This repo contains steganographic embedding/extracting algorithms proposed in
STEAD: Robust Provably Secure Linguistic Steganography with Diffusion Language Model
Yuang Qi · Na Zhao · Qiyi Yao · Benlong Wu · Weiming Zhang · Nenghai Yu · Kejiang Chen
Anhui Province Key Laboratory of Digital Security, University of Science and Technology of China
The code is implemented based on Dream. If you want to reproduce the experiments in our paper, you need to deploy Dream-7B first. THANK THEM! The implementation of Dream is based on the Huggingface transformers library. You should first install transformers by pip install transformers==4.46.2 and torch==2.5.1 as Dream uses the SdpaAttention built in torch. Other versions of transformers and torch are not been fully tested. Run the model requires a GPU with at least 20GB memory.