--- library_name: audio-interv tags: - ace-step - activation-steering - audio - caa - diffusion - interpretability - music - rock-genre - steering --- # CAA — `rock_genre` (ACE-Step) Steering vectors for the **rock_genre** concept on ACE-Step, computed via contrastive activation addition (CAA). ## Paper TADA! Tuning Audio Diffusion Models through Activation Steering — [https://huggingface.co/papers/2602.11910](https://huggingface.co/papers/2602.11910) ## Quickstart ```python from src.steering import SteerableACEModel, CAASteeringController model = SteerableACEModel(device="cuda") model.pipeline.load() ctrl = CAASteeringController.from_pretrained("lukasz-staniszewski/ace-step-caa-electronic-music", alpha=20.0) with model.steer(ctrl): audio = model.generate( prompt="instrumental music", lyrics="[inst]", audio_duration=10.0, infer_step=30, manual_seed=0, ) ``` ## Generation config ```json { "method": "standard_caa", "concept": "rock_genre", "lyrics": "[inst]", "num_cfg_passes": 2, "save_all_cfg_passes": true, "audio_duration": 30.0, "num_inference_steps": 30, "seed": 10, "device": "cuda", "save_dir": "steering_vectors/caa", "guidance_scale_text": 0.0, "guidance_scale_lyric": 0.0, "guidance_scale": 5.0, "guidance_interval": 1.0, "guidance_interval_decay": 0.0 } ```
