A Causal Framework to Quantify the Robustness of Mathematical Reasoning with Language Models
This work addresses the issue of shallow pattern reliance in language models for mathematical reasoning, which is critical for researchers and developers aiming to build more reliable AI systems, though it is incremental as it builds on existing behavioral testing ideas.
The authors tackled the problem of assessing the robustness of language models in mathematical reasoning by proposing a causal framework to quantify the influence of input factors like surface form and operands on solutions. Their analysis on math word problems found that robustness does not consistently improve with model size, but GPT-3 Davinci (175B) showed dramatic improvements in robustness and sensitivity compared to other GPT variants.
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.