These pipelines are:
* Working out of the box (run with zero modifications)
* End-to-end (from nothing to a Vertex Model)
* Feature multiple ML frameworks (TensorFlow, PyTorch, XGBoost, Scikit-learn)
* Feature multiple training objectives: tabular classification and tabular regression
Co-authored-by: Ivan Cheung <ivans.mailbox@gmail.com>
* Minor fixes for CPR Pytorch sample: Add missing test data, add auth info to readme, scrub private project and bucket names from config, tolerate missing config.json in unit tests.
* Minor fixes for CPR Pytorch sample: Add missing test data, add auth info to readme, scrub private project and bucket names from config, tolerate missing config.json in unit tests.
* Fix merge conflicts
* fix typo
* Point CPR links to main branch of SDK repo.
Co-authored-by: Andrew Ferlitsch <aferlitsch@google.com>
* samples: Add a new sample for pre-built Pytorch deployments. It's
borrowed from the examples in community-content/pytorch_text_classification_using_vertex_sdk_and_gcloud.
* samples: Removed all training related stuff in the notebooks.
* samples: Fixed comments.
* samples: Updated readme.
* samples: Updated emails for Pytorch launch.
less chance for an error and confusion in name clashing with the `datasets` pypi package also used in the notebook.
Co-authored-by: Andrew Ferlitsch <aferlitsch@google.com>
* Add an example use case for custom prediction routines.
* Addressing some PR comments: reworded the readme in a few places, added a 'probe' command to build.py that sends a sample predict request, and pinned versions in requirements. Also fixed a bug where the artifacts_uri passed in during deployment on Vertex AI was not recognized as a directory.
* Autoformat code with black and fix a couple of typing errors.
* Addressing PR comments: Add deployment machine type to the config and add docstring to probe_prediction method.
* Update example to work with new LocalModel interface.
* Update example to work with new LocalModel interface.
Co-authored-by: Karl Weinmeister <11586922+kweinmeister@users.noreply.github.com>
In "Step by Step Guide to Building Reinforcement Learning Applications using Vertex AI", the replay_buffer was unbound if training_data_spec_transformation_fn was provided to the train() function
Co-authored-by: Karl Weinmeister <11586922+kweinmeister@users.noreply.github.com>
* PyTorch on Vertex - Updated to match GCPC v0.2.2 API
* PyTorch on Vertex - Fixes based on review comments
* PyTorch on Vertex - fixes based on review
* PyTorch on Vertex - linter fixes
* PyTorch on Vertex - fixes based on feedback
* PyTorch on Vertex - fixes based on feedback
Co-authored-by: Karl Weinmeister <11586922+kweinmeister@users.noreply.github.com>
Co-authored-by: Andrew Ferlitsch <aferlitsch@google.com>
* PyTorch on Vertex - Updated to match GCPC v0.2.2 API
* PyTorch on Vertex - Fixes based on review comments
Co-authored-by: Karl Weinmeister <11586922+kweinmeister@users.noreply.github.com>
* Original alphafold Dockerfile and notebook as a starting point
* Migrate dependencies from notebook into Dockerfile
* adding build_docker.sh script for building docker image
* remove cell, and replace with accelerator configuration cell. Also remove dependency on google.colab
* dockerfile remove redundant
* Changing text in launch button
* add vertexai.png
* update notebook, including permalink to vertexai.png
* updating notebook markdown and correcting the vertexai.png image display
* add div brackets and fix broken launch link
* table instead of div
* width=40
* resizing vertexai image
* updated FAQ
* updated Licence
* add back in the output_file zip
* launch button at top of notebook
* default workdir aligned with JuptyerLab home directory
* intro paragraph
* update CPU instructions
* exchange notebook title and launch header
* Dockerfile license
* collapsing cells
* splitting out sequences into un-collapsed cell
* Update licence
* update download instructions text
* Remove "double-click" text
* hide cells
* Increasing indent to pass linting for alphafold_on_gcp (#224)
* increasing indent to pass linting
* more linting
* wild
* AMBER relaxation f-string
* hidden cells
* wild commit
* move links to main
* added sklearn-example
* nb formatting
* added readme
* added endpoint to notebook
* nb formatting
* added endpoint prediction example
* nb formatting
* fixed request
* minor woring fixes in docstrings
* added codeowner and included PR feedback
Co-authored-by: Maximilian Engelhardt <maximilian.engelhardt@ing.com>
* Made the Reinforcement Learning components shareable
Also made couple of fixes.
* Updated the sample pipeline
* Replaced .after with data passing
* Formatted the notebook using tensorflow_docs.tools.nbfmt
* Fixed all linter errors
* Updated the component URLs
* pytorch on vertex: initial commit
* pytorch on vertex: reorg dir structure with new repo changes
* pytorch on vertex: add cleanup script and update README files
* pytorch on vertex: remove references to bucket names
* PyTorch on Vertex: Updated with linter suggested changes
* PyTorch on Vertex: dry-run and set resource names consistent with app name
* PyTorch on Vertex: update CODEOWNERS
* Pytorch on Vertex: Reorganized directory structure
* Added step-by-step and pipeline tutorials for reinforcement learning with Vertex AI.
* Add READMEs for the step-by-step and end-to-end MLOps pipeline prototypes.
* Make dir and file names more descriptive, and adjust embedded links. Directly load dataset from remote bucket.
* Add TF-Agents bandits movie RecSys demo owner.
