Google this week has published a new version of its TensorFlow machine learning software that adds support for iOS. Google initially teased that it was working on iOS support for TensorFlow last November, but said it was unable to give a timeline. An early version of TensorFlow version 0.9 was released yesterday on GitHub, however, and it brings iOS support.
For those unfamiliar, TensorFlow is Google’s incredibly powerful artificial intelligence software that powers many of Google’s services and initiatives, including AlphaGo. Google describes TensorFlow as “neural network” software that processes data in a way that’s similar how our brain cells process data (via CNET).
With Google adding iOS support to TensorFlow, apps will be able to integrate the smarter neural network capabilities into their apps, ultimately making them considerably smarter and capable.
Last November, Google open sourced TensorFlow, allowing for it to be adopted into all sorts of difference products and research cases. Google explained at the time that it hopes open source availability will allow for researchers, engineers, and hobbyists to help speed the machine’s learning along and to help get it to a much smarter level in less time.
At this point, it’s unclear when the final version of TensorFlow 0.9 will be released, but the early pre-release version is available now on GitHub. In the release notes, Google points out that because TensorFlow is now open source, 46 people from outside the company contributed to TensorFlow version 0.9.
In addition to adding support for iOS, TensorFlow 0.9 adds a handful of other new features and improvements, as well as plenty of smaller bug fixes and performance enhancements. You can read the full change log below and access TensorFlow on GitHub.
Major Features and Improvements
- Python 3.5 support and binaries
- Added iOS support
- Added support for processing on GPUs on MacOS
- Added makefile for better cross-platform build support (C API only)
- fp16 support for many ops
- Higher level functionality in contrib.{layers,losses,metrics,learn}
- More features to Tensorboard
- Improved support for string embedding and sparse features
- TensorBoard now has an Audio Dashboard, with associated audio summaries.
Big Fixes and Other Changes
- Turned on CuDNN Autotune.
- Added support for using third-party Python optimization algorithms (contrib.opt).
- Google Cloud Storage filesystem support.
- HDF5 support
- Add support for 3d convolutions and pooling.
- Update gRPC release to 0.14.
- Eigen version upgrade.
- Switch to eigen thread pool
- tf.nn.moments() now accepts a shift argument. Shifting by a good estimate of the mean improves numerical stability. Also changes the behavior of the shift argument to tf.nn.sufficient_statistics().
- Performance improvements
- Many bugfixes
- Many documentation fixes
- TensorBoard fixes: graphs with only one data point, Nan values, reload button and auto-reload, tooltips in scalar charts, run filtering, stable colors
- Tensorboard graph visualizer now supports run metadata. Clicking on nodes while viewing a stats for a particular run will show runtime statistics, such as memory or compute usage. Unused nodes will be faded out.
FTC: We use income earning auto affiliate links. More.
Comments