BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:STEAM Academy Seminar: Towards Effective and Efficient Reasoning L
 anguage Models
DTSTART:20260728T140000Z
DTEND:20260728T153000Z
DTSTAMP:20260807T152100Z
UID:indico-event-1696049@indico.cern.ch
CONTACT:steam.academy@cern.ch\;ep-seminars.colloquia@cern.ch
DESCRIPTION:Speakers: Maciej Besta (ETH Zurich)\n\nAbstract\nThe field of 
 natural language processing has witnessed significant progress in recent y
 ears\, with a notable focus on improving language models’ performance 
 through innovative reasoning techniques. Among these\, structure-enhance
 d reasoning has emerged as a very promising paradigm\, with designs such
  as Chain-of-Thought (CoT)\, Tree of Thoughts (ToT)\, and Graph of Thought
 s (GoT)\, in which the LLM reasoning is guided by a structure such as th
 at of a graph. The key idea and primary advantage of such a graph-based r
 easoning paradigm is the ability to model the information generated/used 
 by an LLM as an arbitrary graph\, where units of information ("LLM thought
 s") are vertices\, and edges correspond to dependencies between these vert
 ices. This approach enables combining arbitrary LLM thoughts into synergis
 tic outcomes\, distilling the essence of whole networks of thoughts\, or e
 nhancing thoughts using feedback loops. In this talk\, we overview this v
 ery recent field\, focusing on fundamental classes of harnessed structures
 \, the representations of these structures\, algorithms executed with thes
 e structures\, and relationships to other parts of the generative AI pipel
 ine such as knowledge bases\, databases\, verification pipelines\, and o
 thers. The overview will dive deeper into several representative developme
 nts related to prompting (Graph of Thoughts)\, AI assistants (Knowledge Gr
 aph of Thoughts)\, training of Reasoning Language Models\, and the under
 lying abstractions (Topologies of Reasoning). \nSpeaker's Bio:\n Maciej
  Besta leads research on large language models and broad graph computation
 s at the Scalable Parallel Computing Lab at ETH Zurich and at the ETH Futu
 re Computing Lab\; he also works on interconnects\, general sparse computi
 ng\, and others. Maciej published\, as the main leading author\, around 40
  papers at top conferences and journals. He won\, among others\, the IEEE 
 TCHPC Award for Excellence in High-Performance Computing Early Career (202
 4)\, the HiPEAC Tech Transfer Award (2024)\, the OlympusMons Award for con
 tributions to scalable storage systems (2024)\, the IEEE TCSC Award for Ex
 cellence in Scalable Computing Early Career (2023)\, ACM Research Highligh
 ts (2018)\, the ACM/IEEE-CS High-Performance Computing Fellowship (2015)\,
  the first Google Fellowship in Parallel Computing (2013)\, and the compet
 ition for the Best Student of Poland (2012). His doctoral dissertation on 
 irregular computations received awards from ETH (2021)\, IEEE (2021)\, SPE
 C (2022)\, and ACM (2022). Maciej also won Best Paper awards and nominatio
 ns at ACM/IEEE Supercomputing 2013\, 2014\, 2019 (for 2 different papers)\
 , 2022\, and 2023 (for 2 different papers)\; at ACM HPDC 2015 and 2016\, a
 nd others. Finally\, as an explorer working at the overlap of exploration 
 and science\, Maciej is Fellow in The Explorers Club (2022). More detailed
  information on: https://people.inf.ethz.ch/bestam/ .\n\nThis seminar is 
 part of the CERN STEAM Academy Seminar Series. \nNetworking cocktail will
  follow the seminar. \nWith the support of CERN's Next Generation Trigge
 rs Project.\n\n\n\nhttps://indico.cern.ch/event/1696049/
LOCATION:500/1-001 - Main Auditorium (CERN)
URL:https://indico.cern.ch/event/1696049/
END:VEVENT
END:VCALENDAR