arXiv AI
Readout Scaling Shapes Sparse Feature Recovery
Read the original on arXiv AI →The Flow has not summarised this story yet — read it at arXiv AI.
Related stories
arXiv AI
4d ago
How Language Models Organize Task Information
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 AI
Jul 24
Scaling Interpretable Transformers with Parity Bottleneck Layers
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 Machine Learning
Oct 16
Diminishing Returns of Data in Sparse Autoencoders: Functional Behavior does not Guarantee Feature Reproducibility
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 AI
Jun 16
Rational Sparse Autoencoder
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.
arXiv Machine Learning
Oct 5
Fisher-Guided Submodular Data Selection for Continual Pre-Training of Large Language Models
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 AI
Jun 18
SAE Interventions are Unreliable: Post-Intervention Recovery of Suppressed Behavior
arXiv:2606. 18322v1 Announce Type: cross Abstract: Sparse Autoencoders (SAEs) decompose residual-stream activations into interpretable features.