[RyanPCo]

PoseCoordinateFrameTransform gains per_step_target: with target_history K,
step k's keypoints are expressed in step k's OWN wrist frame instead of the
current step's. Re-expressing older steps in the current frame smears wrist
motion into the articulation channels, and the palm-rigid MANO knuckles
(1/5/13/17) have a very tight legitimate range, so they land far outside their
per-cell bounds and worsen with age: measured 2026-09-17, 10.3 / 6.9 / 3.3
out-of-range cells per sample at history steps 0/1/2 against 0.0 at the
current step. That cost 22,500 loader retries on one run and starved rank 0
into an NCCL watchdog kill.

Per-step framing keeps every step a genuine state -- keypoint 0 is 0 at every
step, so the norm stats and the bounds check describe all of them. The ACTION
chunk keeps the current-step frame: its leading axis is the forward horizon,
not history.

Off by default, and the per-step path additionally requires a leading axis on
the target, which a keymap without proprio_history never emits -- so no
existing call site changes. Both halves are now tested: the origin property at
K=4, and bit-identical output at K=1 with the flag on and off.

The human keypoint eef-frame keymap turns it on. CartesianYPRToRot6D also
picks up the note about applying it to the proprio ee_pose, which the
embodiment comment used to carry.

🤖 Generated with Claude Code

https://claude.ai/code/session_01HjCQBqQCr2igFEMR7yUovw

[RyanPCo]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• pi: initialize from PaliGemma with a scratch action expert #677 : 2 dependent PRs (#678 , #680 )
• perf: vectorized image augs and a torch.compile knob for HPT and Pi; Pi clips gradients like openpi #676
• loader: one zarr read per key, not one per keymap entry #674
• eval: build the pinned-episode video batches only where they are used #673
• checkpoints: last.ckpt as a symlink, not a second full copy #672
• trainHydra: metric_frames_per_episode: auto, and warn on empty prompts #671
• transforms: per-step wrist framing for proprio history #670 👈 (View in Graphite)
• viz: one keypoint renderer, the horizon trace #669
• configs: history and backbone-LR knobs on the keypoint wrist HPT model #668
• trainer: ddp_bf16_compress, DDP with a bf16 gradient all-reduce #667
• eval: overlay videos play at the source rate under DDP #666
• hpt: build the policy on CPU and let Lightning place it #664 : 1 other dependent PR (#665 )
• hpt: never re-initialise pretrained stems; hash-check them; mask text padding #645 : 1 other dependent PR (#646 )
• obs: proprio history (K proprio steps, one token each, learned time embedding) #650
• loss: mask the repeat-last action padding out of the HPT loss #649
• resnet: keep a frozen backbone in eval mode; give backbones their own LR group #652
• augs: per-sample image augmentation for HPT train-mode augs #651
• eval: deterministic validation (seeded prompts + flow sampling); log-only grad-norm MAD #644
• data: mecka_fold_flagship_top3_hpt_6d, the 3-operator flagship train split #643
• val: subsampled metric loaders + pinned contiguous video loaders #642
• infra: mecka 6D / keypoint recipes, 300M models, defaults, and the rollout revert #640
• eval: layout-aware frame-invariant metrics, and a train-split viz head #659
• data: stop letting one bad episode or a rotation channel take a run down #658
• norm: guard degenerate channels, horizon pooling, and no rot6d normalization #657
• keypoints: the 144-D wrist-first hand-keypoint action becomes the human default #656
• mecka: fix the double-mirrored LEFT wrist frame, and correct existing zarrs #655
• 6d: PI goes 6D-only; Eva and Human gain the continuous-rotation modes #654
• 6d: continuous rotation representation, bimanual layouts, shared norm helpers #653
• train: give each Slurm slurm job gets its own cache dir to avoid conflict #638 : 1 other dependent PR (#639 )
• write git version into the run dir for reproducibility #637
• train: don't load PI base weights on resume and verify that checkpoint path exists #636
• configs: declare proprio and actions dims once per HPT model config #634 : 1 other dependent PR (#635 )
• cache norm stats for data configs to minimize recomputation #633
• import cloud packages lazily to support local training #631 : 1 other dependent PR (#632 )
• cache dataset resolver for faster training initialization #630
• validate episode hashes in data config before training #628
• modify pi camera map to match data #629
• add tests to ensure training works between every model and data vendor #626
• run CI tests on every PR #625
• add env variables for user-specific paths #624
• build: set pyproject and uv as the dependency managers #623
• fix(configs): fix paths for data hydra yaml configs #622
• refactor(algo): delete ACT #627 : 1 other dependent PR (#621 )
• tests: create dedicated test folder with unit and integration tests #620
• add pyright type checker #619
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[RyanPCo]

Automated adversarial review (read-only; findings were checked by a second skeptical pass before posting): 2 nit.

[RyanPCo]

nit (dead-code): test_per_step_target_is_the_default only reads the signature default through inspect and re-imports PoseCoordinateFrameTransform, which the module already imports at the top. Drop it. The rationale is now written out three times (this docstring, the human.py comment block, and the test docstrings), and the human.py copy carries the 2026-09-17 out-of-range counts. Keep it once, here, without the measurement.

[RyanPCo]

nit (consistency): Pass the already-bound target_world here, as the per-step branch does, instead of re-reading batch[self.target_world].

[RyanPCo]

Review findings: status after the fixup pass. The fixes are squashed into this PR's commits, and the whole stack was force-pushed.

Fixed

• Fallback path re-reads batch[self.target_world] instead of the local: Both branches now pass the bound target_world.

Partial

• per_step_target flag has no production caller; default-value test is tautological: Kept the flag, because the tests use it to compare both framings. Deleted the inspect-based test. The rationale now lives only in the init docstring, with no 2026-09-17 numbers. Removed the human.py comment block and trimmed the test docstring.