Co-creator and lead of @pytorch at Meta A.I.
I also dabble in robotics these days.
AI is delicious when it is open-source and easy-to-use.
If you installed PyTorch-nightly on Linux between Dec. 25 and Dec. 30, uninstall it and torchtriton immediately and use the latest nightly binaries.
Read the security advisory here: https://pytorch.org/blog/compromised-nightly-dependency/
OpenAI just released a 3D version of DALL-E called Point-E
github: https://github.com/openai/point-e
arXiv: https://arxiv.org/abs/2212.08751
I've seen a few 3D point cloud generators over the last year. I don't know the area well. Anyone know if there is anything significantly different going on here?
[EDIT: and "bigger and faster" is an okay answer, though I suspect there is more to it]
Just created @pytorch here.
Start following it for updates (will setup syndication shortly).
the Tooot app has been pretty good for Mastodon on iOS. Fast, responsive, small.
tooot is an open source, simple yet elegant Mastodon mobile client. A Mastodon (https://joinmastodon.org/) account is required to use this app. tooot supports: - Cross platform, including iPadOS and MacOS - Multiple accounts - Dark mode or adapt to system - Adjustable toot font size - Push notifica…
CLIP-Fields won the Outstanding Paper Award at the LangRob workshop @ CoRL 2022. (congrats @notmahi@twitter ! )
This is a starting point of NLP-powered spatial memory. Strictly better than using each pre-trained model separately.
The framework makes it easy to add in more signals (like GraspNet).
Read more about our work here:
https://mahis.life/clip-fields/
Teaching robots in the real world to respond to natural language queries with zero human labels — using pretrained large language models (LLMs), visual language models (VLMs), and neural fields.
PyTorch
Mark Riedl