Publish Crawler 60k continuous policy

Publish Pyramids 60k policy

Restructure the model card to match the MuJoCo repository

Restructure the model card to match the MuJoCo repository

Publish validated Unity ONNX context variants

Publish Unity safetensors policy bundles

Add unified supported environments table

Add final DungeonEscape policy

docs(pyramids): sync the dataset reference to the re-recorded files

Add expert-only PushBlock pc5 ONNX policy

docs: refresh pyramids dataset-reference table to live re-collected tier stats

Publish Pyramids-only Unity policy

Remove legacy Crawler policy and documentation

Add Pyramids ONNX policy and 128-episode benchmark

fix(card): correct SoccerTwos dataset table to shipped ground truth; drop ml-agents tag (broken load widget)

Document SoccerTwos dataset generation

Correct SoccerTwos BOS-discard evaluation

Report final SoccerTwos checkpoint evaluation

Replace SoccerTwos policy with final training checkpoint

Document intermediate SoccerTwos policy

Add intermediate SoccerTwos batch-16 ONNX policy

Clarify DungeonEscape checkpoint is provisional

Add DungeonEscape batch-36 policy and evaluation card

Update pushblock.onnx to newer checkpoint (Discrete(7), self-contained)

Add pushblock.onnx policy (Discrete(7), self-contained, RAW obs in)

Card: note per-sensor obs channels (126 + 32) to match the migrated dataset

Add crawler.onnx policy; fill measured return in card

Add model/dataset card and license

initial commit