Replace HalfCheetah-v5 policy bundle

Sync README: clarify autoregressive structure, move license to footer

Sync README: clarify autoregressive structure, move license to footer

Sync README: clarify autoregressive structure, move license to footer

Fix HalfCheetah-v5 package metadata

Add Hopper-v5 normalized scores

Document Hopper-v5 policy bundle

Add Hopper-v5 policy bundle

Document expert-free training data

Reorder model card links

Fix PolyForm license link

Fix model card GitHub links

Compact evaluation table formatting

Simplify evaluation table columns

Update evaluation table with normalized scores

Update Ant and Humanoid policy bundles

Add model card metadata

Update Walker2d-v5 README metrics

Update Walker2d-v5 bundle

Fix evaluation table plus-minus symbols

Update MuJoCo evaluation results

Correct Humanoid-v5 eval score

Update Humanoid-v5 score

Replace Humanoid-v5 bundle

Upgrade hero GIF: 480p 15fps with palette optimization (1080p source)

Replace replay.mp4 with 1080p source re-encode (CRF 20, ~1.3MB)

Add replay.mp4 for HF RL Video Preview widget (Hopper-v5 Llama)

Add Hopper-v5 Llama variant rollout demo

Update HalfCheetah-v5 bundle

Add root config.json as repo index for HF download tracking

Update halfcheetah-v5 bundle: context 24->32, return 1768±2624 -> 2680±2059 (50 ep, seed=0)

Add tagline, model details (decoder), training data, and ctx column

Update bundle return table (re-measured 5ep seed=0 CPU; reflects new ant-v5 & humanoid-v5 bundles)

Update humanoid-v5 bundle (50ep eval: mean 3058 -> 4571, median 2330 -> 4882, surv 3/50 -> 15/50; context_length 24 -> 32)

Update ant-v5 bundle (50ep eval: mean 1914 -> 2265, median 1562 -> 1969; context_length 24 -> 16)

Sync README with repo: drop device=cpu, drop ?nw, add return numbers

Add model card

Add humanoid-v5 bundle

Add walker2d-v5 bundle

Add halfcheetah-v5 bundle

Upload ant-v5 test bundle

initial commit