{ "best_metric": 13.93560791015625, "best_model_checkpoint": "./qwen2.5-0.5b/qwen2.5-0.5b-expo-L1EXPO-ES-1/checkpoint-450", "epoch": 1.9839395370807746, "eval_steps": 50, "global_step": 700, "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": 368.8500800599453, "learning_rate": 2.840909090909091e-08, "logits": -1.359458565711975, "logps": -84.69721221923828, "loss": 0.0051, "objective": 0.0046141319908201694, "ranking_idealized": 0.5833333134651184, "ranking_idealized_expo": 0.5833333134651184, "ranking_simple": 0.5833333134651184, "regularize": 0.0046141319908201694, "step": 1, "wo_beta": 14.840873718261719 }, { "dpo_loss": 0.7411907315254211, "epoch": 0.14170996693434104, "grad_norm": 326.1369710639882, "learning_rate": 1.4204545454545458e-06, "logits": -1.4541096687316895, "logps": -84.35031127929688, "loss": 0.4306, "objective": 0.4069998860359192, "ranking_idealized": 0.5289115905761719, "ranking_idealized_expo": 0.5221088528633118, "ranking_simple": 0.5225340127944946, "regularize": 0.4069998860359192, "step": 50, "wo_beta": 15.642387390136719 }, { "epoch": 0.14170996693434104, "eval_dpo_loss": 0.7632195949554443, "eval_logits": -1.428945779800415, "eval_logps": -90.4263687133789, "eval_loss": 0.5493111610412598, "eval_objective": 0.5433010458946228, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5212215185165405, "eval_regularize": 0.5433010458946228, "eval_runtime": 308.9459, "eval_samples_per_second": 18.741, "eval_steps_per_second": 1.563, "eval_wo_beta": 16.223697662353516, "step": 50 }, { "dpo_loss": 1.189667820930481, "epoch": 0.2834199338686821, "grad_norm": 271.670123611671, "learning_rate": 2.8409090909090916e-06, "logits": -1.3798158168792725, "logps": -82.86579895019531, "loss": 1.748, "objective": 1.777085542678833, "ranking_idealized": 0.5241666436195374, "ranking_idealized_expo": 0.5137500166893005, "ranking_simple": 0.5174999833106995, "regularize": 1.777085542678833, "step": 100, "wo_beta": 15.245094299316406 }, { "epoch": 0.2834199338686821, "eval_dpo_loss": 1.1354432106018066, "eval_logits": -1.2535114288330078, "eval_logps": -88.04913330078125, "eval_loss": 1.6975041627883911, "eval_objective": 1.6864262819290161, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.522774338722229, "eval_regularize": 1.6864262819290161, "eval_runtime": 308.5261, "eval_samples_per_second": 18.767, "eval_steps_per_second": 1.566, "eval_wo_beta": 15.683449745178223, "step": 100 }, { "dpo_loss": 1.6895211935043335, "epoch": 0.42512990080302315, "grad_norm": 221.9711378891906, "learning_rate": 4.2613636363636365e-06, "logits": -1.2630772590637207, "logps": -79.18020629882812, "loss": 2.8697, "objective": 2.862943172454834, "ranking_idealized": 0.5333333611488342, "ranking_idealized_expo": 0.527916669845581, "ranking_simple": 0.5199999809265137, "regularize": 2.862943172454834, "step": 150, "wo_beta": 15.206230163574219 }, { "epoch": 0.42512990080302315, "eval_dpo_loss": 1.6845829486846924, "eval_logits": -1.252400279045105, "eval_logps": -82.49666595458984, "eval_loss": 2.9623751640319824, "eval_objective": 2.892326593399048, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5243270993232727, "eval_regularize": 2.892326593399048, "eval_runtime": 308.7897, "eval_samples_per_second": 18.751, "eval_steps_per_second": 1.564, "eval_wo_beta": 15.196976661682129, "step": 150 }, { "dpo_loss": 2.04150390625, "epoch": 0.5668398677373642, "grad_norm": 196.4932514581971, "learning_rate": 4.997168347957521e-06, "logits": -1.0472546815872192, "logps": -75.09263610839844, "loss": 3.5268, "objective": 3.5377087593078613, "ranking_idealized": 0.5204166769981384, "ranking_idealized_expo": 0.51541668176651, "ranking_simple": 0.5066666603088379, "regularize": 3.5377087593078613, "step": 200, "wo_beta": 15.402359962463379 }, { "epoch": 0.5668398677373642, "eval_dpo_loss": 2.158970594406128, "eval_logits": -0.9581480622291565, "eval_logps": -75.9715576171875, "eval_loss": 4.030220031738281, "eval_objective": 3.9596574306488037, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.523809552192688, "eval_regularize": 3.9596574306488037, "eval_runtime": 308.8005, "eval_samples_per_second": 18.75, "eval_steps_per_second": 1.564, "eval_wo_beta": 14.579182624816895, "step": 200 }, { "dpo_loss": 2.0967283248901367, "epoch": 0.7085498346717053, "grad_norm": 186.71108857655122, "learning_rate": 4.973122855144066e-06, "logits": -0.8511989116668701, "logps": -75.66019439697266, "loss": 3.7241, "objective": 3.6993343830108643, "ranking_idealized": 0.5249999761581421, "ranking_idealized_expo": 0.5162500143051147, "ranking_simple": 0.5104166865348816, "regularize": 3.6993343830108643, "step": 250, "wo_beta": 15.4879789352417 }, { "epoch": 0.7085498346717053, "eval_dpo_loss": 2.3309643268585205, "eval_logits": -0.7680178880691528, "eval_logps": -81.3046875, "eval_loss": 4.269361972808838, "eval_objective": 4.272838115692139, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5258799195289612, "eval_regularize": 4.272838115692139, "eval_runtime": 308.4403, "eval_samples_per_second": 18.772, "eval_steps_per_second": 1.566, "eval_wo_beta": 14.561494827270508, "step": 250 }, { "dpo_loss": 1.9671478271484375, "epoch": 0.8502598016060463, "grad_norm": 175.55668911243956, "learning_rate": 4.924776641419513e-06, "logits": -0.6268683671951294, "logps": -79.04060363769531, "loss": 3.6109, "objective": 3.6469531059265137, "ranking_idealized": 0.5062500238418579, "ranking_idealized_expo": 0.4950000047683716, "ranking_simple": 0.5049999952316284, "regularize": 3.6469531059265137, "step": 300, "wo_beta": 15.15662956237793 }, { "epoch": 0.8502598016060463, "eval_dpo_loss": 2.4072353839874268, "eval_logits": -0.6388247609138489, "eval_logps": -83.98152923583984, "eval_loss": 4.49080228805542, "eval_objective": 4.45733642578125, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5263975262641907, "eval_regularize": 4.45733642578125, "eval_runtime": 308.8236, "eval_samples_per_second": 18.749, "eval_steps_per_second": 1.564, "eval_wo_beta": 14.346380233764648, "step": 300 }, { "dpo_loss": 1.8654496669769287, "epoch": 0.9919697685403873, "grad_norm": 165.9769670968443, "learning_rate": 4.8526047530778175e-06, "logits": -0.5482619404792786, "logps": -77.96940612792969, "loss": 3.36, "objective": 3.306919574737549, "ranking_idealized": 0.5354166626930237, "ranking_idealized_expo": 0.5254166722297668, "ranking_simple": 0.5224999785423279, "regularize": 3.306919574737549, "step": 350, "wo_beta": 15.03686237335205 }, { "epoch": 0.9919697685403873, "eval_dpo_loss": 2.499128580093384, "eval_logits": -0.5030297636985779, "eval_logps": -80.74913024902344, "eval_loss": 4.658600807189941, "eval_objective": 4.6211676597595215, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5212215185165405, "eval_regularize": 4.6211676597595215, "eval_runtime": 314.3306, "eval_samples_per_second": 18.42, "eval_steps_per_second": 1.537, "eval_wo_beta": 14.346673965454102, "step": 350 }, { "dpo_loss": 1.732424259185791, "epoch": 1.1336797354747283, "grad_norm": 163.81688300949705, "learning_rate": 4.757316345716554e-06, "logits": -0.5219219326972961, "logps": -77.66575622558594, "loss": 3.112, "objective": 3.121299982070923, "ranking_idealized": 0.5412499904632568, "ranking_idealized_expo": 0.5320833325386047, "ranking_simple": 0.53125, "regularize": 3.121299982070923, "step": 400, "wo_beta": 15.248611450195312 }, { "epoch": 1.1336797354747283, "eval_dpo_loss": 2.5402870178222656, "eval_logits": -0.5664270520210266, "eval_logps": -82.49735260009766, "eval_loss": 4.724371433258057, "eval_objective": 4.729303359985352, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5186335444450378, "eval_regularize": 4.729303359985352, "eval_runtime": 310.1606, "eval_samples_per_second": 18.668, "eval_steps_per_second": 1.557, "eval_wo_beta": 14.403800964355469, "step": 400 }, { "dpo_loss": 1.6898671388626099, "epoch": 1.2753897024090695, "grad_norm": 190.1252094592616, "learning_rate": 4.639847716126855e-06, "logits": -0.5678730010986328, "logps": -78.37519836425781, "loss": 2.9448, "objective": 3.0088882446289062, "ranking_idealized": 0.5245833396911621, "ranking_idealized_expo": 0.5191666483879089, "ranking_simple": 0.518750011920929, "regularize": 3.0088882446289062, "step": 450, "wo_beta": 15.990630149841309 }, { "epoch": 1.2753897024090695, "eval_dpo_loss": 2.573472738265991, "eval_logits": -0.6935593485832214, "eval_logps": -80.17532348632812, "eval_loss": 4.835423946380615, "eval_objective": 4.811424732208252, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5248447060585022, "eval_regularize": 4.811424732208252, "eval_runtime": 308.2089, "eval_samples_per_second": 18.786, "eval_steps_per_second": 1.567, "eval_wo_beta": 13.93560791015625, "step": 450 }, { "dpo_loss": 1.5981500148773193, "epoch": 1.4170996693434104, "grad_norm": 156.9344817637022, "learning_rate": 4.501353102310901e-06, "logits": -0.58225017786026, "logps": -76.96984100341797, "loss": 2.8517, "objective": 2.828920602798462, "ranking_idealized": 0.5054166913032532, "ranking_idealized_expo": 0.4970833361148834, "ranking_simple": 0.502916693687439, "regularize": 2.828920602798462, "step": 500, "wo_beta": 15.1870756149292 }, { "epoch": 1.4170996693434104, "eval_dpo_loss": 2.6782028675079346, "eval_logits": -0.597258448600769, "eval_logps": -80.76759338378906, "eval_loss": 5.004448413848877, "eval_objective": 5.005820274353027, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5269151329994202, "eval_regularize": 5.005820274353027, "eval_runtime": 308.5472, "eval_samples_per_second": 18.765, "eval_steps_per_second": 1.565, "eval_wo_beta": 14.262606620788574, "step": 500 }, { "dpo_loss": 1.574170708656311, "epoch": 1.5588096362777515, "grad_norm": 157.3420281051892, "learning_rate": 4.34319334202531e-06, "logits": -0.5697833299636841, "logps": -77.83871459960938, "loss": 2.632, "objective": 2.6769964694976807, "ranking_idealized": 0.5199999809265137, "ranking_idealized_expo": 0.5108333230018616, "ranking_simple": 0.5104166865348816, "regularize": 2.6769964694976807, "step": 550, "wo_beta": 15.042858123779297 }, { "epoch": 1.5588096362777515, "eval_dpo_loss": 2.5752413272857666, "eval_logits": -0.6148820519447327, "eval_logps": -80.52192687988281, "eval_loss": 4.877658843994141, "eval_objective": 4.884354114532471, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5222567319869995, "eval_regularize": 4.884354114532471, "eval_runtime": 309.1726, "eval_samples_per_second": 18.727, "eval_steps_per_second": 1.562, "eval_wo_beta": 14.146936416625977, "step": 550 }, { "dpo_loss": 1.4848108291625977, "epoch": 1.7005196032120926, "grad_norm": 158.42202757367144, "learning_rate": 4.16692250129073e-06, "logits": -0.5375781655311584, "logps": -77.73197937011719, "loss": 2.5208, "objective": 2.518798828125, "ranking_idealized": 0.5220833420753479, "ranking_idealized_expo": 0.5149999856948853, "ranking_simple": 0.5058333277702332, "regularize": 2.518798828125, "step": 600, "wo_beta": 15.039664268493652 }, { "epoch": 1.7005196032120926, "eval_dpo_loss": 2.5973868370056152, "eval_logits": -0.5874757766723633, "eval_logps": -80.17750549316406, "eval_loss": 4.925836086273193, "eval_objective": 4.962131500244141, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5243270993232727, "eval_regularize": 4.962131500244141, "eval_runtime": 308.6109, "eval_samples_per_second": 18.761, "eval_steps_per_second": 1.565, "eval_wo_beta": 14.266931533813477, "step": 600 }, { "dpo_loss": 1.4892569780349731, "epoch": 1.8422295701464337, "grad_norm": 161.21497067375822, "learning_rate": 3.974272604254906e-06, "logits": -0.4858429431915283, "logps": -77.9110107421875, "loss": 2.4198, "objective": 2.448315143585205, "ranking_idealized": 0.5408333539962769, "ranking_idealized_expo": 0.527916669845581, "ranking_simple": 0.527916669845581, "regularize": 2.448315143585205, "step": 650, "wo_beta": 15.80264663696289 }, { "epoch": 1.8422295701464337, "eval_dpo_loss": 2.6345465183258057, "eval_logits": -0.5440609455108643, "eval_logps": -81.05504608154297, "eval_loss": 5.032698154449463, "eval_objective": 5.045432090759277, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5269151329994202, "eval_regularize": 5.045432090759277, "eval_runtime": 315.2786, "eval_samples_per_second": 18.365, "eval_steps_per_second": 1.532, "eval_wo_beta": 14.247919082641602, "step": 650 }, { "dpo_loss": 1.3786677122116089, "epoch": 1.9839395370807746, "grad_norm": 165.2682511618685, "learning_rate": 3.767136614452458e-06, "logits": -0.4978749454021454, "logps": -77.35163116455078, "loss": 2.2699, "objective": 2.302328586578369, "ranking_idealized": 0.5170833468437195, "ranking_idealized_expo": 0.5112500190734863, "ranking_simple": 0.5095833539962769, "regularize": 2.302328586578369, "step": 700, "wo_beta": 15.442806243896484 }, { "epoch": 1.9839395370807746, "eval_dpo_loss": 2.629160165786743, "eval_logits": -0.559353232383728, "eval_logps": -79.73757934570312, "eval_loss": 4.9659104347229, "eval_objective": 4.995144844055176, "eval_ranking_idealized": 0.5295031070709229, "eval_ranking_idealized_expo": 0.5212215185165405, "eval_ranking_simple": 0.5212215185165405, "eval_regularize": 4.995144844055176, "eval_runtime": 308.0921, "eval_samples_per_second": 18.793, "eval_steps_per_second": 1.568, "eval_wo_beta": 14.175456047058105, "step": 700 }, { "epoch": 1.9839395370807746, "step": 700, "total_flos": 0.0, "train_loss": 2.7151880131661894, "train_runtime": 19357.2061, "train_samples_per_second": 13.122, "train_steps_per_second": 0.091 } ], "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 }