In the training run for Astra, we observed only 14 cases of difficulty ending summaries, and no jailbreak-style instructions in the summary using our general monitor.
【数据】在Astra的实际训练运行中,只观察到14个摘要结束困难的案例,且没有检测到越狱式指令。这一对比数据表明,通过修复相关bug和改进监控,研究团队成功减轻了这一问题。