BentoML-0.5.2

• New improved AWS Lambda support!

  • Support uploading large model files to s3 when deploying to AWS Lambda
  • Support trimming down the size of bundled python dependencies
  • Support setting memory size up to 3008MB for Lambda function
  • Support updating Lambda deployment to a newer version of saved BentoService bundle

• Fixed an issue when installing BentoService saved bundle as PyPI package, the setup.py file failed to parse requirements.txt as install_requires filed.

BentoML-0.5.0

• Improved Clipper.ai deployment support
  • Work seemlessly with clipper v0.4.1 release, updated deployment guide
  • https://github.com/bentoml/BentoML/blob/master/guides/deployment/deploy-with-clipper/bentoml-clipper-deployment-guide.ipynb
• New S3 based repository
  • BentoML users can now save to, load from BentoService bundle on S3 storage and deploy those bundles directly
• Deployment python APIs are now available in Beta
  • from bentoml.yatai.python_api import create_deployment

BentoML-0.4.9

• Added Tensorflow SavedModel format support
• Added support for s3 based model repository
• New syntax for BentoService#pack, making it easier to work with multiple models
• Fixed REST API server docker image build issue with new release of gunicorn

BentoML-0.4.8

• Fixed an issue with loading Fastai model in FastaiModelArtifact, when the basic_learn submodule is not already imported
• Fixed an issue with creating AWS SageMaker deployment, previously it will fail with KeyError in certain condition

BentoML-0.4.7

• Fixed an issue where SQLAlchemy alembic files are not found in PyPI distribution
• Fixed an issue with SQLAlchemy alembic overwriting BentoML default logging configuration

BentoML-0.4.5

• Improved BentoML module import time by around 3-4x
• List deployments command now shows "age" column denoting how long the deployment has been created
• Fixed a bug where serverless deployment failed to install required plugins

Docs:

• Updated documentation site https://bentoml.readthedocs.io/

BentoML-0.4.4

New Features:

• Support for both Keras and tensorflow.keras module in KerasModelArtifact
• New serialization option for KerasModelArtifact that stores model in json and weights files (by @ghunkins )
• bentoml list deployments provides clean table outputs now
• Support for AWS S3 based BentoML repository (Beta)

Bug fixes:

• Fixed error with old click version #335
• Fixed REST API Server issue on Windows platform #333

BentoML-0.4.3

• Enhancement to Serverless deployment and SageMaker deployment
• Updated default version string format for user-defined BentoService
• Added the versioneer interface on BentoService for users to define a customized versioning format
• Added '--force' option to bentoml deployment delete command
• Updated clipper base image to 0.4.1

For BentoML developers:

• BentoML now packages local BentoML dev branch when bundling a BentoService for deployment

BentoML-0.4.2

• Introduced SklearnModelArtifact, adding more scikit-learn specific optimizations over previous general PickleArtifact
• Fixed a number of issues with AWS Lambda Serverless deployment
• Improved error message and CLI outputs of AWS SageMaker deployment

BentoML-0.4.1

• Fixed an issue with initializing BentoML logging and repository file direcotry