Describes the three top-level folders and the path mapping from the former ManiLadder_Models and ManiLadder_PI05_Models layouts. Adds *.mp4 to .gitattributes for the evaluation videos under small_flow_matching_policy/.
Add twelve further Astra frozen policies, astra/shared, astra/manifest.json and astra/README.md
Completes astra/ to 112 task folders: 8 re-explored policies (reexplore8_20261104), BagDropIn-v1 and BagHanging-v1 (added3_20261105), DualArmAssembleBracket-v1 and PressFitSoftPeg-v1 (continue2_20261108). Every file was checked against its freeze_manifest.json before upload. astra/manifest.json replaces agent_d20_e20/frozen_20261102/manifest.json with the new paths; the leftover agent_d20_e20/ folder (old manifest and README) is removed.
Add PI0.5 DualArmRubikCube-v1 single-task run, the all-112-task policy and the level-2 rigid data-scaling policies
pi05/single_task/DualArmRubikCube-v1 (campaign pi05_rubikcube_20261107, 0/50 at every evaluated checkpoint), pi05/all_tasks/all112 (snapshots at 60k, 140k and 200k steps with their 50-episode evaluations), pi05/suite_data_scaling/level2_rigid_n{20,50,100,200} (final 100k checkpoints with 50-episode evaluations).
Add small_flow_matching_policy/ (RADIO + Flow Matching checkpoints from Linslab/ManiLadder_Models)
All 111 task folders of Linslab/ManiLadder_Models (revision e4f231e8e836f29720b6101ca05ae357964f54b0) now live under small_flow_matching_policy/<task>/. File contents are unchanged: 220 LFS files were re-uploaded and checked by SHA-256, 215 small files by Git blob hash.
Update pi05 index files to the new paths and add pi05/README.md
results.csv, manifests and per-run metadata/eval files now reference pi05/single_task, pi05/level, pi05/level_transfer, pi05/suite and pi05/suite_transfer. Adds the DualArmRubikCube-v1 rows to the single-task index files (weights follow in a separate commit) and moves the PI0.5 documentation to pi05/README.md.
Move frozen Astra policies to astra/<task>/ (RingTossing-v1 .. WrapRope-v1)
Repository reorganisation: the frozen Agent-D20-E20 (GPT-6 Astra) script policies now live under astra/<task>/ (previously agent_d20_e20/frozen_20261102/tasks/<task>/frozen/). File contents are unchanged.
Move frozen Astra policies to astra/<task>/ (GraspTwoBlocks-v1 .. ReorientPen-v1)
Repository reorganisation: the frozen Agent-D20-E20 (GPT-6 Astra) script policies now live under astra/<task>/ (previously agent_d20_e20/frozen_20261102/tasks/<task>/frozen/). File contents are unchanged.
Move frozen Astra policies to astra/<task>/ (DualArmPushObject-v1 .. GraspCylinder-v1)
Repository reorganisation: the frozen Agent-D20-E20 (GPT-6 Astra) script policies now live under astra/<task>/ (previously agent_d20_e20/frozen_20261102/tasks/<task>/frozen/). File contents are unchanged.
Move frozen Astra policies to astra/<task>/ (DualArmHammerHandover-v1 .. DualArmPushBox-v1)
Repository reorganisation: the frozen Agent-D20-E20 (GPT-6 Astra) script policies now live under astra/<task>/ (previously agent_d20_e20/frozen_20261102/tasks/<task>/frozen/). File contents are unchanged.
Move frozen Astra policies to astra/<task>/ (BuildTent-v1 .. DualArmFoldHandKerchief-v1)
Repository reorganisation: the frozen Agent-D20-E20 (GPT-6 Astra) script policies now live under astra/<task>/ (previously agent_d20_e20/frozen_20261102/tasks/<task>/frozen/). File contents are unchanged.
Split multi_task/ into pi05/level, pi05/level_transfer, pi05/suite and pi05/suite_transfer
Repository reorganisation: PI0.5 checkpoints now live under pi05/. File contents are unchanged. multi_task/ is split by training scope: level_<N> -> pi05/level/, level<B>_from_level<A> -> pi05/level_transfer/, level2_<suite> -> pi05/suite/, level2_<x>_from_<y> -> pi05/suite_transfer/, results.csv -> pi05/multi_task_results.csv.