---
title: "CIRCE Divines LLM Task Complexity from Constraint Structure"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2610.41089v1"
permalink: "https://feed.astralor.com/items/itm_67da48608e1cf00a94477de9"
published: "2026-10-30T00:08:01.000Z"
score: 53
channel: "tech"
tags: ["LLM","AI Agent","RAG","CIRCE Divines LLM","Task Complexity","Constraint Structure Estimating","Estimating","While"]
---

# CIRCE Divines LLM Task Complexity from Constraint Structure

## At a glance

Estimating whether a problem is too complex for a large language model (LLM) before the model attempts it can guide model routing, how to allocate test-time compute, and suggest the need for new tools to aid the agent. …

## Key points

- Estimating whether a problem is too complex for a large language model (LLM) before the model attempts it can guide model routing, how to allocate test-time compute, and suggest the need for new tools to aid the agent.
- While existing estimators rely on the problem text, LLM-annotated task difficulty labels, or signals obtained after a model attempts to solve the problem, such as its hidden states, none explicitly leverage the constrai…
- We introduce CIRCE, which extracts a problem's structure by using a cheap LLM to construct a Constraint Intermediate Representation that deterministically compiles into the constraint graph.

## Full text

Estimating whether a problem is too complex for a large language model (LLM) before the model attempts it can guide model routing, how to allocate test-time compute, and suggest the need for new tools to aid the agent. While existing estimators rely on the problem text, LLM-annotated task difficulty labels, or signals obtained after a model attempts to solve the problem, such as its hidden states, none explicitly leverage the constraint structure inherent in the problem: the variables to assign and how the constraints over them interact. We introduce CIRCE, which extracts a problem's structure by using a cheap LLM to construct a Constraint Intermediate Representation that deterministically compiles into the constraint graph. Across seven reasoning benchmarks (including SAT, logic, and math) and four LLMs, even a randomly initialized GNN encoder improves predictions of LLM response correctness, reducing negative log-likelihood from 0.465 to 0.423 and increasing AUROC from 0.691 to 0.727 relative to the strongest baseline for each metric.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_67da48608e1cf00a94477de9