Skip to content
Snippets Groups Projects
Commit baa10a40 authored by Maxime Morge's avatar Maxime Morge :construction_worker:
Browse files

PyGAAMAS: Add Anonymous ICTAI Paper

parent 13b9994a
No related branches found
No related tags found
No related merge requests found
......@@ -25,11 +25,14 @@ this study assesses the capabilities of models such as
\texttt{Llama3}~\cite{grattafiori24arxiv},
\texttt{Mistral-Small}~\cite{jiang24arxiv},
\texttt{DeepSeek-R1}~\cite{deepseekai25arxiv}, and
\texttt{Qwen3}~\cite{bai23arxiv}. While Morge~\cite{morge25paams} evaluates GAs
on economic rationality and strategic reasoning, we focus on their ability to
\texttt{Qwen3}~\cite{bai23arxiv}. %While Morge~\cite{morge25paams} evaluates GAs
%on economic rationality and strategic reasoning,
We focus on their ability to
make credible one-shot decisions, generate human-like strategies, adapt to their
environment, and coordinate in social interactions\footnote{All code, prompts,
and data traces are available in a public repository~\cite{pygaamas}.}.
and data traces will be available in a public repository.}.
%All code, prompts,
% and data traces are available in a public repository~\cite{pygaamas}.
%These capabilities are evaluated through a series of
%tightly controlled and theoretically well-understood games.
The contributions of this work are as follows:
......
No preview for this file type
......@@ -12,17 +12,24 @@
% use a multiple column layout for up to three different
% affiliations
\author{\IEEEauthorblockN{St\'ephane Bonnevay}
\IEEEauthorblockA{\textit{Lizeo, UCBL, CNRS, INSA Lyon, ERIC} \\
F-69007 Lyon, France \\
Stephane.Bonnevay@univ-lyon1.fr}
\and
\IEEEauthorblockN{Maxime Morge}
\IEEEauthorblockA{\textit{UCBL, CNRS, INSA Lyon, UMR 5205 LIRIS} \\
F-69622 Villeurbanne, France \\
Maxime.Morge@univ-lyon1.fr}
\author{\IEEEauthorblockN{Anonymous}
\IEEEauthorblockA{\textit{Affiliation} \\
Address\\
Mail}
}
% \author{\IEEEauthorblockN{St\'ephane Bonnevay}
% \IEEEauthorblockA{\textit{Lizeo, UCBL, CNRS, INSA Lyon, ERIC} \\
% F-69007 Lyon, France \\
% Stephane.Bonnevay@univ-lyon1.fr}
% \and
% \IEEEauthorblockN{Maxime Morge}
% \IEEEauthorblockA{\textit{UCBL, CNRS, INSA Lyon, UMR 5205 LIRIS} \\
% F-69622 Villeurbanne, France \\
% Maxime.Morge@univ-lyon1.fr}
% }
% make the title area
\maketitle
......
......@@ -71,11 +71,11 @@ failing, for instance, to adopt basic conventions such as alternation in the
Battle of the Sexes game. To address this, they propose prompting agents to
imagine possible actions and their consequences before deciding. However, this
conditional reasoning proves effective mainly for smaller models and may degrade
performance in larger ones due to added complexity~\cite{pygaamas}. While Akata
\textit{et al.} attribute these failures to limited predictive ability and a
tendency to rigidly favor preferred options, we argue that the most fundamental
cause is GAs' inability to incorporate their beliefs into the decision-making
process when selecting actions.
performance in larger ones due to added complexity. %~\cite{pygaamas}
While Akata \textit{et al.} attribute these failures to limited predictive
ability and a tendency to rigidly favor preferred options, we argue that the
most fundamental cause is GAs' inability to incorporate their beliefs into the
decision-making process when selecting actions.
% Akata \textit{et al.}~\cite{akata23arxiv} study the behavior of GAs playing
% finitely repeated games, confronting them with simple yet credible strategies or
% with other GAs. They identify a major behavioral flaw: GAs lack coordination. In
......@@ -105,13 +105,18 @@ Matching Pennies game, where opponent strategies are both minimal and credible.
Moreover, their study is not
reproducible: the code is unavailable, and the experiments rely exclusively on
proprietary LLMs. In contrast, we make our code, prompts,
and datasets openly available~\cite{pygaamas}, and we focus on open-weight
models that can run on standard hardware.
and datasets openly available, and we focus on open-weight
models that can run on standard hardware. %~\cite{pygaamas}
% , which required significant computational resources and resulted in
% substantial carbon costs. Furthermore, we reduce environmental impact by
% prompting LLMs to generate algorithmic strategies, as in~\cite{willis25arxiv},
% rather than issuing multiple one-shot queries.
While Morge~\cite{morge25paams} evaluates GAs
on economic rationality and strategic reasoning, we focus on their ability to
make credible one-shot decisions, generate human-like strategies, adapt to their
environment, and coordinate in social interactions.
Hua \textit{et al.}~\cite{hua24arxiv} show that GAs deviate from rationality as
game complexity increases, and highlight the role of communication in fostering
coordination. We find that while communication may boost short-term
......
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment