Jumping Ahead: Improving Reconstruction Fidelity with JumpReLU Sparse Autoencoders
This improves feature decomposition for interpretability in language models, but is incremental as it modifies an existing method.
The paper tackled the tension between reconstruction fidelity and sparsity in sparse autoencoders for language model activations by introducing JumpReLU SAEs, which achieved state-of-the-art reconstruction fidelity at a given sparsity level on Gemma 2 9B activations compared to other methods like Gated and TopK SAEs.
Sparse autoencoders (SAEs) are a promising unsupervised approach for identifying causally relevant and interpretable linear features in a language model's (LM) activations. To be useful for downstream tasks, SAEs need to decompose LM activations faithfully; yet to be interpretable the decomposition must be sparse -- two objectives that are in tension. In this paper, we introduce JumpReLU SAEs, which achieve state-of-the-art reconstruction fidelity at a given sparsity level on Gemma 2 9B activations, compared to other recent advances such as Gated and TopK SAEs. We also show that this improvement does not come at the cost of interpretability through manual and automated interpretability studies. JumpReLU SAEs are a simple modification of vanilla (ReLU) SAEs -- where we replace the ReLU with a discontinuous JumpReLU activation function -- and are similarly efficient to train and run. By utilising straight-through-estimators (STEs) in a principled manner, we show how it is possible to train JumpReLU SAEs effectively despite the discontinuous JumpReLU function introduced in the SAE's forward pass. Similarly, we use STEs to directly train L0 to be sparse, instead of training on proxies such as L1, avoiding problems like shrinkage.