zwhe99/X-SIR

[ACL 2024] Can Watermarks Survive Translation? On the Cross-lingual Consistency of Text Watermark for Large Language Models

29
/ 100
Experimental

This project helps evaluate how well watermarks in text generated by Large Language Models (LLMs) persist, especially after the text has been translated into different languages. It takes text output from LLMs (either watermarked or not) and measures the detectability of the watermark, even after the text has been paraphrased or translated. This tool is for researchers and developers working on LLM watermarking techniques and their resilience.

No commits in the last 6 months.

Use this if you need to assess the robustness of text watermarks embedded in LLM-generated content against common adversarial transformations like translation and paraphrasing.

Not ideal if you are a casual user looking for a simple tool to watermark your documents or detect AI-generated content without deep technical analysis.

AI-safety large-language-models content-attribution digital-watermarking text-analysis
No License Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 8 / 25
Maturity 8 / 25
Community 13 / 25

How are scores calculated?

Stars

42

Forks

6

Language

Python

License

Last pushed

Jun 04, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/transformers/zwhe99/X-SIR"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.