Extracting Prompts by Inverting LLM Outputs
Paper
•
2405.15012
•
Published
Collin Zhang
collinzrj
AI & ML interests
PhD Student at Cornell University, interested in ml security, ai safety, red teaming, etc
Organizations
authored
5
papers
3 months ago
Harnessing the Universal Geometry of Embeddings
Paper
•
2505.12540
•
Published
•
9
Universal Zero-shot Embedding Inversion
Paper
•
2504.00147
•
Published
Adversarial Decoding: Generating Readable Documents for Adversarial Objectives
Paper
•
2410.02163
•
Published
Language Confusion Gate: Language-Aware Decoding Through Model Self-Distillation
Paper
•
2510.17555
•
Published
•
1