3 ms·
Beyond Surface Statistics: Scene Representations in a Latent Diffusion Model
- satvikpendem 3y agoLDMs are autonomously encoding depth information when training. It really feels like as time goes on, we understand our own brain analogously based on what we learn from neural networks even if they are not exactly the same.