XMark: Reliable Multi-Bit Watermarking for LLM-Generated Texts

1. 1Identify the need for reliable multi-bit watermarking in LLM-generated texts
2. 2Analyze existing methods and their limitations, such as computational infeasibility and poor trade-offs between text quality and decoding accuracy
3. 3Develop and evaluate XMark, a new watermarking method that addresses these limitations
4. 4Apply XMark to LLM-generated texts to enable reliable attribution and tracing of malicious usage

AI researchers and engineers working on LLMs and watermarking techniques can benefit from XMark, as it enables reliable attribution and tracing of malicious usage of LLMs

💡 XMark addresses the limitations of existing multi-bit watermarking methods for LLM-generated texts, providing a reliable solution for attribution and tracing

🔍 XMark: Reliable multi-bit watermarking for LLM-generated texts, enabling attribution and tracing of malicious usage