Byte-level generative predictions for forensics multimedia carving | ResearchPod