GPT Can Solve Mathematical Problems Without a Calculator

09/06/2023
by   Zhen Yang, et al.
0

Previous studies have typically assumed that large language models are unable to accurately perform arithmetic operations, particularly multiplication of >8 digits, and operations involving decimals and fractions, without the use of calculator tools. This paper aims to challenge this misconception. With sufficient training data, a 2 billion-parameter language model can accurately perform multi-digit arithmetic operations with almost 100 data leakage, significantly surpassing GPT-4 (whose multi-digit multiplication accuracy is only 4.3 GLM-10B on a dataset with additional multi-step arithmetic operations and math problems described in text, achieves similar performance to GPT-4 on a 5,000-samples Chinese math problem test set. Our code and data are public at https://github.com/THUDM/MathGLM.

READ FULL TEXT

page 15

page 16

page 18

research
04/04/2023

Neural Comprehension: Language Models with Compiled Neural Networks

Language models have achieved impressive results in natural language pro...
research
06/26/2023

Composing Parameter-Efficient Modules with Arithmetic Operations

As an efficient alternative to conventional full finetuning, parameter-e...
research
04/21/2023

Evaluating Transformer Language Models on Arithmetic Operations Using Number Decomposition

In recent years, Large Language Models such as GPT-3 showed remarkable c...
research
08/02/2023

Arithmetic with Language Models: from Memorization to Computation

A better understanding of the emergent computation and problem-solving c...
research
02/25/2021

Investigating the Limitations of Transformers with Simple Arithmetic Tasks

The ability to perform arithmetic tasks is a remarkable trait of human i...
research
05/09/2019

A multidimensional analog to the Burrows-Wheeler transform

We show how to perform multidimensional pattern matching over an n-dimen...
research
07/07/2023

Teaching Arithmetic to Small Transformers

Large language models like GPT-4 exhibit emergent capabilities across ge...

Please sign up or login with your details

Forgot password? Click here to reset