Encourage or Inhibit Monosemanticity? Revisit Monosemanticity from a Feature Decorrelation Perspective

• URL: http://arxiv.org/abs/2406.17969v2
• Date: Tue, 15 Oct 2024 22:37:55 GMT
• Title: Encourage or Inhibit Monosemanticity? Revisit Monosemanticity from a Feature Decorrelation Perspective
• Authors: Hanqi Yan, Yanzheng Xiang, Guangyi Chen, Yifei Wang, Lin Gui, Yulan He,
• Abstract summary: A monosemantic neuron is dedicated to a single and specific concept, which forms a one-to-one correlation between neurons and concepts. Despite extensive research in monosemanticity probing, it remains unclear whether monosemanticity is beneficial or harmful to model capacity.
• Score: 30.290777756014748
• License:
• Abstract: To better interpret the intrinsic mechanism of large language models (LLMs), recent studies focus on monosemanticity on its basic units. A monosemantic neuron is dedicated to a single and specific concept, which forms a one-to-one correlation between neurons and concepts. Despite extensive research in monosemanticity probing, it remains unclear whether monosemanticity is beneficial or harmful to model capacity. To explore this question, we revisit monosemanticity from the feature decorrelation perspective and advocate for its encouragement. We experimentally observe that the current conclusion by wang2024learning, which suggests that decreasing monosemanticity enhances model performance, does not hold when the model changes. Instead, we demonstrate that monosemanticity consistently exhibits a positive correlation with model capacity, in the preference alignment process. Consequently, we apply feature correlation as a proxy for monosemanticity and incorporate a feature decorrelation regularizer into the dynamic preference optimization process. The experiments show that our method not only enhances representation diversity and activation sparsity but also improves preference alignment performance.

Related papers

• Beyond Interpretability: The Gains of Feature Monosemanticity on Model Robustness [68.69369585600698]
  Deep learning models often suffer from a lack of interpretability due to polysemanticity. Recent advances in monosemanticity, where neurons correspond to consistent and distinct semantics, have significantly improved interpretability. We show that monosemantic features not only enhance interpretability but also bring concrete gains in model performance.
  arXiv  Detail & Related papers  (2024-10-27T18:03:20Z)
• MonoKAN: Certified Monotonic Kolmogorov-Arnold Network [48.623199394622546]
  In certain applications, model predictions must align with expert-imposed requirements, sometimes exemplified by partial monotonicity constraints. We introduce a novel ANN architecture called MonoKAN, based on the KAN architecture and achieves certified partial monotonicity while enhancing interpretability. Our experiments demonstrate that MonoKAN not only enhances interpretability but also improves predictive performance across the majority of benchmarks, outperforming state-of-the-art monotonic approaches.
  arXiv  Detail & Related papers  (2024-09-17T11:10:59Z)
• InterHandGen: Two-Hand Interaction Generation via Cascaded Reverse Diffusion [53.90516061351706]
  We present InterHandGen, a novel framework that learns the generative prior of two-hand interaction. For sampling, we combine anti-penetration and synthesis-free guidance to enable plausible generation. Our method significantly outperforms baseline generative models in terms of plausibility and diversity.
  arXiv  Detail & Related papers  (2024-03-26T06:35:55Z)
• Confronting Reward Overoptimization for Diffusion Models: A Perspective of Inductive and Primacy Biases [76.9127853906115]
  Bridging the gap between diffusion models and human preferences is crucial for their integration into practical generative. We propose Temporal Diffusion Policy Optimization with critic active neuron Reset (TDPO-R), a policy gradient algorithm that exploits the temporal inductive bias of diffusion models. Empirical results demonstrate the superior efficacy of our methods in mitigating reward overoptimization.
  arXiv  Detail & Related papers  (2024-02-13T15:55:41Z)
• Learning from Emergence: A Study on Proactively Inhibiting the Monosemantic Neurons of Artificial Neural Networks [10.390475063385756]
  We propose a new metric to measure the monosemanticity of neurons with the guarantee of efficiency for online computation. We validate our conjecture that monosemanticity brings about performance change at different model scales.
  arXiv  Detail & Related papers  (2023-12-17T14:42:46Z)
• Curve Your Enthusiasm: Concurvity Regularization in Differentiable Generalized Additive Models [5.519653885553456]
  Generalized Additive Models (GAMs) have recently experienced a resurgence in popularity due to their interpretability. We show how concurvity can severly impair the interpretability of GAMs. We propose a remedy: a conceptually simple, yet effective regularizer which penalizes pairwise correlations of the non-linearly transformed feature variables.
  arXiv  Detail & Related papers  (2023-05-19T06:55:49Z)
• On the Generalization and Adaption Performance of Causal Models [99.64022680811281]
  Differentiable causal discovery has proposed to factorize the data generating process into a set of modules. We study the generalization and adaption performance of such modular neural causal models. Our analysis shows that the modular neural causal models outperform other models on both zero and few-shot adaptation in low data regimes.
  arXiv  Detail & Related papers  (2022-06-09T17:12:32Z)
• Modeling Implicit Bias with Fuzzy Cognitive Maps [0.0]
  This paper presents a Fuzzy Cognitive Map model to quantify implicit bias in structured datasets. We introduce a new reasoning mechanism equipped with a normalization-like transfer function that prevents neurons from saturating.
  arXiv  Detail & Related papers  (2021-12-23T17:04:12Z)
• Decomposing Natural Logic Inferences in Neural NLI [9.606462437067984]
  We investigate whether neural NLI models capture the crucial semantic features central to natural logic: monotonicity and concept inclusion. We find that monotonicity information is notably weak in the representations of popular NLI models which achieve high scores on benchmarks.
  arXiv  Detail & Related papers  (2021-12-15T17:35:30Z)
• Efficient Causal Inference from Combined Observational and Interventional Data through Causal Reductions [68.6505592770171]
  Unobserved confounding is one of the main challenges when estimating causal effects. We propose a novel causal reduction method that replaces an arbitrary number of possibly high-dimensional latent confounders. We propose a learning algorithm to estimate the parameterized reduced model jointly from observational and interventional data.
  arXiv  Detail & Related papers  (2021-03-08T14:29:07Z)

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.