We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.CL

Change to browse by:

cs

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computation and Language

Title: On the Pitfalls of Analyzing Individual Neurons in Language Models

Abstract: While many studies have shown that linguistic information is encoded in hidden word representations, few have studied individual neurons, to show how and in which neurons it is encoded. Among these, the common approach is to use an external probe to rank neurons according to their relevance to some linguistic attribute, and to evaluate the obtained ranking using the same probe that produced it. We show two pitfalls in this methodology: 1. It confounds distinct factors: probe quality and ranking quality. We separate them and draw conclusions on each. 2. It focuses on encoded information, rather than information that is used by the model. We show that these are not the same. We compare two recent ranking methods and a simple one we introduce, and evaluate them with regard to both of these aspects.
Subjects: Computation and Language (cs.CL)
Cite as: arXiv:2110.07483 [cs.CL]
  (or arXiv:2110.07483v1 [cs.CL] for this version)

Submission history

From: Omer Antverg [view email]
[v1] Thu, 14 Oct 2021 15:57:07 GMT (1995kb,D)

Link back to: arXiv, form interface, contact.