When Does Tool Use Increase the Expressive Power of Finite-Precision Recurrent Models?
Modern sequence models are increasingly deployed as agents that interleave token generation with calls to external tools. We give an exact, architecture-level account of when such tool access increases computational expressivity. We model any fixed finite-precision recurrent sequence model, including finite-precision state-space models (SSMs) with $B$ bits of internal state, as a deterministic finite-state controller interacting with an oracle through a finite command/observation interface. Our results form a sharp dichotomy. First, tools that are themselves finite-state add essentially nothin
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
- Linked via arxiv authorNikola Zubić →
When Does Tool Use Increase the Expressive Power of Finite-Precision Recurrent Models?
- Linked via arxiv authorQian Li →
When Does Tool Use Increase the Expressive Power of Finite-Precision Recurrent Models?
- Linked via arxiv authorYuyi Wang →
When Does Tool Use Increase the Expressive Power of Finite-Precision Recurrent Models?
- Linked via arxiv authorDavide Scaramuzza →
When Does Tool Use Increase the Expressive Power of Finite-Precision Recurrent Models?