mlsanityData to 14 Oct 2026

Which Sparse Autoencoder Features Are Real? Model-X Knockoffs for False Discovery Rate Control

arXiv 2511.11711 ↗First posted 12 Nov 2025cs.LGcited by 2

On the map: Mechanistic interpretability → Sparse autoencoder training and evaluation

Abstract (arXiv)

Although sparse autoencoders (SAEs) are crucial for identifying interpretable features in neural networks, it is still challenging to distinguish between real computational patterns and erroneous correlations. We introduce Model-X knockoffs to SAE feature selection, using knock-off+ to control the false discovery rate (FDR) with finite-sample guarantees under the standard Model-X assumptions (in our case, via a Gaussian surrogate for the latent distribution). We select 129 features at a target FDR q=0.1 after analyzing 512 high-activity SAE latents for sentiment classification using Pythia-70M. About 25% of the latents under examination carry task-relevant signal, whereas 75% do not, according to the chosen set, which displays a 5.40x separation in knockoff statistics compared to non-selected features. Our method offers a re-producible and principled framework for reliable feature discovery by combining SAEs with multiple-testing-aware inference, advancing the foundations of mechanistic interpretability.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in Sparse autoencoder training and evaluation, one of 13 directions in Mechanistic interpretability. Builds on 11 of its 18 references that are in the map. First posted on arXiv 12 Nov 2025.

Builds on

11 of 18 references.

Closest ancestors · 1

Same direction, newest first

  1. 2022Toy Models of Superposition21 Sep

Other references · 10

  1. 2018The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networkscited by 2,262
  2. 2023Pythia: A Suite for Analyzing Large Language Models Across Training and Scalingcited by 1,786
  3. 2023Sparse Autoencoders Find Highly Interpretable Features in Language Modelscited by 1,420
  4. 2022Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 smallcited by 1,165
  5. 2023Towards Automated Circuit Discovery for Mechanistic Interpretabilitycited by 724
  6. 2024Scaling and evaluating sparse autoencoderscited by 553
  7. 2016Panning for Gold: Model-X Knockoffs for High-dimensional Controlled Variable Selectioncited by 121
  8. 2014Controlling the false discovery rate via knockoffscited by 97
  9. 2018DeepPINK: reproducible feature selection in deep neural networkscited by 41
  10. 2020Fast and Powerful Conditional Randomization Testing via Distillationcited by 19

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 1.

2 papers.

  1. 2025Learning from Negative Examples: Why Warning-Framed Training Data Teaches What It Warns Againstcited by 0
  2. 2026What Do Sparse Autoencoder Feature Discovery Scores Identify? From Activations to Interventionscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Sanity Checks for Sparse Autoencoders: Do SAEs Beat Random Baselines?
  2. 2025Sparse Autoencoders Trained on the Same Data Learn Different Features
  3. 2026Are Sparse Autoencoder Benchmarks Reliable?
  4. 2024Evaluating Sparse Autoencoders on Targeted Concept Erasure Tasks
  5. 2025Sparse Autoencoders Do Not Find Canonical Units of Analysis
  6. 2024SAGE: Scalable Ground Truth Evaluations for Large Sparse Autoencoders
  7. 2025Are Sparse Autoencoders Useful? A Case Study in Sparse Probing
  8. 2025Automated Interpretability Metrics Do Not Distinguish Trained and Random Transformers