Skip to content
NewA calmer, faster ZeroHour

We Query, Therefore We Compute: On Oracle Computation beyond the Machine, with an Application to Agents

Hugging Face daily papersby Kefan Liu, Fengning Ou, Yelin Luo ()1 min read3 · 1 read
AI summary · grok-4.7Summaries by Vani AI

A two-stack oracle machine unifies LLM workflows and agents, with ArchNights proposed on RISC-V.

The paper treats an LLM as an Oracle inside a two-stack pushdown automaton extended by one instruction that submits a whole stack as a query. A Turing-complete companion computation, called the Priestess, acts as the operating system of programs the Oracle runs, with Agent and Workflow as two placements of a task. ArchNights, an extended RISC-V ISA and Linux-style operating system, implements the machine; ArchNights-SE runs on gem5 and is planned to be open source.

Key facts
  • One instruction submits an entire stack as a query to an LLM oracle.
  • Two symmetry breakings place a task as either an Agent or a Workflow.
  • Answer-end synchronization is used to model caching and scheduling.
  • ArchNights-SE runs on gem5 and is intended to be open source.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

1 min read at huggingface.co

Agentic systems use large language models (LLMs) to carry out concrete tasks. Prior work often borrows abstractions such as scheduling, caching or isolation piecemeal from operating systems, so the mechanisms it builds share little common ground, and the shared view of the two forms of agentic system, Workflows and Agents, is limited. We construct an abstract machine that provides both.

Hugging Face daily papers

By Kefan Liu, Fengning Ou, Yelin Luo · huggingface.co

Read the full article at Hugging Face daily papers

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.