Skip to content

Docs: Improve LEAPP clean-clone export guidance - #7156

Merged
kellyguo11 merged 4 commits into
isaac-sim:developfrom
frlai:frlai/better_leapp_guidance_from_clean_clone
Aug 20, 2026
Merged

Docs: Improve LEAPP clean-clone export guidance#7156
kellyguo11 merged 4 commits into
isaac-sim:developfrom
frlai:frlai/better_leapp_guidance_from_clean_clone

Conversation

@frlai

@frlai frlai commented Aug 18, 2026

Copy link
Copy Markdown
Collaborator

Summary

  • Align LEAPP export/deploy docs with the train/play backend-extra pattern (Newton / OV PhysX / Isaac Sim) instead of a standalone uv sync --extra leapp setup.
  • Point readers at isaaclab train and --checkpoint pretrained so a clean clone can get from docs to a working export without relying on removed per-backend train.py scripts.

@frlai
frlai requested a review from a team August 18, 2026 23:57
@github-actions github-actions Bot added the documentation Improvements or additions to documentation label Aug 18, 2026
@greptile-apps

greptile-apps Bot commented Aug 18, 2026

Copy link
Copy Markdown
Contributor

Greptile Summary

The PR updates LEAPP export and deployment guidance to use the current backend-extra and physics-selection model, and adds current training and pretrained-checkpoint references.

  • Documents Newton, OV PhysX, and Isaac Sim export/deployment command variants.
  • Replaces standalone LEAPP synchronization guidance with per-command uv extras.
  • Updates the Direct-workflow tutorial to use the unified training CLI.

Confidence Score: 4/5

The documentation should be corrected before merging because its isaaclab.sh and isaaclab.bat alternatives fail on a clean clone without a separate LEAPP/backend installation step.

The uv examples activate the required extras, but the newly presented wrapper alternatives only execute an existing interpreter and therefore leave required LEAPP and backend packages unavailable.

Files Needing Attention: docs/source/policy_deployment/05_leapp/exporting_policies_with_leapp.rst; docs/source/tutorials/06_exporting/exporting_direct_workflow_policies_with_leapp.rst

Important Files Changed

Filename Overview
docs/source/policy_deployment/05_leapp/exporting_policies_with_leapp.rst Adds backend-specific export/deployment guidance, but the wrapper alternatives omit the dependency installation required on a clean clone.
docs/source/tutorials/06_exporting/exporting_direct_workflow_policies_with_leapp.rst Aligns the Direct-workflow tutorial with unified training and backend selection, while inheriting the wrapper-extra gap in its alternative commands.

Reviews (1): Last reviewed commit: "Docs: Point LEAPP export at train and --..." | Re-trigger Greptile

Comment on lines +114 to +117
# Newton backend (kitless)
./isaaclab.sh -p \
scripts/reinforcement_learning/leapp/<rl_library>/export.py \
--task <TASK_NAME> physics=newton_mjwarp

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P1 Wrapper commands omit required extras

When a clean-clone user selects an isaaclab.sh or isaaclab.bat example, the wrapper only invokes the existing Python interpreter and does not install the specialized LEAPP or backend extras, causing export or deployment to fail because the required packages are unavailable.

@isaaclab-review-bot isaaclab-review-bot Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Isaac Lab Review Bot

The documentation aligns LEAPP export and deployment with the backend-extra and physics=... selection pattern used by training. However, removing the persistent leapp installation step leaves the retained isaaclab.sh and isaaclab.bat workflows without documented instructions for installing an extra that is explicitly excluded from --extra all.

  • Design and architecture: The Newton, OV PhysX, and Isaac Sim PhysX command structure is consistent across export and deployment. The wrapper tabs remain incompatible with the new exclusively per-command extras guidance because those wrappers cannot express uv run --extra.
  • API: No code or CLI contract changes are introduced. The documented checkpoint, visualization, export-path, and physics-selection surfaces are presented consistently, but the installation contract is incomplete for wrapper users.
  • Implementation: The uv commands supply leapp and the appropriate backend extras directly. In contrast, the isaaclab.sh and isaaclab.bat commands rely on an already-installed environment, while the patch removes the only persistent installation instruction; a wrapper-specific installation step should be retained or added.

Minor fixes needed. Posted 1 actionable finding inline.

Automated review; human maintainers own approval decisions.

``leapp`` is a specialized optional extra (it is not part of ``--extra all``).

.. code-block:: bash
Select extras the same way as ``isaaclab train``: add ``--extra leapp`` on every

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🟡 Warning · Implementation — Wrapper tabs lose LEAPP install instruction

This rewrite removed the only persistent install step (uv sync --inexact --extra leapp) and now explains extras solely as per-command uv run --extra flags. The retained and newly added isaaclab.sh / isaaclab.bat tabs (e.g. lines 114-127, 444-463) execute in the already-installed environment and cannot pass --extra, so readers on that path have no documented way to install leapp, which the same section states is excluded from --extra all. Keep an install step for the wrapper path.

@kellyguo11
kellyguo11 merged commit 01f01ed into isaac-sim:develop Aug 20, 2026
45 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

documentation Improvements or additions to documentation

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants