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

Download:

Current browse context:

cs.CL

Change to browse by:

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: Investigating the Benefits of Free-Form Rationales

Abstract: Free-form rationales aim to aid model interpretability by supplying the background knowledge that can help understand model decisions. Crowdsourced rationales are provided for commonsense QA instances in popular datasets such as CoS-E and ECQA, but their utility remains under-investigated. We present human studies which show that ECQA rationales indeed provide additional background information to understand a decision, while over 88% of CoS-E rationales do not. Inspired by this finding, we ask: can the additional context provided by free-form rationales benefit models, similar to human users? We investigate the utility of rationales as an additional source of supervision, by varying the quantity and quality of rationales during training. After controlling for instances where rationales leak the correct answer while not providing additional background knowledge, we find that incorporating only 5% of rationales during training can boost model performance by 47.22% for CoS-E and 57.14% for ECQA during inference. Moreover, we also show that rationale quality matters: compared to crowdsourced rationales, T5-generated rationales provide not only weaker supervision to models, but are also not helpful for humans in aiding model interpretability.
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Cite as: arXiv:2206.11083 [cs.CL]
  (or arXiv:2206.11083v1 [cs.CL] for this version)

Submission history

From: Jiao Sun [view email]
[v1] Wed, 25 May 2022 11:46:11 GMT (7488kb,D)

Link back to: arXiv, form interface, contact.