BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:CECAM workshop: "L2M3: Large language models for materials\, molec
 ules and beyond"
DTSTART;VALUE=DATE:20240709
DTSTAMP:20260408T122124Z
UID:491b22a0f1c3428728d611c20ce855ef31ffc429f275ad19b505e135
CATEGORIES:Conferences - Seminars
DESCRIPTION:You can apply to participate and find all the relevant informa
 tion (speakers\, abstracts\, program\,...) on the event website: https://
 www.cecam.org/workshop-details/l2m3-large-language-models-for-materials-mo
 lecules-and-beyond-1291\n\n*** REGISTRATION DEADLINE *** : 3rd May 2024
 \n\nDescription\nLarge language models (LLMs) have significantly impacted 
 various scientific fields. This has led to special issues on AI in science
  being published by major scientific journals (e.g.\, Science 381 (6654) 2
 023). Our organizing committee and participants have played a crucial role
  in this movement by exploring the applications of LLMs in chemistry and m
 aterials science and contributing to the development of open-source soluti
 ons.\nFor example\, it has been demonstrated that LLMs can be fine-tuned t
 o achieve impressive performance on chemistry and materials science benchm
 arks [1–4]. Some participants have even provided LLMs with access to ext
 ernal tools like Google Search and cloud robotics\, enabling automated che
 mical synthesis. However\, this has raised safety concerns [5\,6].\nDespit
 e the rapid advances and attention to this field\, a fundamental question 
 remains: "What is hype\, and what is reality?" In a recent hackathon [7]\,
  we organized over 150 participants to build prototypes to better understa
 nd the potential applications of LLMs in chemistry and materials science. 
 This collaborative effort has brought forth several open questions that re
 quire intense collaboration across the community.\n\n	What are the safety 
 and dual-use concerns? How can we assess and mitigate them? Some prominen
 t figures have raised serious warnings about the potentially devastatin
 g impacts of such models\, while others have dismissed these concerns as 
 exaggerated.\n	 \n	How should we approach the use of LLMs in science\, pa
 rticularly in chemistry and materials science? There are several challeng
 es associated with using LLMs in a scientific setting. Many powerful model
 s\, such as GPT-4\, have been trained by for-profit companies on proprieta
 ry data\, making it difficult to evaluate them scientifically. Additionall
 y\, the evolving nature of these systems and the lack of systematic evalua
 tions pose further obstacles. For example\, one of the biggest benchmark s
 uites for LLMs\, BIG-bench (maintained by Google)\, contains only two (su
 perficial) chemistry tests.\n	Furthermore\, the role of academic research 
 is being questioned due to the limited access to computational resources\,
  which are predominantly available to a few industrial players.\n	 \n	How
  can we maximize the benefits of these models? What does our community re
 quire to leverage these advancements effectively? Most applications of LLM
 s in chemistry and materials science are still in the prototype or demo st
 age. There is no consensus on the most promising applications in the short
 \, medium\, and long term. Moreover\, there is a lack of agreement on the 
 necessary changes in science governance\, safety measures\, and education 
 to facilitate progress in these areas.\n\nThe proposed CECAM workshop aims
  to bring together academia\, industry\, and non-profits. Our goal is to d
 iscuss future directions\, create a roadmap\, develop new benchmarks and e
 valuations\, and establish a framework for ongoing collaboration.\n \n\nR
 eference\n[1] A. White\, Nat. Rev. Chem.\, 7\, 457-458 (2023)\n[2] G. Hoc
 ky\, A. White\, Digital Discovery\, 1\, 79-83 (2022)\n[3] A. White\, G. H
 ocky\, H. Gandhi\, M. Ansari\, S. Cox\, G. Wellawatte\, S. Sasmal\, Z. Yan
 g\, K. Liu\, Y. Singh\, W. Peña Ccoa\, Digital Discovery\, 2\, 368-376 (
 2023)\n[4] K. Jablonka\, P. Schwaller\, A. Ortega-Guerrero\, B. Smit\, Is 
 GPT all you need for low-data discovery in chemistry?\, 2023\n[5] A. M. Br
 an\, S. Cox\, A. D. White\, and P. Schwaller\, “ChemCrow: Augmenting lar
 ge-language models with chemistry tools”\, arXiv e-prints\, 2023. doi:10
 .48550/arXiv.2304.05376.\n[6] D. A. Boiko\, R. MacKnight\, and G. Gomes\, 
 “Emergent autonomous scientific research capabilities of large language 
 models”\, arXiv e-prints\, 2023. doi:10.48550/arXiv.2304.05332.\n[7] K. 
 M\, Jablonka\, et al. “14 Examples of How LLMs Can Transform Materials S
 cience and Chemistry: A Reflection on a Large Language Model Hackathon”\
 , arXiv e-prints\, 2023. doi:10.48550/arXiv.2306.06283.
LOCATION:BCH 2103 https://plan.epfl.ch/?room==BCH%202103
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR