会员服务
Visuals Lie, Consistency Speaks: Disentangling Spatial Attention from Reliability in Vision-Language Models
Arxiv
0+阅读 · 6月16日
The Faithfulness Gap: Certifying Semantic Equivalence Between Natural-Language and Formal Mathematical Statements
Arxiv
0+阅读 · 6月15日
"OpenBloom": A Stigma-Sensitive LLM Design Probe for Reproductive Well-Being
Arxiv
0+阅读 · 6月14日
Medical Heuristic Learning: An LLM-Driven Framework for Interpretable and Auditable Clinical Decision Rules
Arxiv
0+阅读 · 6月15日
Do Safety Monitors Stay Reliable After an Update? Benchmarking and Predicting Activation-Monitor Staleness
Arxiv
0+阅读 · 6月14日
Probing and graph coloring techniques for trace estimation in Lattice QCD
Arxiv
0+阅读 · 5月29日
Coherent Swap Regret and Channel-Proof Learning
Arxiv
0+阅读 · 6月1日
Statistical Guarantees for Reasoning Probes on Looped Boolean Circuits
Arxiv
0+阅读 · 5月31日
Quality Is Not a Safety Proxy Under Quantization
Arxiv
0+阅读 · 6月8日
Substrate Asymmetry in User-Side Memory: A Diagnostic Framework
Arxiv
0+阅读 · 6月10日
Where's the Plan? Locating Latent Planning in Language Models with Lightweight Mechanistic Interventions
Arxiv
0+阅读 · 6月11日
When Does Routing Become Interpretable? Causal Probes on Block Attention Residuals
Arxiv
0+阅读 · 6月11日
Mirage Probes: How Vision Models Fake Visual Understanding
Arxiv
0+阅读 · 6月11日
Document-Authored Control-Signal Impersonation: A Low-Cost Indirect Prompt Attack on RAG Safety Boundaries
Arxiv
0+阅读 · 6月8日
CUBE: Contrastive Understanding by Balanced Experiments
Arxiv
0+阅读 · 6月4日
参考链接
微信扫码咨询专知VIP会员
Top