Date & Time:
10:00 - 14:00,
Monday 2 December, 2024
10:00 - 14:00,
Monday 2 December, 2024
Location:
1 St Giles
1 St Giles
About
The Oxford LLM Workshop (2 December 2024) aims to bring together researchers from across the University of Oxford who are exploring various aspects of large language models (LLMs). The event will feature two sessions of short talks (8+2 minutes each), providing an opportunity for participants to showcase their work and engage with others in the LLM research. We encourage all participants to prepare a short presentation to help show the breadth of research at Oxford. The event is organised by Dr Adam Mahdi and hosted at the Oxford Internet Institute (OII).
This event is fully booked.
Agenda
Speakers
TIME
TYPE
PRESENTER
TITLE
10.00
Tea/Coffee (15 min)
10.15
Welcome (5 min)
Adam Mahdi
10.20
Talk (8+2 min)
Adel Bibi & Alasdair Paren
Domain Certification in LLMs
10.30
Talk (8+2 min)
Scott Hale
LLMs, Narratives, and Misinformation Response
10.40
Talk (8+2 min)
Luc Rocher & Sofia Hafner
Research of the Synthetic Society
10.50
Talk (8+2 min)
Rachel Barnhard
Appearance-Based Discrimination in Politics
11.00
Talk (8+2 min)
Bernie Hogan
Sidestepping anthropomorphism: Engaging LLMs autoreferentially
11.10
Talk (8+2 min)
Lingyi Yang
Improving In-Context Learning with Small Language Model Ensembles
11.20
Talk (8+2 min)
Chris Russell
Careless Speech and Large Language Models
11.30
Talk (8+2 min)
Zihao Fu
Trustworthy Watermarking for Large Language Models
11.40
Tea/Coffee (20 min)
12.00
Talk (4+1 min)
Thomas Foster
Approaches for self-improving LLM reasoning
12.05
Talk (4+1 min)
Benjamin Walker
Theoretical Foundations of Deep Selective State-Space Models
12.10
Talk (4+1 min)
Yushi Yang
Decode safety fine-tuning through harmful feature reduction
12.15
Talk (4+1 min)
Harry Mayne
Controlling LLMs through their internal representations
12.20
Talk (4+2 min)
Junde Wu
Medical Graph RAG: Towards Safe Medical Large Language Model via Graph Retrieval-Augmented Generation
12.25
Talk (4+1 min)
Andrew Bean
LingOly: A Benchmark of Olympiad-Level Linguistic Reasoning Puzzles in Low-Resource and Extinct Languages
12.30
Talk (4+1 min)
Lukas Aichberger
Attacking Multimodal Agents with Malicious Image Patches
12.35
Talk (4+1 min)
Andrei Lupu
The Decrypto Benchmark for Multi-Agent Reasoning and Theory of Mind
12.40
Summary
Adam Mahdi
12.45
Lunch + Networking
Attendees
NAME
DEPARTMENT/RESEARCH GROUP
Eltayeb Ahmed
Engineering / FLAIR
Lukas Aichberger
Department of Computer Science / OATML / TVG
Andrew Bean
Oxford Internet Institute
Uljad Berdica
Engineering / FLAIR
Adel Bibi
Engineering Science
Piotr Blaszczyk
Cybersmart
Calvin Cheng
Oxford Internet Institute
Cornelius Emde
Department of Computer Science
Thomas Foster
Engineering / FLAIR
Zihao Fu
OII / Governance of Emerging Technology
Sofia Hafner
OII / Synthetic Society
Scott Hale
OII / Equitable Access to Quality Information (eaqilab)
Jude Khouja
OII / Equitable Access to Quality Information (eaqilab)
Hazel Kim
CS, TVG and OATML
Karolina Korgul
Oxford Internet Institute
Fangru Lin
Linguistics
Andrei Lupu
Engineering / FLAIR
Adam Mahdi
Oxford Internet Institute
Harry Mayne
Oxford Internet Institute
Robert McCraith
Engineering Science
Jessica Rodrigues
Oxford Internet Institute
Chris Russell
OII / Governance of Emerging Technology
Julia Sepúlveda
OII / Equitable Access to Quality Information (eaqilab)
Philip Torr
Engineering / TVG
Benjamin Walker
Mathematical Institute
Junde Wu
Engineering Department
Lingyi Yang
Mathematical Institute
William Gitta Lugoloobi
Oxford Internet Institute
Bernie Hogan
Oxford Internet Institute
Ryan Kearns
OII / HAI Lab
Rachel Bernhard
Politics and International Relations
Alasdair Paren
Engineering / TVG
Guy Parsons
Oxford Internet Institute
Luc Rocher
OII / Synthetic Society
Yushi Yang
Oxford Internet Institute