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

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

14 Dec 2022arXiv:2212.07249archive 2025-07-28

Jiashuo Sun, Hang Zhang, Chen Lin, Xiangdong Su, Yeyun Gong, Jian Guo

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 in lower training accuracy and diversity. To solve these problems, we proposed APOLLO to improve the long-form numerical reasoning framework. For the retriever, we adopt a number-aware negative sampling strategy to enable the retriever to be more discriminative on key numerical facts. For the generator, we design consistency-based reinforcement learning and target program augmentation strategy based on the consistency of program execution results. Experimental results on the FinQA and ConvFinQA leaderboard verify the effectiveness of our proposed method, achieving the new state-of-the-art.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

gasolsun36/apollo officialmentioned in papermentioned on GitHubpytorch report
gasolsun36/dynamicrag mentioned on GitHubpytorch report
gasolsun36/iter-cot mentioned on GitHub report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Conversational Question AnsweringDiversityFinancial AnalysisFormQuestion Answering

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Conversational Question Answering ConvFinQA APOLLO Execution Accuracy 78.76 #1 of 2 Archive leaderboard report
Conversational Question Answering ConvFinQA APOLLO Program Accuracy 77.19 #1 of 2 Archive leaderboard report
Question Answering FinQA APOLLO Execution Accuracy 71.07 #1 of 6 Archive leaderboard report
Question Answering FinQA APOLLO Program Accuracy 68.94 #1 of 6 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Apollo

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections