- add model_version_policy to model config
- change single quotes to double quotes
- update MME Pre/Post-Processing model and script paths
- increasing max_retry for model availability check
- multi-model-endpoint support
- update README for multi-model endpoint
- add issue templates
- error code files for TFS2.2 (copied from pip install tf)
- Test against py37 in buildspec
- TF EIA 1.15 and 2.0
- Replaced deprecated aws ecr get-login by get-login-password
- Updating Pyyaml and Awscli version for tf1.15
- upgrade Pillow version in test example
- Allowing arguments for deep_learning_container.py for tf
- Adding of deep_learning_container.py
- update: Remove multi-model label from CPU containers
- update r2.0.1 dockerfiles
- add 2.1 dockerfiles and tensorflow artifacts
- update for 1.15.2
- return on_delete method when model successfully deleted
- validate tensorflow model version number in model path
- document that pre-/post-processing is not supported with multi-model
- Add init.py to fix unit test
- update: Update buildspec for TF 1.15 and 2.0
- update copyright year in license header
- update Readme with correct TF versions.
- update: Release TF 1.15 and TF 2.0 dockerfiles
- check container is ready in tests
- increase attempts to allow for large gpu images
- ping check and model status check
- Update EI Dockerfile 1.14 with New Health check, and new binaries
- update publish-all.sh to match the versions in build-all.sh
- upgrade pillow to 6.2.0 in requirements.txt test
- use regional endpoint for STS in builds and tests
- merge dockerfiles
- update instance type region availability
- mme improvements
- "change: merge asimov branch to master branch (#80)"
- merge asimov branch to master branch
- TFS_SHORT_VERSION explicitly defined in dockerfile
- add Dockerfile for 1.14 in py3 with EIA
- require defaul model env var
- default SAGEMAKER_TFS_DEFAULT_MODEL_NAME to string value
- remove unused fixture
- Update no-p2 and no-p3 regions
- delete models
- different invocation url
- list models
- add model
- update request/response model and url key names
- README changes
- add batching to readme
- use nvidia-docker for local gpu tests
- add 1.13 EIA containers
- fix broken ei tests
- add batch transform integration test
- move SageMaker tests to release build
- use p2.xlarge by default in tests
- add Tensorflow 1.13
- make tox run any pytest tests
- support jsonlines output
- update README.md for EI image
- change Dockerfile directory structure
- allow local test against single container
- update README.md
- add pre/post-processing usage examples
- add pre-post-processing documentation
- add build-all.sh, publish-all.sh scripts
- add tensorflow serving batching
- install requirements.txt in writable dir
- make njs code handle missing custom attributes header
- add python service for pre/post-processing
- improve handling of ei binary during builds
- add data generator and perf tests
- remove per-line parsing
- add additional csv test case
- handle zero values correctly
- update EI binary directory.
- Support payloads with many csv rows and change CSV parsing behavior
- update EI binary location
- remove tfs deployment tests
- create bucket during test
- fix argname in deployment test
- fix repository name in buildspec
- add deployment tests and run them concurrently
- report error for missing ei version
- remove extra commma in buildspec
- add eia images to release build
- update buildspec to output deployments.json
- Modify EI image repository and tag to match Python SDK.
- Change test directory to be consistent with PDT pipeline.
- Add EI support to TFS container.
- simplify tfs versioning
- add buildspec.yml for codebuild
- add tox, pylint, flake8, jshint