Infinigence AI open-sources APXInf for Jetson Thor
Infinigence AI Open-Sources APXInf for Embodied Edge Inference on Jetson Thor
This open-source model is designed for edge devices like Jetson Thor, offering significant latency reduction for AI applications on these platforms.
Infinigence AI has open-sourced APXInf, a model designed for embodied edge inference. The PI 0.5 FP8 version on Jetson Thor reduced end-to-end latency from about 278 ms to about 26 ms. Orin/Thor/RTX 4090 support and a domestic-chip roadmap are included.
Infinigence AI Open-Sources APXInf for Embodied Edge Inference on Jetson Thor
Infinigence AI open-sources APXInf with Tsinghua and SJTU: PI 0.5 FP8 on Jetson Thor drops end-to-end latency from ~278 ms to ~26 ms (~38.46 Hz), with Orin/Thor/RTX 4090 support and a domestic-chip roadmap.
