RT-1 robotics transformer and SayCan
VIMA General Robot Manipulation with Multimodal Prompts
magma a GPT-style multimodal model that can understand any combination of images and language
Versatile-Diffusion Text, Images and Variations All in One Diffusion Model
I think hacking requires certain level of live coding, like fzf, interactive jq, interactive regex and repl, or my modded hy.
Interactive searching is one such kind of live coding. It can also be manual labeling, training a model.
on the other hand, media convey the idea of live coding, mostly by showing off the benefits, but hardly getting to the point when you really need it. media is not coding (requiring coding skill), it is just somehow interactive.
[percol](https://github.com/mooz/percol] supports pinyin
ugrep use ugrep -Q for interactive tui