REVIEW 2 cited by
A Causal Framework to Quantify the Robustness of Mathematical Reasoning with Language Models
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
We have recently witnessed a number of impressive results on hard mathematical reasoning problems with language models. At the same time, the robustness of these models has also been called into question; recent works have shown that models can rely on shallow patterns in the problem description when generating a solution. Building on the idea of behavioral testing, we propose a novel framework, which pins down the causal effect of various factors in the input, e.g., the surface form of the problem text, the operands, and math operators on the output solution. By grounding the behavioral analysis in a causal graph describing an intuitive reasoning process, we study the behavior of language models in terms of robustness and sensitivity to direct interventions in the input space. We apply our framework on a test bed of math word problems. Our analysis shows that robustness does not appear to continuously improve as a function of size, but the GPT-3 Davinci models (175B) achieve a dramatic improvement in both robustness and sensitivity compared to all other GPT variants.
Forward citations
Cited by 2 Pith papers
-
CodeSCM: Causal Analysis for Multi-Modal Code Generation
A causal framework with latent mediators quantifies how prompt modalities affect code LLMs, finding that input-output examples and function-header names are influential beyond natural language instructions.
-
Evaluating and Improving Robustness in Large Language Models: A Survey and Future Directions
LLM robustness research is organized into adversarial robustness, out-of-distribution robustness, and evaluation, with an accompanying GitHub collection of papers.
Discussion (0). Continue with ORCID to comment.