TTRL-Guard mitigates the Correct-Answer Extinction Window in test-time RL via flip-rate-aware reward scaling, minority-preserving sampling, and risk-conditioned sparse updates, yielding best average pass@1 on Qwen models and +54% relative gain on AIME 2025.
arXiv preprint arXiv:2503.00735 , year=
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
representative citing papers
The paper delivers the first systematic review of self-evolving agents, structured around what components evolve, when adaptation occurs, and how it is implemented.
citing papers explorer
-
When the Majority Votes Wrong, the Intervention Timing for Test-Time Reinforcement Learning Hides in the Extinction Window
TTRL-Guard mitigates the Correct-Answer Extinction Window in test-time RL via flip-rate-aware reward scaling, minority-preserving sampling, and risk-conditioned sparse updates, yielding best average pass@1 on Qwen models and +54% relative gain on AIME 2025.
-
A Survey of Self-Evolving Agents: What, When, How, and Where to Evolve on the Path to Artificial Super Intelligence
The paper delivers the first systematic review of self-evolving agents, structured around what components evolve, when adaptation occurs, and how it is implemented.