SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2204.09722 · NAACL · 2022

When Does Syntax Mediate Neural Language Model Performance? Evidence from Dropout Probes

Roger Levy, Julie Shah, Mycal Tucker, Peng Qian

arXiv · PDF · Open in the Atlas

Code that ran

We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. The repositories linked to it are listed below.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Recent causal probing literature reveals when language models and syntactic probes use similar representations. Such techniques may yield "false negative" causality results: models may use representations of syntax, but probes may have learned to use redundant encodings of the same syntactic information. We demonstrate that models do encode syntactic information redundantly and introduce a new probe design that guides probes to consider all syntactic information present in embeddings. Using these probes, we find evidence for the use of syntax in models where prior methods did not, allowing us to boost model performance by injecting syntactic information into representations. Here, we identify a limitation of prior causal probing art in which redundant information in embeddings could lead to probes and models using different representations of the same information, which in turn could lead to uninformative causal analysis results. We propose a new probe architecture that addresses this limitation by encouraging probes to use all sources of information in embeddings.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2204.09722")
get_code_for_paper("2204.09722")
have("2204.09722")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2204.09722.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2204.09722.svg)](https://syntology.ai/paper/2204.09722)

Connect an agent — have() is free.