<div dir="auto">This is highly relevant to at least a few of us.<div dir="auto"><br></div><div dir="auto">Artur</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">---------- Forwarded message ---------<br>From: <strong class="gmail_sendername" dir="auto">Catherine Copetas</strong> <span dir="auto"><<a href="mailto:copetas@cs.cmu.edu">copetas@cs.cmu.edu</a>></span><br>Date: Fri, Nov 8, 2024, 2:29 PM<br>Subject: SCS Katayanagi Distinguished Lecture: Thursday, November 21 - 4:30 pm - GHC 4401<br>To: <<a href="mailto:scs-all@cs.cmu.edu">scs-all@cs.cmu.edu</a>><br></div><br><br><div dir="ltr"><div><i><br></i></div><div><i>Please join us for the next...</i></div><div><div><b><br></b></div><div><b><font size="4">SCS Katayanagi Distinguished Lecture</font></b></div><div>Thursday,<b> 21 November 2024</b></div><div><b>4:30 pm </b><br></div>Rashid Auditorium, <b>Gates Hillman 4401</b><br><div><br></div><div><i>as we welcome<br></i></div><div style="margin-left:40px"><i><br></i></div><div><div style="margin-left:40px"><b><font size="4"><a href="https://noambrown.github.io/" target="_blank" rel="noreferrer">NOAM BROWN</a></font></b><br>Research Scientist<br>OpenAI</div><div><div style="margin-left:40px"><h2 id="m_3314408787396912983gmail-headline2">Learning to Reason with LLMs</h2></div><div><div><div>Large
language models (LLMs) have demonstrated remarkable capabilities in
generating coherent text and completing various natural language tasks.
Nevertheless, their ability to perform complex, general reasoning has
remained limited. In this talk, I will describe OpenAI's new o1 model,
an LLM trained via reinforcement learning to generate a hidden chain of
thought before its response. We have found that the performance of o1
consistently improves with more reinforcement learning compute and with
more inference compute. o1 surpasses previous state-of-the-art models in
a variety of benchmarks that require reasoning, including mathematics
competitions, programming contests, and advanced science question sets. I
will discuss the implications of scaling this paradigm even further.</div><div><br></div><div>—</div><div><b><br></b></div><div><b><a href="https://noambrown.github.io/" rel="noopener noreferrer" target="_blank">Noam Brown</a></b>
is a research scientist at OpenAI investigating reasoning and
multi-agent AI. He co-created <a href="https://en.wikipedia.org/wiki/Libratus" target="_blank" rel="noreferrer">Libratus</a> and <a href="https://www.nytimes.com/2019/07/11/science/poker-robot-ai-artificial-intelligence.html" target="_blank" rel="noreferrer">Pluribus</a>, the first AIs to
defeat top humans in two-player no-limit poker and multiplayer no-limit
poker, respectively, and Cicero, the first AI to achieve human-level
performance in the natural language strategy game Diplomacy. He has
received the Marvin Minsky Medal for Outstanding Achievements in AI, was
named one of <i>MIT Tech Review's 35 Innovators Under 35</i>, and his work on
Pluribus was named by <i>Science</i> as one of the top 10 scientific
breakthroughs of 2019. Noam received his Ph.D. <i>(CS)</i> from Carnegie Mellon
University. </div><div><br></div><div><em><br></em></div><div><em><b>About the Lecture:</b> The Katayanagi Lectures recognize the best and
the brightest in the field of computer science and are presented by the
School of Computer Science at Carnegie Mellon University in cooperation
with the Tokyo University of Technology (TUT). The lectures recognize both senior and junior talent. The series were established through a
gift from Japanese entrepreneur and education advocate, Mr. Koh
Katayanagi, who founded TUT and other technical institutions in
Japan over many multiple decades.</em></div><div><em><br></em></div><div><a href="mailto:scs-dls@cs.cmu.edu" target="_blank" rel="noreferrer">Questions</a>?</div><div><br></div></div></div></div></div></div></div>
</div>