0.291015625, | |
| "rewards/rejected": -0.3365722596645355, | |
| "step": 3805 | |
| }, | |
| { | |
| "epoch": 0.9809474768280123, | |
| "grad_norm": 242.0, | |
| "learning_rate": 9.52626158599382e-09, | |
| "logits/chosen": -3.190624952316284, | |
| "logits/rejected": -3.2406249046325684, | |
| "logps/chosen": -236.1999969482422, | |
| "logps/rejected": -205.1999969482422, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.512499988079071, | |
| "rewards/chosen": 0.05050048977136612, | |
| "rewards/margins": 0.11796875298023224, | |
| "rewards/rejected": -0.06728515774011612, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 0.9822348094747683, | |
| "grad_norm": 187.0, | |
| "learning_rate": 8.88259526261586e-09, | |
| "logits/chosen": -3.1937499046325684, | |
| "logits/rejected": -3.246875047683716, | |
| "logps/chosen": -228.1999969482422, | |
| "logps/rejected": -227.10000610351562, | |
| "loss": 0.6328, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.19858399033546448, | |
| "rewards/margins": 0.20854492485523224, | |
| "rewards/rejected": -0.009326172061264515, | |
| "step": 3815 | |
| }, | |
| { | |
| "epoch": 0.9835221421215242, | |
| "grad_norm": 235.0, | |
| "learning_rate": 8.238928939237899e-09, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -3.0875000953674316, | |
| "logps/chosen": -254.0, | |
| "logps/rejected": -275.79998779296875, | |
| "loss": 0.6648, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.09697265923023224, | |
| "rewards/margins": 0.16191406548023224, | |
| "rewards/rejected": -0.06552734225988388, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.9848094747682801, | |
| "grad_norm": 278.0, | |
| "learning_rate": 7.595262615859938e-09, | |
| "logits/chosen": -3.1312499046325684, | |
| "logits/rejected": -3.140625, | |
| "logps/chosen": -246.1999969482422, | |
| "logps/rejected": -271.79998779296875, | |
| "loss": 0.7031, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.01826171949505806, | |
| "rewards/margins": 0.06210937350988388, | |
| "rewards/rejected": -0.08037109673023224, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.986096807415036, | |
| "grad_norm": 274.0, | |
| "learning_rate": 6.951596292481977e-09, | |
| "logits/chosen": -3.137500047683716, | |
| "logits/rejected": -3.200000047683716, | |
| "logps/chosen": -323.20001220703125, | |
| "logps/rejected": -273.0, | |
| "loss": 0.6441, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.12016601860523224, | |
| "rewards/margins": 0.23681640625, | |
| "rewards/rejected": -0.11748047173023224, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 0.987384140061792, | |
| "grad_norm": 248.0, | |
| "learning_rate": 6.307929969104016e-09, | |
| "logits/chosen": -3.246875047683716, | |
| "logits/rejected": -3.003124952316284, | |
| "logps/chosen": -310.6000061035156, | |
| "logps/rejected": -293.20001220703125, | |
| "loss": 0.6062, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.03066406212747097, | |
| "rewards/margins": 0.2754882872104645, | |
| "rewards/rejected": -0.24458007514476776, | |
| "step": 3835 | |
| }, | |
| { | |
| "epoch": 0.9886714727085479, | |
| "grad_norm": 232.0, | |
| "learning_rate": 5.664263645726055e-09, | |
| "logits/chosen": -3.125, | |
| "logits/rejected": -3.2249999046325684, | |
| "logps/chosen": -321.6000061035156, | |
| "logps/rejected": -283.6000061035156, | |
| "loss": 0.6445, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.2958007752895355, | |
| "rewards/margins": 0.2117919921875, | |
| "rewards/rejected": 0.08476562798023224, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.9899588053553038, | |
| "grad_norm": 200.0, | |
| "learning_rate": 5.020597322348095e-09, | |
| "logits/chosen": -3.106250047683716, | |
| "logits/rejected": -3.2874999046325684, | |
| "logps/chosen": -364.79998779296875, | |
| "logps/rejected": -269.3999938964844, | |
| "loss": 0.543, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.36962890625, | |
| "rewards/margins": 0.4359374940395355, | |
| "rewards/rejected": -0.06679687649011612, | |
| "step": 3845 | |
| }, | |
| { | |
| "epoch": 0.9912461380020597, | |
| "grad_norm": 247.0, | |
| "learning_rate": 4.376930998970134e-09, | |
| "logits/chosen": -3.2750000953674316, | |
| "logits/rejected": -3.237499952316284, | |
| "logps/chosen": -336.0, | |
| "logps/rejected": -253.0, | |
| "loss": 0.5836, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.2777343690395355, | |
| "rewards/margins": 0.3472656309604645, | |
| "rewards/rejected": -0.06899261474609375, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.9925334706488157, | |
| "grad_norm": 169.0, | |
| "learning_rate": 3.733264675592173e-09, | |
| "logits/chosen": -3.043750047683716, | |
| "logits/rejected": -3.0687499046325684, | |
| "logps/chosen": -240.8000030517578, | |
| "logps/rejected": -263.79998779296875, | |
| "loss": 0.6031, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.130859375, | |
| "rewards/margins": 0.31132811307907104, | |
| "rewards/rejected": -0.18056640028953552, | |
| "step": 3855 | |
| }, | |
| { | |
| "epoch": 0.9938208032955715, | |
| "grad_norm": 173.0, | |
| "learning_rate": 3.089598352214212e-09, | |
| "logits/chosen": -3.2281250953674316, | |
| "logits/rejected": -3.265625, | |
| "logps/chosen": -312.79998779296875, | |
| "logps/rejected": -274.20001220703125, | |
| "loss": 0.6016, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.15561524033546448, | |
| "rewards/margins": 0.2874999940395355, | |
| "rewards/rejected": -0.13229981064796448, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.9951081359423275, | |
| "grad_norm": 211.0, | |
| "learning_rate": 2.445932028836251e-09, | |
| "logits/chosen": -3.0999999046325684, | |
| "logits/rejected": -3.1656250953674316, | |
| "logps/chosen": -326.0, | |
| "logps/rejected": -293.6000061035156, | |
| "loss": 0.6191, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.21220703423023224, | |
| "rewards/margins": 0.2823242247104645, | |
| "rewards/rejected": -0.07011719048023224, | |
| "step": 3865 | |
| }, | |
| { | |
| "epoch": 0.9963954685890835, | |
| "grad_norm": 234.0, | |
| "learning_rate": 1.8022657054582903e-09, | |
| "logits/chosen": -3.2437500953674316, | |
| "logits/rejected": -3.184375047683716, | |
| "logps/chosen": -300.79998779296875, | |
| "logps/rejected": -284.3999938964844, | |
| "loss": 0.5766, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.14130859076976776, | |
| "rewards/margins": 0.3531250059604645, | |
| "rewards/rejected": -0.21187743544578552, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 0.9976828012358393, | |
| "grad_norm": 231.0, | |
| "learning_rate": 1.1585993820803295e-09, | |
| "logits/chosen": -3.1312499046325684, | |
| "logits/rejected": -3.2437500953674316, | |
| "logps/chosen": -303.6000061035156, | |
| "logps/rejected": -237.8000030517578, | |
| "loss": 0.6398, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.23046875, | |
| "rewards/margins": 0.22646483778953552, | |
| "rewards/rejected": 0.0038085938431322575, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.9989701338825953, | |
| "grad_norm": 192.0, | |
| "learning_rate": 5.149330587023687e-10, | |
| "logits/chosen": -3.174999952316284, | |
| "logits/rejected": -3.078125, | |
| "logps/chosen": -276.3999938964844, | |
| "logps/rejected": -206.8000030517578, | |
| "loss": 0.6508, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.16433104872703552, | |
| "rewards/margins": 0.16865234076976776, | |
| "rewards/rejected": -0.004150390625, | |
| "step": 3880 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 3884, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 3000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |