Migrate devsite notebooks to gemma-cookbook #276

bebechien · 2026-01-05T09:36:05Z

No description provided.

review-notebook-app · 2026-01-05T09:36:11Z

Check out this pull request on

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

gemini-code-assist · 2026-01-05T09:37:05Z

Summary of Changes

Hello @bebechien, 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 involves the migration of a substantial collection of educational and example notebooks related to various Gemma models and their applications. The primary goal is to centralize these resources within the gemma-cookbook repository, making them more accessible and discoverable for developers. The notebooks demonstrate a broad spectrum of functionalities, from basic text generation and image understanding to advanced fine-tuning techniques and integration with diverse machine learning frameworks, ensuring a rich learning experience for users exploring the Gemma ecosystem.

Highlights

Notebook Migration: A comprehensive set of Gemma-related notebooks have been migrated from the DevSite to the gemma-cookbook repository, consolidating examples and tutorials in a central location.
Diverse Gemma Models: New notebooks introduce various Gemma model variants, including CodeGemma, EmbeddingGemma, FunctionGemma, PaliGemma, and RecurrentGemma, showcasing their unique capabilities.
Framework Integration: The migrated content covers integration with multiple popular AI/ML frameworks such as Keras, Hugging Face Transformers, JAX/Flax, PyTorch, and Sentence Transformers.
Advanced Use Cases: Tutorials now include advanced topics like fine-tuning (LoRA, QLoRA, full fine-tuning), distributed training, embedding generation, function calling, image analysis, code assistance, and chatbot development.
Specific Notebook Additions: Key additions include video analysis with Gemma, AI-assisted programming with CodeGemma and KerasNLP, distributed tuning with Gemma on TPUs, prompting with images using the Gemma library, fine-tuning EmbeddingGemma for domain-specific tasks, and full function calling sequences with FunctionGemma.

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.

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

This pull request is a significant contribution, migrating a large number of dev-site notebooks into the gemma-cookbook. The notebooks cover a wide range of topics, from basic inference to fine-tuning for various Gemma family models. While the migration is a great step, I've found several issues across the notebooks, including broken links, typos, grammatical errors, and some critical code errors that would prevent users from running the tutorials successfully. These include incorrect model identifiers, broken image URLs, and flawed code logic in examples. I've provided detailed comments and suggestions for each issue to help improve the quality and usability of these valuable resources.

gemini-code-assist · 2026-01-05T09:40:28Z

docs/core/huggingface_text_finetune_qlora.ipynb

+        "processor = AutoTokenizer.from_pretrained(args.output_dir)\n",
+        "processor.save_pretrained(\"merged_model\")"


There's a NameError here. The variable processor is used, but it was defined as tokenizer earlier in the notebook. This should be corrected to tokenizer to avoid an error when saving.

tokenizer = AutoTokenizer.from_pretrained(args.output_dir) processor.save_pretrained("merged_model")

gemini-code-assist · 2026-01-05T09:40:28Z

docs/core/huggingface_text_finetune_qlora.ipynb

+        "if model_id == \"google/gemma-3-1b-pt\":\n",
+        "    model_class = AutoModelForCausalLM\n",
+        "else:\n",
+        "    model_class = AutoModelForImageTextToText\n",


The code incorrectly selects AutoModelForImageTextToText for text-only models like google/gemma-3-4b-pt. This will cause an error. The correct class for causal language models is AutoModelForCausalLM. Please update the logic to use the correct model class.

if model_id == "google/gemma-3-1b-pt": model_class = AutoModelForCausalLM else: model_class = AutoModelForCausalLM

gemini-code-assist · 2026-01-05T09:40:28Z

docs/paligemma/inference-with-keras.ipynb

+        "id": "QITD66qJvCTO"
+      },


The image URLs point to localhost. This will fail for any user running this notebook in a cloud environment like Colab or Kaggle, as the images won't be accessible. To make this tutorial runnable, please host these images publicly and update the URLs, or include them in the repository and load them from a local path.

gemini-code-assist · 2026-01-05T09:40:28Z

docs/core/pytorch_gemma.ipynb

+        "model_config.dtype = \"float32\" if MACHINE_TYPE == \"cpu\" else \"float16\"\n",
+        "model_config.tokenizer = tokenizer_path"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "zSp7oLLwjKCa"


The @contextlib.contextmanager decorator is used, but contextlib is not imported. This will raise a NameError. Please add import contextlib at the beginning of the notebook to resolve this.

gemini-code-assist · 2026-01-05T09:40:29Z

docs/core/keras_inference.ipynb

+        "dog_a = read_image(\"http://localhost/images/dog-a.jpg\")\n",
+        "dog_b = read_image(\"http://localhost/images/dog-b.jpg\")\n",


The image URLs point to localhost. This will fail for any user running this notebook in a cloud environment like Colab or Kaggle, as the images won't be accessible. To make this tutorial runnable, please host these images publicly and update the URLs, or include them in the repository and load them from a local path.

gemini-code-assist · 2026-01-05T09:40:29Z

DevSite/docs/core/huggingface_inference.ipynb

+        "    <a target=\"_blank\" href=\"https://ai.google.dev/gemma/docs/core/huggingface_inference\"><img src=\"https://ai.google.dev/static/site-assets/images/docs/notebook-site-button.png\" height=\"32\" width=\"32\" />View on ai.google.dev</a>\n",
+        "  </td>\n",
+        "  <td>\n",
+        "    <a target=\"_blank\" href=\"https://colab.research.google.com/github/google-gemini/gemma-cookbook/blob/main/DevSite/docs/core/huggingface_inference.ipynb\"\"><img src=\"https://www.tensorflow.org/images/colab_logo_32px.png\" />Run in Google Colab</a>\n",


There's an extra double quote at the end of the href attribute, which makes the HTML invalid and might cause rendering issues or broken links. Please remove it.

<a target="_blank" href="https://colab.research.google.com/github/google-gemini/gemma-cookbook/blob/main/DevSite/docs/core/huggingface_inference.ipynb"><img src="https://www.tensorflow.org/images/colab_logo_32px.png" />Run in Google Colab</a>

gemini-code-assist · 2026-01-05T09:40:29Z

DevSite/docs/capabilities/vision/video-understanding.ipynb

+      "source": [
+        "# Video analysis with Gemma\n",
+        "\n",
+        "Video data is a rich source of information that can help you accomplish tasks and understand the world around you. Using Gemma with video data can help you understand spacial relationships, interpret human interactions, and assist with situational awareness. This tutorial shows you how to get started processing video data with Gemma using Hugging Face Transformers. The Transformers Python library provides a API for accessing pre-trained generative AI models, including Gemma. For more information, see the [Transformers](https://huggingface.co/docs/transformers/en/index) documentation.\n",


I found a couple of typos and a grammatical error in this sentence. "spacial" should be "spatial", and "a API" should be "an API". Correcting these will improve the clarity of the documentation.

Video data is a rich source of information that can help you accomplish tasks and understand the world around you. Using Gemma with video data can help you understand spatial relationships, interpret human interactions, and assist with situational awareness. This tutorial shows you how to get started processing video data with Gemma using Hugging Face Transformers. The Transformers Python library provides an API for accessing pre-trained generative AI models, including Gemma. For more information, see the [Transformers](https://huggingface.co/docs/transformers/en/index) documentation.

gemini-code-assist · 2026-01-05T09:40:29Z

DevSite/docs/capabilities/vision/video-understanding.ipynb

+        "    <a target=\"_blank\" href=\"https://ai.google.dev/gemma/docs/capabilities/vision/video-understanding\"><img src=\"https://ai.google.dev/static/site-assets/images/docs/notebook-site-button.png\" height=\"32\" width=\"32\" />View on ai.google.dev</a>\n",
+        "  </td>\n",
+        "  <td>\n",
+        "    <a target=\"_blank\" href=\"https://colab.research.google.com/github/google-gemini/gemma-cookbook/blob/main/DevSite/docs/capabilities/vision/video-understanding.ipynb\"\"><img src=\"https://www.tensorflow.org/images/colab_logo_32px.png\" />Run in Google Colab</a>\n",


There's an extra double quote at the end of the href attribute, which makes the HTML invalid and might cause rendering issues or broken links. Please remove it.

<a target="_blank" href="https://colab.research.google.com/github/google-gemini/gemma-cookbook/blob/main/DevSite/docs/capabilities/vision/video-understanding.ipynb"><img src="https://www.tensorflow.org/images/colab_logo_32px.png" />Run in Google Colab</a>

gemini-code-assist · 2026-01-05T09:40:29Z

DevSite/docs/paligemma/inference-with-keras.ipynb

+        "id": "3_lX1k54KKrx"
+      },
+      "source": [
+        "Copyright 2024 Google LLC."


The copyright header is missing the markdown ##### prefix, which is inconsistent with the other notebooks in this PR. For consistency, please add it.

##### Copyright 2024 Google LLC.

gemini-code-assist · 2026-01-05T09:40:29Z

docs/recurrentgemma/recurrentgemma_jax_finetune.ipynb

+      "source": [
+        "Try the `MTNTDatasetBuilder` out by instantiating the custom `GriffinTokenizer` again, then applying it on the MTNT dataset, and sampling two examples:"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,


The output includes a JAX UserWarning: Some donated buffers were not usable. While this may not be a critical error, it can be confusing for users. It would be helpful to add a markdown cell explaining that this warning is expected and can be safely ignored, to improve the user experience.

irbg

Suggest dropping the DevSite directory and merging the README note into the root README.md

I'm fine with docs/ being lowercase as it works better with the URL that will be used in Vertex/Kaggle/Colab etc

DevSite/README.md

Migrate devsite notebooks to gemma-cookbook

a3c24fe

github-actions bot added the status:awaiting review label Jan 5, 2026

gemini-code-assist bot reviewed Jan 5, 2026

View reviewed changes

bebechien and others added 3 commits January 5, 2026 20:11

fix minor issues like double quote at the end of the colab link

5954204

add README

cb13ab5

Merge branch 'main' into main

ffa8611

bebechien requested review from irbg, jethac and osanseviero January 6, 2026 01:45

irbg requested changes Jan 6, 2026

View reviewed changes

DevSite/README.md Outdated Show resolved Hide resolved

bebechien added 5 commits January 7, 2026 10:42

add responsible ai notebooks

a8046f4

revert link

19e83c9

fix button links

5fa6325

remove DevSite folder, move docs and responsible to the top level

1683855

fix button links

489d7e7

bebechien requested a review from irbg January 7, 2026 02:04

bebechien and others added 5 commits January 7, 2026 11:08

fix doc link for responsible ai

6d9e782

fix links

64c1794

apply google/generative-ai-docs#616

872795a

Merge branch 'main' into main

e0ff5df

Merge branch 'main' into main

c55fdac

irbg approved these changes Jan 14, 2026

View reviewed changes

Merge branch 'main' into main

9c1a314

bebechien merged commit d359a0e into google-gemini:main Jan 15, 2026
3 checks passed

		"processor = AutoTokenizer.from_pretrained(args.output_dir)\n",
		"processor.save_pretrained(\"merged_model\")"

		"dog_a = read_image(\"http://localhost/images/dog-a.jpg\")\n",
		"dog_b = read_image(\"http://localhost/images/dog-b.jpg\")\n",

Migrate devsite notebooks to gemma-cookbook #276

Migrate devsite notebooks to gemma-cookbook #276

Uh oh!

Conversation

bebechien commented Jan 5, 2026

Uh oh!

review-notebook-app bot commented Jan 5, 2026

Uh oh!

gemini-code-assist bot commented Jan 5, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist bot Jan 5, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jan 5, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jan 5, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jan 5, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jan 5, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jan 5, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jan 5, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jan 5, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jan 5, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jan 5, 2026

Choose a reason for hiding this comment

Uh oh!

irbg left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants