From fd79cf2ede012566e2087026b52a8e5fe78133ef Mon Sep 17 00:00:00 2001
From: Maxime MORGE <maxime.morge@univ-lille.fr>
Date: Sun, 16 Feb 2025 15:21:50 +0100
Subject: [PATCH] LLM4AAMAS: entry description
---
README.md | 200 +++++++++++++++++++++++++++++++++++++++---------------
1 file changed, 146 insertions(+), 54 deletions(-)
diff --git a/README.md b/README.md
index 93585ae..f204cb0 100644
--- a/README.md
+++ b/README.md
@@ -10,75 +10,123 @@ to generative AAMAS. This list is a work in progress and will be regularly updat
## Artificial Intelligence
-- **[Intelligence artificielle : une approche moderne (4e édition)](https://hal.archives-ouvertes.fr/hal-04245057)**
- *Stuart Russell, Peter Norvig, Fabrice Popineau, Laurent Miclet, Claire Cadet (2021)*
- Publisher: Pearson France
+- Artificial Intelligence (AI) involves the analysis, design, implementation,
+ and optimization of methods to enable machines to reproduce or simulate human
+ intelligence.
-- **[Apprentissage artificiel - 3e édition : Deep learning, concepts et algorithmes](https://www.eyrolles.com/)**
- *Antoine Cornuéjols, Laurent Miclet, Vincent Barra (2018)*
- Publisher: Eyrolles
+ **[Intelligence artificielle : une approche moderne (4e
+ édition)](https://hal.archives-ouvertes.fr/hal-04245057)** Stuart Russell,
+ Peter Norvig, Fabrice Popineau, Laurent Miclet, Claire Cadet (2021)
+ Publisher: Pearson France
+
+- Machine learning aims to give machines the ability to improve their
+ performance in solving tasks.
+
+ **[Apprentissage artificiel - 3e édition : Deep learning, concepts et
+ algorithmes](https://www.eyrolles.com/)** *Antoine Cornuéjols, Laurent
+ Miclet, Vincent Barra (2018)* Publisher: Eyrolles
## Neural networks (RNN, Transformers)
-- **[Learning representations by back-propagating errors](https://doi.org/10.1038/323533a0)**
- *David E. Rumelhart, Geoffrey E. Hinton, Ronald J. Williams (1986)*
- Published in *Nature*
+- The back-propagation method adjusts the connection weights by propagating
+ errors backward from the output layer to the input layer, aiming to minimize
+ errors and achieve a classification as close as possible to the optimum.
-- **[ImageNet Classification with Deep Convolutional Neural Networks](https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks)**
- *Alex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton (2012)*
- Presented at *NeurIPS*
+ **[Learning representations by back-propagating
+ errors](https://doi.org/10.1038/323533a0)** *David E. Rumelhart, Geoffrey E.
+ Hinton, Ronald J. Williams (1986)* Published in *Nature*
+
+- This approach has halved the image classification error rate on the ImageNet dataset.
+
+ **[ImageNet Classification with Deep Convolutional Neural
+ Networks](https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks)**
+ Alex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton (2012) Presented at *NeurIPS*
## Large Language Models
-- **[A Survey of Large Language Models](https://arxiv.org/abs/2303.18223)**
- *Wayne Xin Zhao, Kun Zhou, Junyi Li, et al. (2024)*
- Published on *arXiv*
+- The literature review of the recent advances in LLMs shown that scaling can
+ largely improve the model capacity
+
+ **[A Survey of Large Language Models](https://arxiv.org/abs/2303.18223)**
+ Wayne Xin Zhao, Kun Zhou, Junyi Li, et al. (2024) Published on *arXiv*
+
+- Based on the planning and reasoning abilities of LLM, the paper consider
+ LLM-based multi-agent systems for complex problem-solving and world
+ simulation.
+
+ **[Large Language Model based Multi-Agents: A Survey of Progress and
+ Challenges](https://arxiv.org/abs/2402.01680)** Taicheng Guo et al. (2024)
+ Published on *arXiv* arXiv:2402.01680 [cs.CL]
-- **[Large Language Model based Multi-Agents: A Survey of Progress and
- Challenges](https://arxiv.org/abs/2402.01680)** *Taicheng Guo et al. (2024)*
- Published on *arXiv* arXiv:2402.01680 [cs.CL]
+- A framework for achieving strong natural language understanding with a single
+ task-agnostic model through generative pre-training and discriminative
+ fine-tuning.
-- **[Improving language understanding by generative
+ **[Improving language understanding by generative
pre-training](https://cdn.openai.com/research-covers/language-unsupervised/language_understanding_paper.pdf)**
- *Alec Radford, Karthik Narasimhan, Tim Salimans, Ilya Sutskever (2018)*
+ Alec Radford, Karthik Narasimhan, Tim Salimans, Ilya Sutskever (2018)
Published by OpenAI
-- **[BERT: Pre-training of Deep Bidirectional Transformers for Language
- Understanding](https://www.aclweb.org/anthology/N19-1423/)** *Jacob Devlin,
- Ming-Wei Chang, Kenton Lee, Kristina Toutanova (2019)* Presented at
- *NAACL-HLT*
+- A language model pre-trained on large unlabeled corpora.
-- **[Sequence to Sequence Learning with Neural
- Networks](https://arxiv.org/abs/1409.3215)** *Ilya Sutskever, Oriol Vinyals,
- Quoc V. Le (2014)* Published on *arXiv*
+ **[BERT: Pre-training of Deep Bidirectional Transformers for Language
+ Understanding](https://www.aclweb.org/anthology/N19-1423/)** *Jacob Devlin,
+ Ming-Wei Chang, Kenton Lee, Kristina Toutanova (2019)* Presented at
+ *NAACL-HLT*
-- **[Learning Phrase Representations using RNN Encoder-Decoder for Statistical
- Machine Translation](https://arxiv.org/abs/1406.1078)** *Kyunghyun Cho, Bart
- van Merrienboer, Caglar Gulcehre, et al. (2014)* Published on *arXiv*
+- Recurrent Neural Networks (RNNs), specifically designed to process sequential data,
+ can capture contextual relationships between elements of a text, known as
+ tokens.
+
+ **[Sequence to Sequence Learning with Neural
+ Networks](https://arxiv.org/abs/1409.3215)** *Ilya Sutskever, Oriol Vinyals,
+ Quoc V. Le (2014)* Published on *arXiv*
+
+- The flexibility of RNN allows for the alignment of contextual representations,
+ thus overcoming the limitations of word-for-word translation.
+
+ **[Learning Phrase Representations using RNN Encoder-Decoder for Statistical
+ Machine Translation](https://arxiv.org/abs/1406.1078)** *Kyunghyun Cho,
+ Bartvan Merrienboer, Caglar Gulcehre, et al. (2014)* Published on *arXiv*
## Tuning
### Instruction tuning
-- **[LoRA: Low-Rank Adaptation of Large Language Models](https://arxiv.org/abs/2106.09685)**
- *Edward J. Hu, Yelong Shen, Phillip Wallis, et al. (2021)*
- Published on *arXiv*
+- The fine-tuning of a pre-trained language model requires significantly fewer
+ data and computational resources, especially when parameter-efficient
+ approaches such as Low-Rank Adaptation (LoRA) are used.
-- **[Language Models are Few-Shot
- Learners](https://papers.nips.cc/paper/2020/file/fc2c7f9a3f3f86cde5d8ad2c7f7e57b2-Paper.pdf)**
- *Tom Brown, Benjamin Mann, Nick Ryder, et al. (2020)* Presented at *NeurIPS*
+ **[LoRA: Low-Rank Adaptation of Large Language
+ Models](https://arxiv.org/abs/2106.09685)** Edward J. Hu, Yelong Shen,
+ Phillip Wallis, et al. (2021)* Published on *arXiv*
+
+- The apparent mastery of textual understanding by LLMs closely resembles human
+ performance.
+
+ **[Language Models are Few-Shot
+ Learners](https://papers.nips.cc/paper/2020/file/fc2c7f9a3f3f86cde5d8ad2c7f7e57b2-Paper.pdf)**
+ Tom Brown, Benjamin Mann, Nick Ryder, et al. (2020)* Presented at *NeurIPS*
### Alignement tuning
-- **[Training language models to follow instructions with human
+- Instruction tuning aims to bridge the gap between the model’s original
+ objective — generating text — and user expectations, where users want the
+ model to follow their instructions and perform specific tasks.
+
+ **[Training language models to follow instructions with human
feedback](https://papers.nips.cc/paper/2022/hash/17f4c5f98073d1fb95f7e53f5c7fdb64-Abstract.html)**
*Long Ouyang, Jeffrey Wu, Xu Jiang, et al. (2022)* Presented at *NeurIPS*
-- [Strong and weak alignment of large language models with human
- value](https://doi.org/10.1038/s41598-024-70031-3). Khamassi, M., Nahon, M. &
- Chatila, R. *Sci Rep* **14**, 19399 (2024).
+- Strong alignment requires cognitive abilities such as understanding and
+ reasoning about agents’ intentions and their ability to causally produce
+ desired effects.
+
+ **[Strong and weak alignment of large language models with human
+ value](https://doi.org/10.1038/s41598-024-70031-3)** Khamassi, M., Nahon, M.
+ & Chatila, R. *Sci Rep** **14**, 19399 (2024).
## Existing LLMs
@@ -130,6 +178,9 @@ Many models are available at the following URLs:
### ICL
+In-context learning involves providing the model with specific information
+without requiring additional training.
+
- **A Survey on In-context Learning** *Qingxiu Dong, Lei Li, Damai Dai, Ce
Zheng, Jingyuan Ma, Rui Li, Heming Xia, Jingjing Xu, Zhiyong Wu, Baobao Chang,
Xu Sun, Lei Li, Zhifang Sui (2024)* Presented at the *Conference on Empirical
@@ -138,14 +189,22 @@ Many models are available at the following URLs:
### CoT
+Chain-of-thought is a prompting strategy that, instead of being limited to
+input-output pairs, incorporates intermediate reasoning steps that serve as a
+link between the inputs and the output.
+
- **[Chain-of-Thought Prompting Elicits Reasoning in Large Language
Models](https://papers.nips.cc/paper/52604-chain-of-thought-prompting-elicits-reasoning-in-large-language-models)**
*Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, et al. (2022)*
Presented at *NeurIPS*
-
### RAG
+Retrieval-Augmented Generation (RAG) is a prompting strategy that involves
+integrating relevant information from external data sources into the
+instructions to enhance the model’s responses using specific and/or recent
+knowledge.
+
- **[Retrieval-Augmented Generation for Large Language Models: A
Survey](https://arxiv.org/abs/2312.10997)** *Yunfan Gao, Yun Xiong, Xinyu Gao,
Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yi Dai, Jiawei Sun, Meng Wang, Haofen Wang
@@ -153,11 +212,22 @@ Many models are available at the following URLs:
## Generative Autonomous Agents
-- **[A Survey on Large Language Model Based Autonomous Agents](https://arxiv.org/abs/2308.11432)** Lei Wang, Chen
- Ma, Xueyang Feng, Zeyu Zhang, Hao Yang, Jingsen Zhang, Zhiyuan Chen, Jiakai
- Tang, Xu Chen, Yankai Lin, Wayne Xin Zhao, Zhewei Wei, Jirong Wen (2024)*
- Published in *Frontiers of Computer Science* (Volume 18, Issue 6, Pages
- 186345) Publisher: Springer
+Leveraging the commonsense knowledge integrated into LLMs represents a promising
+solution to equip autonomous agents with the capabilities necessary to adapt to
+new tasks, while reducing reliance on knowledge engineering or trial-and-error
+learning.
+
+- **[A Survey on Large Language Model Based Autonomous
+ Agents](https://arxiv.org/abs/2308.11432)** Lei Wang, Chen Ma, Xueyang Feng,
+ Zeyu Zhang, Hao Yang, Jingsen Zhang, Zhiyuan Chen, Jiakai Tang, Xu Chen,
+ Yankai Lin, Wayne Xin Zhao, Zhewei Wei, Jirong Wen (2024) Published in
+ *Frontiers of Computer Science* (Volume 18, Issue 6, Pages 186345)
+ Publisher: Springer
+
+Multiple works aim to equip LLMs with the ability to use external tools, such as
+a calculator, a calendar, a DBMS, a code interpreter, a search engine, a machine
+translation tool, a question-answering system, or an AI tool.
+
- **[HuggingGPT: Solving AI Tasks with ChatGPT and its Friends in Hugging
Face](https://papers.nips.cc/paper/2023/hash/38154-hugginggpt-solving-ai-tasks-with-chatgpt-and-its-friends-in-hugging-face.pdf)**
@@ -169,25 +239,33 @@ Many models are available at the following URLs:
*Timo Schick, Jane Dwivedi-Yu, Roberto Dessi, Roberta Raileanu, et al. (2023)*
Presented at *NeurIPS*
+
+To react autonomously in an environment, a generative agent must interpret its
+perceptions (e.g., a user request) based on the knowledge stored in its memory,
+reason, and plan actions. It must execute the plan step by step with the help of
+tools and refine the plan based on feedback from the environment.
+
- **[Cognitive Architectures for Language Agents](https://arxiv.org/abs/2309.02427)**
*Theodore R. Sumers, Shunyu Yao, Karthik Narasimhan, Thomas L. Griffiths (2024)*
Published on *arXiv*
-- **[Large language models as simulated economic agents: What can we learn from
+LLMs have the ability to emulate a real human in certain experiments in
+experimental economics or social psychology.
+
+- **[Large language models as simulated economic agents: What can we learn from
homo silicus?](https://www.nber.org/papers/w31122)** Horton, J. J. (2023).
- National Bureau of Economic Research.
+ National Bureau of Economic Research.
+
+AGENTBENCH is a systematically designed multi-dimensional evolving benchmark for
+evaluating LLMs as agents which measures a significant performance gap between
+these top-tier models and their OSS competitors.
- ***[AgentBench: Evaluating LLMs as
Agents](https://openreview.net/forum?id=zAdUB0aCTQ)**. Xiao Li et al. Poster.
Proc. of 12th International Conference on Learning Representations (ICLR),
Vienna, Austria, May 7-11, 2024.
- AGENTBENCH a systematically designed multi-dimensional evolving benchmark
- for evaluating LLMs as agents which measure a significant performance gap
- between these top-tier models and their OSS competitors.
-
-
### Generative Autonomous Agents on the shelf
- [LangChain](https://www.langchain.com) is an open-source framework for
@@ -207,6 +285,10 @@ Many models are available at the following URLs:
## Generative MAS
+LLMs can simulate realistic perceptions, reasoning, and decision-making, react
+adaptively to environments without predefined explicit instructions by adjusting
+their responses through contextual learning mechanisms, autonomously generate
+objectives, and interact and communicate in natural language.
- **[Large language models empowered agent-based modeling and simulation: A
survey and perspectives](https://doi.org/10.1057/s41599-024-01235-9)** **Chen
@@ -216,6 +298,9 @@ Many models are available at the following URLs:
[repository](https://github.com/tsinghua-fib-lab/LLM-Agent-Based-Modeling-and-Simulation).
+Simulacra studies the emergent social behaviors of a generative multi-agent
+simulation in an environment inspired by The Sims.
+
- **[Social Simulacra: Creating Populated Prototypes for Social Computing
Systems](https://dl.acm.org/doi/10.1145/3526110.3545617)** *Joon Sung Park,
Lindsay Popowski, Carrie Cai, Meredith Ringel Morris, Percy Liang, Michael S.
@@ -232,17 +317,24 @@ Many models are available at the following URLs:
Symposium on User Interface Software and Technology* Articleno: 2, Pages: 22,
Location: San Francisco, CA, USA, Series: UIST '23
+AGENTVERSE is a general multi-agent framework that simulates problem-solving
+procedures of human groups.
+
- **[Agentverse: Facilitating multi-agent collaboration and exploring emergent
behaviors](https://openreview.net/forum?id=HywBMyh6JGR)** *Weize Chen, Yusheng
Su, Jingwei Zuo, Cheng Yang, Chenfei Yuan, Chi-Min Chan, Heyang Yu, Yaxi Lu,
Yi-Hsin Hung, Chen Qian, et al. (2023)* Published in *The Twelfth
International Conference on Learning Representations (ICLR 2023)*
+An open-source platform to simulate a human society.
+
- **[Training socially aligned language models on simulated social
interactions](https://arxiv.org/abs/2305.16960)** *Ruibo Liu, Ruixin Yang,
Chenyan Jia, Ge Zhang, Denny Zhou, Andrew M. Dai, Diyi Yang, Soroush Vosoughi
(2023)* Published on *arXiv* arXiv:2305.16960
+A simulation of the propagation processes in a social network.
+
- **[S3: Social-network Simulation System with Large Language Model-Empowered
Agents](https://arxiv.org/abs/2307.14984)** *Chen Gao, Xiaochong Lan, Zhihong
Lu, Jinzhu Mao, Jinghua Piao, Huandong Wang, Depeng Jin, Yong Li (2023)*
@@ -271,7 +363,7 @@ Many models are available at the following URLs:
where a lead Orchestrator agent is responsible for high-level planning,
directing other agents and tracking task progress.
-- [CrewAI](https://github.com/crewAIInc/crewAI) combines LLM-based agent with precise control flow.
+- [CrewAI](https://github.com/crewAIInc/crewAI) combines LLM-based agent with precise control flow.
- [Agno](https://github.com/agno-agi/agno) is a lightweight framework for
building generative multi-agent systems with workflows.
--
GitLab