Neuron SDK Release 2.25.0

Neuron SDK Release 2.25.0

Author: awsjoshir

.github/pull_request_template.md

@@ -1,50 +1,73 @@
-*Description:*
+**IMPORTANT!** _If this is a documentation PR for a specific release, this PR must go the corresponding release branch_ (`release-X.XX.X`). _If it is an "out-of-band" doc update, the PR must go to the_ `master` _branch_.
 
-**MANDATORY: PR needs test run output**
 
-**Test Run Output:**
-Please specify the release version, instance size and type, OS type and test output.
+## Required PR information
 
-Training tutorial:
-Convergence graph for training tutorials
-Performance metrics average_throughput, latency_p50, latency_p99 and MFU% if available
+To expedite approvals and merges for releases, provide the following information (select the `...` button to the right at the top of your PR message to edit it):
 
+> **AWS email alias**: {_your-name_}@amazon.com
 
-Please make sure this PR contains correct classification terms (Alpha, Beta, and Stable).
+>**Description**: {_What this documentation change is and why you made it. If you have a corresponding Jira ticket or content plan, link it here. The more details you provide around any decisions you made when preparing the docs, the less annoying comments you'll get preparing to release it._}
 
-If possible, provide your results or a link to them for the reviewer to check your work.
+> **Date this must be published by**: {_If empty, we will assume the release date for the branch you're merging into._}
 
+> **Link to ReadTheDocs staging for this branch's doc changes**: https://awsdocs-neuron-staging.readthedocs-hosted.com/en/{YOUR_BRANCH_NAME_HERE}/
 
-*Issue #, sim, or t.corp if available:*
+> **Set the `docs-review-needed` label on the PR for tracking.**
 
+## Before you request approvals
 
-*Link to RTD for my changes:* 
-https://awsdocs-neuron-staging.readthedocs-hosted.com/en/YOUR_BRANCH_NAME/
+> Run a spelling and grammar check over your prose and make the changes it suggests. VSCode has a number of extensions (cSpell, LTeX) that you can use. You can also provide the rendered HTML for (or a cut-and-paste of) your pages to an AI and have it correct your spelling, grammar, and formatting issues. If you need an advanced prompt, contact @erickson-doug.
 
+## Approvers
 
-*Additional context:*
+We require 3-4 approvers to merge for non-trivial content changes (where a "trivial" change is a typo/grammar fix or a minor update to the format syntax):
 
+1. A senior+ engineer who will review your documentation for technical accuracy and clarity in communicating the technical concepts in your work
+2. A product manager for your Neuron component area who will review it for customer relevance and product/component/feature messaging
+3. The lead tech writer (@erickson-doug) who will review your work for overall doc design and quality, and perform the merge when all approvals are met
+4. (For PRs with code/notebook submissions) A QA/test engineer who can run your code and confirm the results.
 
+Make sure you get a commitment from these reviewers in advance! It's hard to get good quality doc reviews in order in the 11th hour of a release.
 
+**Note**: For trivial changes, you only need @erickson-doug's approval. He will merge your content once he's confirmed the fixes on staging.
 
+## Doc review checklist
 
-## PR Checklist
-- [ ] I've completely filled out the form above!
-- [ ] (If applicable) I've automated a test to safegaurd my changes from regression.
-- [ ] (If applicable) I've posted test collateral to prove my change was effective and not harmful.
-- [ ] (If applicable) I've added someone from QA to the list of reviewers.  Do this if you didn't make an automated test or feel it's appropriate for another reason.
-- [ ] (If applicable) I've reviewed the licenses of updated and new binaries and their dependencies to make sure all licenses are on the pre-approved Amazon license list.  See https://inside.amazon.com/en/services/legal/us/OpenSource/Pages/BlessedOpenSourceLicenses.aspx.
+### Engineering reviewer checklist
+
+- [ ] I've confirmed that the contributions in this PR meet the current  [AWS Neuron writing guidelines](https://quip-amazon.com/m97CAO0kQFEU/Writing-for-AWS-Neuron).
+- [ ] I've confirmed that the documentation submitted is technically correct to the best of my knowledge.
+- [ ] I've confirmed that the documentation submitted has no spelling or grammar errors or use of internal jargon/terminology.
+- [ ] I've verified the changes render correctly on RTD (link above).
+- [ ] (If code is included) I've run tests to verify the contents of the change.
+
+---
+
+## For PRs that include code or notebook examples
+
+**MANDATORY: PR must include test run output**
 
+Provide this information for the QA reviewer in order to expedite their review.
 
-## Pytest Marker Checklist
-(Coming soon...)
+**Test run output:**
+Specify the release version, instance size and type, OS type and test output.
 
+**For Training tutorials:**
 
-## Reviewer Checklist
-* For reviewer, before merging, please cross-check:
-- [ ] I've verified the changes render correctly on RTD (link above)
-- [ ] I've ensured the submitter completed the form 
-- [ ] (If appropriate) I've run tests to verify the contents of the change
+{Convergence graph for training tutorials}
 
+{Performance metrics `average_throughput`, `latency_p50`, `latency_p99` and MFU% if available}
+
+Make sure this PR contains correct classification terms (Alpha, Beta, and Stable).
+
+If possible, provide your results or a link to them for the reviewer to check your work.
+
+## Code example/notebook content PR checklist
+
+- [ ] (If applicable) I've automated a test to safeguard my changes from regression.
+- [ ] (If applicable) I've posted test collateral to prove my change was effective and not harmful.
+- [ ] (If applicable) I've added someone from QA to the list of reviewers.  Do this if you didn't make an automated test or feel it's appropriate for another reason.
+- [ ] (If applicable) I've reviewed the licenses of updated and new binaries and their dependencies to make sure all licenses are on the pre-approved Amazon license list.  See https://inside.amazon.com/en/services/legal/us/OpenSource/Pages/BlessedOpenSourceLicenses.aspx.
 
-By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.
+By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

.gitignore

@@ -10,4 +10,5 @@ src/examples/pytorch/libtorch_demo.tar.gz
 *-checkpoint.ipynb
 .idea/
 .vscode/
-*/nki/*/generated/
+*/nki/*/generated/
+uncommitted/

CONTRIBUTING.md

@@ -19,7 +19,6 @@ reported the issue. Please try to include as much information as you can. Detail
 * Any modifications you've made relevant to the bug
 * Anything unusual about your environment or deployment
 
-
 ## Contributing Workflow (via Pull Requests)
 Contributions via pull requests are much appreciated. Before sending us a pull request, please ensure that:
 
@@ -29,28 +28,44 @@ Contributions via pull requests are much appreciated. Before sending us a pull r
 
 To send us a pull request, please:
 
-1. Fork the repository (See: [fork guide](https://help.github.com/articles/fork-a-repo/))
 2. Clone the repository locally:
+
     ```bash
-    git clone [email protected]:YOUR-USERNAME/aws-neuron-sdk.git
+    git clone [email protected]:YOUR-USERNAME/private-aws-neuron-sdk-staging.git
     ```
+
 3. Install the build dependencies. This requires a Python 3 installation:
+
     ```bash
-    cd aws-neuron-sdk/
+    cd .. # The root folder where you have your cloned Git repos
+    python3 -m venv venv && . venv/bin/activate
+    pip install -U pip
+    cd private-aws-neuron-sdk-staging
     pip install -r requirements.txt
     ```
+
 4. Build the documentation into HTML. This command will allow you to view the
    rendered documentation by opening the generated `_build/html/index.html`.
+
+   Run:
+
+   ```bash
+   sphinx-build -b html . _build/html
+   ```
+
+   If this doesn't work, try this command:
+
    ```bash
-   make html
+   sphinx-build -C -b html . _build/html
    ```
+
 2. Modify the source; please focus on the specific change you are contributing. If you also reformat all the code, it will be hard for us to focus on your change.
-3. Rebuild the documentation (`make html`). Always ensure that this builds 
-   correctly and that your contribution is rendered as expected.
-4. Commit to your fork using clear commit messages.
-5. Send us a pull request (See: [pull request guide](https://help.github.com/articles/creating-a-pull-request/)), answering any default questions in the pull request interface.
+3. Rebuild the documentation (`sphinx-build -b html . _build/html`). Always ensure that the docs build without errors and that your changes look correct before pushing your changes to remote.
+4. Commit your changes to your branch with a clear commit messages. 
+5. Push your changes to remote (`git push origin`) and create a PR from your branch into `master`. (See: [pull request guide](https://help.github.com/articles/creating-a-pull-request/)). Answer any default questions in the pull request interface.
 6. Pay attention to any automated CI failures reported in the pull request, and stay involved in the conversation.
 
+Updated process documentation can be found here: [Neuron Sphinx/Readthedocs FAQ](https://quip-amazon.com/ebN0AYH7lAKW/Neuron-SphinxReadthedocs-FAQ).
 
 ## Finding contributions to work on
 Looking at the existing issues is a great way to find something to contribute on. As our projects, by default, use the default GitHub issue labels (enhancement/bug/duplicate/help wanted/invalid/question/wontfix), looking at any 'help wanted' issues is a great place to start.

Makefile

@@ -12,9 +12,13 @@ BUILDDIR      = _build
 help:
 	@$(SPHINXBUILD) -M help "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
 
-.PHONY: help Makefile
+.PHONY: help Makefile clean
 
 # Catch-all target: route all unknown targets to Sphinx using the new
 # "make mode" option.  $(O) is meant as a shortcut for $(SPHINXOPTS).
 %: Makefile
 	@$(SPHINXBUILD) -M $@ "$(SOURCEDIR)" "$(BUILDDIR)" $(SPHINXOPTS) $(O)
+
+clean:
+	-rm -rf $(BUILDDIR)/*
+

_templates/search-field-google.html

@@ -10,25 +10,29 @@ <h1>{{ _("Search") }}</h1>
           functionality.{% endtrans %}</p>
       </div>
     </noscript>
-    {% block search_form %}
-        {% include "search-field-google.html" %}
-    {% endblock %}
 
     <h2>{{ _('Search Results') }}</h2> 
+
     <script async src="https://cse.google.com/cse.js?cx=657ffaecc36684ee1">
     </script>
     <div class="gcse-searchresults-only"></div> 
-
+    <div id="search-results"></div>
   </div>
-  <script>
-    // Activate the search field on page load
-    let searchInput = document.querySelector("form.bd-search input");
-    if (searchInput) {
-      searchInput.focus();
-      searchInput.select();
-      console.log("[PST]: Set focus on search field.");
-    }
-  </script>
+<script>
+
+// Activate the search field on page load
+let searchForm2 = document.getElementById('search-form');
+let searchFormInput2 = searchForm2.querySelector('input[type="search"]');
+
+if (searchFormInput2) {
+  searchFormInput2.focus();
+  searchFormInput2.select();
+  console.log("[PST]: Set focus on search field.");
+  searchFormInput2.value = localStorage.getItem('lastSearchQuery');
+} 
+
+</script>
+
 {% endblock docs_body %}
 {# Below sections just re-create the behavior of Sphinx default search #}
 {# Page metadata #}
@@ -38,7 +42,4 @@ <h2>{{ _('Search Results') }}</h2>
 {# Manually include the search JS that Sphinx includes #}
 {% block scripts -%}
   {{ super() }}
-  <script src="{{ pathto('_static/searchtools.js', 1) }}"></script>
-  <script src="{{ pathto('_static/language_data.js', 1) }}"></script>
-  <script src="{{ pathto('searchindex.js', 1) }}"></script>
 {%- endblock scripts %}

_templates/search-field-rtd.html

@@ -0,0 +1,39 @@
+{%- extends "page.html" %}
+{# Over-ride the body to be custom search structure we want #}
+{% block docs_body %}
+  <div class="bd-search-container">
+    <h1>{{ _("Search") }}</h1>
+    <noscript>
+      <div class="admonition error">
+        <p class="admonition-title">{% trans %}Error{% endtrans %}</p>
+        <p>{% trans %}Please activate JavaScript to enable the search functionality.{% endtrans %}</p>
+      </div>
+    </noscript>
+    <div id="search-results"></div>
+  </div>
+  <script>
+// Activate the search field on page load
+let searchForm3 = document.getElementById('search-form');
+let searchFormInput3 = searchForm3.querySelector('input[type="search"]');
+
+
+if (searchFormInput3) {
+  searchFormInput3.focus();
+  searchFormInput3.select();
+  console.log("[PST]: Set focus on search field3.");
+  searchFormInput3.value = localStorage.getItem('lastSearchQuery');
+} 
+  </script>
+{% endblock docs_body %}
+{# Below sections just re-create the behavior of Sphinx default search #}
+{# Page metadata #}
+{%- block htmltitle -%}
+  <title>{{ _("Search") }} - {{ title or docstitle }}</title>
+{%- endblock htmltitle -%}
+{# Manually include the search JS that Sphinx includes #}
+{% block scripts -%}
+  {{ super() }}
+  <script src="{{ pathto('_static/searchtools.js', 1) }}"></script>
+  <script src="{{ pathto('_static/language_data.js', 1) }}"></script>
+  <script src="{{ pathto('searchindex.js', 1) }}"></script>
+{%- endblock scripts %}

_templates/search-results.html renamed to _templates/search-google.html

@@ -0,0 +1,53 @@
+.. meta::
+   :description: Archived AWS Neuron SDK documentation
+   :keywords: AWS Neuron SDK, archived tutorials, legacy documentation
+
+=====================================
+Archived AWS Neuron SDK documentation
+=====================================
+
+.. note::
+
+    This page contains archived tutorials and other documentation for older versions of the AWS Neuron SDK.
+    These pages are no longer actively maintained and may reference deprecated features or APIs. They are provided as-is and may not reflect the current state of the AWS Neuron SDK.
+
+Overview
+--------
+
+The following content has been archived for reference purposes. For the latest documentation and guides, visit the `AWS Neuron SDK documentation <https://awsdocs-neuron.readthedocs-hosted.com/en/latest/>`_.
+
+Archived Content
+----------------
+
+.. list-table::
+   :header-rows: 1
+
+   * - Tutorial
+     - Last release supported
+     - Date archived
+   * - `Fine-tune T5 model on Trn1 <tutorials/finetune_t5>`_
+     - Neuron 2.24.0
+     - Archived on: 7/31/2025
+   * - `Running SSD300 with AWS Neuron <tutorials/ssd300_demo/ssd300_demo>`_
+     - Neuron 2.24.0
+     - Archived on: 7/31/2025
+   * - `Megatron GPT Pretraining <tutorials/megatron_gpt_pretraining>`_
+     - Neuron 2.25.0
+     - Archived on: 7/31/2025
+
+.. toctree::
+    :maxdepth: 1
+    :hidden:
+
+    tutorials/finetune_t5
+    tutorials/ssd300_demo/ssd300_demo
+    tutorials/megatron_gpt_pretraining
+
+Accessing Archived Content
+--------------------------
+
+Each tutorial listed above corresponds to a specific version or feature set of the Neuron SDK that has since been superseded. Use these resources for historical context or migration guidance.
+
+.. warning::
+
+    Archived tutorials may not be compatible with current Neuron SDK releases. Exercise caution when following instructions from these documents.

_templates/search.html

@@ -1,10 +1,15 @@
 .. _torch-hf-t5-finetune:
 
+.. meta::
+   :noindex:
+   :nofollow:
+   :description: The tutorial for the AWS Neuron SDK is currently deprecated and not maintained. It is provided for reference only.
+
 Fine-tune T5 model on Trn1
 ================================
 
 .. note:: 
-   Update 01/03/24: This tutorial is currently broken and the AWS Neuron team is working on the fix.
+   This page was archived on 7/31/2025.
 
 
 In this tutorial, we show how to fine-tune a Hugging Face (HF) T5 model