Enhance README with Customization details

[ram-cherukuri]

Added sections on codebase architecture, customization guide. Included example configuration for adapting GeoTransolver to internal pipe flow problems.

PhysicsNeMo Pull Request

Description

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.
• The CHANGELOG.md is up to date with these changes.
• An issue is linked to this pull request.
• If I am implementing a new model or modifying any existing model, I have followed the Models Implementation Coding Standards.

Dependencies

Review Process

All PRs are reviewed by the PhysicsNeMo team before merging.

Depending on which files are changed, GitHub may automatically assign a maintainer for review.

We are also testing AI-based code review tools (e.g., Greptile), which may add automated comments with a confidence score.
This score reflects the AI’s assessment of merge readiness and is not a qualitative judgment of your work, nor is
it an indication that the PR will be accepted / rejected.

AI-generated feedback should be reviewed critically for usefulness.
You are not required to respond to every AI comment, but they are intended to help both authors and reviewers.
Please react to Greptile comments with 👍 or 👎 to provide feedback on their accuracy.

[greptile-apps]

Greptile Summary

This PR enhances the README for the transformer_models external aerodynamics example with a "Codebase Architecture" section, a multi-part "Customization Guide", an integration checklist for new CFD domains, and a worked example showing how to adapt GeoTransolver to an internal pipe-flow problem. The additions are helpful for onboarding, but two issues in the sample YAML snippet need fixing before this is ready to merge.

• [cite_start] / [cite: ...] artifacts in the YAML block (lines 287–308): Six config keys are prefixed with [cite_start] and/or suffixed with [cite: N] markers — AI citation tokens that were not removed before the PR was opened. In YAML, a bare [ opens a flow sequence, so any line starting with [cite_start] will fail to parse.
• Training config sub-tree doesn't match the real Hydra layout: The sample config places precision, optimizer: "muon", and learning_rate under a training: key. In the actual configs, precision is a root-level key, and the optimizer is configured with Hydra's _target_ pattern (not a string). A copy-paste user would see no effect and be confused.
• fp8 hardware list omits Blackwell: The new integration checklist says "Hopper/Ada Lovelace" for fp8 support, but the existing README already mentions Hopper, Blackwell, and Ada Lovelace — creating a minor but confusing inconsistency.

Important Files Changed

Filename	Overview
examples/cfd/external_aerodynamics/transformer_models/README.md	Adds architecture overview, customization guide, and a sample YAML config for pipe flow adaptation — but the YAML snippet contains [cite_start] / [cite: ...] AI-citation artifacts that make it invalid YAML, and the training sub-tree structure doesn't match the actual Hydra config layout.

_{Last reviewed commit: 4f46f1d}

[greptile-apps]

fp8 hardware list omits Blackwell

The checklist item mentions only "Hopper/Ada Lovelace", but the existing README (Training Precision section, line 62) already states that fp8 is available on "Hopper, Blackwell, Ada Lovelace, and others." Omitting Blackwell here creates an inconsistency that could mislead users with newer hardware.

Suggested change

	* [ ] **Verify Precision Compatibility**: If using **fp8**, ensure your hardware supports it (Hopper/Ada Lovelace).
	* [ ] **Verify Precision Compatibility**: If using **fp8**, ensure your hardware supports it (Hopper, Blackwell, Ada Lovelace, or other fp8-capable GPUs).

[coreyjadams]

Hi @ram-cherukuri,

I think the AI used to make these updates went off the rails here. Overall, I don't think we should be using AI to update our readmes like this without closer supervision, they need to be vetted for technical content quite closely still, almost at the level of our example codes.

A lot of the content generated here is speculative at best, and kinda wrong at worst, and we shouldn't merge this.

How do you want to proceed?

[ram-cherukuri]

@RishikeshRanade Can you take a stab a creating a psuedo config that would adapt this external aero sample to an internal flow sample

[ram-cherukuri]

@ktangsali Can you update this section to indicate how one can add Physics Loss to GeoTransolver training if they want to.