PulseAugur
实时 09:05:07

Transformer arithmetic study reveals disconnect between representation and computation

Researchers have published a paper investigating how Transformers compute algorithmic intermediates, using arithmetic tasks as a testbed. The study found that while a Transformer model achieved high accuracy on base-digit extraction, causal tests revealed that the identified internal representations of intermediates were not actually used in the computation path to the output. This highlights a divergence between what probes suggest a model represents and how it causally uses that information, even when explicit algorithmic hypotheses are available. AI

影响 Challenges current methods for understanding internal model computations, suggesting a need for more robust causal analysis beyond simple probing.

排序理由 The cluster contains an academic paper detailing novel research findings.

在 arXiv cs.LG 阅读 →

AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →

报道来源 [2]

  1. arXiv cs.LG TIER_1 English(EN) · Ishita Darade, Sushrut Thorat ·

    Represented Is Not Computed: A Causal Test of Candidate Algorithmic Intermediates in a Transformer

    arXiv:2605.22488v1 Announce Type: new Abstract: Structured prompts require integrating components according to task-relevant relations. How a network implements this integration is often hard to judge in language or vision, where those relations are rarely specified precisely eno…

  2. arXiv cs.LG TIER_1 English(EN) · Sushrut Thorat ·

    Represented Is Not Computed: A Causal Test of Candidate Algorithmic Intermediates in a Transformer

    Structured prompts require integrating components according to task-relevant relations. How a network implements this integration is often hard to judge in language or vision, where those relations are rarely specified precisely enough to define a candidate internal algorithm. Ar…