Releases: tensorflow/serving
Releases · tensorflow/serving
0.5.1
0.5.0
New Features:
- Model Server binary in tensorflow_serving/model_servers with a PredictionService API.
- Support SavedModel format and added ability to upconvert legacy SessionBundle exports. See https://github.com/tensorflow/tensorflow/blob/master/tensorflow/python/saved_model/README.md
- NOTE: SessionBundle is now deprecated and we plan to end official support in the upcoming 1.0 release. Please move to use SavedModel.
- Multi-model batch scheduling: interleaveing batches for different models.
- Registry of servable types for Model Server to handle ones other than SessionBundle/SavedModel, including third-party ones not in TF-Serving codebase.
- Resource management: using model size estimates to avoid exceeding server memory capacity.
Concurrent model loading & fast initial load. - Request logging:
- A protobuf based logging API.
- Provides ability for users to log a configurable sample, or all, of the queries served.
- Support querying named signatures.
Other:
- Assembled core manager setup code into a ServerCore object.
- Various bug fixes.
- Documentation updates.
Docker, Inception, and various small fixes & improvements.
This release adds Docker support, and an end-to-end Inception tutorial. It also makes various minor fixes and clean-ups to code and documentation.
New features:
- Dockerfile and README for building a container with a TensorFlow Serving development environment.
- End-to-end example and tutorial for serving an InceptionV3 model in Kubernetes.
- Jenkins continuous integration.
- DynamicManager re-tries failed servable load attempts.
- Utility to wait until a manager has loaded certain servables.
- Utility for tracking the states of servables in a manager, by listening for servable state changes on an event bus.
- Modules for managing servable resources (but not yet integrated into DynamicManager).
- Add GetNamedSignature() to signatures, independent from any signature type.
- Misc. additions to util/.
Compatibility:
- Exporter supports both python 2.7 and python 3.4.
- Migrate to gRPC 0.13.
- Migrate to latest TensorFlow and TF-Models submodules.
Bug fixes:
- Allow importing graphs with no variable nodes.
- Misc small documentation bug-fixes.
Clean-ups:
- Depend on gRPC via a Bazel git repository, rather than via a git submodule.
- Eliminate the batching sub-namespace.
- Misc small code clean-ups.
- Misc minor documentation clean-ups.