Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning

• URL: http://arxiv.org/abs/2405.12241v2
• Date: Fri, 24 May 2024 13:16:32 GMT
• Title: Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
• Authors: Dan Braun, Jordan Taylor, Nicholas Goldowsky-Dill, Lee Sharkey,
• Abstract summary: Identifying the features learned by neural networks is a core challenge in mechanistic interpretability. We propose end-to-end sparse dictionary learning, a method for training SAEs that ensures the features learned are functionally important. We explore geometric and qualitative differences between e2e SAE features and standard SAE features.
• Score: 0.9374652839580183
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Identifying the features learned by neural networks is a core challenge in mechanistic interpretability. Sparse autoencoders (SAEs), which learn a sparse, overcomplete dictionary that reconstructs a network's internal activations, have been used to identify these features. However, SAEs may learn more about the structure of the datatset than the computational structure of the network. There is therefore only indirect reason to believe that the directions found in these dictionaries are functionally important to the network. We propose end-to-end (e2e) sparse dictionary learning, a method for training SAEs that ensures the features learned are functionally important by minimizing the KL divergence between the output distributions of the original model and the model with SAE activations inserted. Compared to standard SAEs, e2e SAEs offer a Pareto improvement: They explain more network performance, require fewer total features, and require fewer simultaneously active features per datapoint, all with no cost to interpretability. We explore geometric and qualitative differences between e2e SAE features and standard SAE features. E2e dictionary learning brings us closer to methods that can explain network behavior concisely and accurately. We release our library for training e2e SAEs and reproducing our analysis at https://github.com/ApolloResearch/e2e_sae

Related papers

• Enhancing Neural Network Interpretability with Feature-Aligned Sparse Autoencoders [8.003244901104111]
  We propose a regularization technique for improving feature learning by encouraging SAEs trained in parallel to learn similar features. textscMFR can improve the reconstruction loss of SAEs by up to 21.21% on GPT-2 Small, and 6.67% on EEG data.
  arXiv  Detail & Related papers  (2024-11-02T11:42:23Z)
• Llama Scope: Extracting Millions of Features from Llama-3.1-8B with Sparse Autoencoders [115.34050914216665]
  Sparse Autoencoders (SAEs) have emerged as a powerful unsupervised method for extracting sparse representations from language models. We introduce a suite of 256 SAEs, trained on each layer and sublayer of the Llama-3.1-8B-Base model, with 32K and 128K features. We assess the generalizability of SAEs trained on base models to longer contexts and fine-tuned models.
  arXiv  Detail & Related papers  (2024-10-27T17:33:49Z)
• Automatically Interpreting Millions of Features in Large Language Models [1.8035046415192353]
  sparse autoencoders (SAEs) can be used to transform activations into a higher-dimensional latent space. We build an open-source pipeline to generate and evaluate natural language explanations for SAE features. Our large-scale analysis confirms that SAE latents are indeed much more interpretable than neurons.
  arXiv  Detail & Related papers  (2024-10-17T17:56:01Z)
• Towards Principled Evaluations of Sparse Autoencoders for Interpretability and Control [43.860799289234755]
  We propose a framework for evaluating feature dictionaries in the context of specific tasks, by comparing them against emphmagnitude feature dictionaries. First, we demonstrate that supervised dictionaries achieve excellent approximation, control, and interpretability of model computations on the task. We apply this framework to the indirect object identification (IOI) task using GPT-2 Small, with sparse autoencoders (SAEs) trained on either the IOI or OpenWebText datasets.
  arXiv  Detail & Related papers  (2024-05-14T07:07:13Z)
• Beyond Prototypes: Semantic Anchor Regularization for Better Representation Learning [82.29761875805369]
  One of the ultimate goals of representation learning is to achieve compactness within a class and well-separability between classes. We propose a novel perspective to use pre-defined class anchors serving as feature centroid to unidirectionally guide feature learning. The proposed Semantic Anchor Regularization (SAR) can be used in a plug-and-play manner in the existing models.
  arXiv  Detail & Related papers  (2023-12-19T05:52:38Z)
• Learning Decoupling Features Through Orthogonality Regularization [55.79910376189138]
  Keywords spotting (KWS) and speaker verification (SV) are two important tasks in speech applications. We develop a two-branch deep network (KWS branch and SV branch) with the same network structure. A novel decoupling feature learning method is proposed to push up the performance of KWS and SV simultaneously.
  arXiv  Detail & Related papers  (2022-03-31T03:18:13Z)
• Learning distinct features helps, provably [98.78384185493624]
  We study the diversity of the features learned by a two-layer neural network trained with the least squares loss. We measure the diversity by the average $L$-distance between the hidden-layer features.
  arXiv  Detail & Related papers  (2021-06-10T19:14:45Z)
• Toward Understanding the Feature Learning Process of Self-supervised Contrastive Learning [43.504548777955854]
  We study how contrastive learning learns the feature representations for neural networks by analyzing its feature learning process. We prove that contrastive learning using textbfReLU networks provably learns the desired sparse features if proper augmentations are adopted.
  arXiv  Detail & Related papers  (2021-05-31T16:42:09Z)
• An Explainable Machine Learning-based Network Intrusion Detection System for Enabling Generalisability in Securing IoT Networks [0.0]
  Machine Learning (ML)-based network intrusion detection systems bring many benefits for enhancing the security posture of an organisation. Many systems have been designed and developed in the research community, often achieving a perfect detection rate when evaluated using certain datasets. This paper tightens the gap by evaluating the generalisability of a common feature set to different network environments and attack types.
  arXiv  Detail & Related papers  (2021-04-15T00:44:45Z)
• PIN: A Novel Parallel Interactive Network for Spoken Language Understanding [68.53121591998483]
  In the existing RNN-based approaches, ID and SF tasks are often jointly modeled to utilize the correlation information between them. The experiments on two benchmark datasets, i.e., SNIPS and ATIS, demonstrate the effectiveness of our approach. More encouragingly, by using the feature embedding of the utterance generated by the pre-trained language model BERT, our method achieves the state-of-the-art among all comparison approaches.
  arXiv  Detail & Related papers  (2020-09-28T15:59:31Z)
• Fully Convolutional Networks for Continuous Sign Language Recognition [83.85895472824221]
  Continuous sign language recognition is a challenging task that requires learning on both spatial and temporal dimensions. We propose a fully convolutional network (FCN) for online SLR to concurrently learn spatial and temporal features from weakly annotated video sequences.
  arXiv  Detail & Related papers  (2020-07-24T08:16:37Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.