Read original ↗
newsReddit r/MachineLearningTrust 72 · CommunityPublished 6d agoLive · 6d ago

Hiding messages in the least significant mantissa bits of fine-tuned ONNX model weights [P]