The role of positional encodings in the ARC benchmark

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Costa, Guilherme H. Bandeira, Freire, Miguel, Oliveira, Arlindo L.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916594730401792
author Costa, Guilherme H. Bandeira
Freire, Miguel
Oliveira, Arlindo L.
author_facet Costa, Guilherme H. Bandeira
Freire, Miguel
Oliveira, Arlindo L.
contents The Abstraction and Reasoning Corpus challenges AI systems to perform abstract reasoning with minimal training data, a task intuitive for humans but demanding for machine learning models. Using CodeT5+ as a case study, we demonstrate how limitations in positional encoding hinder reasoning and impact performance. This work further examines the role of positional encoding across transformer architectures, highlighting its critical influence on models of varying sizes and configurations. Comparing several strategies, we find that while 2D positional encoding and Rotary Position Embedding offer competitive performance, 2D encoding excels in data-constrained scenarios, emphasizing its effectiveness for ARC tasks
format Preprint
id arxiv_https___arxiv_org_abs_2502_00174
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle The role of positional encodings in the ARC benchmark
Costa, Guilherme H. Bandeira
Freire, Miguel
Oliveira, Arlindo L.
Artificial Intelligence
Machine Learning
The Abstraction and Reasoning Corpus challenges AI systems to perform abstract reasoning with minimal training data, a task intuitive for humans but demanding for machine learning models. Using CodeT5+ as a case study, we demonstrate how limitations in positional encoding hinder reasoning and impact performance. This work further examines the role of positional encoding across transformer architectures, highlighting its critical influence on models of varying sizes and configurations. Comparing several strategies, we find that while 2D positional encoding and Rotary Position Embedding offer competitive performance, 2D encoding excels in data-constrained scenarios, emphasizing its effectiveness for ARC tasks
title The role of positional encodings in the ARC benchmark
topic Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2502.00174