Schedule
| Time | Event |
|---|---|
| 08:00–09:00 | Registration |
| 09:00–09:10 | Opening Remarks: Welcome address and workshop overview |
| 09:10–09:40 | Keynote Talk 1: Yuandong Tian, Meta (confirmed) |
| 09:40–10:10 | Keynote Talk 2: Lisa Li, Stanford & UW (confirmed) |
| 10:10–10:40 | Oral Presentations from submissions |
| 10:40–11:10 | Coffee Break |
| 11:10–11:40 | Keynote Talk 3: Samy Bengio, Apple (confirmed) |
| 11:40–12:30 | Poster Session 1 |
| 12:30–13:30 | Lunch Break |
| 13:30–14:00 | Keynote Talk 4: Beidi Chen, CMU (confirmed) |
| 14:00–14:30 | Keynote Talk 5: Tom Goldstein, UMD (confirmed) |
| 14:30–15:00 | Keynote Talk 6: Lingpeng Kong, HKU (confirmed) |
| 15:00–16:00 | Panel Discussion: What are the key challenges in enabling large language models to reason effectively in latent space? |
| 16:00–16:30 | Coffee Break |
| 16:30–17:20 | Poster Session 2 |
| 17:20–17:30 | Award Announcement and Closing Remarks |
Format
To encourage meaningful dialogue, each keynote and contributed talk will be followed by a moderated Q&A period (5-10 min), with discussion questions collected both in-person and online. Dedicated poster sessions and a final panel discussion will provide structured venues for audience engagement. Additionally, we will organize an open “lightning round” after lunch, where participants can share 1-minute research ideas, encouraging spontaneous and inclusive participation.
While the workshop is in-person, all talks and poster abstracts will be made publicly available via our website and YouTube channel within two weeks after the event. We will also host a moderated discussion board for remote participants to post comments and questions. This ensures that researchers facing visa or financial barriers can still meaningfully engage.