Deterministic core of wickra-gym: precompute a candle dataset into a fixed feature tensor once, then step a Gymnasium-style RL environment in O(1) with…
Densify and denoise metric depth images: Depth Anything V2 (via candle) affine-anchored to the trusted parts of a raw sensor depth image, filling holes and…
Reference command-line rollout runner over wickra-gym-core: load a spec + candle dataset, drive a fixed deterministic policy through the environment, and print…
Reference command-line feature store over wickra-feature-store-core: fold a FeatureSpec over per-symbol CSV candle files and emit the feature matrix as JSON,…