Proving membership in LLM pretraining data via data watermarks.
Johnny Tian-Zheng WeiRyan Yixiang WangRobin JiaPublished in: ACL (Findings) (2024)
Keyphrases
- data sets
- raw data
- data structure
- historical data
- database
- image data
- data sources
- data points
- data collection
- computer systems
- synthetic data
- sensor data
- xml documents
- relational databases
- probability distribution
- knowledge discovery
- high dimensional
- small number
- data mining techniques
- multiscale
- complex data
- high quality