[CaCL] Reading for 3/10

[Cheung, Willy]

Hi CaCLers,

Next Thursday we will discuss the following paper: Discourse structure interacts with reference but not syntax in neural language models by Davis and van Schijndel (2020).

Paper link: https://arxiv.org/pdf/2010.04887.pdf
arXiv:2010.04887v1 [cs.CL] 10 Oct 2020<https://arxiv.org/pdf/2010.04887.pdf>
two pretrained unidirectional transformer LMs: TransformerXL (Dai et al.,2019) and GPT-2 XL (Radford et al.,2019).2 TransformerXL was trained on Wikitext-103,
arxiv.org

Zoom link:
https://osu.zoom.us/j/95536921111?pwd=TG9YdVZ0Wk45R2hCdHhTYk5ubkhIQT09

Abstract: Language models (LMs) trained on large quantities of text have been claimed to acquire abstract linguistic representations. Our work tests the robustness of these abstractions by focusing on the ability of LMs to learn interactions between different linguistic representations. In particular, we utilized stimuli from psycholinguistic studies showing that humans can condition reference (i.e. coreference resolution) and syntactic processing on the same discourse structure (implicit causality). We compared both transformer and long short-term memory LMs to find that, contrary to humans, implicit causality only influences LM behavior for reference, not syntax, despite model representations that encode the necessary discourse information. Our results further suggest that LM behavior can contradict not only learned representations of discourse but also syntactic agreement, pointing to shortcomings of standard language modeling.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osu.edu/pipermail/cacl/attachments/20220303/7c9f2684/attachment.html>