The workshop will be held in hybrid mode in Vienna, Austria, as part of the ICLR 2024 conference, on May 11, 2024. Check out ICLR.cc for in-person/virtual workshop registeration, and our official ICLR workshop page to view the live-stream for virtual registrants.
We aim to showcase the latest in AGI research, delve into breakthroughs from LLMs and foundation models, and assess the current gaps and challenges towards AGI. With insights from renowned AI experts, the workshop seeks to spark new ideas, foster discussions, and compile a comprehensive summary to guide the broader AI and ML community. We have confirmed with many renowned researchers to give the keynote talks. Schedule below is shown in Vienna time zone (CEST, GMT+2)
- 8:20 AM - 8:30 AM: Poster setup for authors (optional)
- 8:30 AM - 8:45 AM: Opening remarks: Jiaxuan You
- 8:45 AM - 9:45 AM: Oral presentations from submissions
- On the Shape of Brainscores for Large Language Models, presenter: Jingkai Li
- Do LLM Agents Have Regret? A Case Study in Online Learning and Games, presenter: Chanwoo Park
- Understanding the Reasoning Ability of Language Models From the Perspective of Reasoning Paths Aggregation, presenter: Xinyi Wang
- AI Alignment with Changing and Influenceable Reward Functions, presenter: Micah Carroll
- DeFT: Flash Tree-attention with IO-Awareness for Efficient Tree-search-based LLM Inference, presenter: Jinwei Yao
- Pitfalls of Next-token prediction, presenter: Vaishnavh Nagarajan
- In-person oral presentation starts around 9:15am
- Generative Representational Instruction Tuning (Best Paper Award, in person), presenter: Tao Yu
- Self-Alignment of Large Language Models via Monopolylogue-based Social Scene Simulation (in person), presenter: Rui Ye
- SceneCraft: An LLM Agent for Synthesizing 3D Scene as Blender Code (in person), presenter: Ziniu Hu / Alireza Fathi
- A language model’s guide through latent space (in person), presenter: Dimitri von Rutte
- 9:45 AM - 10:00 AM: Invited talk: UK Government AI Safety Institute
- 10:00 AM - 10:30 AM: Keynote talk: Oriol Vinyals, Google Deepmind: From AI to AGI
- 10:30 AM - 10:45 AM: Coffee break and social
- 10:45 AM - 11:30 AM: Keynote talk: Yejin Choi, University of Washington
- 11:30 AM - 12:00 PM: Keynote talk: Andrew Gordon Wilson, New York University
- 12:00 PM - 12:30 PM: Keynote talk: Song Han, MIT
- 12:30 PM - 1:30 PM: Lunch break and social: Another opportunity for poster setup
- 1:30 PM - 2:30 PM: Keynote talk: Yoshua Bengio, MILA: Towards deep learning for amortized inference of AGI-strength safety guarantees
- 2:35 PM - 4:00 PM: Panel discussions: How far are we from AGI? Panelists: Yoshua Bengio, Yejin Choi, Andrew Gordon Wilson, Song Han, Dawn Song; Moderators: Jiaxuan You, Ge Liu
- 4:00 PM - 5:00 PM: Poster session and social
Should you have any questions, please reach out to us via email:
agiworkshop@googlegroups.com