Published on:

14 October 2023

Primary Category:

Computation and Language

Paper Authors:

Wenqi Zhang,

Yongliang Shen,

Qingpeng Nong,

Zeqi Tan,

Yanna Ma,

Weiming Lu

•

Proposes a tree-structured decoding strategy for math equation generation

•

Decodes equations efficiently in parallel steps based on tree structure

•

Achieves state-of-the-art performance on math word problem benchmarks

•

Reduces the number of decoding steps compared to prior methods

•

Provides a new perspective for structured decoding

Tree-structured decoding of math equations

This paper introduces a new tree-structured decoding strategy for generating mathematical equations from text descriptions. It can capture the inherent tree structure of equations and decode them efficiently in parallel steps.

Process-free mathematical reasoning via Monte Carlo Tree Search

Demystifying complex systems: A clear, approachable guide to analyzing intricate equations

Parallel decoding for faster language model inference

Navigating the Forest of Thought: How Language Models Can Deliberately Reason and Problem-Solve

Enabling LLMs to solve math problems with code

Multimodal reasoning for video question answering

No comments yet, be the first to start the conversation...

Sign up to comment on this paper