References
Core Papers
- I-JEPA (arXiv:2301.08243) β Self-supervised learning with JEPA
- Image World Models (arXiv:2403.00504) β Action-conditioned prediction in latent space
- DreamerV3 (arXiv:2301.04104) β RSSM world model with discrete latents
- HRM (arXiv:2506.21734) β Hierarchical reasoning with two-timescale processing
- TTT for ARC (arXiv:2411.07279) β Test-time training achieving 53% on ARC-AGI-1
ARC Benchmark
- ARC-AGI-3 (arXiv:2603.24621) β Competition specification
- ARC-AGI-2 (arXiv:2505.11831)
- ARC Prize 2025 (arXiv:2601.10904) β SOTA overview
- ARCLE (arXiv:2407.20806) β RL environment
- ARCTraj (arXiv:2511.11079) β Human trajectories
- GPAR (arXiv:2401.07426) β Planning for ARC
Competition
- ARC-AGI-3 on Kaggle
- Prize pool: $850,000
- Deadline: November 2, 2026