| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9990600241708071, | |
| "eval_steps": 100000, | |
| "global_step": 465, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.002148516181012488, | |
| "grad_norm": 4.815295005445406, | |
| "learning_rate": 1.0638297872340425e-08, | |
| "logits/chosen": -11.7161283493042, | |
| "logits/rejected": -11.149711608886719, | |
| "logps/chosen": -155.27786254882812, | |
| "logps/rejected": -143.72726440429688, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.010742580905062441, | |
| "grad_norm": 4.665144316910627, | |
| "learning_rate": 5.3191489361702123e-08, | |
| "logits/chosen": -10.514410018920898, | |
| "logits/rejected": -10.558334350585938, | |
| "logps/chosen": -176.54898071289062, | |
| "logps/rejected": -177.0060577392578, | |
| "loss": 0.6932, | |
| "rewards/accuracies": 0.390625, | |
| "rewards/chosen": 0.0011347520630806684, | |
| "rewards/margins": 0.0005755239981226623, | |
| "rewards/rejected": 0.0005592280067503452, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.021485161810124883, | |
| "grad_norm": 4.631928897966055, | |
| "learning_rate": 1.0638297872340425e-07, | |
| "logits/chosen": -11.231101989746094, | |
| "logits/rejected": -11.33354377746582, | |
| "logps/chosen": -200.58383178710938, | |
| "logps/rejected": -198.61624145507812, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.0011275215074419975, | |
| "rewards/margins": 0.001252257265150547, | |
| "rewards/rejected": -0.00012473580136429518, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.03222774271518732, | |
| "grad_norm": 4.861132978246578, | |
| "learning_rate": 1.5957446808510638e-07, | |
| "logits/chosen": -10.90822696685791, | |
| "logits/rejected": -10.716678619384766, | |
| "logps/chosen": -216.2873992919922, | |
| "logps/rejected": -203.54527282714844, | |
| "loss": 0.6932, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": -0.0005270905676297843, | |
| "rewards/margins": 4.5846936700399965e-05, | |
| "rewards/rejected": -0.0005729374242946506, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.042970323620249766, | |
| "grad_norm": 4.809482719935914, | |
| "learning_rate": 2.127659574468085e-07, | |
| "logits/chosen": -11.042614936828613, | |
| "logits/rejected": -10.997823715209961, | |
| "logps/chosen": -172.80630493164062, | |
| "logps/rejected": -172.10458374023438, | |
| "loss": 0.693, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": -0.00028278256650082767, | |
| "rewards/margins": 2.6181707653449848e-05, | |
| "rewards/rejected": -0.000308964226860553, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05371290452531221, | |
| "grad_norm": 4.825836096297444, | |
| "learning_rate": 2.659574468085106e-07, | |
| "logits/chosen": -11.554967880249023, | |
| "logits/rejected": -11.35639762878418, | |
| "logps/chosen": -191.2166290283203, | |
| "logps/rejected": -184.11294555664062, | |
| "loss": 0.6928, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.0014816813636571169, | |
| "rewards/margins": -0.0005183071480132639, | |
| "rewards/rejected": 0.0019999886862933636, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.06445548543037465, | |
| "grad_norm": 4.717146773464201, | |
| "learning_rate": 3.1914893617021275e-07, | |
| "logits/chosen": -11.667179107666016, | |
| "logits/rejected": -11.451859474182129, | |
| "logps/chosen": -185.59962463378906, | |
| "logps/rejected": -185.79769897460938, | |
| "loss": 0.6923, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.0023945302236825228, | |
| "rewards/margins": 0.0009661655058152974, | |
| "rewards/rejected": 0.0014283646596595645, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07519806633543709, | |
| "grad_norm": 4.359788436218923, | |
| "learning_rate": 3.7234042553191484e-07, | |
| "logits/chosen": -10.990551948547363, | |
| "logits/rejected": -10.911511421203613, | |
| "logps/chosen": -160.88067626953125, | |
| "logps/rejected": -162.3435821533203, | |
| "loss": 0.6919, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.0046579958871006966, | |
| "rewards/margins": 0.0034069078974425793, | |
| "rewards/rejected": 0.0012510876404121518, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08594064724049953, | |
| "grad_norm": 4.682184048048129, | |
| "learning_rate": 4.25531914893617e-07, | |
| "logits/chosen": -10.937858581542969, | |
| "logits/rejected": -10.48929500579834, | |
| "logps/chosen": -167.89256286621094, | |
| "logps/rejected": -166.53628540039062, | |
| "loss": 0.6905, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.008901619352400303, | |
| "rewards/margins": 0.00515585346147418, | |
| "rewards/rejected": 0.0037457651924341917, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09668322814556198, | |
| "grad_norm": 4.77155478266984, | |
| "learning_rate": 4.787234042553192e-07, | |
| "logits/chosen": -11.142602920532227, | |
| "logits/rejected": -10.850725173950195, | |
| "logps/chosen": -175.64431762695312, | |
| "logps/rejected": -174.0598602294922, | |
| "loss": 0.689, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.008818638511002064, | |
| "rewards/margins": 0.006885365583002567, | |
| "rewards/rejected": 0.0019332747906446457, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10742580905062442, | |
| "grad_norm": 5.101867612304127, | |
| "learning_rate": 4.999364550115248e-07, | |
| "logits/chosen": -12.041606903076172, | |
| "logits/rejected": -11.635477066040039, | |
| "logps/chosen": -182.34841918945312, | |
| "logps/rejected": -178.40005493164062, | |
| "loss": 0.6876, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.008523710072040558, | |
| "rewards/margins": 0.016866037622094154, | |
| "rewards/rejected": -0.008342329412698746, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11816838995568685, | |
| "grad_norm": 5.096625071764424, | |
| "learning_rate": 4.995482415049122e-07, | |
| "logits/chosen": -11.534649848937988, | |
| "logits/rejected": -11.349666595458984, | |
| "logps/chosen": -186.76443481445312, | |
| "logps/rejected": -180.79470825195312, | |
| "loss": 0.6831, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.0006141850026324391, | |
| "rewards/margins": 0.01560161542147398, | |
| "rewards/rejected": -0.014987431466579437, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.1289109708607493, | |
| "grad_norm": 5.256195598008515, | |
| "learning_rate": 4.98807664757556e-07, | |
| "logits/chosen": -12.30699634552002, | |
| "logits/rejected": -11.902242660522461, | |
| "logps/chosen": -209.91812133789062, | |
| "logps/rejected": -200.37374877929688, | |
| "loss": 0.6758, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.011014172807335854, | |
| "rewards/margins": 0.033601436764001846, | |
| "rewards/rejected": -0.04461560770869255, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.13965355176581173, | |
| "grad_norm": 5.332372855512194, | |
| "learning_rate": 4.977157704671584e-07, | |
| "logits/chosen": -11.759092330932617, | |
| "logits/rejected": -11.736629486083984, | |
| "logps/chosen": -191.42971801757812, | |
| "logps/rejected": -190.9018096923828, | |
| "loss": 0.6745, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.05662290379405022, | |
| "rewards/margins": 0.05431179329752922, | |
| "rewards/rejected": -0.11093469709157944, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.15039613267087418, | |
| "grad_norm": 5.384194766315848, | |
| "learning_rate": 4.962741003933742e-07, | |
| "logits/chosen": -12.996328353881836, | |
| "logits/rejected": -12.800023078918457, | |
| "logps/chosen": -205.3074188232422, | |
| "logps/rejected": -207.24588012695312, | |
| "loss": 0.6696, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.10499341785907745, | |
| "rewards/margins": 0.028122153133153915, | |
| "rewards/rejected": -0.13311555981636047, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.16113871357593662, | |
| "grad_norm": 6.074413377005886, | |
| "learning_rate": 4.944846901808396e-07, | |
| "logits/chosen": -13.731378555297852, | |
| "logits/rejected": -13.588099479675293, | |
| "logps/chosen": -207.2079620361328, | |
| "logps/rejected": -211.04074096679688, | |
| "loss": 0.6575, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.15880520641803741, | |
| "rewards/margins": 0.08157539367675781, | |
| "rewards/rejected": -0.24038061499595642, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.17188129448099906, | |
| "grad_norm": 6.452938390183295, | |
| "learning_rate": 4.923500664848326e-07, | |
| "logits/chosen": -13.624473571777344, | |
| "logits/rejected": -13.508392333984375, | |
| "logps/chosen": -218.89053344726562, | |
| "logps/rejected": -221.2880401611328, | |
| "loss": 0.6605, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.2407272756099701, | |
| "rewards/margins": 0.07505510747432709, | |
| "rewards/rejected": -0.3157823979854584, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1826238753860615, | |
| "grad_norm": 7.438055763179647, | |
| "learning_rate": 4.898732434036243e-07, | |
| "logits/chosen": -14.367892265319824, | |
| "logits/rejected": -14.180076599121094, | |
| "logps/chosen": -186.84841918945312, | |
| "logps/rejected": -204.40371704101562, | |
| "loss": 0.6488, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2892145812511444, | |
| "rewards/margins": 0.13178381323814392, | |
| "rewards/rejected": -0.4209984242916107, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.19336645629112395, | |
| "grad_norm": 7.432291163652407, | |
| "learning_rate": 4.870577182225589e-07, | |
| "logits/chosen": -14.431337356567383, | |
| "logits/rejected": -14.528360366821289, | |
| "logps/chosen": -235.01089477539062, | |
| "logps/rejected": -242.1006622314453, | |
| "loss": 0.6544, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.3304561972618103, | |
| "rewards/margins": 0.0850444883108139, | |
| "rewards/rejected": -0.4155007302761078, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2041090371961864, | |
| "grad_norm": 7.26568179641241, | |
| "learning_rate": 4.839074664758704e-07, | |
| "logits/chosen": -15.507716178894043, | |
| "logits/rejected": -15.54358196258545, | |
| "logps/chosen": -219.27432250976562, | |
| "logps/rejected": -232.99215698242188, | |
| "loss": 0.6392, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": -0.425987184047699, | |
| "rewards/margins": 0.14367976784706116, | |
| "rewards/rejected": -0.5696669816970825, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.21485161810124884, | |
| "grad_norm": 8.511941780068044, | |
| "learning_rate": 4.804269363332112e-07, | |
| "logits/chosen": -14.985662460327148, | |
| "logits/rejected": -14.705190658569336, | |
| "logps/chosen": -198.18209838867188, | |
| "logps/rejected": -206.1742401123047, | |
| "loss": 0.6385, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.440185546875, | |
| "rewards/margins": 0.12437309324741364, | |
| "rewards/rejected": -0.5645586252212524, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.22559419900631128, | |
| "grad_norm": 8.344983653724734, | |
| "learning_rate": 4.7662104231881575e-07, | |
| "logits/chosen": -15.85981273651123, | |
| "logits/rejected": -15.64745807647705, | |
| "logps/chosen": -245.8246612548828, | |
| "logps/rejected": -250.5571746826172, | |
| "loss": 0.6295, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.6445275545120239, | |
| "rewards/margins": 0.055053092539310455, | |
| "rewards/rejected": -0.6995806694030762, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.2363367799113737, | |
| "grad_norm": 8.805058588475875, | |
| "learning_rate": 4.724951583721707e-07, | |
| "logits/chosen": -16.028610229492188, | |
| "logits/rejected": -16.066186904907227, | |
| "logps/chosen": -226.5513153076172, | |
| "logps/rejected": -243.8558807373047, | |
| "loss": 0.6242, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.6525581479072571, | |
| "rewards/margins": 0.1965942531824112, | |
| "rewards/rejected": -0.8491523861885071, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.24707936081643614, | |
| "grad_norm": 9.436386913158618, | |
| "learning_rate": 4.68055110259988e-07, | |
| "logits/chosen": -17.072982788085938, | |
| "logits/rejected": -17.03071403503418, | |
| "logps/chosen": -255.8236541748047, | |
| "logps/rejected": -276.999267578125, | |
| "loss": 0.6015, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.7960177659988403, | |
| "rewards/margins": 0.268364280462265, | |
| "rewards/rejected": -1.0643821954727173, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2578219417214986, | |
| "grad_norm": 9.856903496627183, | |
| "learning_rate": 4.633071673501959e-07, | |
| "logits/chosen": -16.772096633911133, | |
| "logits/rejected": -16.724201202392578, | |
| "logps/chosen": -267.2545471191406, | |
| "logps/rejected": -286.8907470703125, | |
| "loss": 0.6037, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.9270988702774048, | |
| "rewards/margins": 0.17965544760227203, | |
| "rewards/rejected": -1.1067544221878052, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.26856452262656105, | |
| "grad_norm": 11.574882560301482, | |
| "learning_rate": 4.5825803375956355e-07, | |
| "logits/chosen": -18.174747467041016, | |
| "logits/rejected": -17.954605102539062, | |
| "logps/chosen": -286.30426025390625, | |
| "logps/rejected": -311.0242919921875, | |
| "loss": 0.612, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.0375328063964844, | |
| "rewards/margins": 0.28692322969436646, | |
| "rewards/rejected": -1.324455976486206, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.27930710353162347, | |
| "grad_norm": 12.175022480779115, | |
| "learning_rate": 4.529148388874576e-07, | |
| "logits/chosen": -18.297924041748047, | |
| "logits/rejected": -18.335847854614258, | |
| "logps/chosen": -281.75067138671875, | |
| "logps/rejected": -314.5852966308594, | |
| "loss": 0.6002, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.0217819213867188, | |
| "rewards/margins": 0.3027891218662262, | |
| "rewards/rejected": -1.324571132659912, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.29004968443668594, | |
| "grad_norm": 118.26976750176874, | |
| "learning_rate": 4.472851273490984e-07, | |
| "logits/chosen": -18.722200393676758, | |
| "logits/rejected": -18.794912338256836, | |
| "logps/chosen": -339.0028991699219, | |
| "logps/rejected": -374.34576416015625, | |
| "loss": 0.5702, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -1.2525867223739624, | |
| "rewards/margins": 0.4258871078491211, | |
| "rewards/rejected": -1.6784738302230835, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.30079226534174835, | |
| "grad_norm": 12.772135480895352, | |
| "learning_rate": 4.4137684832252913e-07, | |
| "logits/chosen": -18.832077026367188, | |
| "logits/rejected": -18.659208297729492, | |
| "logps/chosen": -345.35015869140625, | |
| "logps/rejected": -388.00726318359375, | |
| "loss": 0.5654, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.3465111255645752, | |
| "rewards/margins": 0.5241008400917053, | |
| "rewards/rejected": -1.8706119060516357, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.31153484624681077, | |
| "grad_norm": 12.383899725959822, | |
| "learning_rate": 4.351983443243409e-07, | |
| "logits/chosen": -19.82056999206543, | |
| "logits/rejected": -19.986614227294922, | |
| "logps/chosen": -354.8511657714844, | |
| "logps/rejected": -392.865966796875, | |
| "loss": 0.571, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.5482107400894165, | |
| "rewards/margins": 0.39296191930770874, | |
| "rewards/rejected": -1.9411723613739014, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.32227742715187324, | |
| "grad_norm": 14.735365105108821, | |
| "learning_rate": 4.287583394300015e-07, | |
| "logits/chosen": -18.688817977905273, | |
| "logits/rejected": -18.646465301513672, | |
| "logps/chosen": -351.5058288574219, | |
| "logps/rejected": -397.3396911621094, | |
| "loss": 0.5707, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -1.7229034900665283, | |
| "rewards/margins": 0.48934563994407654, | |
| "rewards/rejected": -2.2122488021850586, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33302000805693566, | |
| "grad_norm": 13.312063222247302, | |
| "learning_rate": 4.220659269554216e-07, | |
| "logits/chosen": -19.77194595336914, | |
| "logits/rejected": -19.732973098754883, | |
| "logps/chosen": -348.5200500488281, | |
| "logps/rejected": -388.44189453125, | |
| "loss": 0.5553, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.6245940923690796, | |
| "rewards/margins": 0.4843548834323883, | |
| "rewards/rejected": -2.1089487075805664, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.34376258896199813, | |
| "grad_norm": 12.019324231367111, | |
| "learning_rate": 4.1513055661715206e-07, | |
| "logits/chosen": -19.84881591796875, | |
| "logits/rejected": -19.640766143798828, | |
| "logps/chosen": -377.59075927734375, | |
| "logps/rejected": -410.1768493652344, | |
| "loss": 0.5591, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.8603988885879517, | |
| "rewards/margins": 0.37856483459472656, | |
| "rewards/rejected": -2.2389636039733887, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.35450516986706054, | |
| "grad_norm": 11.824585303309874, | |
| "learning_rate": 4.0796202118934096e-07, | |
| "logits/chosen": -19.047245025634766, | |
| "logits/rejected": -19.169963836669922, | |
| "logps/chosen": -342.62176513671875, | |
| "logps/rejected": -398.24237060546875, | |
| "loss": 0.5394, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.6620702743530273, | |
| "rewards/margins": 0.6188061237335205, | |
| "rewards/rejected": -2.280876636505127, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.365247750772123, | |
| "grad_norm": 11.817356726675012, | |
| "learning_rate": 4.0057044267629303e-07, | |
| "logits/chosen": -18.676464080810547, | |
| "logits/rejected": -18.591510772705078, | |
| "logps/chosen": -377.8108825683594, | |
| "logps/rejected": -435.5389709472656, | |
| "loss": 0.5568, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.849261999130249, | |
| "rewards/margins": 0.5185360312461853, | |
| "rewards/rejected": -2.367798328399658, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.37599033167718543, | |
| "grad_norm": 13.96823700531248, | |
| "learning_rate": 3.9296625802015354e-07, | |
| "logits/chosen": -19.861886978149414, | |
| "logits/rejected": -19.66740608215332, | |
| "logps/chosen": -365.8138122558594, | |
| "logps/rejected": -428.0379943847656, | |
| "loss": 0.5511, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.8249616622924805, | |
| "rewards/margins": 0.604793131351471, | |
| "rewards/rejected": -2.4297547340393066, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.3867329125822479, | |
| "grad_norm": 13.667628227402297, | |
| "learning_rate": 3.851602043638994e-07, | |
| "logits/chosen": -20.17488670349121, | |
| "logits/rejected": -20.171833038330078, | |
| "logps/chosen": -392.9968566894531, | |
| "logps/rejected": -458.20660400390625, | |
| "loss": 0.5262, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.052845001220703, | |
| "rewards/margins": 0.6375163793563843, | |
| "rewards/rejected": -2.690361499786377, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3974754934873103, | |
| "grad_norm": 15.39892517008571, | |
| "learning_rate": 3.771633038904446e-07, | |
| "logits/chosen": -20.305858612060547, | |
| "logits/rejected": -20.228055953979492, | |
| "logps/chosen": -416.68560791015625, | |
| "logps/rejected": -482.61676025390625, | |
| "loss": 0.5027, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.168687105178833, | |
| "rewards/margins": 0.6908921599388123, | |
| "rewards/rejected": -2.859579086303711, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.4082180743923728, | |
| "grad_norm": 12.128334417208155, | |
| "learning_rate": 3.689868482592684e-07, | |
| "logits/chosen": -20.484973907470703, | |
| "logits/rejected": -20.257816314697266, | |
| "logps/chosen": -466.16424560546875, | |
| "logps/rejected": -569.1168212890625, | |
| "loss": 0.5224, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -2.359057903289795, | |
| "rewards/margins": 1.0049209594726562, | |
| "rewards/rejected": -3.363978862762451, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.4189606552974352, | |
| "grad_norm": 15.400085837818166, | |
| "learning_rate": 3.606423826625414e-07, | |
| "logits/chosen": -20.399295806884766, | |
| "logits/rejected": -20.145183563232422, | |
| "logps/chosen": -396.54296875, | |
| "logps/rejected": -456.8973693847656, | |
| "loss": 0.5273, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.9560238122940063, | |
| "rewards/margins": 0.6280576586723328, | |
| "rewards/rejected": -2.5840814113616943, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.4297032362024977, | |
| "grad_norm": 14.374891311569105, | |
| "learning_rate": 3.5214168952326203e-07, | |
| "logits/chosen": -20.65167808532715, | |
| "logits/rejected": -20.682342529296875, | |
| "logps/chosen": -410.12579345703125, | |
| "logps/rejected": -451.20355224609375, | |
| "loss": 0.5324, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -2.331077814102173, | |
| "rewards/margins": 0.4432566165924072, | |
| "rewards/rejected": -2.77433443069458, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.4404458171075601, | |
| "grad_norm": 14.839772397078658, | |
| "learning_rate": 3.4349677185842243e-07, | |
| "logits/chosen": -20.4150333404541, | |
| "logits/rejected": -20.26569938659668, | |
| "logps/chosen": -389.5531311035156, | |
| "logps/rejected": -442.98388671875, | |
| "loss": 0.4873, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -2.1742894649505615, | |
| "rewards/margins": 0.5466232895851135, | |
| "rewards/rejected": -2.720912456512451, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.45118839801262256, | |
| "grad_norm": 73.54965957378546, | |
| "learning_rate": 3.347198363306941e-07, | |
| "logits/chosen": -20.8847713470459, | |
| "logits/rejected": -20.70589828491211, | |
| "logps/chosen": -467.5254821777344, | |
| "logps/rejected": -525.5687255859375, | |
| "loss": 0.5156, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -2.7870867252349854, | |
| "rewards/margins": 0.5678809881210327, | |
| "rewards/rejected": -3.3549678325653076, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.461930978917685, | |
| "grad_norm": 13.712145057455235, | |
| "learning_rate": 3.2582327601256564e-07, | |
| "logits/chosen": -21.396650314331055, | |
| "logits/rejected": -21.34550666809082, | |
| "logps/chosen": -439.74151611328125, | |
| "logps/rejected": -528.05908203125, | |
| "loss": 0.52, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -2.560230255126953, | |
| "rewards/margins": 0.9471774101257324, | |
| "rewards/rejected": -3.5074074268341064, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.4726735598227474, | |
| "grad_norm": 13.556889594752073, | |
| "learning_rate": 3.168196528872682e-07, | |
| "logits/chosen": -20.23183250427246, | |
| "logits/rejected": -19.94950294494629, | |
| "logps/chosen": -433.8387756347656, | |
| "logps/rejected": -496.75946044921875, | |
| "loss": 0.52, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -2.558835506439209, | |
| "rewards/margins": 0.7217434048652649, | |
| "rewards/rejected": -3.28057861328125, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.48341614072780986, | |
| "grad_norm": 22.472329896519035, | |
| "learning_rate": 3.077216801111989e-07, | |
| "logits/chosen": -20.4947452545166, | |
| "logits/rejected": -20.315448760986328, | |
| "logps/chosen": -434.6400451660156, | |
| "logps/rejected": -493.80181884765625, | |
| "loss": 0.4996, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.2407140731811523, | |
| "rewards/margins": 0.6915763020515442, | |
| "rewards/rejected": -2.932290554046631, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.4941587216328723, | |
| "grad_norm": 14.506146770106906, | |
| "learning_rate": 2.9854220406288666e-07, | |
| "logits/chosen": -20.191608428955078, | |
| "logits/rejected": -19.9714412689209, | |
| "logps/chosen": -414.80535888671875, | |
| "logps/rejected": -482.10015869140625, | |
| "loss": 0.4813, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": -2.2866439819335938, | |
| "rewards/margins": 0.7474405169487, | |
| "rewards/rejected": -3.0340847969055176, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5049013025379347, | |
| "grad_norm": 13.028250670943683, | |
| "learning_rate": 2.892941862038475e-07, | |
| "logits/chosen": -19.92314910888672, | |
| "logits/rejected": -20.13300895690918, | |
| "logps/chosen": -414.1224670410156, | |
| "logps/rejected": -509.6715393066406, | |
| "loss": 0.5218, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.3021442890167236, | |
| "rewards/margins": 0.8769405484199524, | |
| "rewards/rejected": -3.1790850162506104, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.5156438834429972, | |
| "grad_norm": 16.027232766680346, | |
| "learning_rate": 2.7999068477694325e-07, | |
| "logits/chosen": -20.204256057739258, | |
| "logits/rejected": -20.092927932739258, | |
| "logps/chosen": -400.50579833984375, | |
| "logps/rejected": -485.25225830078125, | |
| "loss": 0.4754, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -2.113093376159668, | |
| "rewards/margins": 0.857138454914093, | |
| "rewards/rejected": -2.970231771469116, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5263864643480596, | |
| "grad_norm": 14.352076936568274, | |
| "learning_rate": 2.706448363680831e-07, | |
| "logits/chosen": -20.70675277709961, | |
| "logits/rejected": -20.695838928222656, | |
| "logps/chosen": -476.7577209472656, | |
| "logps/rejected": -539.758544921875, | |
| "loss": 0.4996, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -2.558915615081787, | |
| "rewards/margins": 0.7049631476402283, | |
| "rewards/rejected": -3.2638790607452393, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.5371290452531221, | |
| "grad_norm": 13.840765525316273, | |
| "learning_rate": 2.6126983735730555e-07, | |
| "logits/chosen": -20.357967376708984, | |
| "logits/rejected": -20.35165023803711, | |
| "logps/chosen": -415.8984375, | |
| "logps/rejected": -537.2420654296875, | |
| "loss": 0.4602, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -2.3948779106140137, | |
| "rewards/margins": 1.2330429553985596, | |
| "rewards/rejected": -3.6279208660125732, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5478716261581845, | |
| "grad_norm": 22.763085198341653, | |
| "learning_rate": 2.518789252854305e-07, | |
| "logits/chosen": -20.021839141845703, | |
| "logits/rejected": -19.900529861450195, | |
| "logps/chosen": -421.52142333984375, | |
| "logps/rejected": -513.4830322265625, | |
| "loss": 0.481, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.304274797439575, | |
| "rewards/margins": 0.8148962259292603, | |
| "rewards/rejected": -3.119170665740967, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.5586142070632469, | |
| "grad_norm": 13.166723636543091, | |
| "learning_rate": 2.424853601625913e-07, | |
| "logits/chosen": -20.84389877319336, | |
| "logits/rejected": -20.853782653808594, | |
| "logps/chosen": -437.2691345214844, | |
| "logps/rejected": -521.557861328125, | |
| "loss": 0.4885, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -2.4156320095062256, | |
| "rewards/margins": 0.8837593197822571, | |
| "rewards/rejected": -3.299391269683838, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5693567879683094, | |
| "grad_norm": 15.08081211674444, | |
| "learning_rate": 2.3310240574504182e-07, | |
| "logits/chosen": -20.66632652282715, | |
| "logits/rejected": -20.677610397338867, | |
| "logps/chosen": -436.674560546875, | |
| "logps/rejected": -518.9525146484375, | |
| "loss": 0.4856, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -2.5705347061157227, | |
| "rewards/margins": 0.8193251490592957, | |
| "rewards/rejected": -3.389859676361084, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.5800993688733719, | |
| "grad_norm": 14.09838406164426, | |
| "learning_rate": 2.2374331080667165e-07, | |
| "logits/chosen": -21.183074951171875, | |
| "logits/rejected": -21.082101821899414, | |
| "logps/chosen": -473.5470275878906, | |
| "logps/rejected": -534.5736083984375, | |
| "loss": 0.4661, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.6635870933532715, | |
| "rewards/margins": 0.7532528638839722, | |
| "rewards/rejected": -3.4168403148651123, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5908419497784343, | |
| "grad_norm": 13.797790240550018, | |
| "learning_rate": 2.1442129043167873e-07, | |
| "logits/chosen": -20.22403335571289, | |
| "logits/rejected": -20.4445858001709, | |
| "logps/chosen": -487.107666015625, | |
| "logps/rejected": -583.4486083984375, | |
| "loss": 0.4801, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.895622491836548, | |
| "rewards/margins": 1.0642651319503784, | |
| "rewards/rejected": -3.959887742996216, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.6015845306834967, | |
| "grad_norm": 17.151814735789866, | |
| "learning_rate": 2.0514950735481053e-07, | |
| "logits/chosen": -20.70838165283203, | |
| "logits/rejected": -20.784690856933594, | |
| "logps/chosen": -465.6155700683594, | |
| "logps/rejected": -568.5462036132812, | |
| "loss": 0.483, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -2.638089418411255, | |
| "rewards/margins": 1.0538232326507568, | |
| "rewards/rejected": -3.6919121742248535, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.6123271115885591, | |
| "grad_norm": 15.353425386955319, | |
| "learning_rate": 1.959410533755232e-07, | |
| "logits/chosen": -20.860107421875, | |
| "logits/rejected": -20.810924530029297, | |
| "logps/chosen": -461.963623046875, | |
| "logps/rejected": -551.6959228515625, | |
| "loss": 0.4484, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -2.7611382007598877, | |
| "rewards/margins": 0.9577289819717407, | |
| "rewards/rejected": -3.7188668251037598, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.6230696924936215, | |
| "grad_norm": 14.20571732032185, | |
| "learning_rate": 1.8680893087230203e-07, | |
| "logits/chosen": -21.382736206054688, | |
| "logits/rejected": -21.493896484375, | |
| "logps/chosen": -479.07830810546875, | |
| "logps/rejected": -589.9129638671875, | |
| "loss": 0.4659, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -2.985625743865967, | |
| "rewards/margins": 1.1087753772735596, | |
| "rewards/rejected": -4.0944013595581055, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6338122733986841, | |
| "grad_norm": 14.026183532380196, | |
| "learning_rate": 1.7776603444324442e-07, | |
| "logits/chosen": -21.12714958190918, | |
| "logits/rejected": -20.915111541748047, | |
| "logps/chosen": -475.962158203125, | |
| "logps/rejected": -582.6109008789062, | |
| "loss": 0.4644, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -2.6778082847595215, | |
| "rewards/margins": 1.093707799911499, | |
| "rewards/rejected": -3.7715163230895996, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.6445548543037465, | |
| "grad_norm": 15.588364098928517, | |
| "learning_rate": 1.6882513269882913e-07, | |
| "logits/chosen": -20.675750732421875, | |
| "logits/rejected": -20.60934829711914, | |
| "logps/chosen": -463.9253845214844, | |
| "logps/rejected": -531.9859008789062, | |
| "loss": 0.4895, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -2.831554889678955, | |
| "rewards/margins": 0.7405329942703247, | |
| "rewards/rejected": -3.572087526321411, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6552974352088089, | |
| "grad_norm": 14.269595453262387, | |
| "learning_rate": 1.5999885023258097e-07, | |
| "logits/chosen": -21.330524444580078, | |
| "logits/rejected": -21.496753692626953, | |
| "logps/chosen": -497.48052978515625, | |
| "logps/rejected": -608.5450439453125, | |
| "loss": 0.4797, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -2.899573802947998, | |
| "rewards/margins": 1.18546724319458, | |
| "rewards/rejected": -4.085041046142578, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.6660400161138713, | |
| "grad_norm": 15.040839375830227, | |
| "learning_rate": 1.5129964979508792e-07, | |
| "logits/chosen": -21.34568214416504, | |
| "logits/rejected": -21.257770538330078, | |
| "logps/chosen": -451.4082946777344, | |
| "logps/rejected": -546.2640991210938, | |
| "loss": 0.4786, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -2.661827325820923, | |
| "rewards/margins": 1.007796049118042, | |
| "rewards/rejected": -3.669623613357544, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.6767825970189338, | |
| "grad_norm": 15.632549985107628, | |
| "learning_rate": 1.4273981469654093e-07, | |
| "logits/chosen": -21.15200424194336, | |
| "logits/rejected": -21.08346176147461, | |
| "logps/chosen": -492.291259765625, | |
| "logps/rejected": -575.3035888671875, | |
| "loss": 0.445, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -2.991098403930664, | |
| "rewards/margins": 0.8909767270088196, | |
| "rewards/rejected": -3.882075071334839, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.6875251779239963, | |
| "grad_norm": 14.077151016136973, | |
| "learning_rate": 1.343314314626449e-07, | |
| "logits/chosen": -21.24916648864746, | |
| "logits/rejected": -21.25559425354004, | |
| "logps/chosen": -479.4493103027344, | |
| "logps/rejected": -581.2391357421875, | |
| "loss": 0.4654, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.8373022079467773, | |
| "rewards/margins": 1.0110161304473877, | |
| "rewards/rejected": -3.848318099975586, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6982677588290587, | |
| "grad_norm": 17.9696416705504, | |
| "learning_rate": 1.2608637276838984e-07, | |
| "logits/chosen": -20.854616165161133, | |
| "logits/rejected": -20.866744995117188, | |
| "logps/chosen": -488.8270568847656, | |
| "logps/rejected": -580.0584716796875, | |
| "loss": 0.4757, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -3.009934663772583, | |
| "rewards/margins": 0.9066306948661804, | |
| "rewards/rejected": -3.9165656566619873, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.7090103397341211, | |
| "grad_norm": 17.609393256344518, | |
| "learning_rate": 1.1801628067378031e-07, | |
| "logits/chosen": -21.025402069091797, | |
| "logits/rejected": -21.025859832763672, | |
| "logps/chosen": -479.85076904296875, | |
| "logps/rejected": -567.4064331054688, | |
| "loss": 0.4834, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.9113426208496094, | |
| "rewards/margins": 0.9323997497558594, | |
| "rewards/rejected": -3.8437423706054688, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7197529206391836, | |
| "grad_norm": 14.184878018533347, | |
| "learning_rate": 1.1013255018519424e-07, | |
| "logits/chosen": -20.682275772094727, | |
| "logits/rejected": -20.78445053100586, | |
| "logps/chosen": -497.1404724121094, | |
| "logps/rejected": -582.8640747070312, | |
| "loss": 0.492, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -3.0111889839172363, | |
| "rewards/margins": 0.9917904138565063, | |
| "rewards/rejected": -4.002979278564453, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.730495501544246, | |
| "grad_norm": 19.63828439419386, | |
| "learning_rate": 1.0244631316558266e-07, | |
| "logits/chosen": -20.81680679321289, | |
| "logits/rejected": -20.792213439941406, | |
| "logps/chosen": -459.59619140625, | |
| "logps/rejected": -587.2064819335938, | |
| "loss": 0.4753, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -2.8939058780670166, | |
| "rewards/margins": 1.304612398147583, | |
| "rewards/rejected": -4.1985182762146, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7412380824493084, | |
| "grad_norm": 14.039431719587414, | |
| "learning_rate": 9.49684226162292e-08, | |
| "logits/chosen": -21.063154220581055, | |
| "logits/rejected": -21.0861873626709, | |
| "logps/chosen": -486.67181396484375, | |
| "logps/rejected": -604.0262451171875, | |
| "loss": 0.4649, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.923081874847412, | |
| "rewards/margins": 1.2030454874038696, | |
| "rewards/rejected": -4.12612771987915, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.7519806633543709, | |
| "grad_norm": 13.629114451388812, | |
| "learning_rate": 8.770943735226302e-08, | |
| "logits/chosen": -20.562801361083984, | |
| "logits/rejected": -20.783218383789062, | |
| "logps/chosen": -468.8253479003906, | |
| "logps/rejected": -557.6900024414062, | |
| "loss": 0.476, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -2.7901599407196045, | |
| "rewards/margins": 0.8934658765792847, | |
| "rewards/rejected": -3.6836256980895996, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7627232442594333, | |
| "grad_norm": 15.10932047374051, | |
| "learning_rate": 8.067960709356478e-08, | |
| "logits/chosen": -20.748620986938477, | |
| "logits/rejected": -20.654150009155273, | |
| "logps/chosen": -437.6307067871094, | |
| "logps/rejected": -533.923828125, | |
| "loss": 0.4898, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.6699411869049072, | |
| "rewards/margins": 1.0374929904937744, | |
| "rewards/rejected": -3.7074341773986816, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.7734658251644958, | |
| "grad_norm": 14.97714076175326, | |
| "learning_rate": 7.388885799211572e-08, | |
| "logits/chosen": -20.947423934936523, | |
| "logits/rejected": -20.915830612182617, | |
| "logps/chosen": -459.2757873535156, | |
| "logps/rejected": -527.4080200195312, | |
| "loss": 0.486, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -2.7610130310058594, | |
| "rewards/margins": 0.6833056211471558, | |
| "rewards/rejected": -3.4443182945251465, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.7842084060695582, | |
| "grad_norm": 16.18171761321377, | |
| "learning_rate": 6.73467786162265e-08, | |
| "logits/chosen": -21.14028549194336, | |
| "logits/rejected": -21.127323150634766, | |
| "logps/chosen": -493.4454650878906, | |
| "logps/rejected": -586.5130004882812, | |
| "loss": 0.4738, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -3.0033657550811768, | |
| "rewards/margins": 0.9621575474739075, | |
| "rewards/rejected": -3.9655234813690186, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.7949509869746206, | |
| "grad_norm": 13.115509215618868, | |
| "learning_rate": 6.106260641143546e-08, | |
| "logits/chosen": -20.768413543701172, | |
| "logits/rejected": -20.86481475830078, | |
| "logps/chosen": -466.8462829589844, | |
| "logps/rejected": -595.0626220703125, | |
| "loss": 0.4302, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -2.7288107872009277, | |
| "rewards/margins": 1.3605787754058838, | |
| "rewards/rejected": -4.089389801025391, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.805693567879683, | |
| "grad_norm": 14.69508011494516, | |
| "learning_rate": 5.5045214657193914e-08, | |
| "logits/chosen": -20.987638473510742, | |
| "logits/rejected": -20.953441619873047, | |
| "logps/chosen": -478.612548828125, | |
| "logps/rejected": -593.9140014648438, | |
| "loss": 0.4562, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -2.7526895999908447, | |
| "rewards/margins": 1.1709507703781128, | |
| "rewards/rejected": -3.923640489578247, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.8164361487847456, | |
| "grad_norm": 15.279291329204895, | |
| "learning_rate": 4.930309993775578e-08, | |
| "logits/chosen": -20.60492515563965, | |
| "logits/rejected": -20.59804916381836, | |
| "logps/chosen": -476.54034423828125, | |
| "logps/rejected": -611.9324951171875, | |
| "loss": 0.4789, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -2.751157760620117, | |
| "rewards/margins": 1.3292760848999023, | |
| "rewards/rejected": -4.0804338455200195, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.827178729689808, | |
| "grad_norm": 13.850627883318026, | |
| "learning_rate": 4.384437014496214e-08, | |
| "logits/chosen": -20.922626495361328, | |
| "logits/rejected": -20.863100051879883, | |
| "logps/chosen": -500.1202697753906, | |
| "logps/rejected": -598.0996704101562, | |
| "loss": 0.4577, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.9632060527801514, | |
| "rewards/margins": 0.9760335087776184, | |
| "rewards/rejected": -3.9392402172088623, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.8379213105948704, | |
| "grad_norm": 11.603127675586604, | |
| "learning_rate": 3.8676733029861604e-08, | |
| "logits/chosen": -20.93654441833496, | |
| "logits/rejected": -20.852642059326172, | |
| "logps/chosen": -520.106201171875, | |
| "logps/rejected": -597.531982421875, | |
| "loss": 0.4372, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -3.235527515411377, | |
| "rewards/margins": 0.8415508270263672, | |
| "rewards/rejected": -4.077078342437744, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8486638914999328, | |
| "grad_norm": 18.764205897014417, | |
| "learning_rate": 3.380748531933103e-08, | |
| "logits/chosen": -20.87473487854004, | |
| "logits/rejected": -20.917461395263672, | |
| "logps/chosen": -531.99365234375, | |
| "logps/rejected": -618.3016357421875, | |
| "loss": 0.4429, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -3.4170734882354736, | |
| "rewards/margins": 0.9161173105239868, | |
| "rewards/rejected": -4.33319091796875, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.8594064724049953, | |
| "grad_norm": 14.247674482174348, | |
| "learning_rate": 2.9243502413064362e-08, | |
| "logits/chosen": -21.378040313720703, | |
| "logits/rejected": -21.21609115600586, | |
| "logps/chosen": -446.82244873046875, | |
| "logps/rejected": -543.5723876953125, | |
| "loss": 0.4515, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.90150785446167, | |
| "rewards/margins": 1.0038450956344604, | |
| "rewards/rejected": -3.905353546142578, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8701490533100578, | |
| "grad_norm": 17.314036056734587, | |
| "learning_rate": 2.4991228675477288e-08, | |
| "logits/chosen": -20.581409454345703, | |
| "logits/rejected": -20.422592163085938, | |
| "logps/chosen": -497.35101318359375, | |
| "logps/rejected": -586.8109130859375, | |
| "loss": 0.5112, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -3.250107526779175, | |
| "rewards/margins": 0.9303083419799805, | |
| "rewards/rejected": -4.180416107177734, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.8808916342151202, | |
| "grad_norm": 16.039138434077582, | |
| "learning_rate": 2.1056668336235623e-08, | |
| "logits/chosen": -21.51209259033203, | |
| "logits/rejected": -21.335285186767578, | |
| "logps/chosen": -479.401611328125, | |
| "logps/rejected": -569.4190673828125, | |
| "loss": 0.4632, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -2.925386428833008, | |
| "rewards/margins": 0.9599825143814087, | |
| "rewards/rejected": -3.885369062423706, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.8916342151201826, | |
| "grad_norm": 17.110617421758416, | |
| "learning_rate": 1.7445377012256123e-08, | |
| "logits/chosen": -21.36821174621582, | |
| "logits/rejected": -21.375337600708008, | |
| "logps/chosen": -483.33367919921875, | |
| "logps/rejected": -577.9762573242188, | |
| "loss": 0.465, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -3.0296521186828613, | |
| "rewards/margins": 0.9877541661262512, | |
| "rewards/rejected": -4.017406463623047, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.9023767960252451, | |
| "grad_norm": 13.910969737585662, | |
| "learning_rate": 1.4162453863150181e-08, | |
| "logits/chosen": -21.081308364868164, | |
| "logits/rejected": -21.06283950805664, | |
| "logps/chosen": -493.18524169921875, | |
| "logps/rejected": -594.3995361328125, | |
| "loss": 0.4338, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -2.9149138927459717, | |
| "rewards/margins": 1.035150408744812, | |
| "rewards/rejected": -3.9500644207000732, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.9131193769303075, | |
| "grad_norm": 17.52833661463194, | |
| "learning_rate": 1.1212534391187689e-08, | |
| "logits/chosen": -20.905872344970703, | |
| "logits/rejected": -20.837444305419922, | |
| "logps/chosen": -438.93048095703125, | |
| "logps/rejected": -538.4364013671875, | |
| "loss": 0.4813, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -2.69669246673584, | |
| "rewards/margins": 0.979594349861145, | |
| "rewards/rejected": -3.6762871742248535, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.92386195783537, | |
| "grad_norm": 23.048838465895624, | |
| "learning_rate": 8.59978389594676e-09, | |
| "logits/chosen": -20.701757431030273, | |
| "logits/rejected": -20.72020149230957, | |
| "logps/chosen": -465.57501220703125, | |
| "logps/rejected": -549.2703857421875, | |
| "loss": 0.4546, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -2.8212685585021973, | |
| "rewards/margins": 0.8496788740158081, | |
| "rewards/rejected": -3.670947313308716, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9346045387404324, | |
| "grad_norm": 15.846431503688864, | |
| "learning_rate": 6.327891592892126e-09, | |
| "logits/chosen": -20.888439178466797, | |
| "logits/rejected": -20.962871551513672, | |
| "logps/chosen": -478.03790283203125, | |
| "logps/rejected": -602.384765625, | |
| "loss": 0.4658, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -2.8073556423187256, | |
| "rewards/margins": 1.2782353162765503, | |
| "rewards/rejected": -4.0855913162231445, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.9453471196454948, | |
| "grad_norm": 14.390183815864935, | |
| "learning_rate": 4.400065404186276e-09, | |
| "logits/chosen": -20.80968475341797, | |
| "logits/rejected": -20.828323364257812, | |
| "logps/chosen": -491.7540588378906, | |
| "logps/rejected": -579.70703125, | |
| "loss": 0.4437, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -2.9315555095672607, | |
| "rewards/margins": 0.9275356531143188, | |
| "rewards/rejected": -3.8590915203094482, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.9560897005505573, | |
| "grad_norm": 16.74015785431614, | |
| "learning_rate": 2.8190274290888215e-09, | |
| "logits/chosen": -21.063623428344727, | |
| "logits/rejected": -21.043384552001953, | |
| "logps/chosen": -504.7342224121094, | |
| "logps/rejected": -598.46044921875, | |
| "loss": 0.4766, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -3.1327033042907715, | |
| "rewards/margins": 0.9961974024772644, | |
| "rewards/rejected": -4.128900527954102, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.9668322814556197, | |
| "grad_norm": 13.546941096732384, | |
| "learning_rate": 1.5870101003402081e-09, | |
| "logits/chosen": -20.875934600830078, | |
| "logits/rejected": -20.832775115966797, | |
| "logps/chosen": -518.4654541015625, | |
| "logps/rejected": -618.839111328125, | |
| "loss": 0.4585, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -3.1628730297088623, | |
| "rewards/margins": 1.0721877813339233, | |
| "rewards/rejected": -4.235061168670654, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9775748623606821, | |
| "grad_norm": 14.614694225953357, | |
| "learning_rate": 7.057530319564409e-10, | |
| "logits/chosen": -20.87482261657715, | |
| "logits/rejected": -20.740245819091797, | |
| "logps/chosen": -467.67559814453125, | |
| "logps/rejected": -538.7994384765625, | |
| "loss": 0.4798, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -2.8199124336242676, | |
| "rewards/margins": 0.7901204824447632, | |
| "rewards/rejected": -3.610032558441162, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.9883174432657446, | |
| "grad_norm": 12.357402004120742, | |
| "learning_rate": 1.7650056288651127e-10, | |
| "logits/chosen": -20.734127044677734, | |
| "logits/rejected": -20.775203704833984, | |
| "logps/chosen": -483.3863220214844, | |
| "logps/rejected": -600.2500610351562, | |
| "loss": 0.4388, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -3.0567145347595215, | |
| "rewards/margins": 1.2218314409255981, | |
| "rewards/rejected": -4.278546333312988, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.9990600241708071, | |
| "grad_norm": 15.198640734812669, | |
| "learning_rate": 0.0, | |
| "logits/chosen": -21.103975296020508, | |
| "logits/rejected": -21.10438346862793, | |
| "logps/chosen": -485.390625, | |
| "logps/rejected": -577.9816284179688, | |
| "loss": 0.4514, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -2.860701084136963, | |
| "rewards/margins": 0.9835926294326782, | |
| "rewards/rejected": -3.8442935943603516, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.9990600241708071, | |
| "step": 465, | |
| "total_flos": 0.0, | |
| "train_loss": 0.5369526706716066, | |
| "train_runtime": 9531.2046, | |
| "train_samples_per_second": 6.25, | |
| "train_steps_per_second": 0.049 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 465, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000000, | |
| "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": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |