Compress to Remember: Learning Compact Memory via On-Policy Distillation for Long Video Generation
arXiv:2609.36364v1 Announce Type: new Abstract: Standard video generators do not natively compact historical context into reusable memory tokens. As generation continues, the growing history makes it increasingly difficult to retain information from earlier frames due to long-context degradation. K…
Read the full story at arXiv cs.CV ↗
Timeline · 1 report
- 2026-09-30 04:00 · arXiv cs.CV
Compress to Remember: Learning Compact Memory via On-Policy Distillation for Long Video Generation