Saved in:
Bibliographic Details
Main Authors: Lim, Henry, Lim, Kwan Hui
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2510.17388
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917027510222848
author Lim, Henry
Lim, Kwan Hui
author_facet Lim, Henry
Lim, Kwan Hui
contents Instruction-tuned large language models (IT-LLMs) exhibit strong zero-shot reasoning, yet their ability to execute simple, self-contained instructions remains underexplored, despite this being foundational to complex instruction-following. We evaluate 20 IT-LLMs on modified MMLU and MMLU-Pro benchmarks, by systematically varying the format of option labels (alphabetic, numeric, Roman) while keeping their meaning identical under four paradigms, namely: (1) With explicit instructions, label changes cause large performance shifts (e.g., -30.45\% for Roman vs. numeric), revealing instruction-format bias. (2) Without instructions, performance drops further (up to -10.84\%) and label sensitivity intensifies, underscoring the role of explicit guidance. (3) When option contents are removed, models fail random-choice baselines except with numeric labels, suggesting weak adherence to atomic directives. (4) Three-shot exemplars yield no significant gains in robustness or fidelity, and generation analyses show persistent label errors, especially for non-numeric formats. Across model sizes, larger LLMs achieve higher accuracy but remain inconsistent in instruction adherence. These results expose the insufficiencies of current instruction-tuning paradigms and highlight the need for evaluation methods and training strategies that explicitly target atomic instruction-following.
format Preprint
id arxiv_https___arxiv_org_abs_2510_17388
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle The Atomic Instruction Gap: Instruction-Tuned LLMs Struggle with Simple, Self-Contained Directives
Lim, Henry
Lim, Kwan Hui
Computation and Language
Instruction-tuned large language models (IT-LLMs) exhibit strong zero-shot reasoning, yet their ability to execute simple, self-contained instructions remains underexplored, despite this being foundational to complex instruction-following. We evaluate 20 IT-LLMs on modified MMLU and MMLU-Pro benchmarks, by systematically varying the format of option labels (alphabetic, numeric, Roman) while keeping their meaning identical under four paradigms, namely: (1) With explicit instructions, label changes cause large performance shifts (e.g., -30.45\% for Roman vs. numeric), revealing instruction-format bias. (2) Without instructions, performance drops further (up to -10.84\%) and label sensitivity intensifies, underscoring the role of explicit guidance. (3) When option contents are removed, models fail random-choice baselines except with numeric labels, suggesting weak adherence to atomic directives. (4) Three-shot exemplars yield no significant gains in robustness or fidelity, and generation analyses show persistent label errors, especially for non-numeric formats. Across model sizes, larger LLMs achieve higher accuracy but remain inconsistent in instruction adherence. These results expose the insufficiencies of current instruction-tuning paradigms and highlight the need for evaluation methods and training strategies that explicitly target atomic instruction-following.
title The Atomic Instruction Gap: Instruction-Tuned LLMs Struggle with Simple, Self-Contained Directives
topic Computation and Language
url https://arxiv.org/abs/2510.17388