| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1465, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00682681230532918, | |
| "grad_norm": 20.491548678692148, | |
| "learning_rate": 6.122448979591837e-08, | |
| "logits/chosen": 0.03672148287296295, | |
| "logits/rejected": 0.041521187871694565, | |
| "logps/chosen": -191.74862670898438, | |
| "logps/rejected": -189.4052276611328, | |
| "loss": 0.6921, | |
| "rewards/accuracies": 0.41874998807907104, | |
| "rewards/chosen": 0.0014678842853754759, | |
| "rewards/margins": 0.0024292597081512213, | |
| "rewards/rejected": -0.0009613755391910672, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01365362461065836, | |
| "grad_norm": 21.860852469835415, | |
| "learning_rate": 1.2925170068027211e-07, | |
| "logits/chosen": 0.04523754119873047, | |
| "logits/rejected": 0.05510401353240013, | |
| "logps/chosen": -187.8703155517578, | |
| "logps/rejected": -187.6009979248047, | |
| "loss": 0.6937, | |
| "rewards/accuracies": 0.4859375059604645, | |
| "rewards/chosen": -0.0003124059294350445, | |
| "rewards/margins": -0.0007655444787815213, | |
| "rewards/rejected": 0.0004531386948656291, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02048043691598754, | |
| "grad_norm": 20.278529512570657, | |
| "learning_rate": 1.9727891156462583e-07, | |
| "logits/chosen": 0.020983930677175522, | |
| "logits/rejected": 0.04532231390476227, | |
| "logps/chosen": -185.85728454589844, | |
| "logps/rejected": -188.9866180419922, | |
| "loss": 0.6936, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": 0.00064073596149683, | |
| "rewards/margins": -0.0005829028668813407, | |
| "rewards/rejected": 0.001223638653755188, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02730724922131672, | |
| "grad_norm": 19.626379046619967, | |
| "learning_rate": 2.653061224489796e-07, | |
| "logits/chosen": 0.03043345920741558, | |
| "logits/rejected": 0.032446593046188354, | |
| "logps/chosen": -193.6338653564453, | |
| "logps/rejected": -190.4232635498047, | |
| "loss": 0.6913, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.002521326532587409, | |
| "rewards/margins": 0.004052319563925266, | |
| "rewards/rejected": -0.0015309930313378572, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0341340615266459, | |
| "grad_norm": 21.08295374738999, | |
| "learning_rate": 3.333333333333333e-07, | |
| "logits/chosen": 0.04947035759687424, | |
| "logits/rejected": 0.06372452527284622, | |
| "logps/chosen": -188.39315795898438, | |
| "logps/rejected": -190.05992126464844, | |
| "loss": 0.6942, | |
| "rewards/accuracies": 0.46406251192092896, | |
| "rewards/chosen": 0.0021625806111842394, | |
| "rewards/margins": -0.0017312343697994947, | |
| "rewards/rejected": 0.003893814980983734, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04096087383197508, | |
| "grad_norm": 20.25039554823623, | |
| "learning_rate": 4.0136054421768705e-07, | |
| "logits/chosen": 0.053825099021196365, | |
| "logits/rejected": 0.0521962009370327, | |
| "logps/chosen": -189.28480529785156, | |
| "logps/rejected": -184.31430053710938, | |
| "loss": 0.6937, | |
| "rewards/accuracies": 0.5015624761581421, | |
| "rewards/chosen": 0.004196351859718561, | |
| "rewards/margins": -0.0006979627651162446, | |
| "rewards/rejected": 0.0048943147994577885, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04778768613730426, | |
| "grad_norm": 22.505298366939336, | |
| "learning_rate": 4.693877551020408e-07, | |
| "logits/chosen": 0.03855639323592186, | |
| "logits/rejected": 0.041457682847976685, | |
| "logps/chosen": -189.49111938476562, | |
| "logps/rejected": -190.42034912109375, | |
| "loss": 0.6933, | |
| "rewards/accuracies": 0.4937499761581421, | |
| "rewards/chosen": 0.008006598800420761, | |
| "rewards/margins": 4.7756126150488853e-05, | |
| "rewards/rejected": 0.007958842441439629, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.05461449844263344, | |
| "grad_norm": 19.99809543741437, | |
| "learning_rate": 5.374149659863945e-07, | |
| "logits/chosen": 0.026321567595005035, | |
| "logits/rejected": 0.013571225106716156, | |
| "logps/chosen": -189.8534393310547, | |
| "logps/rejected": -187.626708984375, | |
| "loss": 0.6878, | |
| "rewards/accuracies": 0.6031249761581421, | |
| "rewards/chosen": 0.018732454627752304, | |
| "rewards/margins": 0.011271494440734386, | |
| "rewards/rejected": 0.007460957858711481, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.06144131074796262, | |
| "grad_norm": 22.176568391543768, | |
| "learning_rate": 6.054421768707482e-07, | |
| "logits/chosen": 0.020383019000291824, | |
| "logits/rejected": 0.02592673897743225, | |
| "logps/chosen": -186.662841796875, | |
| "logps/rejected": -189.3004608154297, | |
| "loss": 0.6876, | |
| "rewards/accuracies": 0.582812488079071, | |
| "rewards/chosen": 0.027650414034724236, | |
| "rewards/margins": 0.011809633113443851, | |
| "rewards/rejected": 0.01584078185260296, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0682681230532918, | |
| "grad_norm": 20.53234701755388, | |
| "learning_rate": 6.734693877551019e-07, | |
| "logits/chosen": 0.02966993674635887, | |
| "logits/rejected": 0.05219441279768944, | |
| "logps/chosen": -190.25782775878906, | |
| "logps/rejected": -189.80935668945312, | |
| "loss": 0.6858, | |
| "rewards/accuracies": 0.5718749761581421, | |
| "rewards/chosen": 0.040990687906742096, | |
| "rewards/margins": 0.01583397202193737, | |
| "rewards/rejected": 0.025156717747449875, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07509493535862098, | |
| "grad_norm": 21.19602898096358, | |
| "learning_rate": 7.414965986394558e-07, | |
| "logits/chosen": -0.007384412921965122, | |
| "logits/rejected": -0.016086794435977936, | |
| "logps/chosen": -189.52395629882812, | |
| "logps/rejected": -192.64816284179688, | |
| "loss": 0.6817, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.05572628974914551, | |
| "rewards/margins": 0.024477079510688782, | |
| "rewards/rejected": 0.031249215826392174, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.08192174766395016, | |
| "grad_norm": 20.08862529877448, | |
| "learning_rate": 8.095238095238095e-07, | |
| "logits/chosen": -0.04889947175979614, | |
| "logits/rejected": -0.049361489713191986, | |
| "logps/chosen": -197.39492797851562, | |
| "logps/rejected": -192.8791046142578, | |
| "loss": 0.6828, | |
| "rewards/accuracies": 0.5843750238418579, | |
| "rewards/chosen": 0.059998854994773865, | |
| "rewards/margins": 0.023517701774835587, | |
| "rewards/rejected": 0.03648114949464798, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.08874855996927934, | |
| "grad_norm": 19.78186965312465, | |
| "learning_rate": 8.775510204081632e-07, | |
| "logits/chosen": -0.022162066772580147, | |
| "logits/rejected": -0.02603471651673317, | |
| "logps/chosen": -192.2538604736328, | |
| "logps/rejected": -190.6973876953125, | |
| "loss": 0.6782, | |
| "rewards/accuracies": 0.651562511920929, | |
| "rewards/chosen": 0.07047584652900696, | |
| "rewards/margins": 0.03453099727630615, | |
| "rewards/rejected": 0.035944852977991104, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.09557537227460852, | |
| "grad_norm": 21.72668562860521, | |
| "learning_rate": 9.45578231292517e-07, | |
| "logits/chosen": -0.028122998774051666, | |
| "logits/rejected": -0.0023567965254187584, | |
| "logps/chosen": -193.58602905273438, | |
| "logps/rejected": -189.49517822265625, | |
| "loss": 0.6721, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 0.08927410840988159, | |
| "rewards/margins": 0.04811044782400131, | |
| "rewards/rejected": 0.04116365686058998, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.1024021845799377, | |
| "grad_norm": 20.630914226397604, | |
| "learning_rate": 9.98482549317147e-07, | |
| "logits/chosen": -0.07732997089624405, | |
| "logits/rejected": -0.08366119861602783, | |
| "logps/chosen": -203.80441284179688, | |
| "logps/rejected": -202.51812744140625, | |
| "loss": 0.666, | |
| "rewards/accuracies": 0.6500000357627869, | |
| "rewards/chosen": 0.10389578342437744, | |
| "rewards/margins": 0.06255247443914413, | |
| "rewards/rejected": 0.04134330898523331, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.10922899688526688, | |
| "grad_norm": 20.25669433495337, | |
| "learning_rate": 9.908952959028832e-07, | |
| "logits/chosen": -0.09441889822483063, | |
| "logits/rejected": -0.08870529383420944, | |
| "logps/chosen": -185.63307189941406, | |
| "logps/rejected": -186.53253173828125, | |
| "loss": 0.6654, | |
| "rewards/accuracies": 0.6500000357627869, | |
| "rewards/chosen": 0.08515263348817825, | |
| "rewards/margins": 0.0660884901881218, | |
| "rewards/rejected": 0.019064147025346756, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.11605580919059606, | |
| "grad_norm": 20.384593980794733, | |
| "learning_rate": 9.833080424886191e-07, | |
| "logits/chosen": -0.08715031296014786, | |
| "logits/rejected": -0.05636933073401451, | |
| "logps/chosen": -188.3374481201172, | |
| "logps/rejected": -190.37437438964844, | |
| "loss": 0.659, | |
| "rewards/accuracies": 0.6343749761581421, | |
| "rewards/chosen": 0.04690036177635193, | |
| "rewards/margins": 0.08634677529335022, | |
| "rewards/rejected": -0.03944641351699829, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.12288262149592524, | |
| "grad_norm": 21.86056528276187, | |
| "learning_rate": 9.75720789074355e-07, | |
| "logits/chosen": -0.07912790030241013, | |
| "logits/rejected": -0.07271625846624374, | |
| "logps/chosen": -197.11959838867188, | |
| "logps/rejected": -197.41287231445312, | |
| "loss": 0.6528, | |
| "rewards/accuracies": 0.6749999523162842, | |
| "rewards/chosen": 0.04622086510062218, | |
| "rewards/margins": 0.10496747493743896, | |
| "rewards/rejected": -0.058746613562107086, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.12970943380125444, | |
| "grad_norm": 22.24802422589698, | |
| "learning_rate": 9.68133535660091e-07, | |
| "logits/chosen": -0.07506565004587173, | |
| "logits/rejected": -0.05108420550823212, | |
| "logps/chosen": -190.35340881347656, | |
| "logps/rejected": -195.009521484375, | |
| "loss": 0.6441, | |
| "rewards/accuracies": 0.6812500357627869, | |
| "rewards/chosen": 0.052541881799697876, | |
| "rewards/margins": 0.12386594712734222, | |
| "rewards/rejected": -0.07132406532764435, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1365362461065836, | |
| "grad_norm": 22.419822765649933, | |
| "learning_rate": 9.60546282245827e-07, | |
| "logits/chosen": -0.11874101310968399, | |
| "logits/rejected": -0.08336825668811798, | |
| "logps/chosen": -193.62611389160156, | |
| "logps/rejected": -196.01084899902344, | |
| "loss": 0.6249, | |
| "rewards/accuracies": 0.7046875357627869, | |
| "rewards/chosen": 0.03949081152677536, | |
| "rewards/margins": 0.17370560765266418, | |
| "rewards/rejected": -0.13421478867530823, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.1433630584119128, | |
| "grad_norm": 22.915739502006815, | |
| "learning_rate": 9.52959028831563e-07, | |
| "logits/chosen": -0.17365601658821106, | |
| "logits/rejected": -0.15520283579826355, | |
| "logps/chosen": -203.1890869140625, | |
| "logps/rejected": -200.14974975585938, | |
| "loss": 0.6287, | |
| "rewards/accuracies": 0.6687500476837158, | |
| "rewards/chosen": -0.01979774236679077, | |
| "rewards/margins": 0.18479280173778534, | |
| "rewards/rejected": -0.2045905441045761, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.15018987071724196, | |
| "grad_norm": 20.769969852017695, | |
| "learning_rate": 9.453717754172988e-07, | |
| "logits/chosen": -0.1847243756055832, | |
| "logits/rejected": -0.15192236006259918, | |
| "logps/chosen": -198.33010864257812, | |
| "logps/rejected": -200.56228637695312, | |
| "loss": 0.6015, | |
| "rewards/accuracies": 0.7109375, | |
| "rewards/chosen": -0.03269830346107483, | |
| "rewards/margins": 0.25339096784591675, | |
| "rewards/rejected": -0.2860892415046692, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.15701668302257116, | |
| "grad_norm": 21.597574913870996, | |
| "learning_rate": 9.377845220030348e-07, | |
| "logits/chosen": -0.21274694800376892, | |
| "logits/rejected": -0.19206659495830536, | |
| "logps/chosen": -197.59228515625, | |
| "logps/rejected": -200.42283630371094, | |
| "loss": 0.611, | |
| "rewards/accuracies": 0.6734374761581421, | |
| "rewards/chosen": -0.09015801548957825, | |
| "rewards/margins": 0.24926723539829254, | |
| "rewards/rejected": -0.3394252359867096, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.16384349532790032, | |
| "grad_norm": 24.09497342960952, | |
| "learning_rate": 9.301972685887707e-07, | |
| "logits/chosen": -0.2293986827135086, | |
| "logits/rejected": -0.19997453689575195, | |
| "logps/chosen": -191.1751251220703, | |
| "logps/rejected": -196.63511657714844, | |
| "loss": 0.6125, | |
| "rewards/accuracies": 0.690625011920929, | |
| "rewards/chosen": -0.15094764530658722, | |
| "rewards/margins": 0.24523335695266724, | |
| "rewards/rejected": -0.39618098735809326, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.17067030763322952, | |
| "grad_norm": 22.186402685803138, | |
| "learning_rate": 9.226100151745068e-07, | |
| "logits/chosen": -0.23599499464035034, | |
| "logits/rejected": -0.20987126231193542, | |
| "logps/chosen": -191.61639404296875, | |
| "logps/rejected": -197.80091857910156, | |
| "loss": 0.6205, | |
| "rewards/accuracies": 0.6546875238418579, | |
| "rewards/chosen": -0.22373469173908234, | |
| "rewards/margins": 0.2635762691497803, | |
| "rewards/rejected": -0.4873109459877014, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.17749711993855868, | |
| "grad_norm": 23.30196457741843, | |
| "learning_rate": 9.150227617602428e-07, | |
| "logits/chosen": -0.2195354700088501, | |
| "logits/rejected": -0.19019638001918793, | |
| "logps/chosen": -190.50746154785156, | |
| "logps/rejected": -195.74331665039062, | |
| "loss": 0.6056, | |
| "rewards/accuracies": 0.7046875357627869, | |
| "rewards/chosen": -0.2523514926433563, | |
| "rewards/margins": 0.29894089698791504, | |
| "rewards/rejected": -0.5512923002243042, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.18432393224388788, | |
| "grad_norm": 23.437160399579792, | |
| "learning_rate": 9.074355083459787e-07, | |
| "logits/chosen": -0.2144363671541214, | |
| "logits/rejected": -0.19538246095180511, | |
| "logps/chosen": -194.883056640625, | |
| "logps/rejected": -202.83575439453125, | |
| "loss": 0.595, | |
| "rewards/accuracies": 0.7078125476837158, | |
| "rewards/chosen": -0.27382633090019226, | |
| "rewards/margins": 0.3095867931842804, | |
| "rewards/rejected": -0.5834130644798279, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.19115074454921704, | |
| "grad_norm": 23.67928529051871, | |
| "learning_rate": 8.998482549317147e-07, | |
| "logits/chosen": -0.2671777606010437, | |
| "logits/rejected": -0.23835715651512146, | |
| "logps/chosen": -189.7034912109375, | |
| "logps/rejected": -194.55117797851562, | |
| "loss": 0.589, | |
| "rewards/accuracies": 0.7000000476837158, | |
| "rewards/chosen": -0.2815781235694885, | |
| "rewards/margins": 0.34006255865097046, | |
| "rewards/rejected": -0.621640682220459, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.19797755685454624, | |
| "grad_norm": 26.3785919721159, | |
| "learning_rate": 8.922610015174506e-07, | |
| "logits/chosen": -0.2851921319961548, | |
| "logits/rejected": -0.2668570280075073, | |
| "logps/chosen": -202.77801513671875, | |
| "logps/rejected": -207.8894805908203, | |
| "loss": 0.59, | |
| "rewards/accuracies": 0.7046875357627869, | |
| "rewards/chosen": -0.33676964044570923, | |
| "rewards/margins": 0.35969871282577515, | |
| "rewards/rejected": -0.6964683532714844, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.2048043691598754, | |
| "grad_norm": 23.715391013722297, | |
| "learning_rate": 8.846737481031866e-07, | |
| "logits/chosen": -0.2776036262512207, | |
| "logits/rejected": -0.24332435429096222, | |
| "logps/chosen": -201.10296630859375, | |
| "logps/rejected": -203.72195434570312, | |
| "loss": 0.6111, | |
| "rewards/accuracies": 0.715624988079071, | |
| "rewards/chosen": -0.3781723380088806, | |
| "rewards/margins": 0.3227519989013672, | |
| "rewards/rejected": -0.700924277305603, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.2116311814652046, | |
| "grad_norm": 21.57268816738927, | |
| "learning_rate": 8.770864946889226e-07, | |
| "logits/chosen": -0.29242080450057983, | |
| "logits/rejected": -0.2669425308704376, | |
| "logps/chosen": -204.4817352294922, | |
| "logps/rejected": -214.0943603515625, | |
| "loss": 0.5794, | |
| "rewards/accuracies": 0.7265625, | |
| "rewards/chosen": -0.3647349178791046, | |
| "rewards/margins": 0.4395143985748291, | |
| "rewards/rejected": -0.8042493462562561, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.21845799377053376, | |
| "grad_norm": 25.227342019618998, | |
| "learning_rate": 8.694992412746586e-07, | |
| "logits/chosen": -0.27386438846588135, | |
| "logits/rejected": -0.2711098790168762, | |
| "logps/chosen": -198.40101623535156, | |
| "logps/rejected": -204.6220703125, | |
| "loss": 0.5727, | |
| "rewards/accuracies": 0.7281250357627869, | |
| "rewards/chosen": -0.3862449824810028, | |
| "rewards/margins": 0.41143903136253357, | |
| "rewards/rejected": -0.7976840734481812, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.22528480607586296, | |
| "grad_norm": 24.00522520700325, | |
| "learning_rate": 8.619119878603945e-07, | |
| "logits/chosen": -0.3334537744522095, | |
| "logits/rejected": -0.3187546730041504, | |
| "logps/chosen": -208.01986694335938, | |
| "logps/rejected": -212.91488647460938, | |
| "loss": 0.5913, | |
| "rewards/accuracies": 0.6703125238418579, | |
| "rewards/chosen": -0.4798099398612976, | |
| "rewards/margins": 0.37955817580223083, | |
| "rewards/rejected": -0.8593681454658508, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.23211161838119213, | |
| "grad_norm": 23.49360024665317, | |
| "learning_rate": 8.543247344461305e-07, | |
| "logits/chosen": -0.30438894033432007, | |
| "logits/rejected": -0.28073978424072266, | |
| "logps/chosen": -203.7110595703125, | |
| "logps/rejected": -211.83615112304688, | |
| "loss": 0.56, | |
| "rewards/accuracies": 0.7328125238418579, | |
| "rewards/chosen": -0.3902357518672943, | |
| "rewards/margins": 0.5086088180541992, | |
| "rewards/rejected": -0.8988445401191711, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.23893843068652132, | |
| "grad_norm": 23.086500001623612, | |
| "learning_rate": 8.467374810318663e-07, | |
| "logits/chosen": -0.3257724940776825, | |
| "logits/rejected": -0.2853447198867798, | |
| "logps/chosen": -204.09765625, | |
| "logps/rejected": -212.38494873046875, | |
| "loss": 0.5515, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -0.4673992991447449, | |
| "rewards/margins": 0.5267953872680664, | |
| "rewards/rejected": -0.9941946864128113, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.24576524299185049, | |
| "grad_norm": 24.60129579583855, | |
| "learning_rate": 8.391502276176023e-07, | |
| "logits/chosen": -0.3029869794845581, | |
| "logits/rejected": -0.2718327045440674, | |
| "logps/chosen": -196.5174560546875, | |
| "logps/rejected": -204.4929656982422, | |
| "loss": 0.5809, | |
| "rewards/accuracies": 0.7046875357627869, | |
| "rewards/chosen": -0.4800136089324951, | |
| "rewards/margins": 0.43177759647369385, | |
| "rewards/rejected": -0.9117912650108337, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.25259205529717965, | |
| "grad_norm": 23.03353178121409, | |
| "learning_rate": 8.315629742033384e-07, | |
| "logits/chosen": -0.28175657987594604, | |
| "logits/rejected": -0.2525416612625122, | |
| "logps/chosen": -197.58517456054688, | |
| "logps/rejected": -210.83853149414062, | |
| "loss": 0.5675, | |
| "rewards/accuracies": 0.7234375476837158, | |
| "rewards/chosen": -0.5489044785499573, | |
| "rewards/margins": 0.4759043753147125, | |
| "rewards/rejected": -1.0248088836669922, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.2594188676025089, | |
| "grad_norm": 21.702116754195792, | |
| "learning_rate": 8.239757207890743e-07, | |
| "logits/chosen": -0.3090224266052246, | |
| "logits/rejected": -0.2872709333896637, | |
| "logps/chosen": -204.044921875, | |
| "logps/rejected": -214.3769989013672, | |
| "loss": 0.5414, | |
| "rewards/accuracies": 0.7265625, | |
| "rewards/chosen": -0.501671552658081, | |
| "rewards/margins": 0.5782625675201416, | |
| "rewards/rejected": -1.0799341201782227, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.26624567990783804, | |
| "grad_norm": 22.690534272455945, | |
| "learning_rate": 8.163884673748103e-07, | |
| "logits/chosen": -0.2652078866958618, | |
| "logits/rejected": -0.22916777431964874, | |
| "logps/chosen": -206.28855895996094, | |
| "logps/rejected": -217.3023681640625, | |
| "loss": 0.532, | |
| "rewards/accuracies": 0.7343750596046448, | |
| "rewards/chosen": -0.47486239671707153, | |
| "rewards/margins": 0.6135950684547424, | |
| "rewards/rejected": -1.088457465171814, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.2730724922131672, | |
| "grad_norm": 24.587498727216616, | |
| "learning_rate": 8.088012139605462e-07, | |
| "logits/chosen": -0.28489071130752563, | |
| "logits/rejected": -0.23875750601291656, | |
| "logps/chosen": -202.77565002441406, | |
| "logps/rejected": -216.6030731201172, | |
| "loss": 0.5272, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -0.5735213756561279, | |
| "rewards/margins": 0.6210550665855408, | |
| "rewards/rejected": -1.194576382637024, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2798993045184964, | |
| "grad_norm": 24.707605897401567, | |
| "learning_rate": 8.012139605462822e-07, | |
| "logits/chosen": -0.3593894839286804, | |
| "logits/rejected": -0.3138624429702759, | |
| "logps/chosen": -202.06204223632812, | |
| "logps/rejected": -208.73065185546875, | |
| "loss": 0.5575, | |
| "rewards/accuracies": 0.721875011920929, | |
| "rewards/chosen": -0.604642391204834, | |
| "rewards/margins": 0.5522481203079224, | |
| "rewards/rejected": -1.156890630722046, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.2867261168238256, | |
| "grad_norm": 24.754070000277498, | |
| "learning_rate": 7.936267071320181e-07, | |
| "logits/chosen": -0.3502323627471924, | |
| "logits/rejected": -0.3173756003379822, | |
| "logps/chosen": -207.6633758544922, | |
| "logps/rejected": -216.3917236328125, | |
| "loss": 0.5265, | |
| "rewards/accuracies": 0.7484375238418579, | |
| "rewards/chosen": -0.6551162004470825, | |
| "rewards/margins": 0.6169639229774475, | |
| "rewards/rejected": -1.2720801830291748, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.29355292912915476, | |
| "grad_norm": 23.564476771066985, | |
| "learning_rate": 7.860394537177542e-07, | |
| "logits/chosen": -0.3500007092952728, | |
| "logits/rejected": -0.32545575499534607, | |
| "logps/chosen": -211.29928588867188, | |
| "logps/rejected": -227.12037658691406, | |
| "loss": 0.5223, | |
| "rewards/accuracies": 0.7421875596046448, | |
| "rewards/chosen": -0.7528213262557983, | |
| "rewards/margins": 0.739406943321228, | |
| "rewards/rejected": -1.492228388786316, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3003797414344839, | |
| "grad_norm": 21.091018091079327, | |
| "learning_rate": 7.784522003034901e-07, | |
| "logits/chosen": -0.35516998171806335, | |
| "logits/rejected": -0.3074837327003479, | |
| "logps/chosen": -203.1188507080078, | |
| "logps/rejected": -212.15496826171875, | |
| "loss": 0.5055, | |
| "rewards/accuracies": 0.7765625715255737, | |
| "rewards/chosen": -0.6801650524139404, | |
| "rewards/margins": 0.7159599661827087, | |
| "rewards/rejected": -1.396125078201294, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3072065537398131, | |
| "grad_norm": 30.178688833532316, | |
| "learning_rate": 7.708649468892261e-07, | |
| "logits/chosen": -0.3771928548812866, | |
| "logits/rejected": -0.34754854440689087, | |
| "logps/chosen": -208.95216369628906, | |
| "logps/rejected": -225.38938903808594, | |
| "loss": 0.5226, | |
| "rewards/accuracies": 0.746874988079071, | |
| "rewards/chosen": -0.7856850624084473, | |
| "rewards/margins": 0.6984450817108154, | |
| "rewards/rejected": -1.4841301441192627, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.3140333660451423, | |
| "grad_norm": 22.73508892423378, | |
| "learning_rate": 7.632776934749621e-07, | |
| "logits/chosen": -0.40090760588645935, | |
| "logits/rejected": -0.3806273937225342, | |
| "logps/chosen": -208.29766845703125, | |
| "logps/rejected": -223.73020935058594, | |
| "loss": 0.5013, | |
| "rewards/accuracies": 0.765625, | |
| "rewards/chosen": -0.7431963086128235, | |
| "rewards/margins": 0.8224382400512695, | |
| "rewards/rejected": -1.5656344890594482, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.3208601783504715, | |
| "grad_norm": 24.65367082247547, | |
| "learning_rate": 7.55690440060698e-07, | |
| "logits/chosen": -0.41392359137535095, | |
| "logits/rejected": -0.3990693688392639, | |
| "logps/chosen": -211.69845581054688, | |
| "logps/rejected": -222.681884765625, | |
| "loss": 0.4896, | |
| "rewards/accuracies": 0.7671874761581421, | |
| "rewards/chosen": -0.7812504768371582, | |
| "rewards/margins": 0.8228715062141418, | |
| "rewards/rejected": -1.6041220426559448, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.32768699065580065, | |
| "grad_norm": 26.060565630616303, | |
| "learning_rate": 7.481031866464339e-07, | |
| "logits/chosen": -0.4470677673816681, | |
| "logits/rejected": -0.4043146073818207, | |
| "logps/chosen": -201.87158203125, | |
| "logps/rejected": -216.65240478515625, | |
| "loss": 0.5178, | |
| "rewards/accuracies": 0.7406250238418579, | |
| "rewards/chosen": -0.8349807858467102, | |
| "rewards/margins": 0.7298619151115417, | |
| "rewards/rejected": -1.564842700958252, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.3345138029611298, | |
| "grad_norm": 24.867787006387463, | |
| "learning_rate": 7.405159332321699e-07, | |
| "logits/chosen": -0.4602758288383484, | |
| "logits/rejected": -0.4031441807746887, | |
| "logps/chosen": -215.20541381835938, | |
| "logps/rejected": -234.6583251953125, | |
| "loss": 0.5155, | |
| "rewards/accuracies": 0.7484375238418579, | |
| "rewards/chosen": -0.9265861511230469, | |
| "rewards/margins": 0.8055697679519653, | |
| "rewards/rejected": -1.7321559190750122, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.34134061526645904, | |
| "grad_norm": 32.86790243336268, | |
| "learning_rate": 7.329286798179059e-07, | |
| "logits/chosen": -0.4144153594970703, | |
| "logits/rejected": -0.3892706036567688, | |
| "logps/chosen": -216.45887756347656, | |
| "logps/rejected": -225.97056579589844, | |
| "loss": 0.5274, | |
| "rewards/accuracies": 0.7328125238418579, | |
| "rewards/chosen": -0.9314414262771606, | |
| "rewards/margins": 0.7752954363822937, | |
| "rewards/rejected": -1.7067368030548096, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3481674275717882, | |
| "grad_norm": 29.0406209714796, | |
| "learning_rate": 7.253414264036418e-07, | |
| "logits/chosen": -0.4518946707248688, | |
| "logits/rejected": -0.4360005855560303, | |
| "logps/chosen": -210.40875244140625, | |
| "logps/rejected": -227.6586456298828, | |
| "loss": 0.4918, | |
| "rewards/accuracies": 0.7640624642372131, | |
| "rewards/chosen": -0.7644888162612915, | |
| "rewards/margins": 0.8264600038528442, | |
| "rewards/rejected": -1.5909489393234253, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.35499423987711737, | |
| "grad_norm": 29.792037648827193, | |
| "learning_rate": 7.177541729893778e-07, | |
| "logits/chosen": -0.46055272221565247, | |
| "logits/rejected": -0.41955289244651794, | |
| "logps/chosen": -203.9451904296875, | |
| "logps/rejected": -225.48402404785156, | |
| "loss": 0.5137, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.7920026779174805, | |
| "rewards/margins": 0.7750235795974731, | |
| "rewards/rejected": -1.5670262575149536, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.36182105218244653, | |
| "grad_norm": 28.48324275582042, | |
| "learning_rate": 7.101669195751137e-07, | |
| "logits/chosen": -0.44266417622566223, | |
| "logits/rejected": -0.4136849045753479, | |
| "logps/chosen": -217.11045837402344, | |
| "logps/rejected": -232.384521484375, | |
| "loss": 0.5059, | |
| "rewards/accuracies": 0.7500000596046448, | |
| "rewards/chosen": -0.9079422950744629, | |
| "rewards/margins": 0.8452929258346558, | |
| "rewards/rejected": -1.7532353401184082, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.36864786448777576, | |
| "grad_norm": 24.346858846505146, | |
| "learning_rate": 7.025796661608497e-07, | |
| "logits/chosen": -0.4453073740005493, | |
| "logits/rejected": -0.39773428440093994, | |
| "logps/chosen": -199.64686584472656, | |
| "logps/rejected": -217.36294555664062, | |
| "loss": 0.5282, | |
| "rewards/accuracies": 0.745312511920929, | |
| "rewards/chosen": -0.7881425023078918, | |
| "rewards/margins": 0.745051383972168, | |
| "rewards/rejected": -1.533193826675415, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.3754746767931049, | |
| "grad_norm": 23.88017645464549, | |
| "learning_rate": 6.949924127465857e-07, | |
| "logits/chosen": -0.4227825701236725, | |
| "logits/rejected": -0.3899107873439789, | |
| "logps/chosen": -218.3785400390625, | |
| "logps/rejected": -230.14222717285156, | |
| "loss": 0.5021, | |
| "rewards/accuracies": 0.7703125476837158, | |
| "rewards/chosen": -0.704402506351471, | |
| "rewards/margins": 0.8275265693664551, | |
| "rewards/rejected": -1.5319291353225708, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.3823014890984341, | |
| "grad_norm": 23.672046628232867, | |
| "learning_rate": 6.874051593323217e-07, | |
| "logits/chosen": -0.42757853865623474, | |
| "logits/rejected": -0.394180566072464, | |
| "logps/chosen": -208.079345703125, | |
| "logps/rejected": -228.22598266601562, | |
| "loss": 0.4667, | |
| "rewards/accuracies": 0.770312488079071, | |
| "rewards/chosen": -0.8188365697860718, | |
| "rewards/margins": 0.9388971328735352, | |
| "rewards/rejected": -1.7577338218688965, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.38912830140376325, | |
| "grad_norm": 27.539677366232738, | |
| "learning_rate": 6.798179059180577e-07, | |
| "logits/chosen": -0.4404156506061554, | |
| "logits/rejected": -0.3975413739681244, | |
| "logps/chosen": -208.03125, | |
| "logps/rejected": -224.20956420898438, | |
| "loss": 0.5004, | |
| "rewards/accuracies": 0.7593750357627869, | |
| "rewards/chosen": -0.8374041318893433, | |
| "rewards/margins": 0.7886074781417847, | |
| "rewards/rejected": -1.6260114908218384, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.3959551137090925, | |
| "grad_norm": 25.29375987198196, | |
| "learning_rate": 6.722306525037936e-07, | |
| "logits/chosen": -0.4404994249343872, | |
| "logits/rejected": -0.40123340487480164, | |
| "logps/chosen": -213.8634490966797, | |
| "logps/rejected": -234.7059326171875, | |
| "loss": 0.497, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.8731653094291687, | |
| "rewards/margins": 0.9025252461433411, | |
| "rewards/rejected": -1.7756905555725098, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.40278192601442164, | |
| "grad_norm": 23.196272876570017, | |
| "learning_rate": 6.646433990895296e-07, | |
| "logits/chosen": -0.42805609107017517, | |
| "logits/rejected": -0.3933747410774231, | |
| "logps/chosen": -210.49766540527344, | |
| "logps/rejected": -230.8019256591797, | |
| "loss": 0.472, | |
| "rewards/accuracies": 0.7718750238418579, | |
| "rewards/chosen": -0.9004274606704712, | |
| "rewards/margins": 0.9456923604011536, | |
| "rewards/rejected": -1.84611976146698, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.4096087383197508, | |
| "grad_norm": 26.112729497646914, | |
| "learning_rate": 6.570561456752655e-07, | |
| "logits/chosen": -0.419676810503006, | |
| "logits/rejected": -0.3932231068611145, | |
| "logps/chosen": -212.6820831298828, | |
| "logps/rejected": -230.1705322265625, | |
| "loss": 0.4551, | |
| "rewards/accuracies": 0.776562511920929, | |
| "rewards/chosen": -0.9198075532913208, | |
| "rewards/margins": 1.0145457983016968, | |
| "rewards/rejected": -1.9343533515930176, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.41643555062508, | |
| "grad_norm": 26.539025702964505, | |
| "learning_rate": 6.494688922610015e-07, | |
| "logits/chosen": -0.47971057891845703, | |
| "logits/rejected": -0.43692541122436523, | |
| "logps/chosen": -212.31594848632812, | |
| "logps/rejected": -234.5380859375, | |
| "loss": 0.4563, | |
| "rewards/accuracies": 0.770312488079071, | |
| "rewards/chosen": -1.0747839212417603, | |
| "rewards/margins": 1.078429937362671, | |
| "rewards/rejected": -2.1532137393951416, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.4232623629304092, | |
| "grad_norm": 26.2859842178028, | |
| "learning_rate": 6.418816388467374e-07, | |
| "logits/chosen": -0.4652007818222046, | |
| "logits/rejected": -0.4464990496635437, | |
| "logps/chosen": -212.9930419921875, | |
| "logps/rejected": -230.19207763671875, | |
| "loss": 0.4778, | |
| "rewards/accuracies": 0.7906250357627869, | |
| "rewards/chosen": -1.1166890859603882, | |
| "rewards/margins": 0.9617180228233337, | |
| "rewards/rejected": -2.0784072875976562, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.43008917523573836, | |
| "grad_norm": 27.943160005363282, | |
| "learning_rate": 6.342943854324734e-07, | |
| "logits/chosen": -0.507358968257904, | |
| "logits/rejected": -0.46083295345306396, | |
| "logps/chosen": -211.0389404296875, | |
| "logps/rejected": -234.06576538085938, | |
| "loss": 0.4689, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -1.125166654586792, | |
| "rewards/margins": 1.1086124181747437, | |
| "rewards/rejected": -2.233778953552246, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.43691598754106753, | |
| "grad_norm": 27.031702699703523, | |
| "learning_rate": 6.267071320182093e-07, | |
| "logits/chosen": -0.5109987854957581, | |
| "logits/rejected": -0.4727884531021118, | |
| "logps/chosen": -216.13302612304688, | |
| "logps/rejected": -241.88287353515625, | |
| "loss": 0.4635, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -1.1550945043563843, | |
| "rewards/margins": 1.1085400581359863, | |
| "rewards/rejected": -2.263634443283081, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.4437427998463967, | |
| "grad_norm": 26.49416191451856, | |
| "learning_rate": 6.191198786039453e-07, | |
| "logits/chosen": -0.5049822330474854, | |
| "logits/rejected": -0.46804797649383545, | |
| "logps/chosen": -220.15802001953125, | |
| "logps/rejected": -241.11386108398438, | |
| "loss": 0.4646, | |
| "rewards/accuracies": 0.770312488079071, | |
| "rewards/chosen": -1.1580806970596313, | |
| "rewards/margins": 1.064436435699463, | |
| "rewards/rejected": -2.222517490386963, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4505696121517259, | |
| "grad_norm": 28.052993928802096, | |
| "learning_rate": 6.115326251896813e-07, | |
| "logits/chosen": -0.5224714875221252, | |
| "logits/rejected": -0.496852308511734, | |
| "logps/chosen": -217.48992919921875, | |
| "logps/rejected": -234.48318481445312, | |
| "loss": 0.5188, | |
| "rewards/accuracies": 0.7671874761581421, | |
| "rewards/chosen": -1.1128088235855103, | |
| "rewards/margins": 0.9438337087631226, | |
| "rewards/rejected": -2.056642532348633, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.4573964244570551, | |
| "grad_norm": 32.11947138128127, | |
| "learning_rate": 6.039453717754173e-07, | |
| "logits/chosen": -0.4993141293525696, | |
| "logits/rejected": -0.4682856798171997, | |
| "logps/chosen": -206.40176391601562, | |
| "logps/rejected": -231.08042907714844, | |
| "loss": 0.4953, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -1.0732593536376953, | |
| "rewards/margins": 1.1166470050811768, | |
| "rewards/rejected": -2.189906358718872, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.46422323676238425, | |
| "grad_norm": 24.595239877463356, | |
| "learning_rate": 5.963581183611533e-07, | |
| "logits/chosen": -0.5414324998855591, | |
| "logits/rejected": -0.5145028233528137, | |
| "logps/chosen": -219.66567993164062, | |
| "logps/rejected": -236.0765380859375, | |
| "loss": 0.4608, | |
| "rewards/accuracies": 0.7781250476837158, | |
| "rewards/chosen": -0.9715930819511414, | |
| "rewards/margins": 1.0554088354110718, | |
| "rewards/rejected": -2.0270018577575684, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.47105004906771347, | |
| "grad_norm": 27.819824043736283, | |
| "learning_rate": 5.887708649468892e-07, | |
| "logits/chosen": -0.482106477022171, | |
| "logits/rejected": -0.43574321269989014, | |
| "logps/chosen": -211.92596435546875, | |
| "logps/rejected": -234.6639862060547, | |
| "loss": 0.4352, | |
| "rewards/accuracies": 0.792187511920929, | |
| "rewards/chosen": -0.9926649332046509, | |
| "rewards/margins": 1.141036033630371, | |
| "rewards/rejected": -2.1337008476257324, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.47787686137304264, | |
| "grad_norm": 27.82950606174818, | |
| "learning_rate": 5.811836115326252e-07, | |
| "logits/chosen": -0.491192102432251, | |
| "logits/rejected": -0.45507892966270447, | |
| "logps/chosen": -215.52423095703125, | |
| "logps/rejected": -239.1810302734375, | |
| "loss": 0.4534, | |
| "rewards/accuracies": 0.7812500596046448, | |
| "rewards/chosen": -1.0811206102371216, | |
| "rewards/margins": 1.171852469444275, | |
| "rewards/rejected": -2.2529730796813965, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4847036736783718, | |
| "grad_norm": 32.40109215164061, | |
| "learning_rate": 5.735963581183611e-07, | |
| "logits/chosen": -0.48725226521492004, | |
| "logits/rejected": -0.4451846480369568, | |
| "logps/chosen": -211.22933959960938, | |
| "logps/rejected": -236.77740478515625, | |
| "loss": 0.4487, | |
| "rewards/accuracies": 0.7828124761581421, | |
| "rewards/chosen": -1.0895929336547852, | |
| "rewards/margins": 1.1770341396331787, | |
| "rewards/rejected": -2.2666268348693848, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.49153048598370097, | |
| "grad_norm": 27.259651037643604, | |
| "learning_rate": 5.660091047040971e-07, | |
| "logits/chosen": -0.5053711533546448, | |
| "logits/rejected": -0.4444194436073303, | |
| "logps/chosen": -205.80319213867188, | |
| "logps/rejected": -230.7117919921875, | |
| "loss": 0.4743, | |
| "rewards/accuracies": 0.770312488079071, | |
| "rewards/chosen": -1.1533528566360474, | |
| "rewards/margins": 1.07535982131958, | |
| "rewards/rejected": -2.228712797164917, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.4983572982890302, | |
| "grad_norm": 23.45407239305211, | |
| "learning_rate": 5.584218512898331e-07, | |
| "logits/chosen": -0.46755921840667725, | |
| "logits/rejected": -0.41828638315200806, | |
| "logps/chosen": -214.959716796875, | |
| "logps/rejected": -237.14413452148438, | |
| "loss": 0.4451, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -1.1057405471801758, | |
| "rewards/margins": 1.1711297035217285, | |
| "rewards/rejected": -2.2768704891204834, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5051841105943593, | |
| "grad_norm": 24.513672931022274, | |
| "learning_rate": 5.508345978755691e-07, | |
| "logits/chosen": -0.5107758045196533, | |
| "logits/rejected": -0.47158223390579224, | |
| "logps/chosen": -214.1978759765625, | |
| "logps/rejected": -236.34100341796875, | |
| "loss": 0.4356, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": -1.0896263122558594, | |
| "rewards/margins": 1.132210612297058, | |
| "rewards/rejected": -2.221837043762207, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5120109228996885, | |
| "grad_norm": 27.557361902005226, | |
| "learning_rate": 5.432473444613049e-07, | |
| "logits/chosen": -0.47495898604393005, | |
| "logits/rejected": -0.42891502380371094, | |
| "logps/chosen": -215.3628692626953, | |
| "logps/rejected": -240.29644775390625, | |
| "loss": 0.4433, | |
| "rewards/accuracies": 0.8046875596046448, | |
| "rewards/chosen": -1.1231842041015625, | |
| "rewards/margins": 1.1870129108428955, | |
| "rewards/rejected": -2.310196876525879, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5188377352050177, | |
| "grad_norm": 25.763088367806024, | |
| "learning_rate": 5.356600910470409e-07, | |
| "logits/chosen": -0.5234218835830688, | |
| "logits/rejected": -0.46476346254348755, | |
| "logps/chosen": -214.0421142578125, | |
| "logps/rejected": -238.0985565185547, | |
| "loss": 0.4236, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -1.1129274368286133, | |
| "rewards/margins": 1.2772108316421509, | |
| "rewards/rejected": -2.3901383876800537, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.5256645475103469, | |
| "grad_norm": 27.345063991868273, | |
| "learning_rate": 5.280728376327769e-07, | |
| "logits/chosen": -0.5037857294082642, | |
| "logits/rejected": -0.4784386157989502, | |
| "logps/chosen": -210.1291046142578, | |
| "logps/rejected": -236.04969787597656, | |
| "loss": 0.4347, | |
| "rewards/accuracies": 0.8109375238418579, | |
| "rewards/chosen": -1.1615896224975586, | |
| "rewards/margins": 1.2552942037582397, | |
| "rewards/rejected": -2.416883945465088, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.5324913598156761, | |
| "grad_norm": 23.559487104074414, | |
| "learning_rate": 5.204855842185128e-07, | |
| "logits/chosen": -0.5264319777488708, | |
| "logits/rejected": -0.47137507796287537, | |
| "logps/chosen": -218.16024780273438, | |
| "logps/rejected": -245.5438995361328, | |
| "loss": 0.4609, | |
| "rewards/accuracies": 0.7703125476837158, | |
| "rewards/chosen": -1.2951855659484863, | |
| "rewards/margins": 1.170878291130066, | |
| "rewards/rejected": -2.4660637378692627, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.5393181721210053, | |
| "grad_norm": 30.437623350555043, | |
| "learning_rate": 5.128983308042489e-07, | |
| "logits/chosen": -0.4954899251461029, | |
| "logits/rejected": -0.45233067870140076, | |
| "logps/chosen": -213.85757446289062, | |
| "logps/rejected": -242.7041473388672, | |
| "loss": 0.4193, | |
| "rewards/accuracies": 0.8093750476837158, | |
| "rewards/chosen": -1.2700811624526978, | |
| "rewards/margins": 1.2533843517303467, | |
| "rewards/rejected": -2.523465394973755, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.5461449844263344, | |
| "grad_norm": 25.96035380580991, | |
| "learning_rate": 5.053110773899848e-07, | |
| "logits/chosen": -0.49867063760757446, | |
| "logits/rejected": -0.44984591007232666, | |
| "logps/chosen": -218.67074584960938, | |
| "logps/rejected": -247.30982971191406, | |
| "loss": 0.424, | |
| "rewards/accuracies": 0.8046875, | |
| "rewards/chosen": -1.237367868423462, | |
| "rewards/margins": 1.278685212135315, | |
| "rewards/rejected": -2.5160531997680664, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5529717967316636, | |
| "grad_norm": 27.066709483078917, | |
| "learning_rate": 4.977238239757208e-07, | |
| "logits/chosen": -0.4714178144931793, | |
| "logits/rejected": -0.4372885823249817, | |
| "logps/chosen": -218.98892211914062, | |
| "logps/rejected": -242.98770141601562, | |
| "loss": 0.4266, | |
| "rewards/accuracies": 0.7984375357627869, | |
| "rewards/chosen": -1.298151969909668, | |
| "rewards/margins": 1.222092866897583, | |
| "rewards/rejected": -2.520244836807251, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.5597986090369927, | |
| "grad_norm": 28.230804755745105, | |
| "learning_rate": 4.901365705614567e-07, | |
| "logits/chosen": -0.45390385389328003, | |
| "logits/rejected": -0.43030381202697754, | |
| "logps/chosen": -220.013427734375, | |
| "logps/rejected": -241.9390411376953, | |
| "loss": 0.4526, | |
| "rewards/accuracies": 0.796875, | |
| "rewards/chosen": -1.354661464691162, | |
| "rewards/margins": 1.215053677558899, | |
| "rewards/rejected": -2.5697154998779297, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.566625421342322, | |
| "grad_norm": 32.13534664184047, | |
| "learning_rate": 4.825493171471927e-07, | |
| "logits/chosen": -0.475396066904068, | |
| "logits/rejected": -0.43329310417175293, | |
| "logps/chosen": -210.43185424804688, | |
| "logps/rejected": -236.67987060546875, | |
| "loss": 0.4189, | |
| "rewards/accuracies": 0.815625011920929, | |
| "rewards/chosen": -1.264033317565918, | |
| "rewards/margins": 1.377021074295044, | |
| "rewards/rejected": -2.641054153442383, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.5734522336476512, | |
| "grad_norm": 22.262860568714245, | |
| "learning_rate": 4.7496206373292864e-07, | |
| "logits/chosen": -0.4692656993865967, | |
| "logits/rejected": -0.4306912422180176, | |
| "logps/chosen": -211.5372772216797, | |
| "logps/rejected": -246.39736938476562, | |
| "loss": 0.3916, | |
| "rewards/accuracies": 0.8375000357627869, | |
| "rewards/chosen": -1.1525495052337646, | |
| "rewards/margins": 1.4558607339859009, | |
| "rewards/rejected": -2.608410358428955, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.5802790459529803, | |
| "grad_norm": 22.80617456340079, | |
| "learning_rate": 4.673748103186646e-07, | |
| "logits/chosen": -0.46342021226882935, | |
| "logits/rejected": -0.41512057185173035, | |
| "logps/chosen": -221.32496643066406, | |
| "logps/rejected": -251.7954864501953, | |
| "loss": 0.394, | |
| "rewards/accuracies": 0.8218750357627869, | |
| "rewards/chosen": -1.2910584211349487, | |
| "rewards/margins": 1.4217520952224731, | |
| "rewards/rejected": -2.712810516357422, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.5871058582583095, | |
| "grad_norm": 24.868191575194487, | |
| "learning_rate": 4.597875569044006e-07, | |
| "logits/chosen": -0.48653626441955566, | |
| "logits/rejected": -0.4366312623023987, | |
| "logps/chosen": -217.47422790527344, | |
| "logps/rejected": -241.48968505859375, | |
| "loss": 0.4269, | |
| "rewards/accuracies": 0.801562488079071, | |
| "rewards/chosen": -1.3257293701171875, | |
| "rewards/margins": 1.3266490697860718, | |
| "rewards/rejected": -2.652378559112549, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.5939326705636387, | |
| "grad_norm": 27.035059402616938, | |
| "learning_rate": 4.5220030349013654e-07, | |
| "logits/chosen": -0.5033361911773682, | |
| "logits/rejected": -0.4694429039955139, | |
| "logps/chosen": -214.79815673828125, | |
| "logps/rejected": -237.64102172851562, | |
| "loss": 0.4296, | |
| "rewards/accuracies": 0.7921874523162842, | |
| "rewards/chosen": -1.3357491493225098, | |
| "rewards/margins": 1.2649694681167603, | |
| "rewards/rejected": -2.6007187366485596, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6007594828689679, | |
| "grad_norm": 27.746278145893346, | |
| "learning_rate": 4.446130500758725e-07, | |
| "logits/chosen": -0.5227242708206177, | |
| "logits/rejected": -0.4751604497432709, | |
| "logps/chosen": -218.23658752441406, | |
| "logps/rejected": -249.3454132080078, | |
| "loss": 0.4233, | |
| "rewards/accuracies": 0.817187488079071, | |
| "rewards/chosen": -1.3457627296447754, | |
| "rewards/margins": 1.428666591644287, | |
| "rewards/rejected": -2.7744295597076416, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6075862951742971, | |
| "grad_norm": 26.892931653503698, | |
| "learning_rate": 4.370257966616085e-07, | |
| "logits/chosen": -0.5066260099411011, | |
| "logits/rejected": -0.47855502367019653, | |
| "logps/chosen": -214.84915161132812, | |
| "logps/rejected": -240.56436157226562, | |
| "loss": 0.4612, | |
| "rewards/accuracies": 0.7812500596046448, | |
| "rewards/chosen": -1.4467679262161255, | |
| "rewards/margins": 1.3007091283798218, | |
| "rewards/rejected": -2.7474770545959473, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.6144131074796262, | |
| "grad_norm": 32.793455771900234, | |
| "learning_rate": 4.2943854324734444e-07, | |
| "logits/chosen": -0.4987248182296753, | |
| "logits/rejected": -0.4517776668071747, | |
| "logps/chosen": -218.49545288085938, | |
| "logps/rejected": -252.3199462890625, | |
| "loss": 0.4007, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": -1.4036812782287598, | |
| "rewards/margins": 1.497314453125, | |
| "rewards/rejected": -2.9009957313537598, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6212399197849554, | |
| "grad_norm": 30.14977908240741, | |
| "learning_rate": 4.2185128983308036e-07, | |
| "logits/chosen": -0.5123909711837769, | |
| "logits/rejected": -0.456384539604187, | |
| "logps/chosen": -221.94183349609375, | |
| "logps/rejected": -250.5224151611328, | |
| "loss": 0.4185, | |
| "rewards/accuracies": 0.8265625238418579, | |
| "rewards/chosen": -1.3800506591796875, | |
| "rewards/margins": 1.4040327072143555, | |
| "rewards/rejected": -2.784083366394043, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.6280667320902846, | |
| "grad_norm": 23.187149506889586, | |
| "learning_rate": 4.142640364188164e-07, | |
| "logits/chosen": -0.5007960200309753, | |
| "logits/rejected": -0.4656420350074768, | |
| "logps/chosen": -224.66000366210938, | |
| "logps/rejected": -250.5994873046875, | |
| "loss": 0.4194, | |
| "rewards/accuracies": 0.817187488079071, | |
| "rewards/chosen": -1.4466440677642822, | |
| "rewards/margins": 1.3647561073303223, | |
| "rewards/rejected": -2.8114004135131836, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.6348935443956137, | |
| "grad_norm": 26.465496977643166, | |
| "learning_rate": 4.0667678300455234e-07, | |
| "logits/chosen": -0.5095345973968506, | |
| "logits/rejected": -0.44781219959259033, | |
| "logps/chosen": -219.541259765625, | |
| "logps/rejected": -253.14544677734375, | |
| "loss": 0.3631, | |
| "rewards/accuracies": 0.8765624761581421, | |
| "rewards/chosen": -1.3718998432159424, | |
| "rewards/margins": 1.6033210754394531, | |
| "rewards/rejected": -2.9752209186553955, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.641720356700943, | |
| "grad_norm": 21.651167586614733, | |
| "learning_rate": 3.990895295902883e-07, | |
| "logits/chosen": -0.5611530542373657, | |
| "logits/rejected": -0.5065969824790955, | |
| "logps/chosen": -222.84457397460938, | |
| "logps/rejected": -251.35067749023438, | |
| "loss": 0.397, | |
| "rewards/accuracies": 0.8250000476837158, | |
| "rewards/chosen": -1.4304229021072388, | |
| "rewards/margins": 1.4556035995483398, | |
| "rewards/rejected": -2.886026620864868, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.6485471690062722, | |
| "grad_norm": 21.56653990852637, | |
| "learning_rate": 3.915022761760243e-07, | |
| "logits/chosen": -0.575349748134613, | |
| "logits/rejected": -0.5415146350860596, | |
| "logps/chosen": -209.71266174316406, | |
| "logps/rejected": -239.22946166992188, | |
| "loss": 0.4001, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -1.521388053894043, | |
| "rewards/margins": 1.4791213274002075, | |
| "rewards/rejected": -3.000509262084961, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.6553739813116013, | |
| "grad_norm": 23.31036794244746, | |
| "learning_rate": 3.8391502276176024e-07, | |
| "logits/chosen": -0.5698951482772827, | |
| "logits/rejected": -0.5178714394569397, | |
| "logps/chosen": -228.25030517578125, | |
| "logps/rejected": -261.415771484375, | |
| "loss": 0.3891, | |
| "rewards/accuracies": 0.817187488079071, | |
| "rewards/chosen": -1.6143665313720703, | |
| "rewards/margins": 1.645197868347168, | |
| "rewards/rejected": -3.2595643997192383, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.6622007936169305, | |
| "grad_norm": 26.214223596010875, | |
| "learning_rate": 3.763277693474962e-07, | |
| "logits/chosen": -0.5214463472366333, | |
| "logits/rejected": -0.46749287843704224, | |
| "logps/chosen": -218.10549926757812, | |
| "logps/rejected": -251.87442016601562, | |
| "loss": 0.4196, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -1.63547945022583, | |
| "rewards/margins": 1.4419658184051514, | |
| "rewards/rejected": -3.0774452686309814, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.6690276059222596, | |
| "grad_norm": 24.89349466924626, | |
| "learning_rate": 3.687405159332321e-07, | |
| "logits/chosen": -0.5419428944587708, | |
| "logits/rejected": -0.5022714734077454, | |
| "logps/chosen": -223.1068115234375, | |
| "logps/rejected": -255.94949340820312, | |
| "loss": 0.4144, | |
| "rewards/accuracies": 0.815625011920929, | |
| "rewards/chosen": -1.7430050373077393, | |
| "rewards/margins": 1.4413095712661743, | |
| "rewards/rejected": -3.184314489364624, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.6758544182275888, | |
| "grad_norm": 25.914909518247867, | |
| "learning_rate": 3.611532625189681e-07, | |
| "logits/chosen": -0.5115488767623901, | |
| "logits/rejected": -0.4625004827976227, | |
| "logps/chosen": -229.49105834960938, | |
| "logps/rejected": -265.0625, | |
| "loss": 0.3983, | |
| "rewards/accuracies": 0.8218750357627869, | |
| "rewards/chosen": -1.6320453882217407, | |
| "rewards/margins": 1.5734854936599731, | |
| "rewards/rejected": -3.2055306434631348, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.6826812305329181, | |
| "grad_norm": 31.456143694319483, | |
| "learning_rate": 3.5356600910470406e-07, | |
| "logits/chosen": -0.5371730327606201, | |
| "logits/rejected": -0.4974362850189209, | |
| "logps/chosen": -236.7477569580078, | |
| "logps/rejected": -264.0472106933594, | |
| "loss": 0.3827, | |
| "rewards/accuracies": 0.8437500596046448, | |
| "rewards/chosen": -1.620214819908142, | |
| "rewards/margins": 1.5398459434509277, | |
| "rewards/rejected": -3.1600606441497803, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6895080428382472, | |
| "grad_norm": 29.661159656571126, | |
| "learning_rate": 3.459787556904401e-07, | |
| "logits/chosen": -0.5440015196800232, | |
| "logits/rejected": -0.49301889538764954, | |
| "logps/chosen": -224.03494262695312, | |
| "logps/rejected": -254.42193603515625, | |
| "loss": 0.4033, | |
| "rewards/accuracies": 0.8296875357627869, | |
| "rewards/chosen": -1.5924382209777832, | |
| "rewards/margins": 1.5776193141937256, | |
| "rewards/rejected": -3.170057773590088, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.6963348551435764, | |
| "grad_norm": 38.12069128333079, | |
| "learning_rate": 3.3839150227617604e-07, | |
| "logits/chosen": -0.5860447883605957, | |
| "logits/rejected": -0.543270468711853, | |
| "logps/chosen": -228.84930419921875, | |
| "logps/rejected": -262.8966064453125, | |
| "loss": 0.3898, | |
| "rewards/accuracies": 0.8406250476837158, | |
| "rewards/chosen": -1.6053173542022705, | |
| "rewards/margins": 1.590077519416809, | |
| "rewards/rejected": -3.19539475440979, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.7031616674489056, | |
| "grad_norm": 32.08364090632609, | |
| "learning_rate": 3.30804248861912e-07, | |
| "logits/chosen": -0.6051906943321228, | |
| "logits/rejected": -0.5597983598709106, | |
| "logps/chosen": -224.02899169921875, | |
| "logps/rejected": -258.93511962890625, | |
| "loss": 0.396, | |
| "rewards/accuracies": 0.8171875476837158, | |
| "rewards/chosen": -1.7182796001434326, | |
| "rewards/margins": 1.5724890232086182, | |
| "rewards/rejected": -3.290768623352051, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.7099884797542347, | |
| "grad_norm": 25.599680429412086, | |
| "learning_rate": 3.232169954476479e-07, | |
| "logits/chosen": -0.6112679243087769, | |
| "logits/rejected": -0.5801026821136475, | |
| "logps/chosen": -225.71258544921875, | |
| "logps/rejected": -264.3663330078125, | |
| "loss": 0.3637, | |
| "rewards/accuracies": 0.8421875238418579, | |
| "rewards/chosen": -1.4613301753997803, | |
| "rewards/margins": 1.712023138999939, | |
| "rewards/rejected": -3.1733531951904297, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.716815292059564, | |
| "grad_norm": 26.325121380352627, | |
| "learning_rate": 3.156297420333839e-07, | |
| "logits/chosen": -0.6216264963150024, | |
| "logits/rejected": -0.5548665523529053, | |
| "logps/chosen": -226.58059692382812, | |
| "logps/rejected": -263.7754821777344, | |
| "loss": 0.3636, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -1.6483052968978882, | |
| "rewards/margins": 1.7705044746398926, | |
| "rewards/rejected": -3.4188098907470703, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.7236421043648931, | |
| "grad_norm": 23.347203569226366, | |
| "learning_rate": 3.0804248861911986e-07, | |
| "logits/chosen": -0.5403355360031128, | |
| "logits/rejected": -0.49409806728363037, | |
| "logps/chosen": -225.88253784179688, | |
| "logps/rejected": -256.93182373046875, | |
| "loss": 0.393, | |
| "rewards/accuracies": 0.831250011920929, | |
| "rewards/chosen": -1.6747300624847412, | |
| "rewards/margins": 1.6634035110473633, | |
| "rewards/rejected": -3.3381335735321045, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.7304689166702223, | |
| "grad_norm": 26.591582696664684, | |
| "learning_rate": 3.004552352048558e-07, | |
| "logits/chosen": -0.60378497838974, | |
| "logits/rejected": -0.5446761250495911, | |
| "logps/chosen": -222.86285400390625, | |
| "logps/rejected": -254.32901000976562, | |
| "loss": 0.3562, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -1.373286247253418, | |
| "rewards/margins": 1.5952813625335693, | |
| "rewards/rejected": -2.9685676097869873, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.7372957289755515, | |
| "grad_norm": 26.301256433411677, | |
| "learning_rate": 2.928679817905918e-07, | |
| "logits/chosen": -0.575655996799469, | |
| "logits/rejected": -0.5388238430023193, | |
| "logps/chosen": -226.25411987304688, | |
| "logps/rejected": -257.7029724121094, | |
| "loss": 0.3889, | |
| "rewards/accuracies": 0.832812488079071, | |
| "rewards/chosen": -1.5678967237472534, | |
| "rewards/margins": 1.5496362447738647, | |
| "rewards/rejected": -3.1175332069396973, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.7441225412808806, | |
| "grad_norm": 29.1969544488184, | |
| "learning_rate": 2.8528072837632776e-07, | |
| "logits/chosen": -0.563581109046936, | |
| "logits/rejected": -0.4889605939388275, | |
| "logps/chosen": -215.546630859375, | |
| "logps/rejected": -251.0224609375, | |
| "loss": 0.3594, | |
| "rewards/accuracies": 0.854687511920929, | |
| "rewards/chosen": -1.5211578607559204, | |
| "rewards/margins": 1.6970359086990356, | |
| "rewards/rejected": -3.218193531036377, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.7509493535862098, | |
| "grad_norm": 28.75255873182244, | |
| "learning_rate": 2.776934749620637e-07, | |
| "logits/chosen": -0.5607287883758545, | |
| "logits/rejected": -0.5297821760177612, | |
| "logps/chosen": -213.63365173339844, | |
| "logps/rejected": -240.619384765625, | |
| "loss": 0.4057, | |
| "rewards/accuracies": 0.8203125, | |
| "rewards/chosen": -1.6024796962738037, | |
| "rewards/margins": 1.537475347518921, | |
| "rewards/rejected": -3.1399548053741455, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.7577761658915391, | |
| "grad_norm": 28.027697277996715, | |
| "learning_rate": 2.7010622154779964e-07, | |
| "logits/chosen": -0.5775099992752075, | |
| "logits/rejected": -0.5231542587280273, | |
| "logps/chosen": -224.80667114257812, | |
| "logps/rejected": -259.0721435546875, | |
| "loss": 0.4044, | |
| "rewards/accuracies": 0.8140624761581421, | |
| "rewards/chosen": -1.598193883895874, | |
| "rewards/margins": 1.5613579750061035, | |
| "rewards/rejected": -3.1595516204833984, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.7646029781968682, | |
| "grad_norm": 19.772049611357087, | |
| "learning_rate": 2.6251896813353566e-07, | |
| "logits/chosen": -0.5745671987533569, | |
| "logits/rejected": -0.5299438834190369, | |
| "logps/chosen": -225.1347198486328, | |
| "logps/rejected": -255.4309539794922, | |
| "loss": 0.3858, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -1.4894109964370728, | |
| "rewards/margins": 1.691686987876892, | |
| "rewards/rejected": -3.181097984313965, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.7714297905021974, | |
| "grad_norm": 23.20450105175028, | |
| "learning_rate": 2.549317147192716e-07, | |
| "logits/chosen": -0.5600322484970093, | |
| "logits/rejected": -0.5002347230911255, | |
| "logps/chosen": -219.86434936523438, | |
| "logps/rejected": -253.78878784179688, | |
| "loss": 0.3663, | |
| "rewards/accuracies": 0.8453125357627869, | |
| "rewards/chosen": -1.4636483192443848, | |
| "rewards/margins": 1.6817249059677124, | |
| "rewards/rejected": -3.1453733444213867, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.7782566028075265, | |
| "grad_norm": 28.72150866508454, | |
| "learning_rate": 2.473444613050076e-07, | |
| "logits/chosen": -0.6041327118873596, | |
| "logits/rejected": -0.5645285844802856, | |
| "logps/chosen": -216.68939208984375, | |
| "logps/rejected": -247.66275024414062, | |
| "loss": 0.3806, | |
| "rewards/accuracies": 0.8328125476837158, | |
| "rewards/chosen": -1.5338340997695923, | |
| "rewards/margins": 1.5916988849639893, | |
| "rewards/rejected": -3.125532865524292, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.7850834151128557, | |
| "grad_norm": 29.858461214238897, | |
| "learning_rate": 2.3975720789074356e-07, | |
| "logits/chosen": -0.6299252510070801, | |
| "logits/rejected": -0.586955189704895, | |
| "logps/chosen": -231.401611328125, | |
| "logps/rejected": -263.02197265625, | |
| "loss": 0.3998, | |
| "rewards/accuracies": 0.8328125476837158, | |
| "rewards/chosen": -1.6045633554458618, | |
| "rewards/margins": 1.6497775316238403, | |
| "rewards/rejected": -3.2543411254882812, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.791910227418185, | |
| "grad_norm": 26.24413163476253, | |
| "learning_rate": 2.321699544764795e-07, | |
| "logits/chosen": -0.5830259919166565, | |
| "logits/rejected": -0.5397896766662598, | |
| "logps/chosen": -213.19375610351562, | |
| "logps/rejected": -249.24717712402344, | |
| "loss": 0.3717, | |
| "rewards/accuracies": 0.8250000476837158, | |
| "rewards/chosen": -1.6501479148864746, | |
| "rewards/margins": 1.6961115598678589, | |
| "rewards/rejected": -3.346259593963623, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.7987370397235141, | |
| "grad_norm": 31.016581977192125, | |
| "learning_rate": 2.2458270106221546e-07, | |
| "logits/chosen": -0.5983390808105469, | |
| "logits/rejected": -0.5455670952796936, | |
| "logps/chosen": -224.10618591308594, | |
| "logps/rejected": -254.94383239746094, | |
| "loss": 0.3732, | |
| "rewards/accuracies": 0.8296875357627869, | |
| "rewards/chosen": -1.5914267301559448, | |
| "rewards/margins": 1.632917046546936, | |
| "rewards/rejected": -3.2243435382843018, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.8055638520288433, | |
| "grad_norm": 82.84012389678055, | |
| "learning_rate": 2.1699544764795143e-07, | |
| "logits/chosen": -0.6019859910011292, | |
| "logits/rejected": -0.5678104758262634, | |
| "logps/chosen": -222.878662109375, | |
| "logps/rejected": -253.78060913085938, | |
| "loss": 0.4147, | |
| "rewards/accuracies": 0.828125, | |
| "rewards/chosen": -1.6500358581542969, | |
| "rewards/margins": 1.5844390392303467, | |
| "rewards/rejected": -3.2344746589660645, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.8123906643341725, | |
| "grad_norm": 22.55342908794488, | |
| "learning_rate": 2.094081942336874e-07, | |
| "logits/chosen": -0.5819066762924194, | |
| "logits/rejected": -0.5418481826782227, | |
| "logps/chosen": -221.70608520507812, | |
| "logps/rejected": -254.09922790527344, | |
| "loss": 0.3354, | |
| "rewards/accuracies": 0.8609375357627869, | |
| "rewards/chosen": -1.6243677139282227, | |
| "rewards/margins": 1.7402938604354858, | |
| "rewards/rejected": -3.364661455154419, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.8192174766395016, | |
| "grad_norm": 21.249823285036445, | |
| "learning_rate": 2.0182094081942336e-07, | |
| "logits/chosen": -0.5841631889343262, | |
| "logits/rejected": -0.5415323972702026, | |
| "logps/chosen": -225.88800048828125, | |
| "logps/rejected": -254.038818359375, | |
| "loss": 0.3821, | |
| "rewards/accuracies": 0.839062511920929, | |
| "rewards/chosen": -1.6330121755599976, | |
| "rewards/margins": 1.732587456703186, | |
| "rewards/rejected": -3.3655996322631836, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.8260442889448308, | |
| "grad_norm": 22.957761561567523, | |
| "learning_rate": 1.9423368740515933e-07, | |
| "logits/chosen": -0.5876274704933167, | |
| "logits/rejected": -0.5527446866035461, | |
| "logps/chosen": -237.04470825195312, | |
| "logps/rejected": -263.58868408203125, | |
| "loss": 0.3658, | |
| "rewards/accuracies": 0.8531250357627869, | |
| "rewards/chosen": -1.6271567344665527, | |
| "rewards/margins": 1.6703208684921265, | |
| "rewards/rejected": -3.297477960586548, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.83287110125016, | |
| "grad_norm": 26.3109466733547, | |
| "learning_rate": 1.8664643399089527e-07, | |
| "logits/chosen": -0.5855602622032166, | |
| "logits/rejected": -0.5348464846611023, | |
| "logps/chosen": -220.74581909179688, | |
| "logps/rejected": -259.97076416015625, | |
| "loss": 0.392, | |
| "rewards/accuracies": 0.8234375715255737, | |
| "rewards/chosen": -1.666372299194336, | |
| "rewards/margins": 1.7341811656951904, | |
| "rewards/rejected": -3.4005534648895264, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.8396979135554892, | |
| "grad_norm": 32.86005475979103, | |
| "learning_rate": 1.7905918057663124e-07, | |
| "logits/chosen": -0.6146824359893799, | |
| "logits/rejected": -0.5769205093383789, | |
| "logps/chosen": -223.04859924316406, | |
| "logps/rejected": -259.2931213378906, | |
| "loss": 0.3747, | |
| "rewards/accuracies": 0.8484375476837158, | |
| "rewards/chosen": -1.6388548612594604, | |
| "rewards/margins": 1.6829884052276611, | |
| "rewards/rejected": -3.321843147277832, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.8465247258608184, | |
| "grad_norm": 27.824013672905682, | |
| "learning_rate": 1.7147192716236723e-07, | |
| "logits/chosen": -0.5848041772842407, | |
| "logits/rejected": -0.5365484356880188, | |
| "logps/chosen": -224.9688262939453, | |
| "logps/rejected": -253.75857543945312, | |
| "loss": 0.374, | |
| "rewards/accuracies": 0.8343750238418579, | |
| "rewards/chosen": -1.633022427558899, | |
| "rewards/margins": 1.6262296438217163, | |
| "rewards/rejected": -3.2592520713806152, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.8533515381661475, | |
| "grad_norm": 28.870976428951412, | |
| "learning_rate": 1.638846737481032e-07, | |
| "logits/chosen": -0.6266176700592041, | |
| "logits/rejected": -0.5750494003295898, | |
| "logps/chosen": -225.53489685058594, | |
| "logps/rejected": -251.16812133789062, | |
| "loss": 0.3643, | |
| "rewards/accuracies": 0.8421875238418579, | |
| "rewards/chosen": -1.6029326915740967, | |
| "rewards/margins": 1.637751817703247, | |
| "rewards/rejected": -3.2406845092773438, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.8601783504714767, | |
| "grad_norm": 28.44671682958466, | |
| "learning_rate": 1.5629742033383914e-07, | |
| "logits/chosen": -0.5748768448829651, | |
| "logits/rejected": -0.5039246082305908, | |
| "logps/chosen": -229.083740234375, | |
| "logps/rejected": -265.5872802734375, | |
| "loss": 0.3464, | |
| "rewards/accuracies": 0.8515625, | |
| "rewards/chosen": -1.6385741233825684, | |
| "rewards/margins": 2.006284713745117, | |
| "rewards/rejected": -3.6448588371276855, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.867005162776806, | |
| "grad_norm": 26.03554320093484, | |
| "learning_rate": 1.487101669195751e-07, | |
| "logits/chosen": -0.580173671245575, | |
| "logits/rejected": -0.5294475555419922, | |
| "logps/chosen": -225.72938537597656, | |
| "logps/rejected": -262.03546142578125, | |
| "loss": 0.3718, | |
| "rewards/accuracies": 0.8359375, | |
| "rewards/chosen": -1.6143238544464111, | |
| "rewards/margins": 1.8427155017852783, | |
| "rewards/rejected": -3.4570393562316895, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.8738319750821351, | |
| "grad_norm": 22.97729500897279, | |
| "learning_rate": 1.4112291350531107e-07, | |
| "logits/chosen": -0.6003884673118591, | |
| "logits/rejected": -0.5561665296554565, | |
| "logps/chosen": -221.987548828125, | |
| "logps/rejected": -258.51727294921875, | |
| "loss": 0.3686, | |
| "rewards/accuracies": 0.8359375, | |
| "rewards/chosen": -1.4877190589904785, | |
| "rewards/margins": 1.6500287055969238, | |
| "rewards/rejected": -3.1377477645874023, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.8806587873874643, | |
| "grad_norm": 31.37447822214391, | |
| "learning_rate": 1.3353566009104704e-07, | |
| "logits/chosen": -0.6364210844039917, | |
| "logits/rejected": -0.575194239616394, | |
| "logps/chosen": -225.1094207763672, | |
| "logps/rejected": -260.13885498046875, | |
| "loss": 0.3534, | |
| "rewards/accuracies": 0.864062488079071, | |
| "rewards/chosen": -1.6206319332122803, | |
| "rewards/margins": 1.7905977964401245, | |
| "rewards/rejected": -3.4112298488616943, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.8874855996927934, | |
| "grad_norm": 22.936789815076953, | |
| "learning_rate": 1.25948406676783e-07, | |
| "logits/chosen": -0.6323338747024536, | |
| "logits/rejected": -0.6003640294075012, | |
| "logps/chosen": -227.20034790039062, | |
| "logps/rejected": -259.46502685546875, | |
| "loss": 0.3575, | |
| "rewards/accuracies": 0.8406250476837158, | |
| "rewards/chosen": -1.6749684810638428, | |
| "rewards/margins": 1.7170754671096802, | |
| "rewards/rejected": -3.3920438289642334, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.8943124119981226, | |
| "grad_norm": 22.489511604558004, | |
| "learning_rate": 1.1836115326251896e-07, | |
| "logits/chosen": -0.6401182413101196, | |
| "logits/rejected": -0.5833394527435303, | |
| "logps/chosen": -223.30029296875, | |
| "logps/rejected": -262.72998046875, | |
| "loss": 0.3353, | |
| "rewards/accuracies": 0.8593750596046448, | |
| "rewards/chosen": -1.560599446296692, | |
| "rewards/margins": 1.906503677368164, | |
| "rewards/rejected": -3.4671034812927246, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.9011392243034518, | |
| "grad_norm": 37.43162732034228, | |
| "learning_rate": 1.1077389984825493e-07, | |
| "logits/chosen": -0.5761069059371948, | |
| "logits/rejected": -0.5430048108100891, | |
| "logps/chosen": -237.7594757080078, | |
| "logps/rejected": -275.5934753417969, | |
| "loss": 0.3514, | |
| "rewards/accuracies": 0.859375, | |
| "rewards/chosen": -1.6714935302734375, | |
| "rewards/margins": 1.8643473386764526, | |
| "rewards/rejected": -3.5358407497406006, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.907966036608781, | |
| "grad_norm": 22.988879587386872, | |
| "learning_rate": 1.0318664643399089e-07, | |
| "logits/chosen": -0.5806565284729004, | |
| "logits/rejected": -0.5450279116630554, | |
| "logps/chosen": -221.33053588867188, | |
| "logps/rejected": -256.5147705078125, | |
| "loss": 0.3729, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -1.6562050580978394, | |
| "rewards/margins": 1.747424840927124, | |
| "rewards/rejected": -3.403630018234253, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.9147928489141102, | |
| "grad_norm": 19.80848176554877, | |
| "learning_rate": 9.559939301972686e-08, | |
| "logits/chosen": -0.6481366157531738, | |
| "logits/rejected": -0.6148696541786194, | |
| "logps/chosen": -224.6954803466797, | |
| "logps/rejected": -256.4845275878906, | |
| "loss": 0.3775, | |
| "rewards/accuracies": 0.8421875238418579, | |
| "rewards/chosen": -1.7428375482559204, | |
| "rewards/margins": 1.636692762374878, | |
| "rewards/rejected": -3.379530191421509, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.9216196612194394, | |
| "grad_norm": 25.8470434123946, | |
| "learning_rate": 8.801213960546281e-08, | |
| "logits/chosen": -0.6496397852897644, | |
| "logits/rejected": -0.5912147164344788, | |
| "logps/chosen": -223.9413299560547, | |
| "logps/rejected": -259.1372375488281, | |
| "loss": 0.3461, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -1.6729114055633545, | |
| "rewards/margins": 1.7730145454406738, | |
| "rewards/rejected": -3.445925712585449, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.9284464735247685, | |
| "grad_norm": 33.2201336722171, | |
| "learning_rate": 8.042488619119878e-08, | |
| "logits/chosen": -0.645717203617096, | |
| "logits/rejected": -0.6112032532691956, | |
| "logps/chosen": -225.99624633789062, | |
| "logps/rejected": -257.4811706542969, | |
| "loss": 0.4065, | |
| "rewards/accuracies": 0.8218750357627869, | |
| "rewards/chosen": -1.761589527130127, | |
| "rewards/margins": 1.6280558109283447, | |
| "rewards/rejected": -3.389645576477051, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.9352732858300977, | |
| "grad_norm": 27.005710517490183, | |
| "learning_rate": 7.283763277693475e-08, | |
| "logits/chosen": -0.573918342590332, | |
| "logits/rejected": -0.5335432291030884, | |
| "logps/chosen": -225.52552795410156, | |
| "logps/rejected": -255.49449157714844, | |
| "loss": 0.3465, | |
| "rewards/accuracies": 0.8531249761581421, | |
| "rewards/chosen": -1.7273519039154053, | |
| "rewards/margins": 1.7527152299880981, | |
| "rewards/rejected": -3.480067253112793, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.9421000981354269, | |
| "grad_norm": 32.140399259495645, | |
| "learning_rate": 6.525037936267071e-08, | |
| "logits/chosen": -0.6214314103126526, | |
| "logits/rejected": -0.570462167263031, | |
| "logps/chosen": -224.70672607421875, | |
| "logps/rejected": -264.4761962890625, | |
| "loss": 0.3218, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -1.6792542934417725, | |
| "rewards/margins": 1.915861964225769, | |
| "rewards/rejected": -3.595116138458252, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.948926910440756, | |
| "grad_norm": 31.250154294424732, | |
| "learning_rate": 5.766312594840667e-08, | |
| "logits/chosen": -0.6339004635810852, | |
| "logits/rejected": -0.5892723798751831, | |
| "logps/chosen": -220.26611328125, | |
| "logps/rejected": -252.96212768554688, | |
| "loss": 0.3864, | |
| "rewards/accuracies": 0.8312499523162842, | |
| "rewards/chosen": -1.6645467281341553, | |
| "rewards/margins": 1.5790960788726807, | |
| "rewards/rejected": -3.243642807006836, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.9557537227460853, | |
| "grad_norm": 30.068762957187783, | |
| "learning_rate": 5.007587253414264e-08, | |
| "logits/chosen": -0.678811252117157, | |
| "logits/rejected": -0.6359538435935974, | |
| "logps/chosen": -224.49069213867188, | |
| "logps/rejected": -258.3272705078125, | |
| "loss": 0.3447, | |
| "rewards/accuracies": 0.8531250357627869, | |
| "rewards/chosen": -1.575748085975647, | |
| "rewards/margins": 1.9220972061157227, | |
| "rewards/rejected": -3.49784517288208, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.9625805350514144, | |
| "grad_norm": 22.16371068962549, | |
| "learning_rate": 4.2488619119878606e-08, | |
| "logits/chosen": -0.6366287469863892, | |
| "logits/rejected": -0.5852836966514587, | |
| "logps/chosen": -227.71780395507812, | |
| "logps/rejected": -267.0358581542969, | |
| "loss": 0.3718, | |
| "rewards/accuracies": 0.831250011920929, | |
| "rewards/chosen": -1.6289258003234863, | |
| "rewards/margins": 1.7643526792526245, | |
| "rewards/rejected": -3.3932785987854004, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.9694073473567436, | |
| "grad_norm": 31.032456565988113, | |
| "learning_rate": 3.4901365705614566e-08, | |
| "logits/chosen": -0.6306103467941284, | |
| "logits/rejected": -0.5921708345413208, | |
| "logps/chosen": -221.66065979003906, | |
| "logps/rejected": -254.41958618164062, | |
| "loss": 0.3678, | |
| "rewards/accuracies": 0.823437511920929, | |
| "rewards/chosen": -1.5656054019927979, | |
| "rewards/margins": 1.682039499282837, | |
| "rewards/rejected": -3.2476449012756348, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.9762341596620728, | |
| "grad_norm": 26.873435878225383, | |
| "learning_rate": 2.731411229135053e-08, | |
| "logits/chosen": -0.6624563336372375, | |
| "logits/rejected": -0.6294071078300476, | |
| "logps/chosen": -224.36407470703125, | |
| "logps/rejected": -263.2255859375, | |
| "loss": 0.3681, | |
| "rewards/accuracies": 0.8484375476837158, | |
| "rewards/chosen": -1.7730777263641357, | |
| "rewards/margins": 1.7527307271957397, | |
| "rewards/rejected": -3.525808334350586, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.9830609719674019, | |
| "grad_norm": 28.36352572432148, | |
| "learning_rate": 1.9726858877086493e-08, | |
| "logits/chosen": -0.6402366161346436, | |
| "logits/rejected": -0.5960521697998047, | |
| "logps/chosen": -225.24977111816406, | |
| "logps/rejected": -257.8275451660156, | |
| "loss": 0.3734, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -1.7362611293792725, | |
| "rewards/margins": 1.7620372772216797, | |
| "rewards/rejected": -3.498298168182373, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.9898877842727312, | |
| "grad_norm": 30.66526971215358, | |
| "learning_rate": 1.2139605462822458e-08, | |
| "logits/chosen": -0.6005350351333618, | |
| "logits/rejected": -0.5661831498146057, | |
| "logps/chosen": -227.962158203125, | |
| "logps/rejected": -261.6782531738281, | |
| "loss": 0.3924, | |
| "rewards/accuracies": 0.8328125476837158, | |
| "rewards/chosen": -1.720937728881836, | |
| "rewards/margins": 1.5882391929626465, | |
| "rewards/rejected": -3.3091769218444824, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.9967145965780604, | |
| "grad_norm": 36.64240487573334, | |
| "learning_rate": 4.552352048558422e-09, | |
| "logits/chosen": -0.6393886804580688, | |
| "logits/rejected": -0.6115251183509827, | |
| "logps/chosen": -229.70652770996094, | |
| "logps/rejected": -268.06982421875, | |
| "loss": 0.3379, | |
| "rewards/accuracies": 0.8734375238418579, | |
| "rewards/chosen": -1.651149034500122, | |
| "rewards/margins": 1.8959904909133911, | |
| "rewards/rejected": -3.5471396446228027, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1465, | |
| "total_flos": 161167907028992.0, | |
| "train_loss": 0.47723283336431094, | |
| "train_runtime": 14257.9418, | |
| "train_samples_per_second": 6.575, | |
| "train_steps_per_second": 0.103 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1465, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 161167907028992.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |