Improving Dictionary Learning with Gated Sparse Autoencoders
This work addresses a specific bottleneck in unsupervised feature extraction for language models, offering an incremental improvement over existing methods.
The paper tackles the problem of undesirable biases like shrinkage in sparse autoencoders (SAEs) used for interpretable feature discovery in language models, by introducing Gated SAEs that separate direction selection from magnitude estimation, resulting in solving shrinkage, maintaining interpretability, and requiring half as many firing features for comparable reconstruction fidelity.
Recent work has found that sparse autoencoders (SAEs) are an effective technique for unsupervised discovery of interpretable features in language models' (LMs) activations, by finding sparse, linear reconstructions of LM activations. We introduce the Gated Sparse Autoencoder (Gated SAE), which achieves a Pareto improvement over training with prevailing methods. In SAEs, the L1 penalty used to encourage sparsity introduces many undesirable biases, such as shrinkage -- systematic underestimation of feature activations. The key insight of Gated SAEs is to separate the functionality of (a) determining which directions to use and (b) estimating the magnitudes of those directions: this enables us to apply the L1 penalty only to the former, limiting the scope of undesirable side effects. Through training SAEs on LMs of up to 7B parameters we find that, in typical hyper-parameter ranges, Gated SAEs solve shrinkage, are similarly interpretable, and require half as many firing features to achieve comparable reconstruction fidelity.