Add T5gemma2

[jrabary]

Resolves #108 (only t5gemma2)

Reference

• Paper: https://arxiv.org/pdf/2512.14856
• Website: https://blog.google/technology/developers/t5gemma-2/
• Reference code https://github.com/huggingface/transformers/blob/main/src/transformers/models/t5gemma2/modeling_t5gemma2.py

Checklist

• I have read the Contribution Guidelines and used pre-commit hooks to format this commit.
• I have added all the necessary unit tests for my change. (run_model.py for model usage, test_outputs.py and/or model_validation_colab.ipynb for quality).
• (If using an LLM) I have carefully reviewed and removed all superfluous comments or unneeded, commented-out code. Only necessary and functional code remains.
• I have signed the Contributor License Agreement (CLA).

[gemini-code-assist]

Summary of Changes

Hello @jrabary, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request introduces the T5Gemma2 model, a sequence-to-sequence transformer model implemented in JAX. It incorporates both text and vision processing capabilities, utilizing a combination of global and sliding window attention mechanisms. The implementation is designed to support efficient autoregressive decoding through the use of KV caches.

Highlights

• Model Implementation: Adds the T5Gemma2 model in JAX, resolving issue T5Gemma and T5Gemma2 #108. This includes the model's architecture, configuration, and necessary components for both the encoder and decoder.
• Attention Mechanisms: Implements both global and local sliding window attention mechanisms, with specific configurations for different layers within the model.
• Multimodal Support: Includes vision embedding capabilities, allowing the model to process image inputs alongside text. This involves vision-specific layers and utilities for merging vision and text embeddings.
• Autoregressive Decoding: Integrates nnx.Cache for efficient autoregressive decoding, including initialization and management of KV caches for both self-attention and cross-attention.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

The pull request introduces the T5Gemma2 model, including its configuration, various components like attention mechanisms, embeddings, and multimodal utilities. The overall structure is clear, separating encoder and decoder functionalities. However, there are several instances where Gemma3 prefixed classes are used instead of their T5Gemma2 counterparts, indicating potential copy-paste errors or incomplete renaming. Additionally, some type hints are undefined, and a crucial parameter for the feed-forward network in the decoder blocks is not explicitly passed, which could lead to inconsistent behavior compared to the encoder.