Hidden_embeddings

ML Author: MiKi

Our competitors have been baking a new AI model, and I overheard that they've hidden an important 35-character trade secret inside. Unfortunately, they've obfuscated the model by shuffling the few real layers and introducing a bunch of fake layers with random bias to keep it secure.

Find a way to extract it before they get ahead of us and start eating into our market share!