From Templates to Natural Language: Generalization Challenges in Instruction-Tuned LLMs for Spatial Reasoning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kranti, Chalamalasetti, Hakimov, Sherzod, Schlangen, David
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911108668850176
author Kranti, Chalamalasetti
Hakimov, Sherzod
Schlangen, David
author_facet Kranti, Chalamalasetti
Hakimov, Sherzod
Schlangen, David
contents Instruction-tuned large language models (LLMs) have shown strong performance on a variety of tasks; however, generalizing from synthetic to human-authored instructions in grounded environments remains a challenge for them. In this work, we study generalization challenges in spatial grounding tasks where models interpret and translate instructions for building object arrangements on a $2.5$D grid. We fine-tune LLMs using only synthetic instructions and evaluate their performance on a benchmark dataset containing both synthetic and human-written instructions. Our results reveal that while models generalize well on simple tasks, their performance degrades significantly on more complex tasks. We present a detailed error analysis of the gaps in instruction generalization.
format Preprint
id arxiv_https___arxiv_org_abs_2505_14425
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle From Templates to Natural Language: Generalization Challenges in Instruction-Tuned LLMs for Spatial Reasoning
Kranti, Chalamalasetti
Hakimov, Sherzod
Schlangen, David
Computation and Language
Instruction-tuned large language models (LLMs) have shown strong performance on a variety of tasks; however, generalizing from synthetic to human-authored instructions in grounded environments remains a challenge for them. In this work, we study generalization challenges in spatial grounding tasks where models interpret and translate instructions for building object arrangements on a $2.5$D grid. We fine-tune LLMs using only synthetic instructions and evaluate their performance on a benchmark dataset containing both synthetic and human-written instructions. Our results reveal that while models generalize well on simple tasks, their performance degrades significantly on more complex tasks. We present a detailed error analysis of the gaps in instruction generalization.
title From Templates to Natural Language: Generalization Challenges in Instruction-Tuned LLMs for Spatial Reasoning
topic Computation and Language
url https://arxiv.org/abs/2505.14425