“allowing higher acceleration during the alignment phase” @elonmusk Assertive offense ✔️
@Kristennetten We’re close to using NNs all the way from image-space to vector-space to control
@elonmusk @Kristennetten I wonder if in the future, it will be possible to use neural networks in the time-space domain, not just images and vectors?
@Xu0P5C4rMmMKUdu @elonmusk @Kristennetten FSD's NNs do have video context which is images over time. For example, it can predict objects through occlusions if it saw them previously.
@Xu0P5C4rMmMKUdu @elonmusk @Kristennetten It already does