jwyang

8 ms·

1. 1.▲
   by jwyang 2y ago

   very good question, now we are mainly focusing on building the foundtion for multimodal perception and atomic action taking. Of course, integrating the trace-of-mark prediction for robotics and human video data enhances the model's med
2. 2.▲
   by jwyang 2y ago

   ops, a typo, no M from Microsoft.
3. 3.▲
   by jwyang 2y ago

   Thanks for your great interests on our Magma work, everyone! We will gradually roll out the inference/training/evaluation/data preprocessing code on our codebase: https://github.com/microsoft/Magma , and
4. 4.▲
   by jwyang 2y ago

   Good catch! A minor correction: Magma - M(ultimodal) Ag(entic) M(odel) at M(icrosoft) (Rese)A(rch), the last part is similar to how the name Llama came out, :)
5. 5.▲
   by jwyang 3y ago

   Here are other related links: code: https://github.com/microsoft/SoM demo: https://user-images.githubusercontent.com/3894247/281586831-...