---
title: "Test-Time Self-Play: Scaling without a Verifier"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2611.07367v1"
permalink: "https://feed.astralor.com/items/itm_c36a6167875f11a1f01dc01b"
published: "2026-11-05T03:07:02.000Z"
score: 55
channel: "tech"
tags: ["LLM","RAG","Test-Time Self-Play","Scaling","Verifier Test-time","Most","Test-time","TTSP"]
---

# Test-Time Self-Play: Scaling without a Verifier

## At a glance

Test-time scaling methods spend additional compute at deployment to improve a large language model's performance on a hard target problem. Most methods rely on a verifier that scores candidate solutions for correctness,…

## Key points

- Test-time scaling methods spend additional compute at deployment to improve a large language model's performance on a hard target problem.
- Most methods rely on a verifier that scores candidate solutions for correctness, using it to select among or refine the model's outputs.
- In many deployment regimes, however, a strong verifier on the target problem is unavailable.

## Full text

Test-time scaling methods spend additional compute at deployment to improve a large language model's performance on a hard target problem. Most methods rely on a verifier that scores candidate solutions for correctness, using it to select among or refine the model's outputs. In many deployment regimes, however, a strong verifier on the target problem is unavailable. We introduce Test-Time Self-Play (TTSP), an approach to test-time scaling using asymmetric self-play without reliance on a verifier. On a difficulty-calibrated subset of LiveCodeBench v6, TTSP is the best-performing method at every verifier strength we evaluate, and the gap to the strongest baseline widens as the verifier weakens; in the no-verifier limit, TTSP yields a 35\% absolute improvement in mean@8. The gains hold across a range of verifier strengths, except when the base policy has near-zero coverage of the correct answer.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_c36a6167875f11a1f01dc01b