Sep 30 10:19 Remove log / cache / raw-output files flagged by the 2026-09-30 public-repo audit (category a) HI +1: Claude Fable 5.1
Tree-only removal of logs, __pycache__/.pyc, vendored pylibs, TensorBoard events, pid/lock files and parity-test dumps. History untouched; no data, dataset media or checkpoint files affected.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Sep 27 11:53 Remove pi0.5 checkpoints: pi05_ctc502_30k_rawparams, pi05_ctc502_60k_rawparams HI
Sep 23 02:57 eval_stack: point REPLAY_ROOT at the in-repo negmesh8 bank HI
02:56 eval_stack: add the negmesh8 160-episode replay bank (8x20) so transfer is self-contained HI
02:49 eval_stack: make the 160-episode bank self-identifying — object list, ep_id mapping, and the incompatible sibling bank HI
02:40 eval_stack: warn that two different VACE checkpoint sets exist with different norm stats HI
02:28 eval_stack: document precedence — flag that upstream EVAL_SEEDS.md action_horizon=10 is wrong (should be 50) and state norm-stat rules HI
02:27 eval_stack: pi0.5 negmesh8 160-episode replay eval harness, job scripts, protocol docs HI
Sep 22 21:14 transfer.tar rebuilt: actaug data replaced by lerobot_actaug256_neg_pi05 HI
21:13 README: actaug data replaced by the pi0.5 hard-set dataset HI
21:13 actaug: neg-mesh tooling, neg runner, raw dumps of the 256 neg episodes HI
21:09 dataset: lerobot_actaug256_neg_pi05 (pi0.5 hard set, grasps by pi0.5 60k, 8x32) HI
21:08 replace actaug data: remove actaug/raw_dumps/robocasa_aug256_prio HI
21:08 replace actaug data: remove dataset/lerobot_actaug256_prio_pi05 HI
21:08 replace actaug data: remove dataset/lerobot_actaug256_prio HI
19:00 README: add pi0.5-grasped actaug256 dataset row HI
19:00 dataset: lerobot_actaug256_prio_pi05 (grasps by pi0.5 60k, same 8 SAM3D targets) HI
17:41 README: bundle contents incl. actaug dataset + config HI
17:41 dataset: lerobot_actaug256_prio (256 eps, 8 SAM3D targets x 32, LeRobot v2.1) HI
17:41 transfer.tar rebuilt: + dataset/lerobot_actaug256_prio, actaug/ (code, eval, raw dumps), actaug256 train config HI
17:41 actaug: simulator action-augmentation generation code, Table-2 eval protocol, raw dumps of the 256 prio episodes HI
15:27 Reproducibility fix: per-episode env/noise seeding, deterministic XLA, vendored robosuite fork, nested quantile norm_stats, revised EVAL_SEEDS.md + INSTALL.md, validation harness HI
Sep 21 13:37 Document eval seeds (action seed 42; env seed = episode_index + offset) HI
13:16 Add pi0.5 ctc502 60k (59999) non-EMA rawparams checkpoint HI
13:11 Add pi0.5 ctc502 30k (repro29999) non-EMA rawparams checkpoint HI
12:46 Upload transfer.tar with huggingface_hub HI
Sep 20 15:40 Upload transfer.tar with huggingface_hub HI
14:30 initial commit HI
