ARCIDUCA - End of Project Workshop
Queen Mary University, 8th July 2025
Embedded Files
Location
Location
This will be a hybrid event - participation will be possible both in person and online - but the size of the room is limited, so please RSVP to Juexi Shao if you would like to attend in person. Also let us know if you want to attend online so we can add you to the list of invited speakers on Webex.
In-person location: Graduate Centre GC101
All times in British Summer Time.
Schedule
Schedule
10:00 Massimo Poesio (Queen Mary Uni / Uni Utrecht ): Welcome, Summary of the Project
11:00 Coffee break
11:30 Chris Madge (Queen Mary Uni): TBA
Abstract: TBA
12:15 Yujian Gan (Queen Mary Uni, UCL): TBA
Abstract: TBA
13:00 Lunch
14:00 Marc-Alexandre Côté (Microsoft Research): TALES: Text Adventure Learning Environment Suite
Abstract: Reasoning is an essential skill to enable Large Language Models (LLMs) to interact with the world. As tasks become more complex, they demand increasingly sophisticated and diverse reasoning capabilities for sequential decision-making, requiring structured reasoning over the context history to determine the next best action. We introduce TALES, a diverse collection of synthetic and human-written text-adventure games designed to challenge and evaluate diverse reasoning capabilities. We present results over a range of LLMs, open- and closed-weights, performing a qualitative analysis on the top performing models. Despite an impressive showing on synthetic games, even the top LLM-driven agents fail to achieve 20% on games designed for human enjoyment. Code and visualization of the experiments can be found at https://microsoft.github.io/tale-suite.
Bio
Marc-Alexandre Côté is a Principal Researcher at Microsoft Research - Montréal. Since joining Microsoft in 2017, his research focused on reinforcement learning (RL) and natural language understanding. Specifically, he is leading the Microsoft TextWorld project which aims at developing new RL agents capable of navigating and interacting with text environments (e.g. text-based adventure games). Such agents should possess skills like reading and understanding natural language text, information/knowledge-gathering, planning, and dealing with a combinatorially large action-space. Marc-Alexandre Côté is also one of the organizers for the Wordplay: When Language Meets Games workshop at EMNLP 2025.
15:00 Julian Hough (Swansea University): TBA
Abstract: TBA
15:45 Juexi Shao (Queen Mary Uni): TBA
Abstract: TBA
16:20 Tea Break
16:40 Juntao Yu (Queen Mary Uni): TBA
Abstract: TBA
17:20 Final discussion, End
Page updated
Google Sites
Report abuse