Imperial lab prototypes latent tokenization and learnable adaptive-size memory
PontiEdoardo · x · 2026-09-11
Edoardo Ponti elaborates on his lab's research: they have prototyped latent tokenization in decoders and learnable, adaptive-size memory. Next steps are scaling these methods up and improving them, while exploring applications to inference-time hyper-scaling and long-horizon world modeling. The post is part of his AToM project recruiting thread; prototypes exist as Qwen3-8B-DMS-8x and Bolmo-7B.
More from Research
- HABIT-Bench: Best agent memory system scores just 33.61% at inferring user habits — SJTU · 2026-10-28
- MatAnyone-E brings 7.4x faster 4K video matting via sparse memory propagation — mmlab-ntu · 2026-10-28
- CodeCap-Bench: executable code beats natural language as video captions — Jiafu Tang · 2026-10-28
- NVIDIA's ThinkWorld learns robot skills from unlabeled human videos via chain-of-dynamics reasoning — nvidia · 2026-10-28
- Omni-InteractScope benchmarks proactive interaction in continuous omnimodal streams — Qwen · 2026-10-28
- Memory Tax study: dedicated agent memory systems often lose to naive RAG — UIUC-CS · 2026-10-28