THU-BPM/MarkLLM
An open-source toolkit for implementing, evaluating, and benchmarking watermarking algorithms on large language model outputs.
Not currently ranked — collecting fresh signals.
star history
MarkLLM provides researchers and developers with standardized implementations of LLM text watermarking techniques. It offers a modular framework to embed invisible signals into LLM-generated text and detect them later, helping identify AI-generated content. The toolkit includes multiple watermark algorithms, detection utilities, and evaluation metrics. It is designed to accelerate research on trustworthy AI and content provenance.
Frequently asked
- What is THU-BPM/MarkLLM?
- An open-source toolkit for implementing, evaluating, and benchmarking watermarking algorithms on large language model outputs.
- Is MarkLLM open source?
- Yes — THU-BPM/MarkLLM is open source, released under the Apache-2.0 license.
- What language is MarkLLM written in?
- THU-BPM/MarkLLM is primarily written in Python.
- How popular is MarkLLM?
- THU-BPM/MarkLLM has 1k stars on GitHub.
- Where can I find MarkLLM?
- THU-BPM/MarkLLM is on GitHub at https://github.com/THU-BPM/MarkLLM.