Instructions to use cetiennec/mos-routine-v1-20261011 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Microduck
How to use cetiennec/mos-routine-v1-20261011 with Microduck:
# Replace SLOT with the slot specified in the model card (walk, stand, sitstand, ground_pick, kick_left, kick_right, roulade). sudo robotctl policy load SLOT cetiennec/mos-routine-v1-20261011
- MJLab
How to use cetiennec/mos-routine-v1-20261011 with MJLab:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
mos-routine-v1-20261011
Training run of Mjlab-OnSourccey-Routine-Dome-MicroDuck โ a Microduck standing on a moving Sourccey robot
(dome top), tracking head-pose commands. Code: microduck-on-sourccey @ 0df073c.
- Hardware: HF Jobs
l4x1 - Args:
--env.scene.num-envs 4096 --agent.max-iterations 2000 --agent.algorithm.entropy-coef 0.005 --agent.logger tensorboard --agent.resume True --agent.load-run init_from_hub --agent.load-checkpoint model_1999.pt - Checkpoints:
<run>/model_<iter>.pt(pushed every minute while training) - Metrics: see the Training metrics tab (TensorBoard events)
- Policy:
exported/policy.onnx([1, 61] โ [1, 14], obs normalizer baked in)