| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9988571428571429, | |
| "eval_steps": 50, | |
| "global_step": 437, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.022857142857142857, | |
| "grad_norm": 3.2784174727745485, | |
| "learning_rate": 1.1363636363636363e-07, | |
| "logits/chosen": -2.700894832611084, | |
| "logits/rejected": -2.625084400177002, | |
| "logps/chosen": -301.3052062988281, | |
| "logps/rejected": -281.75274658203125, | |
| "loss": 0.0557, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": -0.00019026221707463264, | |
| "rewards/margins": -0.00015981406613718718, | |
| "rewards/rejected": -3.0448136385530233e-05, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.045714285714285714, | |
| "grad_norm": 2.5255601771436815, | |
| "learning_rate": 2.2727272727272726e-07, | |
| "logits/chosen": -2.641096830368042, | |
| "logits/rejected": -2.6057686805725098, | |
| "logps/chosen": -278.879638671875, | |
| "logps/rejected": -254.6970672607422, | |
| "loss": 0.0553, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.002915444318205118, | |
| "rewards/margins": 0.002195452805608511, | |
| "rewards/rejected": 0.0007199913961812854, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06857142857142857, | |
| "grad_norm": 2.762527455327235, | |
| "learning_rate": 3.4090909090909085e-07, | |
| "logits/chosen": -2.6387953758239746, | |
| "logits/rejected": -2.6177260875701904, | |
| "logps/chosen": -263.30303955078125, | |
| "logps/rejected": -263.4232177734375, | |
| "loss": 0.0535, | |
| "rewards/accuracies": 0.606249988079071, | |
| "rewards/chosen": 0.013800328597426414, | |
| "rewards/margins": 0.008101832121610641, | |
| "rewards/rejected": 0.005698497407138348, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.09142857142857143, | |
| "grad_norm": 2.8469620893370813, | |
| "learning_rate": 4.545454545454545e-07, | |
| "logits/chosen": -2.6508777141571045, | |
| "logits/rejected": -2.5878684520721436, | |
| "logps/chosen": -290.2543640136719, | |
| "logps/rejected": -268.0738830566406, | |
| "loss": 0.0502, | |
| "rewards/accuracies": 0.581250011920929, | |
| "rewards/chosen": 0.03742903843522072, | |
| "rewards/margins": 0.04185490682721138, | |
| "rewards/rejected": -0.004425862338393927, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.11428571428571428, | |
| "grad_norm": 7.044347610102239, | |
| "learning_rate": 4.997124959943201e-07, | |
| "logits/chosen": -2.68748140335083, | |
| "logits/rejected": -2.6086513996124268, | |
| "logps/chosen": -290.8190612792969, | |
| "logps/rejected": -249.57925415039062, | |
| "loss": 0.048, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": 0.05491219088435173, | |
| "rewards/margins": 0.08206149935722351, | |
| "rewards/rejected": -0.02714930847287178, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11428571428571428, | |
| "eval_logits/chosen": -2.55232834815979, | |
| "eval_logits/rejected": -2.4544785022735596, | |
| "eval_logps/chosen": -269.3974304199219, | |
| "eval_logps/rejected": -224.57150268554688, | |
| "eval_loss": 0.04283786565065384, | |
| "eval_rewards/accuracies": 0.7025862336158752, | |
| "eval_rewards/chosen": 0.06206508353352547, | |
| "eval_rewards/margins": 0.11701909452676773, | |
| "eval_rewards/rejected": -0.054954010993242264, | |
| "eval_runtime": 91.1445, | |
| "eval_samples_per_second": 20.089, | |
| "eval_steps_per_second": 0.318, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.13714285714285715, | |
| "grad_norm": 3.0003693671132106, | |
| "learning_rate": 4.979579212164186e-07, | |
| "logits/chosen": -2.5842273235321045, | |
| "logits/rejected": -2.476365804672241, | |
| "logps/chosen": -280.49664306640625, | |
| "logps/rejected": -259.8401794433594, | |
| "loss": 0.0438, | |
| "rewards/accuracies": 0.668749988079071, | |
| "rewards/chosen": 9.337197843706235e-05, | |
| "rewards/margins": 0.11297394335269928, | |
| "rewards/rejected": -0.11288057267665863, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 3.2128849852724057, | |
| "learning_rate": 4.946196886175515e-07, | |
| "logits/chosen": -2.5273566246032715, | |
| "logits/rejected": -2.468844175338745, | |
| "logps/chosen": -273.30010986328125, | |
| "logps/rejected": -270.6748962402344, | |
| "loss": 0.0415, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.024781424552202225, | |
| "rewards/margins": 0.13526295125484467, | |
| "rewards/rejected": -0.11048151552677155, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.18285714285714286, | |
| "grad_norm": 2.993790876710757, | |
| "learning_rate": 4.897191188239667e-07, | |
| "logits/chosen": -2.4337241649627686, | |
| "logits/rejected": -2.369915008544922, | |
| "logps/chosen": -269.41571044921875, | |
| "logps/rejected": -276.94500732421875, | |
| "loss": 0.0424, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.023866862058639526, | |
| "rewards/margins": 0.1924942582845688, | |
| "rewards/rejected": -0.2163611203432083, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.2057142857142857, | |
| "grad_norm": 7.3103992020217285, | |
| "learning_rate": 4.832875107981763e-07, | |
| "logits/chosen": -2.3889706134796143, | |
| "logits/rejected": -2.3026340007781982, | |
| "logps/chosen": -275.86724853515625, | |
| "logps/rejected": -275.403076171875, | |
| "loss": 0.0422, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 0.004180185496807098, | |
| "rewards/margins": 0.20273157954216003, | |
| "rewards/rejected": -0.19855138659477234, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.22857142857142856, | |
| "grad_norm": 3.5295110150949056, | |
| "learning_rate": 4.753659419387223e-07, | |
| "logits/chosen": -2.3000099658966064, | |
| "logits/rejected": -2.165864944458008, | |
| "logps/chosen": -289.35467529296875, | |
| "logps/rejected": -259.2148742675781, | |
| "loss": 0.04, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.029682844877243042, | |
| "rewards/margins": 0.22795169055461884, | |
| "rewards/rejected": -0.25763455033302307, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.22857142857142856, | |
| "eval_logits/chosen": -2.1305601596832275, | |
| "eval_logits/rejected": -1.9564552307128906, | |
| "eval_logps/chosen": -283.3714904785156, | |
| "eval_logps/rejected": -250.93673706054688, | |
| "eval_loss": 0.03854820877313614, | |
| "eval_rewards/accuracies": 0.75, | |
| "eval_rewards/chosen": -0.07767525315284729, | |
| "eval_rewards/margins": 0.2409307211637497, | |
| "eval_rewards/rejected": -0.31860601902008057, | |
| "eval_runtime": 91.3911, | |
| "eval_samples_per_second": 20.035, | |
| "eval_steps_per_second": 0.317, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.25142857142857145, | |
| "grad_norm": 3.082269192831021, | |
| "learning_rate": 4.660050057270191e-07, | |
| "logits/chosen": -2.183840036392212, | |
| "logits/rejected": -2.098686456680298, | |
| "logps/chosen": -322.04498291015625, | |
| "logps/rejected": -316.5348205566406, | |
| "loss": 0.0406, | |
| "rewards/accuracies": 0.643750011920929, | |
| "rewards/chosen": -0.05741389840841293, | |
| "rewards/margins": 0.1288377195596695, | |
| "rewards/rejected": -0.18625161051750183, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2742857142857143, | |
| "grad_norm": 3.1619647572029415, | |
| "learning_rate": 4.5526448859687144e-07, | |
| "logits/chosen": -2.122063159942627, | |
| "logits/rejected": -1.943372130393982, | |
| "logps/chosen": -315.7963562011719, | |
| "logps/rejected": -253.3599395751953, | |
| "loss": 0.0377, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.048919014632701874, | |
| "rewards/margins": 0.1675383597612381, | |
| "rewards/rejected": -0.21645736694335938, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.29714285714285715, | |
| "grad_norm": 3.421686175984915, | |
| "learning_rate": 4.432129880904388e-07, | |
| "logits/chosen": -2.2705729007720947, | |
| "logits/rejected": -2.1812658309936523, | |
| "logps/chosen": -313.17596435546875, | |
| "logps/rejected": -274.95806884765625, | |
| "loss": 0.0371, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.08020049333572388, | |
| "rewards/margins": 0.16467759013175964, | |
| "rewards/rejected": -0.24487809836864471, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 3.9283597913931922, | |
| "learning_rate": 4.299274747394055e-07, | |
| "logits/chosen": -2.398083448410034, | |
| "logits/rejected": -2.3765132427215576, | |
| "logps/chosen": -314.16827392578125, | |
| "logps/rejected": -288.8768310546875, | |
| "loss": 0.0375, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.012793675065040588, | |
| "rewards/margins": 0.18244774639606476, | |
| "rewards/rejected": -0.16965410113334656, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.34285714285714286, | |
| "grad_norm": 3.018006960325513, | |
| "learning_rate": 4.1549280046953653e-07, | |
| "logits/chosen": -2.4656407833099365, | |
| "logits/rejected": -2.367140531539917, | |
| "logps/chosen": -295.70556640625, | |
| "logps/rejected": -305.4766845703125, | |
| "loss": 0.0363, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.08719640970230103, | |
| "rewards/margins": 0.204580619931221, | |
| "rewards/rejected": -0.2917770445346832, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.34285714285714286, | |
| "eval_logits/chosen": -2.3440675735473633, | |
| "eval_logits/rejected": -2.195537805557251, | |
| "eval_logps/chosen": -296.12109375, | |
| "eval_logps/rejected": -265.0228271484375, | |
| "eval_loss": 0.03708173334598541, | |
| "eval_rewards/accuracies": 0.7543103694915771, | |
| "eval_rewards/chosen": -0.20517148077487946, | |
| "eval_rewards/margins": 0.25429537892341614, | |
| "eval_rewards/rejected": -0.4594668447971344, | |
| "eval_runtime": 91.4169, | |
| "eval_samples_per_second": 20.029, | |
| "eval_steps_per_second": 0.317, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3657142857142857, | |
| "grad_norm": 3.319581864694512, | |
| "learning_rate": 4.000011566683401e-07, | |
| "logits/chosen": -2.376082181930542, | |
| "logits/rejected": -2.2733752727508545, | |
| "logps/chosen": -320.1603088378906, | |
| "logps/rejected": -300.1729431152344, | |
| "loss": 0.0377, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.20820029079914093, | |
| "rewards/margins": 0.23710182309150696, | |
| "rewards/rejected": -0.4453020989894867, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.38857142857142857, | |
| "grad_norm": 3.33265699836665, | |
| "learning_rate": 3.8355148537705047e-07, | |
| "logits/chosen": -2.5347208976745605, | |
| "logits/rejected": -2.48330020904541, | |
| "logps/chosen": -312.109619140625, | |
| "logps/rejected": -294.4687805175781, | |
| "loss": 0.0377, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.08004748821258545, | |
| "rewards/margins": 0.2073298990726471, | |
| "rewards/rejected": -0.28737738728523254, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.4114285714285714, | |
| "grad_norm": 2.9907171954669534, | |
| "learning_rate": 3.662488473675315e-07, | |
| "logits/chosen": -2.622641086578369, | |
| "logits/rejected": -2.556631565093994, | |
| "logps/chosen": -332.7588806152344, | |
| "logps/rejected": -290.630859375, | |
| "loss": 0.0373, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.02579095959663391, | |
| "rewards/margins": 0.2810118794441223, | |
| "rewards/rejected": -0.3068028390407562, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4342857142857143, | |
| "grad_norm": 3.4010102454112205, | |
| "learning_rate": 3.48203751140067e-07, | |
| "logits/chosen": -2.642368793487549, | |
| "logits/rejected": -2.5668444633483887, | |
| "logps/chosen": -286.96087646484375, | |
| "logps/rejected": -263.45306396484375, | |
| "loss": 0.0366, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.10703293979167938, | |
| "rewards/margins": 0.1825161725282669, | |
| "rewards/rejected": -0.2895490825176239, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.45714285714285713, | |
| "grad_norm": 2.9611644003699062, | |
| "learning_rate": 3.2953144712759537e-07, | |
| "logits/chosen": -2.651844024658203, | |
| "logits/rejected": -2.5362985134124756, | |
| "logps/chosen": -272.91973876953125, | |
| "logps/rejected": -257.00128173828125, | |
| "loss": 0.0373, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.04576968401670456, | |
| "rewards/margins": 0.23543760180473328, | |
| "rewards/rejected": -0.28120729327201843, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.45714285714285713, | |
| "eval_logits/chosen": -2.4903368949890137, | |
| "eval_logits/rejected": -2.384784460067749, | |
| "eval_logps/chosen": -280.1239318847656, | |
| "eval_logps/rejected": -251.76303100585938, | |
| "eval_loss": 0.035302262753248215, | |
| "eval_rewards/accuracies": 0.7715517282485962, | |
| "eval_rewards/chosen": -0.04520006850361824, | |
| "eval_rewards/margins": 0.28166908025741577, | |
| "eval_rewards/rejected": -0.3268691301345825, | |
| "eval_runtime": 91.7563, | |
| "eval_samples_per_second": 19.955, | |
| "eval_steps_per_second": 0.316, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 2.9976100938955375, | |
| "learning_rate": 3.103511916141658e-07, | |
| "logits/chosen": -2.4521584510803223, | |
| "logits/rejected": -2.38016939163208, | |
| "logps/chosen": -272.8417053222656, | |
| "logps/rejected": -276.30755615234375, | |
| "loss": 0.0362, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.09388158470392227, | |
| "rewards/margins": 0.21639461815357208, | |
| "rewards/rejected": -0.31027621030807495, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.5028571428571429, | |
| "grad_norm": 3.3697037289842515, | |
| "learning_rate": 2.9078548506882117e-07, | |
| "logits/chosen": -2.486403226852417, | |
| "logits/rejected": -2.43949556350708, | |
| "logps/chosen": -299.24713134765625, | |
| "logps/rejected": -279.8057556152344, | |
| "loss": 0.0389, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.14181171357631683, | |
| "rewards/margins": 0.19441165030002594, | |
| "rewards/rejected": -0.3362233638763428, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.5257142857142857, | |
| "grad_norm": 3.1432532704884797, | |
| "learning_rate": 2.709592897595191e-07, | |
| "logits/chosen": -2.507004499435425, | |
| "logits/rejected": -2.437407970428467, | |
| "logps/chosen": -295.362060546875, | |
| "logps/rejected": -273.03448486328125, | |
| "loss": 0.0364, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.1430293470621109, | |
| "rewards/margins": 0.19444888830184937, | |
| "rewards/rejected": -0.3374782204627991, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5485714285714286, | |
| "grad_norm": 3.601368929942566, | |
| "learning_rate": 2.509992316440332e-07, | |
| "logits/chosen": -2.5563902854919434, | |
| "logits/rejected": -2.469142436981201, | |
| "logps/chosen": -310.55108642578125, | |
| "logps/rejected": -312.4164733886719, | |
| "loss": 0.0359, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.1184125691652298, | |
| "rewards/margins": 0.2915334701538086, | |
| "rewards/rejected": -0.4099460542201996, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 2.959775188259923, | |
| "learning_rate": 2.3103279163519918e-07, | |
| "logits/chosen": -2.5154330730438232, | |
| "logits/rejected": -2.477905750274658, | |
| "logps/chosen": -291.4244689941406, | |
| "logps/rejected": -302.8068542480469, | |
| "loss": 0.0374, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.1079820767045021, | |
| "rewards/margins": 0.21808166801929474, | |
| "rewards/rejected": -0.32606372237205505, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "eval_logits/chosen": -2.5244851112365723, | |
| "eval_logits/rejected": -2.4306986331939697, | |
| "eval_logps/chosen": -283.6198425292969, | |
| "eval_logps/rejected": -253.70823669433594, | |
| "eval_loss": 0.034400373697280884, | |
| "eval_rewards/accuracies": 0.75, | |
| "eval_rewards/chosen": -0.08015903830528259, | |
| "eval_rewards/margins": 0.26616212725639343, | |
| "eval_rewards/rejected": -0.346321165561676, | |
| "eval_runtime": 91.0071, | |
| "eval_samples_per_second": 20.119, | |
| "eval_steps_per_second": 0.319, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5942857142857143, | |
| "grad_norm": 3.439430942929606, | |
| "learning_rate": 2.1118749140573358e-07, | |
| "logits/chosen": -2.5266551971435547, | |
| "logits/rejected": -2.515558958053589, | |
| "logps/chosen": -293.00262451171875, | |
| "logps/rejected": -303.7609558105469, | |
| "loss": 0.038, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.13215936720371246, | |
| "rewards/margins": 0.17202024161815643, | |
| "rewards/rejected": -0.3041796088218689, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.6171428571428571, | |
| "grad_norm": 3.3473957187400707, | |
| "learning_rate": 1.9159007893272703e-07, | |
| "logits/chosen": -2.563744068145752, | |
| "logits/rejected": -2.4393467903137207, | |
| "logps/chosen": -282.9759216308594, | |
| "logps/rejected": -266.39312744140625, | |
| "loss": 0.0343, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.10450073331594467, | |
| "rewards/margins": 0.2543557584285736, | |
| "rewards/rejected": -0.3588564991950989, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 3.487492966730279, | |
| "learning_rate": 1.7236571898357766e-07, | |
| "logits/chosen": -2.576411724090576, | |
| "logits/rejected": -2.5091185569763184, | |
| "logps/chosen": -281.6929931640625, | |
| "logps/rejected": -297.8419189453125, | |
| "loss": 0.036, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.08654884248971939, | |
| "rewards/margins": 0.2608022391796112, | |
| "rewards/rejected": -0.34735107421875, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.6628571428571428, | |
| "grad_norm": 3.951421197104747, | |
| "learning_rate": 1.5363719371356882e-07, | |
| "logits/chosen": -2.622135877609253, | |
| "logits/rejected": -2.544238328933716, | |
| "logps/chosen": -311.6895751953125, | |
| "logps/rejected": -299.3029479980469, | |
| "loss": 0.0349, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.10783736407756805, | |
| "rewards/margins": 0.2297072857618332, | |
| "rewards/rejected": -0.33754467964172363, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6857142857142857, | |
| "grad_norm": 3.035652974254887, | |
| "learning_rate": 1.3552411848071565e-07, | |
| "logits/chosen": -2.646151304244995, | |
| "logits/rejected": -2.5031981468200684, | |
| "logps/chosen": -313.28375244140625, | |
| "logps/rejected": -289.15740966796875, | |
| "loss": 0.0346, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.11374591290950775, | |
| "rewards/margins": 0.25140708684921265, | |
| "rewards/rejected": -0.3651530146598816, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6857142857142857, | |
| "eval_logits/chosen": -2.5812017917633057, | |
| "eval_logits/rejected": -2.479748487472534, | |
| "eval_logps/chosen": -279.3269958496094, | |
| "eval_logps/rejected": -251.0284881591797, | |
| "eval_loss": 0.03423487767577171, | |
| "eval_rewards/accuracies": 0.7456896305084229, | |
| "eval_rewards/chosen": -0.03723056614398956, | |
| "eval_rewards/margins": 0.2822931110858917, | |
| "eval_rewards/rejected": -0.3195236325263977, | |
| "eval_runtime": 90.8627, | |
| "eval_samples_per_second": 20.151, | |
| "eval_steps_per_second": 0.319, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7085714285714285, | |
| "grad_norm": 2.850744846118877, | |
| "learning_rate": 1.1814217788631473e-07, | |
| "logits/chosen": -2.6023502349853516, | |
| "logits/rejected": -2.559641122817993, | |
| "logps/chosen": -263.7620849609375, | |
| "logps/rejected": -264.8592224121094, | |
| "loss": 0.0369, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.07457949221134186, | |
| "rewards/margins": 0.20205926895141602, | |
| "rewards/rejected": -0.27663877606391907, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.7314285714285714, | |
| "grad_norm": 3.4445308568499575, | |
| "learning_rate": 1.0160238692045331e-07, | |
| "logits/chosen": -2.5764858722686768, | |
| "logits/rejected": -2.4977431297302246, | |
| "logps/chosen": -255.32528686523438, | |
| "logps/rejected": -270.0458984375, | |
| "loss": 0.0364, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.11042918264865875, | |
| "rewards/margins": 0.18014779686927795, | |
| "rewards/rejected": -0.2905769944190979, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.7542857142857143, | |
| "grad_norm": 2.7891304584846908, | |
| "learning_rate": 8.601038193139438e-08, | |
| "logits/chosen": -2.669182300567627, | |
| "logits/rejected": -2.559919595718384, | |
| "logps/chosen": -304.69049072265625, | |
| "logps/rejected": -277.16827392578125, | |
| "loss": 0.0348, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.05472307652235031, | |
| "rewards/margins": 0.23908019065856934, | |
| "rewards/rejected": -0.29380327463150024, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7771428571428571, | |
| "grad_norm": 3.0804434627972372, | |
| "learning_rate": 7.146574594727572e-08, | |
| "logits/chosen": -2.5831470489501953, | |
| "logits/rejected": -2.54848051071167, | |
| "logps/chosen": -275.44757080078125, | |
| "logps/rejected": -277.88311767578125, | |
| "loss": 0.0352, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.09496594220399857, | |
| "rewards/margins": 0.2636275291442871, | |
| "rewards/rejected": -0.3585934638977051, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 3.3935330484921606, | |
| "learning_rate": 5.8061372659157306e-08, | |
| "logits/chosen": -2.6102333068847656, | |
| "logits/rejected": -2.5170912742614746, | |
| "logps/chosen": -299.60601806640625, | |
| "logps/rejected": -284.296630859375, | |
| "loss": 0.0375, | |
| "rewards/accuracies": 0.6812499761581421, | |
| "rewards/chosen": -0.09898806363344193, | |
| "rewards/margins": 0.20100173354148865, | |
| "rewards/rejected": -0.29998978972435, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_logits/chosen": -2.5560529232025146, | |
| "eval_logits/rejected": -2.447376012802124, | |
| "eval_logps/chosen": -283.43243408203125, | |
| "eval_logps/rejected": -256.5389099121094, | |
| "eval_loss": 0.03421488776803017, | |
| "eval_rewards/accuracies": 0.7413793206214905, | |
| "eval_rewards/chosen": -0.07828506827354431, | |
| "eval_rewards/margins": 0.29634276032447815, | |
| "eval_rewards/rejected": -0.37462782859802246, | |
| "eval_runtime": 90.9316, | |
| "eval_samples_per_second": 20.136, | |
| "eval_steps_per_second": 0.319, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.8228571428571428, | |
| "grad_norm": 2.9199488201620456, | |
| "learning_rate": 4.5882873127531614e-08, | |
| "logits/chosen": -2.6238226890563965, | |
| "logits/rejected": -2.5243430137634277, | |
| "logps/chosen": -294.04046630859375, | |
| "logps/rejected": -285.36041259765625, | |
| "loss": 0.0342, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.08635450899600983, | |
| "rewards/margins": 0.2563842833042145, | |
| "rewards/rejected": -0.3427388072013855, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8457142857142858, | |
| "grad_norm": 3.0306421043965925, | |
| "learning_rate": 3.500802900154412e-08, | |
| "logits/chosen": -2.61911940574646, | |
| "logits/rejected": -2.5165693759918213, | |
| "logps/chosen": -277.0403137207031, | |
| "logps/rejected": -271.5109558105469, | |
| "loss": 0.034, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.05978885293006897, | |
| "rewards/margins": 0.2711651027202606, | |
| "rewards/rejected": -0.330953985452652, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.8685714285714285, | |
| "grad_norm": 3.142763842399202, | |
| "learning_rate": 2.550629574310309e-08, | |
| "logits/chosen": -2.68263578414917, | |
| "logits/rejected": -2.584908962249756, | |
| "logps/chosen": -327.3045959472656, | |
| "logps/rejected": -287.1323547363281, | |
| "loss": 0.0352, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.09248252213001251, | |
| "rewards/margins": 0.20484676957130432, | |
| "rewards/rejected": -0.29732924699783325, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8914285714285715, | |
| "grad_norm": 3.6385746368012355, | |
| "learning_rate": 1.7438359028687983e-08, | |
| "logits/chosen": -2.6608521938323975, | |
| "logits/rejected": -2.607428789138794, | |
| "logps/chosen": -316.05438232421875, | |
| "logps/rejected": -321.3371887207031, | |
| "loss": 0.0359, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.04681408405303955, | |
| "rewards/margins": 0.2188311070203781, | |
| "rewards/rejected": -0.26564517617225647, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.9142857142857143, | |
| "grad_norm": 5.02358291913839, | |
| "learning_rate": 1.0855747162029361e-08, | |
| "logits/chosen": -2.590088367462158, | |
| "logits/rejected": -2.570838451385498, | |
| "logps/chosen": -289.6832580566406, | |
| "logps/rejected": -296.08538818359375, | |
| "loss": 0.0367, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.10950014740228653, | |
| "rewards/margins": 0.18561770021915436, | |
| "rewards/rejected": -0.2951178550720215, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.9142857142857143, | |
| "eval_logits/chosen": -2.561089277267456, | |
| "eval_logits/rejected": -2.4522511959075928, | |
| "eval_logps/chosen": -282.986083984375, | |
| "eval_logps/rejected": -256.24615478515625, | |
| "eval_loss": 0.03414054214954376, | |
| "eval_rewards/accuracies": 0.7413793206214905, | |
| "eval_rewards/chosen": -0.07382136583328247, | |
| "eval_rewards/margins": 0.2978789508342743, | |
| "eval_rewards/rejected": -0.3717002868652344, | |
| "eval_runtime": 90.9735, | |
| "eval_samples_per_second": 20.127, | |
| "eval_steps_per_second": 0.319, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.9371428571428572, | |
| "grad_norm": 3.2423722180060786, | |
| "learning_rate": 5.8005019731033615e-09, | |
| "logits/chosen": -2.6259548664093018, | |
| "logits/rejected": -2.5545239448547363, | |
| "logps/chosen": -299.16107177734375, | |
| "logps/rejected": -292.75714111328125, | |
| "loss": 0.0352, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.10981108993291855, | |
| "rewards/margins": 0.2204599827528, | |
| "rewards/rejected": -0.3302710950374603, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 3.0950678320925222, | |
| "learning_rate": 2.3049103053431886e-09, | |
| "logits/chosen": -2.5695414543151855, | |
| "logits/rejected": -2.504147529602051, | |
| "logps/chosen": -286.32177734375, | |
| "logps/rejected": -269.1705627441406, | |
| "loss": 0.0345, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.03573674336075783, | |
| "rewards/margins": 0.31610041856765747, | |
| "rewards/rejected": -0.351837158203125, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.9828571428571429, | |
| "grad_norm": 2.978033708452065, | |
| "learning_rate": 3.9129780600541397e-10, | |
| "logits/chosen": -2.5734434127807617, | |
| "logits/rejected": -2.4971184730529785, | |
| "logps/chosen": -290.81646728515625, | |
| "logps/rejected": -297.24932861328125, | |
| "loss": 0.0344, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.08584621548652649, | |
| "rewards/margins": 0.24995127320289612, | |
| "rewards/rejected": -0.33579742908477783, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9988571428571429, | |
| "step": 437, | |
| "total_flos": 0.0, | |
| "train_loss": 0.03879852061004333, | |
| "train_runtime": 11351.3146, | |
| "train_samples_per_second": 4.933, | |
| "train_steps_per_second": 0.038 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 437, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "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": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |