[KongchangAI]
ConceptTTSP

Test-Time Self-Play

A test-time scaling method that requires no external verifier. The same LLM alternately plays proposer and solver in an asymmetric self-play setup, achieving self-improvement through an automatic curriculum.

Timeline (last 90 days)

Nov 6

The Test-Time Self-Play (TTSP) method proposed in the paper bypasses the dependence on verifiers through asymmetric self-play, offering a new idea for test-time scaling without a verifier.

Unverified50%

All Facts (1)

Source Articles