About ICLR
Representation learning and deep learning broadly: optimisation, generative models, reinforcement learning, theory, and an increasing share of work on large models. Reviews and discussions are public on OpenReview, which makes it unusually easy to see what reviewers pushed on.
In Singapore. ICLR 2025 was held at Singapore EXPO.
What Meridian does at ICLR
- Before: a Venue Report (Meridian Pro) on the orals, themes and authors to watch.
- During: recordings with authors near the venue, and a Notes item each day.
- After: the award and oral papers in Briefs, and invitations to Singapore Sessions.
Season one themes that draw on ICLR
- How models learn to reason
- Alignment and safety research
Put a ICLR paper forward
Authors of accepted main-track papers can apply, and anyone can nominate. Selection is editorial and never for sale.
Sources
Reviewed 26 October 2026. Rules and fees change; confirm with the authority before relying on them.