AI 中文总结
该研究针对ASR与NMT的无信息幻觉问题，通过空令牌探查弃权信号，发现提升空令牌分数可抑制虚构但存在删除有效内容的权衡，推动多维度评估弃权方法。
AI 中文摘要
现代编解码器系统即便输入无可恢复信息，也能生成流畅文本。我们通过模型预留的空令牌研究ASR（自动语音识别）与NMT（神经机器翻译）中的这一故障，探究终止生成的分数是否已携带可用的弃权（不执行）信号。针对语音识别器和翻译模型，我们审计原生空令牌分数与标量logit偏移；在Whisper模型中，我们额外探查解码器状态，对比监督行编辑与传统外部门控。被评估模型常展现有用的弃权信号，但标准解码无法可靠利用该信号。提升空令牌分数可大幅抑制生成内容的虚构，但过度干预也会删除有效语音或缩短合法翻译。这些发现将空令牌转化为幻觉的诊断视角，并推动从抑制效果与删除成本两方面评估弃权方法，而非仅以幻觉减少为依据。
英文摘要
Modern encoder-decoder systems can produce fluent text even when their input contains no recoverable message. We study this failure in ASR and NMT through the models' reserved null tokens, asking whether the score for ending generation already carries a usable abstention signal. Across speech recognizers and translation models, we audit native null-token scores and scalar logit shifts. In Whisper, we additionally probe decoder states and compare supervised row edits with conventional external gates. The evaluated models often expose a useful abstention signal, but stock decoding does not reliably act on it. Raising the null-token score can sharply suppress fabrication, but aggressive intervention also deletes valid speech or shortens legitimate translations. These findings turn the null token into a diagnostic lens on hallucination and motivate evaluating abstention methods by both suppression and deletion costs, rather than by hallucination reduction alone.
CommentsSubmitted to the Thirty-Ninth AAAI Conference on Artificial Intelligence (AAAI-27)
