SO-ARM Starter GR00T N1.7#
GR00T N1.7 checkpoint for the same SO-ARM101 scissor pick-and-place task, loaded by the scissor_pick_and_place policy_n17 mode. Input: room + wrist camera (256×256), 6D joint state, language instruction. Output: 16-step action chunk.
| Property | Details |
|---|---|
| Model size | 3B parameters (GR00T N1.7) |
| Model type | Vision Language Action (VLA); GR00T N1.7 |
| Performance | — |
| Workflow | scissor_ |
| Hugging Face | nvidia/SO_ARM_Starter_Gr00tN17 |