arXiv:2607. 18305v1 Announce Type: cross Abstract: Some limits on what language models know are not gaps in data coverage but structural properties of learning from text.
Matryoshka Attribution (MAttr) is a mask‑learning method that identifies nested subsets of a language model’s internal components by minimizing downstream loss. It uses a differentiable sigmoid top‑k operator and randomizes sparsity during training to produce an attribution ordering of components. MAttr tops the Mechanistic Interpretability Benchmark leaderboard and can be applied via reinforcement learning to pinpoint weight changes that control behaviors such as refusal in Llama 3.1 8B Instruct, where restoring just 1% of weights removes refusals while preserving capabilities.
The paper introduces Concept-Targeted Attribution (CTA), a method that trains attribution graphs to explain the emergence of internal concept representations in language models, rather than just the final token prediction. CTA produces probe-specific circuits that reveal which internal computations drive a linear probe’s accuracy, and cross-layer transcoders demonstrate that these graphs contain predictive structure across multiple concept categories. Causal ablations show that probe-targeted and logit-targeted graphs capture distinct mechanisms, with probe-relevant features affecting internal concept scores and logit-relevant features altering generated tokens.
arXiv:2609.16229v1 Announce Type: new Abstract: Machine unlearning aims to remove specific knowledge from a trained large language model (LLM) without retraining from scratch. Existing methods modify...
The paper introduces Circuit Condensation, a post‑training method that prunes low‑attribution edges from large causal graphs and trains a low‑rank adapter to preserve behavior. Across four behaviors and eight models, the condensed circuits are on average 8.1× smaller than the strongest frozen baseline, with reductions up to 316×. Experiments show that weight updates drive the size reduction, and detailed ablations reveal dependencies among remaining edges and a more focused set of heads for indirect object identification.
arXiv:2605. 29548v2 Announce Type: replace Abstract: Larger models learn tasks smaller models do not.
arXiv:2601. 09624v2 Announce Type: replace-cross Abstract: Machine unlearning is becoming essential for building trustworthy and compliant language models.
arXiv:2606. 18383v1 Announce Type: new Abstract: Sparse autoencoders (SAEs) are increasingly used to extract interpretable features from language models (LMs), yet a central question remains: when can an SAE-based explanation be treated as a faithful view of an underlying frozen LM We study this through a post-hoc generalization framework that certifies the LM via a sparse proxy, obtained by replacing a native hidden activation with its pretrained SAE reconstruction.
arXiv:2607. 20652v1 Announce Type: cross Abstract: Language models are thought to exhibit the phenomenon of superposition, representing many more features than dimensions in their residual streams.
arXiv:2607. 18921v1 Announce Type: cross Abstract: Circuit extraction identifies a small set of model components whose presence preserves a target behavior under ablation, and the resulting circuit is often read as the mechanism behind that behavior.
arXiv:2510. 12957v4 Announce Type: replace-cross Abstract: We treat the internals of generative models as mechanistic objects rather than black boxes.
arXiv:2605.01699v4 Announce Type: replace Abstract: Recent attacks show that behavioural unlearning of large language models leaves internal traces recoverable by adversarial probes. We characterise...