Instructions to use djduecaster/dfreq-tug with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Microduck
How to use djduecaster/dfreq-tug 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 djduecaster/dfreq-tug
- Notebooks
- Google Colab
- Kaggle
Commit History
2 days ago
-
Publish Tug declared-start specialist, local 160.168 cm
Verified signatureApproved own-trajectory specialization over attributed T19 fallback. Five exposed starts, five pit wins, local mean 160.1679945 cm; not fresh robustness or Arena rank evidence. Preserve upstream terms and prior history. Policy SHA256 3d67cd3705bf8c8e12ee27723edc05b378fe8c7274350d61c7e53f19b8cfc362.5ea1210
3 days ago
-
Publish validated T19 Tug feedback policy
Verified signatureFrozen T19 policy SHA256 61707f5f3aca15b0a5bec1ff68ae119bbc840f6fdc3d618953791ab52b7cb6d1. Preserve Nico-robot/Pollen starter attribution and inherited metadata. Local official CLI 146.5520 cm, 5/5 pit wins; frozen fresh paired assessment 146.4636 cm, 128/128 pit wins. These are local results, not an Arena receipt or rank-one claim. Prior source-authored baseline remains in repository history.6bb0a48
4 days ago
-
Publish reviewed source-authored Tug baseline
Verified signatureExact reviewed policy SHA256 4d913bef9eb5b9db45bb9741836ab084e37024a8ba64315eaa5b51bf55bd963e. Original feedback controller and reproducible source, manifest and measured limitations. Technically validated locally; weak behavior and five local rating losses disclosed. License intentionally unspecified. No Arena submission or server score claimed.c234222 -
initial commit
Verified signature83c505d