thickertoofan@lemm.ee to LocalLLaMAEnglish · 4 days agoSpatialLM, a 1B model capable of spatial identification, using 3d point cloud data. The video demo is amazing.manycore-research.github.ioexternal-linkmessage-square6fedilinkarrow-up126arrow-down10
arrow-up126arrow-down1external-linkSpatialLM, a 1B model capable of spatial identification, using 3d point cloud data. The video demo is amazing.manycore-research.github.iothickertoofan@lemm.ee to LocalLLaMAEnglish · 4 days agomessage-square6fedilink
minus-squareKarkitoo@lemmy.mllinkfedilinkEnglisharrow-up6·4 days agoLooks impressive and it’s truly open-source. However, I see it requires CUDA. Could it run anyway: Without this? With AMD hardware? On mobile (as the model is only 1B) ?
minus-squarethickertoofan@lemm.eeOPlinkfedilinkEnglisharrow-up2·4 days agoI think the bigger bottleneck is SLAM, running that is intensive, it wont directly run on video, and SLAM is tough i guess, reading the repo doesn’t give any clues of it being able to run on CPU inference.
Looks impressive and it’s truly open-source.
However, I see it requires CUDA. Could it run anyway:
I think the bigger bottleneck is SLAM, running that is intensive, it wont directly run on video, and SLAM is tough i guess, reading the repo doesn’t give any clues of it being able to run on CPU inference.