Jackyf/feat/lora nnx

[RexBearIU]

Description

This pull request introduces support for HuggingFace (HF) LoRA adapter conversion to MaxText format, expands and clarifies LoRA/QLoRA configuration options. The most significant change is the addition of a new script to convert HF LoRA adapters for use with MaxText, along with the necessary configuration and type system updates to enable this workflow.

Tests

Please describe how you tested this change, and include any instructions and/or
commands to reproduce.

Checklist

Before submitting this PR, please make sure (put X in square brackets):

• I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
• I have necessary comments in my code, particularly in hard-to-understand areas.
• I have run end-to-end tests tests and provided workload links above if applicable.
• I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.

[codecov]

Codecov Report

❌ Patch coverage is 42.06549% with 230 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/maxtext/layers/nnx_decoders.py	42.85%	180 Missing and 36 partials ⚠️
src/maxtext/models/models.py	33.33%	7 Missing and 3 partials ⚠️
src/maxtext/layers/multi_token_prediction.py	0.00%	3 Missing ⚠️
src/maxtext/layers/nnx_wrappers.py	0.00%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!