New AutoGluon Webpage #2924

gidler · 2023-02-15T17:26:15Z

Description of changes:

This change is a major facelift on the AutoGluon webpage. It allows AutoGluon to leave behind the outdated version of d2l-book that the webpage was tied to, along with limitations on the version of Sphinx that could be used. We now use the sphinx furo theme, but can be more easily changed in the future by choosing other sphinx-design themes.

This change touches most files in the docs directory. I intend to split the review of this PR across:

changes in .github directory
changes to website config and build process
changes in tutorials directory which can be inspected by different vertical teams

Tutorials are now maintained in ipynb notebooks rather than md files. This change come with significant pros and cons:
Pros:

easy for users to download and use AutoGluon tutorials directly from repository
easy to maintain notebooks that can be opened in Colab or Studio lab

Cons:

authoring and reviewing ipynb files is more difficult (using this tool can help mitigate this)

When reviewing these changes, it is probably easier to inspect tutorials on the new webpage than it will be to look through raw files and diffs.

This PR is to merge the new website onto master/dev. Once we are comfortable with it, we will merge onto stable as well.

This checklist of items needs to be done before merge:

AutoGluon 0.7 Release
Remove new branch url from Colab/Studio Lab links
Update 0.6.2 to 0.7.0 in new docs
Add/remove pre- from correct locations in Install page

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

review-notebook-app · 2023-02-15T17:26:21Z

Check out this pull request on

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

github-actions · 2023-02-15T19:00:58Z

Job PR-2924-7d38808 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/7d38808/index.html

github-actions · 2023-02-16T23:49:52Z

Job PR-2924-dce8df7 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/dce8df7/index.html

github-actions · 2023-02-17T20:32:33Z

Job PR-2924-c0fc7f8 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/c0fc7f8/index.html

github-actions · 2023-02-18T01:00:26Z

Job PR-2924-2e2e60a is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/2e2e60a/index.html

github-actions · 2023-02-20T20:54:09Z

Job PR-2924-a5fb2a6 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/a5fb2a6/index.html

Innixma

Awesome work! Added comments.

In general, Tabular LGTM!

docs/install.md

Innixma · 2023-02-21T02:16:24Z

docs/install.md

+Modules and optional dependencies:
+
+1. `autogluon.tabular`: functionality for tabular data (TabularPredictor)
+   - Optional dependencies included by default: `lightgbm`,`catboost`,`xgboost`,`fastai`.


These are part of the optional dependency all, which is included by default during pip install autogluon.

Calling pip install autogluon.tabular will not install these dependencies. The user will need to specify pip install autogluon.tabular[all].

I updated the modularized install instructions on my most recent push.

Innixma · 2023-02-21T02:17:03Z

docs/install.md

+
+1. `autogluon.tabular`: functionality for tabular data (TabularPredictor)
+   - Optional dependencies included by default: `lightgbm`,`catboost`,`xgboost`,`fastai`.
+   - Optional dependencies not included by default: `vowpalwabbit`, `skex`. The later will speedup KNN training and inference on CPU by 25x.


later -> latter

Missing mention of optional dependencies imodels and skl2onnx

I updated the modularized install instructions on my most recent push.

docs/install.md

docs/tutorials/tabular/tabular-quick-start.ipynb

Innixma · 2023-02-21T03:06:43Z

docs/tutorials/tabular/tabular-quick-start.ipynb

@@ -0,0 +1,345 @@
+{


This quick start is pretty good. We may want to consider using titanic instead of this example (although we definitely should keep this example for at least 1 tutorial, since it is quite good). For the purposes of an initial merge, I am ok with this tutorial.

The main thing we should add to this tutorial is a comparison of how simple this solution is + how strong the result is compared to DeepMind's original solution that was on the cover of nature (300+ lines of code & 80% accuracy with DeepMind vs 3 lines of code & 95% accuracy with AutoGluon). This comparison shouldn't be present in a quick-start tutorial, which is why we should wait to add that comparison until we make a different quick-start tutorial for Tabular.

docs/api/autogluon.tabular.rst

github-actions · 2023-02-21T07:21:58Z

Job PR-2924-3a58579 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/3a58579/index.html

github-actions · 2023-02-21T23:25:21Z

Job PR-2924-185d50f is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/185d50f/index.html

github-actions · 2023-02-22T00:17:29Z

Job PR-2924-1fe92ce is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/1fe92ce/index.html

github-actions · 2023-02-22T01:58:47Z

Job PR-2924-9226be5 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/9226be5/index.html

github-actions · 2023-02-22T18:17:59Z

Job PR-2924-8cf5222 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/8cf5222/index.html

github-actions · 2023-02-22T23:28:49Z

Job PR-2924-f5541d5 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/f5541d5/index.html

github-actions · 2023-02-27T20:43:07Z

Job PR-2924-4761f3d is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/4761f3d/index.html

github-actions · 2023-02-28T02:17:17Z

Job PR-2924-024484c is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/024484c/index.html

github-actions · 2023-03-01T01:50:46Z

Job PR-2924-29f5bc5 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/29f5bc5/index.html

yinweisu

LGTM in general regarding the workflow change. Had one minor comment

yinweisu · 2023-03-01T18:22:38Z

.github/workflow_scripts/build_all_docs.sh

-sed -i -e "s@###_PLACEHOLDER_WEB_CONTENT_ROOT_###@http://$site@g" docs/config.ini
-sed -i -e "s@###_OTHER_VERSIONS_DOCUMENTATION_LABEL_###@$other_doc_version_text@g" docs/config.ini
-sed -i -e "s@###_OTHER_VERSIONS_DOCUMENTATION_BRANCH_###@$other_doc_version_branch@g" docs/config.ini


With the new website how do we generate link to stable/dev one dynamically?

These ad-hoc replacement strings were used in the nav-bar of the old website to link to things like versions, api, tutorials, etc. These exact links don't exist on the new webpage, but their equivalent sections are handled by the sphinx build process automatically. I believe these only needed to be done this way because the links were hard coded in the config file, not built by sphinx.

github-actions · 2023-03-01T19:05:47Z

Job PR-2924-f44ebae is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/f44ebae/index.html

tonyhoo

LGTM in general. Thanks for the change and the new website looks great! Some minor comments regarding installation guide and build process

.github/workflow_scripts/build_all_docs.sh

docs/ReleaseInstructions.md

tonyhoo · 2023-03-01T18:52:32Z

docs/cheatsheet.md

+
+Download the PDF version with clickable links [multimodal-cheatsheet].
+
+## Time Series


Wondering if we have the cheatsheet to be added for eda?

An EDA Cheat Sheet did not exist last time I synced with Alex. I'm not sure if Alex is planning on making one in the future.

tonyhoo · 2023-03-01T19:05:41Z

docs/install-mac-libomp.md

+# Uninstall libomp if it was previous installed
+brew uninstall -f libomp
+wget https://raw.githubusercontent.com/Homebrew/homebrew-core/fb8323f2b170bd4ae97e1bac9bf3e2983af3fdb0/Formula/libomp.rb
+brew install libomp.rb


We need to add rm libomp.rb after installation

Good catch, I will add in next push

Innixma · 2023-03-01T19:27:03Z

docs/install-gpu-pip.md

+# Install the proper version of PyTorch following https://pytorch.org/get-started/locally/
+pip install torch==1.13.1+cu116 torchvision==0.14.1+cu116 --extra-index-url https://download.pytorch.org/whl/cu116


I thought we didn't need to manually install torch for GPU? Shouldn't it install automatically via pip install autogluon?

My intention was for these to be the exact same as they are on the existing website. I am not sure if your statement about not needing to manually install torch for GPU is correct but can look into it later.

Innixma · 2023-03-01T19:27:26Z

docs/install-gpu-source.md

+# Install the proper version of PyTorch following https://pytorch.org/get-started/locally/
+pip install torch==1.13.1+cu116 torchvision==0.14.1+cu116 --extra-index-url https://download.pytorch.org/whl/cu116


I thought we didn't need to manually install torch for GPU? Shouldn't it install automatically?

See above comment

Innixma · 2023-03-01T19:30:12Z

docs/install.md

+
+:::{dropdown} M1 and M2 Apple Silicon
+
+Apple Silicon is now supported via the `conda` installation instructions outlined above. `conda-forge` will install the GPU version if a user's machine supports it.


@gradientsky please confirm if this is correct

Innixma

In general LGTM! We can make smaller PRs for additional improvements & follow-ups after this is merged. Amazing work!

tonyhoo

LGTM!

github-actions · 2023-03-02T01:00:01Z

Job PR-2924-ad09d55 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/ad09d55/index.html

github-actions · 2023-03-02T01:26:45Z

Job PR-2924-79b4e81 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/79b4e81/index.html

Innixma

Looks great!! Added some minor comments

docs/install-modules.md

github-actions · 2023-03-02T17:27:48Z

Job PR-2924-0ff3285 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/0ff3285/index.html

github-actions · 2023-03-03T18:01:45Z

Job PR-2924-0ff3285 is done.
Docs are uploaded to http://autogluon-staging.s3-website-us-west-2.amazonaws.com/PR-2924/0ff3285/index.html

Co-authored-by: Oleksandr Shchur <shchuro@amazon.com>

gidler added model list checked You have updated the model list after modifying multimodal unit tests/docs API & Doc Improvements or additions to documentation labels Feb 15, 2023

gidler force-pushed the new branch 2 times, most recently from a9176a4 to c0fc7f8 Compare February 17, 2023 18:59

gidler force-pushed the new branch from c0fc7f8 to 2e2e60a Compare February 17, 2023 23:26

gidler force-pushed the new branch from 2397e48 to a5fb2a6 Compare February 20, 2023 19:28

Innixma reviewed Feb 21, 2023

View reviewed changes

Innixma added the priority: 0 Maximum priority label Feb 21, 2023

Innixma added this to the 0.7 Fast-Follow Items milestone Feb 21, 2023

gidler added 6 commits March 1, 2023 09:09

Bring mli ag_docs into autogluon repo

1fa31fd

Remove unused files

58642af

New quickstart notebooks

78a2947

Update gitignore

212a68a

Update landing/index page

6f1f2b9

Update install page

c84c55a

gidler added 3 commits March 1, 2023 09:23

Rebase onto latest master

8fb2139

Remove "new docs" banner

582a07c

Change Colab notebook paths from new to master

f44ebae

gidler force-pushed the new branch from 29f5bc5 to f44ebae Compare March 1, 2023 17:04

yinweisu approved these changes Mar 1, 2023

View reviewed changes

tonyhoo reviewed Mar 1, 2023

View reviewed changes

Innixma reviewed Mar 1, 2023

View reviewed changes

Innixma approved these changes Mar 1, 2023

View reviewed changes

gidler added 4 commits March 1, 2023 14:01

Fix linux cpu source install instructions

039acaa

Docs point to core.TabularDataset rather than tabular.TabularDataset

5b712c9

Add "rm libomp.rb" line to mac install instructions

e7985ee

Replace Gluon favicon png with AutoGluon favicon ico

4955dcd

tonyhoo approved these changes Mar 1, 2023

View reviewed changes

Copy and update install submodule text from existing website

79b4e81

gidler force-pushed the new branch from ad09d55 to 79b4e81 Compare March 1, 2023 23:27

Innixma approved these changes Mar 2, 2023

View reviewed changes

docs/install-modules.md Outdated Show resolved Hide resolved

docs/install-modules.md Outdated Show resolved Hide resolved

docs/install-modules.md Outdated Show resolved Hide resolved

docs/install-modules.md Outdated Show resolved Hide resolved

Customized module install instructions update

0ff3285

Update google analytics tag

4e81aee

gidler merged commit db8a1b1 into master Mar 6, 2023

gidler added a commit to gidler/autogluon that referenced this pull request Mar 7, 2023

New AutoGluon Webpage (autogluon#2924)

aaa2947

Co-authored-by: Oleksandr Shchur <shchuro@amazon.com>

gidler mentioned this pull request Mar 7, 2023

Merge new webpage onto stable #3018

Merged

5 tasks

Innixma modified the milestones: 0.7 Fast-Follow Items, 0.8 Release May 23, 2023

AnirudhDagar mentioned this pull request Nov 9, 2023

[Fix] Fixing CI build docs failure #3676

Merged

Innixma deleted the new branch November 30, 2023 01:57


		Download the PDF version with clickable links [multimodal-cheatsheet].

		## Time Series

		# Install the proper version of PyTorch following https://pytorch.org/get-started/locally/
		pip install torch==1.13.1+cu116 torchvision==0.14.1+cu116 --extra-index-url https://download.pytorch.org/whl/cu116


		:::{dropdown} M1 and M2 Apple Silicon

		Apple Silicon is now supported via the `conda` installation instructions outlined above. `conda-forge` will install the GPU version if a user's machine supports it.

New AutoGluon Webpage #2924

New AutoGluon Webpage #2924

Conversation

gidler commented Feb 15, 2023 • edited

review-notebook-app bot commented Feb 15, 2023

github-actions bot commented Feb 15, 2023

github-actions bot commented Feb 16, 2023

github-actions bot commented Feb 17, 2023

github-actions bot commented Feb 18, 2023

github-actions bot commented Feb 20, 2023

Innixma left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

github-actions bot commented Feb 21, 2023

github-actions bot commented Feb 21, 2023

github-actions bot commented Feb 22, 2023

github-actions bot commented Feb 22, 2023

github-actions bot commented Feb 22, 2023

github-actions bot commented Feb 22, 2023

github-actions bot commented Feb 27, 2023

github-actions bot commented Feb 28, 2023

github-actions bot commented Mar 1, 2023

yinweisu left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

github-actions bot commented Mar 1, 2023

tonyhoo left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Innixma left a comment

Choose a reason for hiding this comment

tonyhoo left a comment

Choose a reason for hiding this comment

github-actions bot commented Mar 2, 2023

github-actions bot commented Mar 2, 2023

Innixma left a comment

Choose a reason for hiding this comment

github-actions bot commented Mar 2, 2023

github-actions bot commented Mar 3, 2023

gidler commented Feb 15, 2023 •

edited