Saved in:
Bibliographic Details
Main Authors: Rakotonirina, Nathanaël Carraz, Hamdy, Mohammed, Campos, Jon Ander, Weber, Lucas, Testoni, Alberto, Fadaee, Marzieh, Pezzelle, Sandro, Del Tredici, Marco
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2502.13791
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910992808542208
author Rakotonirina, Nathanaël Carraz
Hamdy, Mohammed
Campos, Jon Ander
Weber, Lucas
Testoni, Alberto
Fadaee, Marzieh
Pezzelle, Sandro
Del Tredici, Marco
author_facet Rakotonirina, Nathanaël Carraz
Hamdy, Mohammed
Campos, Jon Ander
Weber, Lucas
Testoni, Alberto
Fadaee, Marzieh
Pezzelle, Sandro
Del Tredici, Marco
contents Large Language Models (LLMs) are increasingly used in working environments for a wide range of tasks, excelling at solving individual problems in isolation. However, are they also able to effectively collaborate over long-term interactions? To investigate this, we introduce MemoryCode, a synthetic multi-session dataset designed to test LLMs' ability to track and execute simple coding instructions amid irrelevant information, simulating a realistic setting. While all the models we tested handle isolated instructions well, even the performance of state-of-the-art models like GPT-4o deteriorates when instructions are spread across sessions. Our analysis suggests this is due to their failure to retrieve and integrate information over long instruction chains. Our results highlight a fundamental limitation of current LLMs, restricting their ability to collaborate effectively in long interactions.
format Preprint
id arxiv_https___arxiv_org_abs_2502_13791
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle From Tools to Teammates: Evaluating LLMs in Multi-Session Coding Interactions
Rakotonirina, Nathanaël Carraz
Hamdy, Mohammed
Campos, Jon Ander
Weber, Lucas
Testoni, Alberto
Fadaee, Marzieh
Pezzelle, Sandro
Del Tredici, Marco
Computation and Language
Large Language Models (LLMs) are increasingly used in working environments for a wide range of tasks, excelling at solving individual problems in isolation. However, are they also able to effectively collaborate over long-term interactions? To investigate this, we introduce MemoryCode, a synthetic multi-session dataset designed to test LLMs' ability to track and execute simple coding instructions amid irrelevant information, simulating a realistic setting. While all the models we tested handle isolated instructions well, even the performance of state-of-the-art models like GPT-4o deteriorates when instructions are spread across sessions. Our analysis suggests this is due to their failure to retrieve and integrate information over long instruction chains. Our results highlight a fundamental limitation of current LLMs, restricting their ability to collaborate effectively in long interactions.
title From Tools to Teammates: Evaluating LLMs in Multi-Session Coding Interactions
topic Computation and Language
url https://arxiv.org/abs/2502.13791