arXiv:2610.27724v1 Announce Type: new Abstract: Dense linear probes assess task readability from language model (LM) representations, but do not directly reveal how many features suffice for predicti...
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.
Sparse autoencoders (SAEs) are usually trained on hundreds of millions to billions of activation tokens, but this study shows that the amount of data needed depends on the evaluation metric. Training four SAE architectures on three language models with token budgets from 5 M to 500 M revealed that reconstruction quality can reach near‑full‑data performance with far fewer tokens, while one‑to‑one feature reproducibility continues to improve as more data is used. The research also found that random sampling preserves activation distributions better for reconstruction, whereas diversity‑based sampling can enhance feature interpretability, and provided a theoretical framework distinguishing statistical consistency from the stricter conditions needed for exact feature recovery.
arXiv:2606. 14990v1 Announce Type: cross Abstract: Sparse autoencoders (SAEs) are standard tools for mechanistic interpretability, but current SAE families are constrained by fixed encoder nonlinearities such as ReLU, JumpReLU, and TopK.
The paper introduces a Fisher-guided submodular data selection method for continual pre‑training of large language models, addressing the forgetting problem that arises when a target‑domain corpus overwrites pretrained knowledge. By decomposing candidate gradients into anchor and frontier components and optimizing a log‑determinant submodular objective, the method selects data that both improves target‑domain performance and limits forgetting. Experiments on TinyLlama‑1.1B and Llama‑3.1‑8B show that the selector achieves superior adaptation and forgetting control while using ten times fewer tokens than traditional replay strategies.
arXiv:2606. 18322v1 Announce Type: cross Abstract: Sparse Autoencoders (SAEs) decompose residual-stream activations into interpretable features.
arXiv:2606. 01155v1 Announce Type: cross Abstract: Scaling laws for dense LLMs under infinite data are well explored, but how sparsity interacts with limited data is not.
arXiv:2507. 07754v3 Announce Type: replace-cross Abstract: Machine unlearning is usually evaluated by what the classifier outputs: forget-set accuracy, confidence, membership-inference scores.
arXiv:2606. 04485v1 Announce Type: new Abstract: Tabular foundation models (TFMs) increasingly rival tree ensembles, but their performance is often compute-inefficient: with standard affine scalar tokenization, each feature injects value variation through an essentially one-dimensional channel, and feature IDs/positional signals cannot increase within-feature value degrees of freedom, yielding weak early-layer value sensitivity and redundant hidden states.
arXiv:2610.15514v1 Announce Type: cross Abstract: Structured pruning reduces the inference cost of large language models (LLMs) by permanently removing neurons and weight rows, yielding dense, hardwa...
arXiv:2606. 26620v1 Announce Type: cross Abstract: Sparse autoencoders (SAEs) have emerged as a powerful tool for decomposing superposed language model representations into sparse and interpretable features.
The paper introduces the Redistribution Sensitivity Score (RSS) to detect consistent changes in sparse feature activations of Sparse Autoencoders (SAEs) when prompts are transformed. RSS identifies large, directionally consistent feature responses and evaluates their impact on downstream readouts across 27 pretrained SAEs from eight language models. Experiments show that combining RSS support erosion with the clean margin predicts recurrent safety readout failures with high AUROC, demonstrating RSS’s utility in quantifying representation changes without conflating them with failure.