Faith and Fate: Limits of Transformers on Compositionality
Nouha Dziri · Ximing Lu · Melanie Sclar · Xiang (Lorraine) Li · Liwei Jiang · Bill Yuchen Lin · Sean Welleck · Peter West · Chandra Bhagavatula · Ronan Le Bras · Jena Hwang · Soumya Sanyal · Xiang Ren · Allyson Ettinger · Zaid Harchaoui · Yejin Choi
Great Hall & Hall B1+B2 (level 1) #421
Transformer large language models (LLMs) have sparked admiration for their exceptional performance on tasks that demand intricate multi-step reasoning. Yet, these models simultaneously show failures on surprisingly trivial problems. This begs the question: Are these errors incidental, or do they signal more substantial limitations?In an attempt to demystify transformer LLMs, we investigate the limits of these models across three representative compositional tasks---multi-digit multiplication, logic grid puzzles, and a classic dynamic programming problem. These tasks require breaking problems down into sub-steps and synthesizing these steps into a precise answer. We formulate compositional tasks as computation graphs to systematically quantify the level of complexity, and break down reasoning steps into intermediate sub-procedures. Our empirical findings suggest that transformer LLMs solve compositional tasks by reducing multi-step compositional reasoning into linearized subgraph matching, without necessarily developing systematic problem-solving skills. To round off our empirical study, we provide theoretical arguments on abstract multi-step reasoning problems that highlight how autoregressive generations' performance can rapidly decay with increased task complexity.