Refresh card: tf6/tf7 filename split + paper link

README.md

@@ -4,20 +4,38 @@ tags:
   - ace-step
   - audio
   - feature-selection
-  - guitar_electronic
+  - guitar-electronic
   - interpretability
   - music
   - sae
+  - sparse-autoencoder
+  - steering
 ---
 
 # SAE Feature-Selection Scores — `guitar_electronic` (ACE-Step)
 
-Per-concept feature-importance scores for the ACE-Step SAEs at `transformer_blocks.6.cross_attn` and `transformer_blocks.7.cross_attn`. Consumed at inference time by `SAEController` via `scores_cache_path` (or directly by `FeatureSelector`).
+Per-concept feature-importance scores for the ACE-Step SAEs at `transformer_blocks.6.cross_attn` and `transformer_blocks.7.cross_attn`. Consumed at inference time by `SAESteeringController` via `load_features_from_score_cache` (top-k features per diffusion step).
 
 ## Files
 
-- `all_scores.pkl` — scores for the tf7 SAE.
-- `tf6_all_scores.pkl` — scores for the tf6 SAE.
-- `best_config.pkl` (optional) — best-performing selection config.
+- `tf7_scores.pkl` — scores for the tf7 SAE.
+- `tf6_scores.pkl` — scores for the tf6 SAE.
 
-The raw activations cache (`activations.pkl`, ~9.3 GB) is NOT included — regeneratable via `run_compute.py --scorer sae-activations`.
+Each pickle is a dict keyed by selection method (`tfidf`, `diff`, `mean_pos`, ...); values are tensors of shape `(num_timesteps, num_features)`.
+
+## 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.methods.sae import load_features_from_score_cache
+
+top20_tf7 = load_features_from_score_cache(
+    "lukasz-staniszewski/ace-step-sae-scores-guitar-electronic", score_filename="tf7_scores.pkl", top_k=20,
+)
+top20_tf6 = load_features_from_score_cache(
+    "lukasz-staniszewski/ace-step-sae-scores-guitar-electronic", score_filename="tf6_scores.pkl", top_k=20,
+)
+```