weee
Keep finetuning the weights
A perpetual policy for the microduck (61-D observation, 14 actions, 50 Hz). Runs until told otherwise — a gait, loaded into a policy slot.
Run it on a robot
sudo robotctl policy load <slot> weerawhin/microduck-weee
The observation normalizer is baked into policy.onnx; feed raw observations.
manifest.json follows schema 2 of the microduck policy manifest (docs/policy-manifest.md in the daemon repo).
Training
- task_id:
Mjlab-Sprint2m-MicroDuck - repo:
https://github.com/NIRUN-Weerawit/microduck-challenges - branch:
main - commit:
d6a113332 - checkpoint:
750 - seed:
42 - base:
mjlab-microduck 0.1.0 @ 8d0dbc94c - started:
2026-10-13T13:23:17Z
Reproduce
Same code, same uv.lock, same command, same seed. Training it again yields a comparable policy, not the same weights: GPU reinforcement learning is not bit-reproducible across machines.
git clone https://github.com/NIRUN-Weerawit/microduck-challenges
cd microduck-challenges
git checkout d6a113332
uv sync
uv run train Mjlab-Sprint2m-MicroDuck --env.scene.num-envs 8192 --agent.max-iterations 2500 --agent.seed 42 --agent.logger tensorboard --agent.run-name 50-15-50
- Downloads last month
- 8