| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 100, | |
| "global_step": 288, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.010416666666666666, | |
| "grad_norm": 155.97537788651405, | |
| "learning_rate": 7.395155355091305e-09, | |
| "logits/chosen": -2.590585231781006, | |
| "logits/rejected": -2.5664222240448, | |
| "logps/chosen": -80.29847717285156, | |
| "logps/rejected": -53.10200881958008, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.10416666666666667, | |
| "grad_norm": 140.96156802230377, | |
| "learning_rate": 7.395155355091307e-08, | |
| "logits/chosen": -2.5568971633911133, | |
| "logits/rejected": -2.538863182067871, | |
| "logps/chosen": -87.95579528808594, | |
| "logps/rejected": -81.03325653076172, | |
| "loss": 0.6946, | |
| "rewards/accuracies": 0.1875, | |
| "rewards/chosen": -0.009626655839383602, | |
| "rewards/margins": -0.01086943969130516, | |
| "rewards/rejected": 0.001242783502675593, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.20833333333333334, | |
| "grad_norm": 125.98597666420928, | |
| "learning_rate": 1.4790310710182613e-07, | |
| "logits/chosen": -2.615262269973755, | |
| "logits/rejected": -2.566425323486328, | |
| "logps/chosen": -103.3228759765625, | |
| "logps/rejected": -89.88175964355469, | |
| "loss": 0.6912, | |
| "rewards/accuracies": 0.3062500059604645, | |
| "rewards/chosen": 0.006651306990534067, | |
| "rewards/margins": 0.006033898331224918, | |
| "rewards/rejected": 0.0006174088339321315, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.3125, | |
| "grad_norm": 163.17755441844716, | |
| "learning_rate": 2.1401757670576506e-07, | |
| "logits/chosen": -2.5203137397766113, | |
| "logits/rejected": -2.5346357822418213, | |
| "logps/chosen": -67.3207778930664, | |
| "logps/rejected": -74.93171691894531, | |
| "loss": 0.684, | |
| "rewards/accuracies": 0.30000001192092896, | |
| "rewards/chosen": 0.027593865990638733, | |
| "rewards/margins": 0.028976192697882652, | |
| "rewards/rejected": -0.0013823278713971376, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.4166666666666667, | |
| "grad_norm": 101.98216597077327, | |
| "learning_rate": 2.0959829078693693e-07, | |
| "logits/chosen": -2.5797293186187744, | |
| "logits/rejected": -2.567654848098755, | |
| "logps/chosen": -71.80469512939453, | |
| "logps/rejected": -70.60755920410156, | |
| "loss": 0.6752, | |
| "rewards/accuracies": 0.26249998807907104, | |
| "rewards/chosen": 0.08097527176141739, | |
| "rewards/margins": 0.05593211576342583, | |
| "rewards/rejected": 0.02504315972328186, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.5208333333333334, | |
| "grad_norm": 116.53954973852585, | |
| "learning_rate": 2.0517900486810885e-07, | |
| "logits/chosen": -2.491152048110962, | |
| "logits/rejected": -2.503134250640869, | |
| "logps/chosen": -50.373291015625, | |
| "logps/rejected": -58.62458419799805, | |
| "loss": 0.6651, | |
| "rewards/accuracies": 0.22499999403953552, | |
| "rewards/chosen": 0.1250298023223877, | |
| "rewards/margins": 0.05763009935617447, | |
| "rewards/rejected": 0.06739969551563263, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 118.26991196128117, | |
| "learning_rate": 2.0075971894928074e-07, | |
| "logits/chosen": -2.5747580528259277, | |
| "logits/rejected": -2.556131601333618, | |
| "logps/chosen": -77.9587173461914, | |
| "logps/rejected": -78.81073760986328, | |
| "loss": 0.6597, | |
| "rewards/accuracies": 0.32499998807907104, | |
| "rewards/chosen": 0.26304519176483154, | |
| "rewards/margins": 0.13826420903205872, | |
| "rewards/rejected": 0.12478096783161163, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.7291666666666666, | |
| "grad_norm": 109.77312584060142, | |
| "learning_rate": 1.9634043303045264e-07, | |
| "logits/chosen": -2.552567720413208, | |
| "logits/rejected": -2.5432896614074707, | |
| "logps/chosen": -94.4707260131836, | |
| "logps/rejected": -84.51776885986328, | |
| "loss": 0.6456, | |
| "rewards/accuracies": 0.36250001192092896, | |
| "rewards/chosen": 0.4439457952976227, | |
| "rewards/margins": 0.22308318316936493, | |
| "rewards/rejected": 0.22086259722709656, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 178.1381395653738, | |
| "learning_rate": 1.9192114711162454e-07, | |
| "logits/chosen": -2.552009105682373, | |
| "logits/rejected": -2.497727632522583, | |
| "logps/chosen": -83.81471252441406, | |
| "logps/rejected": -78.81379699707031, | |
| "loss": 0.6258, | |
| "rewards/accuracies": 0.39375001192092896, | |
| "rewards/chosen": 0.4336014688014984, | |
| "rewards/margins": 0.29889458417892456, | |
| "rewards/rejected": 0.13470691442489624, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.9375, | |
| "grad_norm": 158.54231900826267, | |
| "learning_rate": 1.8750186119279643e-07, | |
| "logits/chosen": -2.4831385612487793, | |
| "logits/rejected": -2.477928400039673, | |
| "logps/chosen": -52.954994201660156, | |
| "logps/rejected": -63.29545974731445, | |
| "loss": 0.645, | |
| "rewards/accuracies": 0.2562499940395355, | |
| "rewards/chosen": 0.2427663803100586, | |
| "rewards/margins": 0.20185653865337372, | |
| "rewards/rejected": 0.04090985283255577, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.0416666666666667, | |
| "grad_norm": 59.1130579303193, | |
| "learning_rate": 1.8308257527396833e-07, | |
| "logits/chosen": -2.492121458053589, | |
| "logits/rejected": -2.470026731491089, | |
| "logps/chosen": -71.71488952636719, | |
| "logps/rejected": -72.61717224121094, | |
| "loss": 0.5502, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.42804664373397827, | |
| "rewards/margins": 0.5683026909828186, | |
| "rewards/rejected": -0.14025607705116272, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.0416666666666667, | |
| "eval_logits/chosen": -2.5289955139160156, | |
| "eval_logits/rejected": -2.5124404430389404, | |
| "eval_logps/chosen": -73.31734466552734, | |
| "eval_logps/rejected": -81.00989532470703, | |
| "eval_loss": 0.6403160691261292, | |
| "eval_rewards/accuracies": 0.3253968358039856, | |
| "eval_rewards/chosen": 0.2780722677707672, | |
| "eval_rewards/margins": 0.25001367926597595, | |
| "eval_rewards/rejected": 0.028058595955371857, | |
| "eval_runtime": 142.317, | |
| "eval_samples_per_second": 14.053, | |
| "eval_steps_per_second": 0.443, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.1458333333333333, | |
| "grad_norm": 47.71650195966285, | |
| "learning_rate": 1.7866328935514022e-07, | |
| "logits/chosen": -2.499016284942627, | |
| "logits/rejected": -2.532212495803833, | |
| "logps/chosen": -62.74967575073242, | |
| "logps/rejected": -83.47831726074219, | |
| "loss": 0.4501, | |
| "rewards/accuracies": 0.45625001192092896, | |
| "rewards/chosen": 0.6203856468200684, | |
| "rewards/margins": 1.092772126197815, | |
| "rewards/rejected": -0.4723864197731018, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 34.72113709789245, | |
| "learning_rate": 1.7424400343631214e-07, | |
| "logits/chosen": -2.566894054412842, | |
| "logits/rejected": -2.53535795211792, | |
| "logps/chosen": -105.34260559082031, | |
| "logps/rejected": -97.03143310546875, | |
| "loss": 0.4582, | |
| "rewards/accuracies": 0.5562499761581421, | |
| "rewards/chosen": 0.630843997001648, | |
| "rewards/margins": 1.5061585903167725, | |
| "rewards/rejected": -0.8753145337104797, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.3541666666666667, | |
| "grad_norm": 44.58377424440547, | |
| "learning_rate": 1.69824717517484e-07, | |
| "logits/chosen": -2.4931442737579346, | |
| "logits/rejected": -2.4973056316375732, | |
| "logps/chosen": -85.21261596679688, | |
| "logps/rejected": -93.46207427978516, | |
| "loss": 0.4423, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": 0.3205886781215668, | |
| "rewards/margins": 1.2978527545928955, | |
| "rewards/rejected": -0.9772640466690063, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.4583333333333333, | |
| "grad_norm": 73.130106273938, | |
| "learning_rate": 1.6540543159865593e-07, | |
| "logits/chosen": -2.4969403743743896, | |
| "logits/rejected": -2.513062000274658, | |
| "logps/chosen": -47.37117385864258, | |
| "logps/rejected": -67.8313980102539, | |
| "loss": 0.4504, | |
| "rewards/accuracies": 0.4124999940395355, | |
| "rewards/chosen": 0.4540034234523773, | |
| "rewards/margins": 1.0959008932113647, | |
| "rewards/rejected": -0.641897439956665, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.5625, | |
| "grad_norm": 40.370233820639314, | |
| "learning_rate": 1.609861456798278e-07, | |
| "logits/chosen": -2.5404396057128906, | |
| "logits/rejected": -2.5168280601501465, | |
| "logps/chosen": -69.16145324707031, | |
| "logps/rejected": -76.45948791503906, | |
| "loss": 0.4389, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.7074292302131653, | |
| "rewards/margins": 1.3904778957366943, | |
| "rewards/rejected": -0.6830486059188843, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 35.06923806180356, | |
| "learning_rate": 1.5656685976099972e-07, | |
| "logits/chosen": -2.502528190612793, | |
| "logits/rejected": -2.5127906799316406, | |
| "logps/chosen": -77.15453338623047, | |
| "logps/rejected": -91.57726287841797, | |
| "loss": 0.4299, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": 0.8931992650032043, | |
| "rewards/margins": 1.7554128170013428, | |
| "rewards/rejected": -0.8622136116027832, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.7708333333333335, | |
| "grad_norm": 55.605591268891615, | |
| "learning_rate": 1.521475738421716e-07, | |
| "logits/chosen": -2.4732327461242676, | |
| "logits/rejected": -2.4644830226898193, | |
| "logps/chosen": -71.61512756347656, | |
| "logps/rejected": -86.00133514404297, | |
| "loss": 0.4399, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": 0.8539541363716125, | |
| "rewards/margins": 1.5681732892990112, | |
| "rewards/rejected": -0.7142192125320435, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.875, | |
| "grad_norm": 53.2168247664102, | |
| "learning_rate": 1.4772828792334352e-07, | |
| "logits/chosen": -2.4987406730651855, | |
| "logits/rejected": -2.4796464443206787, | |
| "logps/chosen": -68.8732681274414, | |
| "logps/rejected": -77.2318344116211, | |
| "loss": 0.4149, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.7064772248268127, | |
| "rewards/margins": 1.5638738870620728, | |
| "rewards/rejected": -0.8573967814445496, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.9791666666666665, | |
| "grad_norm": 45.792720550959615, | |
| "learning_rate": 1.433090020045154e-07, | |
| "logits/chosen": -2.5336966514587402, | |
| "logits/rejected": -2.5320465564727783, | |
| "logps/chosen": -85.67411041259766, | |
| "logps/rejected": -96.0043716430664, | |
| "loss": 0.4271, | |
| "rewards/accuracies": 0.518750011920929, | |
| "rewards/chosen": 0.6603367328643799, | |
| "rewards/margins": 1.9354482889175415, | |
| "rewards/rejected": -1.275111436843872, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.0833333333333335, | |
| "grad_norm": 5.83003196296594, | |
| "learning_rate": 1.388897160856873e-07, | |
| "logits/chosen": -2.482923746109009, | |
| "logits/rejected": -2.505110263824463, | |
| "logps/chosen": -86.26466369628906, | |
| "logps/rejected": -109.96630859375, | |
| "loss": 0.377, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.6489728093147278, | |
| "rewards/margins": 2.6458792686462402, | |
| "rewards/rejected": -1.9969062805175781, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.0833333333333335, | |
| "eval_logits/chosen": -2.5251481533050537, | |
| "eval_logits/rejected": -2.5078439712524414, | |
| "eval_logps/chosen": -75.29920959472656, | |
| "eval_logps/rejected": -83.76541137695312, | |
| "eval_loss": 0.6467146873474121, | |
| "eval_rewards/accuracies": 0.3313491940498352, | |
| "eval_rewards/chosen": -0.19139644503593445, | |
| "eval_rewards/margins": 0.43327686190605164, | |
| "eval_rewards/rejected": -0.6246733069419861, | |
| "eval_runtime": 144.2917, | |
| "eval_samples_per_second": 13.861, | |
| "eval_steps_per_second": 0.437, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.1875, | |
| "grad_norm": 8.844561075287848, | |
| "learning_rate": 1.344704301668592e-07, | |
| "logits/chosen": -2.526526927947998, | |
| "logits/rejected": -2.5184459686279297, | |
| "logps/chosen": -60.16546630859375, | |
| "logps/rejected": -74.23641967773438, | |
| "loss": 0.3939, | |
| "rewards/accuracies": 0.42500001192092896, | |
| "rewards/chosen": 0.8121669888496399, | |
| "rewards/margins": 2.1258318424224854, | |
| "rewards/rejected": -1.3136647939682007, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.2916666666666665, | |
| "grad_norm": 5.363217741474801, | |
| "learning_rate": 1.300511442480311e-07, | |
| "logits/chosen": -2.5266761779785156, | |
| "logits/rejected": -2.477085590362549, | |
| "logps/chosen": -98.9388198852539, | |
| "logps/rejected": -102.80915832519531, | |
| "loss": 0.3463, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.736892819404602, | |
| "rewards/margins": 3.152064800262451, | |
| "rewards/rejected": -2.4151718616485596, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.3958333333333335, | |
| "grad_norm": 3.636762082821054, | |
| "learning_rate": 1.25631858329203e-07, | |
| "logits/chosen": -2.5067200660705566, | |
| "logits/rejected": -2.4729416370391846, | |
| "logps/chosen": -57.91691970825195, | |
| "logps/rejected": -63.94114303588867, | |
| "loss": 0.3891, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": 0.45365357398986816, | |
| "rewards/margins": 2.347585678100586, | |
| "rewards/rejected": -1.8939321041107178, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 10.817240165720927, | |
| "learning_rate": 1.212125724103749e-07, | |
| "logits/chosen": -2.4893410205841064, | |
| "logits/rejected": -2.438758373260498, | |
| "logps/chosen": -82.23040771484375, | |
| "logps/rejected": -85.9568862915039, | |
| "loss": 0.3836, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 1.3419866561889648, | |
| "rewards/margins": 2.889305591583252, | |
| "rewards/rejected": -1.547318696975708, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.6041666666666665, | |
| "grad_norm": 13.517868605826433, | |
| "learning_rate": 1.167932864915468e-07, | |
| "logits/chosen": -2.4413411617279053, | |
| "logits/rejected": -2.429288387298584, | |
| "logps/chosen": -70.37754821777344, | |
| "logps/rejected": -79.62659454345703, | |
| "loss": 0.3565, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.542807400226593, | |
| "rewards/margins": 2.719435930252075, | |
| "rewards/rejected": -2.176628589630127, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.7083333333333335, | |
| "grad_norm": 4.587653594042731, | |
| "learning_rate": 1.1237400057271868e-07, | |
| "logits/chosen": -2.440176010131836, | |
| "logits/rejected": -2.458115577697754, | |
| "logps/chosen": -57.05979537963867, | |
| "logps/rejected": -83.13993835449219, | |
| "loss": 0.3849, | |
| "rewards/accuracies": 0.44999998807907104, | |
| "rewards/chosen": 0.25371766090393066, | |
| "rewards/margins": 2.664337635040283, | |
| "rewards/rejected": -2.4106199741363525, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.8125, | |
| "grad_norm": 17.548402893009843, | |
| "learning_rate": 1.0795471465389056e-07, | |
| "logits/chosen": -2.4485597610473633, | |
| "logits/rejected": -2.4837899208068848, | |
| "logps/chosen": -74.96304321289062, | |
| "logps/rejected": -89.0799331665039, | |
| "loss": 0.3625, | |
| "rewards/accuracies": 0.4937500059604645, | |
| "rewards/chosen": 0.405341237783432, | |
| "rewards/margins": 2.9616785049438477, | |
| "rewards/rejected": -2.556337356567383, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.9166666666666665, | |
| "grad_norm": 9.09052295820177, | |
| "learning_rate": 1.0353542873506248e-07, | |
| "logits/chosen": -2.5102744102478027, | |
| "logits/rejected": -2.474376678466797, | |
| "logps/chosen": -81.9388656616211, | |
| "logps/rejected": -92.2305679321289, | |
| "loss": 0.3734, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.8098124265670776, | |
| "rewards/margins": 3.1447136402130127, | |
| "rewards/rejected": -2.3349013328552246, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 288, | |
| "total_flos": 0.0, | |
| "train_loss": 0.0, | |
| "train_runtime": 0.0237, | |
| "train_samples_per_second": 774670.33, | |
| "train_steps_per_second": 12165.606 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 288, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "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 | |
| } | |