APOLLO: An Optimized Training Approach for Long-form Numerical Reasoning

arXiv:2212.07249v3 Announce Type: replace-cross
Abstract: Long-form numerical reasoning in financial analysis aims to generate a reasoning program to calculate the correct answer for a given question. Previous work followed a retriever-generator framework, where the retriever selects key facts from a long-form document, and the generator generates a reasoning program based on retrieved facts. However, they treated all facts equally without considering the different contributions of facts with and without numbers. Meanwhile, the program consistency were ignored under supervised training, resulting …

abstract analysis apollo arxiv cs.cl cs.lg document facts financial form framework generate generator key numerical question reasoning training type work