You can deploy this model and start earning money today!

lukasz-staniszewski/
ace-step-sae-scores-rock-genre

Name: ace-step-sae-scores-rock-genre
Availability: InStock

Views

No views yet

ACE-Step SAE feature scores — rock_genre

Per-timestep feature-importance score tables for the ACE-Step tf6/tf7 cross-attention SAEs: activations collected with the shared CONCEPT_TO_PROMPTS pairs (the same text every data-driven method in the benchmark uses) and encoded with the SAE in fp32. Provenance (seed, git SHA) in the *_scores_config.json sidecars.

python

ctrl = SAESteeringController.from_pretrained(
    "lukasz-staniszewski/ace-step-sae-scores-rock-genre", hookpoint="transformer_blocks.7.cross_attn",
    scores_cache_path="lukasz-staniszewski/ace-step-sae-scores-rock-genre", score_filename="tf7_scores.pkl",
    top_k=20, intervention_mode="steering_vector", alpha=10.0,
)

The original paper-era score tables remain available at revision 53b7f35b1b305e23624a6a87c4f6726d4065de52 of this repo (revision="53b7f35b1b305e23624a6a87c4f6726d4065de52" in hf_hub_download).