Pith. sign in

REVIEW 10 cited by

LEACE: Perfect linear concept erasure in closed form

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2306.03819 v4 pith:YQLGT5C3 submitted 2023-06-06 cs.LG cs.CLcs.CY

classification cs.LGcs.CLcs.CY
keywords concepterasureleaceembeddinggenderinformationlanguagelinear
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

Concept erasure aims to remove specified features from an embedding. It can improve fairness (e.g. preventing a classifier from using gender or race) and interpretability (e.g. removing a concept to observe changes in model behavior). We introduce LEAst-squares Concept Erasure (LEACE), a closed-form method which provably prevents all linear classifiers from detecting a concept while changing the embedding as little as possible, as measured by a broad class of norms. We apply LEACE to large language models with a novel procedure called "concept scrubbing," which erases target concept information from every layer in the network. We demonstrate our method on two tasks: measuring the reliance of language models on part-of-speech information, and reducing gender bias in BERT embeddings. Code is available at https://github.com/EleutherAI/concept-erasure.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 10 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. OpenAlex reports about 12 citations worldwide. Full citation record

  1. Emergent Misalignment Recruits a Pre-existing Persona Subspace

    cs.LG 2026-07 conditional novelty 7.0 of 10

    Fine-tuning on narrow bad data recruits a low-rank persona subspace already present in a frozen instruction-tuned model; holding that subspace out of activations prevents broad misalignment, and injecting it into the ...

  2. Convergent Linear Representations of Emergent Misalignment

    cs.LG 2025-06 conditional novelty 7.0 of 10

    A single activation direction extracted from one misaligned fine-tune can ablate emergent misalignment across models trained on different datasets with different LoRA setups.

  3. Slowing Learning by Erasing Simple Features

    cs.LG 2025-02 conditional novelty 7.0 of 10

    QLEACE removes all quadratically available class information from a representation, reliably slows feedforward networks, but can inject higher-order information that lets stronger architectures learn faster.

  4. ECG-InterpBench: Benchmarking the Interpretability of ECG Foundation Models with Matched-Scale Sparse Autoencoders

    cs.LG 2026-07 conditional novelty 6.0 of 10

    With matched-scale sparse autoencoders, HuBERT-ECG best preserves its ECG representation while ECG-JEPA best exposes clinical measurements through single features — a leader split that repeats on MIMIC-IV-ECG.

  5. Emergent Latent-State Computation under Stochastic Volatility

    cs.LG 2026-07 conditional novelty 6.0 of 10

    Volatility forecasters develop linearly decodable representations of the next hidden log-volatility state; in long cycles this appears immediately after the input projection and ℓ2 normalization.

  6. Improved Representation Steering for Language Models

    cs.CL 2025-05 conditional novelty 6.0 of 10

    RePS, a reference-free bidirectional preference optimization objective, improves representation steering and suppression for Gemma models, outperforming language-modeling objectives and approaching prompting performance.

  7. Unlearning Sensitive Information in Multimodal LLMs: Benchmark and Attack-Defense Evaluation

    cs.CL 2025-05 conditional novelty 6.0 of 10

    UnLOK-VQA adds rephrase and neighborhood samples to OK-VQA to measure generalization and specificity of multimodal unlearning, and its evaluation shows that hiding answer tokens in hidden states beats other deletion o...

  8. Converting MLPs into Polynomials in Closed Form

    cs.LG 2025-02 conditional novelty 6.0 of 10

    Closed-form polynomial approximations of MLPs and GLUs under Gaussian mixture inputs reveal a training phase where networks shift from linear to quadratic behavior.

  9. The Confounder Trap: Treatment-Encoding Representations in Causal Inference with Text

    stat.ME 2026-07 conditional novelty 5.0 of 10

    Masking treatment-defining words before learning adjustment representations preserves overlap and reduces bias in text-as-treatment causal inference.

  10. Perspectives for Direct Interpretability in Multi-Agent Deep Reinforcement Learning

    cs.AI 2025-02 unverdicted novelty 4.0 of 10

    A perspective paper that advocates direct, post hoc interpretability for multi-agent deep reinforcement learning and offers a taxonomy of where those methods might apply.

Pith tools