https://github.com/thu-bpm/markllm

MarkLLM: An Open-Source Toolkit for LLM Watermarking.
https://github.com/thu-bpm/markllm

llm toolkit watermark

Last synced: 3 months ago
JSON representation

MarkLLM: An Open-Source Toolkit for LLM Watermarking.

• Host: GitHub
• URL: https://github.com/thu-bpm/markllm
• Owner: THU-BPM
• Created: 2024-05-16T12:50:40.000Z (6 months ago)
• Default Branch: main
• Last Pushed: 2024-08-02T14:15:26.000Z (3 months ago)
• Last Synced: 2024-08-02T16:00:15.650Z (3 months ago)
• Topics: llm, toolkit, watermark
• Language: Jupyter Notebook
• Homepage: https://arxiv.org/pdf/2405.10051
• Size: 99.2 MB
• Stars: 207
• Watchers: 5
• Forks: 22
• Open Issues: 2
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• AwesomeResponsibleAI - MarkLLM: An Open-Source Toolkit for LLM Watermarking
• StarryDivineSky - thu-bpm/markllm - SIR,EXP,EXP-Edit,ITS-Edit。可视化解决方案：该工具包包括自定义可视化工具，可以清晰而深入地了解不同水印算法在各种场景下的运行方式。这些可视化有助于揭开算法机制的神秘面纱，使用户更容易理解它们。评估模块：MarkLLM 拥有 12 种评估工具，涵盖可检测性、鲁棒性和对文本质量的影响，在其评估水印技术的综合方法中脱颖而出。它还具有可定制的自动化评估管道，可满足不同的需求和场景，从而增强了工具包的实际实用性。 (文本生成、文本对话 / 大语言对话模型及数据)