{"output_dir": "/circuits-benchmark/results", "atol": 0.05, "lr": 0.001, "use_single_loss": true, "iit_weight": 1.0, "behavior_weight": 0.4, "strict_weight": 0.7, "epochs": 1000, "early_stop_accuracy_threshold": 99.9, "act_fn": "gelu", "use_wandb": true, "wandb_project": "iit-train-seed-67-siit-weigth-0.7", "wandb_name": "case-60-seed-67-s-0.7-b-0.4-iit-1", "save_model_to_wandb": true, "device": "cuda", "clip_grad_norm": 0.1, "lr_scheduler": "linear", "model_pair": "strict", "same_size": false, "seed": 67, "batch_size": 256, "include_mlp": false, "detach_while_caching": true, "scheduler_val_metric": ["val/accuracy", "val/IIA", "val/strict_accuracy"], "siit_sampling": "sample_all", "val_iia_sampling": "all"}