Apply for community grant: Academic project
#2
by
Snowball
- opened
Large language models are now demonstrating human-like competencies in various fields, which has led to concerns regarding potential misuse, such as the dissemination of misinformation and facilitation of academic dishonesty. Consequently, detecting text generated by these models becomes paramount. For this reason, we propose a watermarking framework to inject watermarks into already generated text. This will enable third parties utilizing black-box language models to autonomously watermark text for detection or authentication purposes. We are keen to integrate our demo into Hugging Face Spaces. However, to provide a superior experience, we need access to a GPU. Therefore, we would like to request for free GPU usage.