{ "best_metric": 0.5295031070709229, "best_model_checkpoint": "./qwen2.5-0.5b/qwen2.5-0.5b-expo-L2EXPO-ES-10/checkpoint-200", "epoch": 1.7005196032120926, "eval_steps": 50, "global_step": 600, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "dpo_loss": 0.6931471824645996, "epoch": 0.002834199338686821, "grad_norm": 3688.5083619666543, "learning_rate": 2.840909090909091e-08, "logits": -1.359458565711975, "logps": -84.69721221923828, "loss": 0.3913, "objective": 0.3618059456348419, "ranking_idealized": 0.5833333134651184, "ranking_idealized_expo": 0.5833333134651184, "ranking_simple": 0.5833333134651184, "regularize": 0.3618059456348419, "step": 1, "wo_beta": 14.830931663513184 }, { "dpo_loss": 2.238539934158325, "epoch": 0.14170996693434104, "grad_norm": 3276.735561173721, "learning_rate": 1.4204545454545458e-06, "logits": -1.448372483253479, "logps": -84.56832885742188, "loss": 4.4544, "objective": 4.151083946228027, "ranking_idealized": 0.5225340127944946, "ranking_idealized_expo": 0.5216836929321289, "ranking_simple": 0.5199829936027527, "regularize": 4.151083946228027, "step": 50, "wo_beta": 15.634044647216797 }, { "epoch": 0.14170996693434104, "eval_dpo_loss": 3.076866626739502, "eval_logits": -1.4000310897827148, "eval_logps": -90.06948852539062, "eval_loss": 6.113166332244873, "eval_objective": 6.235374927520752, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5243270993232727, "eval_regularize": 6.235374927520752, "eval_runtime": 307.8342, "eval_samples_per_second": 18.809, "eval_steps_per_second": 1.569, "eval_wo_beta": 16.07046890258789, "step": 50 }, { "dpo_loss": 8.729989051818848, "epoch": 0.2834199338686821, "grad_norm": 2869.8187087544343, "learning_rate": 2.8409090909090916e-06, "logits": -1.3595805168151855, "logps": -82.20881652832031, "loss": 17.3779, "objective": 18.13115882873535, "ranking_idealized": 0.5141666531562805, "ranking_idealized_expo": 0.5137500166893005, "ranking_simple": 0.5199999809265137, "regularize": 18.13115882873535, "step": 100, "wo_beta": 15.266292572021484 }, { "epoch": 0.2834199338686821, "eval_dpo_loss": 7.9374237060546875, "eval_logits": -1.3237512111663818, "eval_logps": -85.5257339477539, "eval_loss": 16.1760196685791, "eval_objective": 16.003671646118164, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5258799195289612, "eval_regularize": 16.003671646118164, "eval_runtime": 307.1481, "eval_samples_per_second": 18.851, "eval_steps_per_second": 1.573, "eval_wo_beta": 15.778022766113281, "step": 100 }, { "dpo_loss": 14.852143287658691, "epoch": 0.42512990080302315, "grad_norm": 2245.895525757807, "learning_rate": 4.2613636363636365e-06, "logits": -1.168060302734375, "logps": -79.34931945800781, "loss": 28.1478, "objective": 28.49669647216797, "ranking_idealized": 0.5287500023841858, "ranking_idealized_expo": 0.527916669845581, "ranking_simple": 0.5287500023841858, "regularize": 28.49669647216797, "step": 150, "wo_beta": 15.188308715820312 }, { "epoch": 0.42512990080302315, "eval_dpo_loss": 14.723884582519531, "eval_logits": -1.0823547840118408, "eval_logps": -82.48079681396484, "eval_loss": 28.730850219726562, "eval_objective": 28.130752563476562, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.522774338722229, "eval_regularize": 28.130752563476562, "eval_runtime": 307.3658, "eval_samples_per_second": 18.837, "eval_steps_per_second": 1.571, "eval_wo_beta": 15.409582138061523, "step": 150 }, { "dpo_loss": 17.784215927124023, "epoch": 0.5668398677373642, "grad_norm": 1921.2331717681436, "learning_rate": 4.997168347957521e-06, "logits": -0.9321090579032898, "logps": -74.18836212158203, "loss": 35.2522, "objective": 35.61735534667969, "ranking_idealized": 0.51583331823349, "ranking_idealized_expo": 0.51541668176651, "ranking_simple": 0.5066666603088379, "regularize": 35.61735534667969, "step": 200, "wo_beta": 15.46297550201416 }, { "epoch": 0.5668398677373642, "eval_dpo_loss": 18.911605834960938, "eval_logits": -0.8236067295074463, "eval_logps": -75.27147674560547, "eval_loss": 38.52634048461914, "eval_objective": 37.43656921386719, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5295031070709229, "eval_regularize": 37.43656921386719, "eval_runtime": 307.428, "eval_samples_per_second": 18.834, "eval_steps_per_second": 1.571, "eval_wo_beta": 14.68159294128418, "step": 200 }, { "dpo_loss": 19.207956314086914, "epoch": 0.7085498346717053, "grad_norm": 1961.1411038553429, "learning_rate": 4.973122855144066e-06, "logits": -0.7678465843200684, "logps": -71.71759033203125, "loss": 37.8556, "objective": 38.2454833984375, "ranking_idealized": 0.5166666507720947, "ranking_idealized_expo": 0.5162500143051147, "ranking_simple": 0.5183333158493042, "regularize": 38.2454833984375, "step": 250, "wo_beta": 15.593232154846191 }, { "epoch": 0.7085498346717053, "eval_dpo_loss": 22.749528884887695, "eval_logits": -0.602376401424408, "eval_logps": -76.27977752685547, "eval_loss": 44.81644821166992, "eval_objective": 44.5794677734375, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5222567319869995, "eval_regularize": 44.5794677734375, "eval_runtime": 309.9732, "eval_samples_per_second": 18.679, "eval_steps_per_second": 1.558, "eval_wo_beta": 14.318175315856934, "step": 250 }, { "dpo_loss": 18.925874710083008, "epoch": 0.8502598016060463, "grad_norm": 1894.7438599816405, "learning_rate": 4.924776641419513e-06, "logits": -0.6545333862304688, "logps": -74.65443420410156, "loss": 36.0351, "objective": 36.791263580322266, "ranking_idealized": 0.4962500035762787, "ranking_idealized_expo": 0.4950000047683716, "ranking_simple": 0.5074999928474426, "regularize": 36.791263580322266, "step": 300, "wo_beta": 15.123795509338379 }, { "epoch": 0.8502598016060463, "eval_dpo_loss": 22.145748138427734, "eval_logits": -0.7056564688682556, "eval_logps": -79.1832504272461, "eval_loss": 44.38310623168945, "eval_objective": 43.87770462036133, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5253623127937317, "eval_regularize": 43.87770462036133, "eval_runtime": 307.3056, "eval_samples_per_second": 18.841, "eval_steps_per_second": 1.572, "eval_wo_beta": 14.267467498779297, "step": 300 }, { "dpo_loss": 16.664411544799805, "epoch": 0.9919697685403873, "grad_norm": 1652.397965840821, "learning_rate": 4.8526047530778175e-06, "logits": -0.6539748907089233, "logps": -77.53305053710938, "loss": 32.9882, "objective": 31.92953109741211, "ranking_idealized": 0.5262500047683716, "ranking_idealized_expo": 0.5254166722297668, "ranking_simple": 0.5174999833106995, "regularize": 31.92953109741211, "step": 350, "wo_beta": 15.103469848632812 }, { "epoch": 0.9919697685403873, "eval_dpo_loss": 23.009784698486328, "eval_logits": -0.6344824433326721, "eval_logps": -80.31657409667969, "eval_loss": 46.694580078125, "eval_objective": 45.59531021118164, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5248447060585022, "eval_regularize": 45.59531021118164, "eval_runtime": 307.1817, "eval_samples_per_second": 18.849, "eval_steps_per_second": 1.572, "eval_wo_beta": 14.169025421142578, "step": 350 }, { "dpo_loss": 15.38097858428955, "epoch": 1.1336797354747283, "grad_norm": 1767.7891148738722, "learning_rate": 4.757316345716554e-06, "logits": -0.5037474036216736, "logps": -77.35700225830078, "loss": 30.7247, "objective": 30.997066497802734, "ranking_idealized": 0.5333333611488342, "ranking_idealized_expo": 0.5320833325386047, "ranking_simple": 0.5258333086967468, "regularize": 30.997066497802734, "step": 400, "wo_beta": 15.239886283874512 }, { "epoch": 1.1336797354747283, "eval_dpo_loss": 24.61825180053711, "eval_logits": -0.4809892475605011, "eval_logps": -82.41106414794922, "eval_loss": 48.38045120239258, "eval_objective": 48.06561279296875, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5165631175041199, "eval_regularize": 48.06561279296875, "eval_runtime": 307.5308, "eval_samples_per_second": 18.827, "eval_steps_per_second": 1.571, "eval_wo_beta": 14.105908393859863, "step": 400 }, { "dpo_loss": 15.528545379638672, "epoch": 1.2753897024090695, "grad_norm": 1599.4345499147116, "learning_rate": 4.639847716126855e-06, "logits": -0.4860953390598297, "logps": -78.5374755859375, "loss": 29.6491, "objective": 30.698896408081055, "ranking_idealized": 0.5195833444595337, "ranking_idealized_expo": 0.5191666483879089, "ranking_simple": 0.5208333134651184, "regularize": 30.698896408081055, "step": 450, "wo_beta": 15.879746437072754 }, { "epoch": 1.2753897024090695, "eval_dpo_loss": 24.949462890625, "eval_logits": -0.5861236453056335, "eval_logps": -81.5285415649414, "eval_loss": 48.523677825927734, "eval_objective": 48.84111022949219, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5243270993232727, "eval_regularize": 48.84111022949219, "eval_runtime": 311.2891, "eval_samples_per_second": 18.6, "eval_steps_per_second": 1.552, "eval_wo_beta": 14.479339599609375, "step": 450 }, { "dpo_loss": 14.001568794250488, "epoch": 1.4170996693434104, "grad_norm": 1563.7145297724219, "learning_rate": 4.501353102310901e-06, "logits": -0.5200133323669434, "logps": -76.25775909423828, "loss": 28.3933, "objective": 28.495651245117188, "ranking_idealized": 0.49791666865348816, "ranking_idealized_expo": 0.4970833361148834, "ranking_simple": 0.5062500238418579, "regularize": 28.495651245117188, "step": 500, "wo_beta": 15.288237571716309 }, { "epoch": 1.4170996693434104, "eval_dpo_loss": 24.81561851501465, "eval_logits": -0.5585454106330872, "eval_logps": -79.88430786132812, "eval_loss": 47.81501388549805, "eval_objective": 47.92095947265625, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5212215185165405, "eval_regularize": 47.92095947265625, "eval_runtime": 307.162, "eval_samples_per_second": 18.85, "eval_steps_per_second": 1.572, "eval_wo_beta": 14.345782279968262, "step": 500 }, { "dpo_loss": 13.645363807678223, "epoch": 1.5588096362777515, "grad_norm": 1564.1500708905278, "learning_rate": 4.34319334202531e-06, "logits": -0.5150942802429199, "logps": -77.01350402832031, "loss": 26.3026, "objective": 26.619535446166992, "ranking_idealized": 0.5112500190734863, "ranking_idealized_expo": 0.5104166865348816, "ranking_simple": 0.5120833516120911, "regularize": 26.619535446166992, "step": 550, "wo_beta": 14.997577667236328 }, { "epoch": 1.5588096362777515, "eval_dpo_loss": 24.458288192749023, "eval_logits": -0.5593612194061279, "eval_logps": -79.55674743652344, "eval_loss": 48.00811767578125, "eval_objective": 48.22153854370117, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.522774338722229, "eval_regularize": 48.22153854370117, "eval_runtime": 307.4518, "eval_samples_per_second": 18.832, "eval_steps_per_second": 1.571, "eval_wo_beta": 14.15873908996582, "step": 550 }, { "dpo_loss": 13.081768989562988, "epoch": 1.7005196032120926, "grad_norm": 1567.2994510773917, "learning_rate": 4.16692250129073e-06, "logits": -0.4002828598022461, "logps": -76.4526138305664, "loss": 25.1162, "objective": 25.69132423400879, "ranking_idealized": 0.51541668176651, "ranking_idealized_expo": 0.5149999856948853, "ranking_simple": 0.5137500166893005, "regularize": 25.69132423400879, "step": 600, "wo_beta": 14.990047454833984 }, { "epoch": 1.7005196032120926, "eval_dpo_loss": 25.221858978271484, "eval_logits": -0.4875362813472748, "eval_logps": -79.4244613647461, "eval_loss": 49.42710494995117, "eval_objective": 49.74281311035156, "eval_ranking_idealized": 0.5212215185165405, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5258799195289612, "eval_regularize": 49.74281311035156, "eval_runtime": 311.292, "eval_samples_per_second": 18.6, "eval_steps_per_second": 1.552, "eval_wo_beta": 14.192330360412598, "step": 600 }, { "epoch": 1.7005196032120926, "step": 600, "total_flos": 0.0, "train_loss": 11.682157999674478, "train_runtime": 6847.3936, "train_samples_per_second": 37.096, "train_steps_per_second": 0.257 } ], "logging_steps": 50, "max_steps": 1760, "num_input_tokens_seen": 0, "num_train_epochs": 5, "save_steps": 50, "stateful_callbacks": { "EarlyStoppingCallback": { "args": { "early_stopping_patience": 5, "early_stopping_threshold": 0.0 }, "attributes": { "early_stopping_patience_counter": 0 } }, "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": 4, "trial_name": null, "trial_params": null }