Is the Reversal Curse a Binding Problem? Uncovering Limitations of Transformers from a Basic Generalization Failure
This addresses a fundamental limitation in transformer models for AI researchers, offering a novel solution to enhance generalization and robustness, though it is incremental in building on existing cognitive science concepts.
The paper tackles the Reversal Curse in LLMs, where models fail to learn reversible factual associations, by linking it to the binding problem and proposing a JEPA-based design that breaks this curse, enabling improved performance on arithmetic reasoning tasks and outperforming frontier LLMs.
Despite their impressive capabilities, LLMs exhibit a basic generalization failure known as the Reversal Curse, where they struggle to learn reversible factual associations. Understanding why this occurs could help identify weaknesses in current models and advance their generalization and robustness. In this paper, we conjecture that the Reversal Curse in LLMs is a manifestation of the long-standing binding problem in cognitive science, neuroscience and AI. Specifically, we identify two primary causes of the Reversal Curse stemming from transformers' limitations in conceptual binding: the inconsistency and entanglements of concept representations. We perform a series of experiments that support these conjectures. Our exploration leads to a model design based on JEPA (Joint-Embedding Predictive Architecture) that for the first time breaks the Reversal Curse without side-stepping it with specialized data augmentation or non-causal masking, and moreover, generalization could be further improved by incorporating special memory layers that support disentangled concept representations. We demonstrate that the skill of reversal unlocks a new kind of memory integration that enables models to solve large-scale arithmetic reasoning problems via parametric forward-chaining, outperforming frontier LLMs based on non-parametric memory and prolonged explicit reasoning.