Slot-JEPA: Object-Structured Prediction Interfaces for Latent World Models
China666 · hf · 2026-10-21
Researchers from CAS, BAAI and UCAS propose Slot-JEPA, which decomposes dense JEPA patch representations into reusable semantic slots and spatial masks. This object-structured interface gives downstream predictors explicit entity correspondence over time, improving predictive modeling, planning, and transfer across visual tasks while avoiding redundant recovery of object structure.
More from Research
- Recursive generalization makes neurons 'clean' — agent swarms could crash alignment costs to 7 figures — daniel_mac8 · 2026-10-21
- OXR: 1.3B Open-Source Parser Tops Three Doc Benchmarks at 3.05x GLM-OCR Throughput — rohanpaul_ai · 2026-10-21
- Python 3.15 ships Tachyon profiler with 1ms sampling and live flame graphs — KhuyenTran16 · 2026-10-21
- Stanford's full CS336 LLM course is on YouTube — homework includes FlashAttention in Triton — techNmak · 2026-10-21
- PhysCanvas world model predicts vision and touch, lifting sim-real correlation to 0.89 — stepjamUK · 2026-10-21
- SCAMPER: a virtual rat with realistic motion for RL-based neuroscience experiments — tyrell_turing · 2026-10-21