Watermark Detection on Alpaca instruction-following 52K
61.67TPRHATS
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HATSBackbone=Llama-2-7b-chat-hf, Number of samples=600, Tokens per sample=2502025.12 | 61.67 | — | |
| KGWBackbone=Llama-2-7b-chat-hf, Number of samples=600, Tokens per sample=2502025.12 | 38.83 | — | |
| MarylandzBackbone=Llama-2-7b-chat-hf, Number of samples=600, Tokens per sample=2502025.12 | 20.17 | — | |
| OpenAIBackbone=Llama-2-7b-chat-hf, Number of samples=600, Tokens per sample=2502025.12 | 19.5 | — | |
| KGW2Backbone=Llama-2-7b-chat-hf, Number of samples=600, Tokens per sample=2502025.12 | 16 | — |