lukasz-staniszewski/ace-step-tokemb-rock-genre

Token-Embedding (TokEmb) Direction — rock_genre (ACE-Step)

Per-concept T5 hidden-state direction for steering ACE-Step audio generation toward the rock_genre concept. At inference, TokEmbSteeringController adds alpha * direction to the concept token's hidden state in the neutral prompt embedding before the diffusion process.

The .pt file is a dict with keys: direction (Tensor[hidden_dim]), concept (str), hidden_dim (int).

Paper

TADA! Tuning Audio Diffusion Models through Activation Steering — https://huggingface.co/papers/2602.11910

Quickstart

from src.steering import SteerableACEModel, TokEmbSteeringController

model = SteerableACEModel(device="cuda")
model.pipeline.load()
ctrl = TokEmbSteeringController.from_pretrained(
    "lukasz-staniszewski/ace-step-tokemb-rock-genre", alpha=1.0, te_split_step=3,
)

with model.steer(ctrl):
    audio = model.generate(
        prompt="instrumental music", lyrics="[inst]",
        audio_duration=10.0, infer_step=30, manual_seed=0,
    )

Metadata

{
  "hidden_dim": 768,
  "direction_file": "rock_genre_direction.pt"
}