self-hosted/ai
§01·model · /models

KiMoDo

specializedactiveNVIDIA-OpenLicense

Kinematic motion diffusion text-to-3D-motion by NVIDIA. 282M params; default ~17 GB VRAM with Llama-3-8B encoder, drops to <3 GB with TEXT_ENCODER_DEVICE=cpu.

§02·GPUs that run this model
0 total
No community benchmarks for this model yet.