参考 ClaudePlaysPokemon - Twitch（twitch.tv/claudeplayspokemon） "Honesty over Accuracy: Trustworthy Language Models through Reinforced Hesitation" (arXiv:2511.11500, 2025) "Beyond Binary Rewards: Training LMs to Reason About Their Uncerta…