card: all 12 discrete-env datasets re-recorded for the action-label gap; refresh counts and ladder figures

card: pyramids + soccer-twos expert-v0 re-record (all four discrete experts now fixed)

card: dungeon-escape expert-v0 re-record (35,505 ep, anchor 0.019355, medium norm 0.81)

card: pushblock expert-v0 re-record (53,039 ep, anchor 0.353699) + action-label data-quality notice

docs(card): correct the companion policy-checkpoint list to Pyramids

docs: frame as offline-RL datasets (usable by any offline-RL method); add offline-reinforcement-learning tag

docs: add 'reduced backward from the expert' rationale to Quality ladders intro

fix: replace attached '~' (approx) with ≈ so HF markdown stops rendering it as strikethrough; backtick 'T ~ U' notation

fix: pyramids medium/simple episode counts in dataset table (4267->4159, 3040->3042) to match re-collected tiers

docs: intro describes softmax-temperature degradation for discrete tiers (not just action noise); refresh simple/medium spans

soccer-twos medium/simple: describe per-team per-match softmax-temperature-range self-play tiers (T~U(1.3,2.6)/U(2.1,3.4), side-swap norm ~0.80/0.60), replacing the epsilon-pool description

pushblock medium/simple: describe per-episode softmax-temperature-range tiers (T~U(2.1,3.2)/U(3.25,4.75), norm 0.80/0.60), replacing the fixed-T description

pyramids medium/simple: describe per-episode softmax-temperature-range tiers (T~U(2.7,4.9)/U(4.8,7.1)), replacing the fixed-T description

dungeon-escape medium/simple: describe new per-group softmax-temperature-range tiers (step-reward metric), replacing the old epsilon group-noise description

docs: pushblock/pyramids tiers -> softmax-temperature, step-reward metric

fix(card): companion policy list -> 4 shipped policies; set viewer:false

docs: restructure card (overview + unified ladder summary + collapsible per-env tables); move reproduction recipe to docs/reproduction.md

docs: add crawler/pushblock/3dball-hard/worm/walker medium+simple ladders (per-episode noise pool); pyramids pool re-collect; manifest 14->24

docs: intro simple anchor 0.50 -> 0.60

docs(pyramids): update simple ladder row to eps 0.67 / norm ~0.60 re-collect

docs: restore medium tier (Minari-aligned ~0.8) for dungeon/pyramids/soccer with re-collected numbers

docs: realign quality ladder to Minari levels (simple=former medium; medium re-collecting to 0.75); drop medium rows

docs: add pyramids/simple-v0; convert Pyramids medium note into a quality-ladder table

docs: add pyramids/medium-v0 (contents, tier note, intro ladder mention) + manifest

docs: add DungeonEscape quality ladder table (expert/medium/simple)

docs: add unity/dungeon-escape/simple-v0 to README contents + release_manifest

Add SoccerTwos simple-v0 1M dataset

docs: add unity/dungeon-escape/medium-v0 to README contents + release_manifest

docs: point README/manifest at renamed expert-v0 datasets

Add SoccerTwos medium-v0 1M dataset

Simplify documentation for release-23-only artifacts

Document release-23 3DBallHard expert dataset

Document Pyramids, Worm, and Walker release-23 datasets

Document release-23 Crawler and PushBlock expert datasets

Document ML-Agents release_20/release_23 dataset boundary

Document ML-Agents release_20/release_23 dataset boundary

Redefine DungeonEscape and SoccerTwos as decentralized ego-agent datasets

Document SoccerTwos expert self-play dataset

Retract PushBlockWithInput dataset

Clarify PushBlock dataset provenance and naming

Replace PushBlock sample with 1M real rollouts

Document portable PushBlock dataset download

Add discrete PushBlock expert Minari dataset

Card: link the GitHub collection & measurement recipe

Update card: dataset present (1047 ep, 1M steps)

Add model/dataset card and license

initial commit