FailSAE: Towards Interpretable Failure Prediction for Vision-Language Models via Sparse Autoencoders
英文标题：FailSAE: Towards Interpretable Failure Prediction for Vision-Language Models via Sparse Autoencoders
arXiv:2609.04276v1 Announce Type: new Abstract: Vision-language models (VLMs), such as CLIP, have achieved strong performance across multimodal tasks by aligning visual and textual representations in a shared embedding space. As VLMs are increasingly used for high-stakes domains, failure prediction
打开原文相关 2.4 · 机会 0
2.1 KB
612 B
380 B
9.1 KB
32.7 KB
22.8 KB
24.2 KB
34.7 KB
中文
arXiv:2609.04276v1 Announce Type: new Abstract: Vision-language models (VLMs), such as CLIP, have achieved strong performance across multimodal tasks by aligning visual and textual representations in a shared embedding space. As VLMs are increasingly used for high-stakes domains, failure prediction
中文全文翻译正在补齐，可先查看下方英文原文。
English / 原文
BibTeX formatted citation
×