Daisy Circuit Shadow Teacher v1

This repository is a parameter-free, position-indexed Shadow driver bundle for the Vlab Mario Kart Wii environment. It is not a neural-network checkpoint.

The driver replays the expert RKG only through frame 279 for the race start, then follows the bundled Daisy Circuit references in closed loop. Items are always suppressed. The intended match is 150cc, CPU Normal, Solo Race, Daisy on Mach Bike, Manual drift, and Item None.

Use it with a Vlab-Kart-env revision that supports vlab-shadow-driver-v1:

python visual.py vlabki/daisy-circuit-shadow-v1 -n 1 --temp 0

Discord:

/match_bot name:vlabki/daisy-circuit-shadow-v1 branch:feat/daisy-shadow-driver

The tiny model.safetensors contains only a format-version marker. It exists for compatibility with the Discord bot's model-source preflight; all driving behavior comes from shadow_driver.json and the reference files.

Downloads last month
4
Safetensors
Model size
1 params
Tensor type
I32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support