| commit | f7a6bb9e02b3d9e2b2f32a2b901bc9d571593b06 | [log] [tgz] |
|---|---|---|
| author | Qiao Zhang <zhangqiaorjc@google.com> | Tue May 18 11:42:40 2021 -0700 |
| committer | TensorFlower Gardener <gardener@tensorflow.org> | Tue May 18 11:58:29 2021 -0700 |
| tree | 35530e2a44bd78a4807ed0bb8e01623e5a6e966b | |
| parent | 1807e61724ef014ee14ad374636bcd21f1aa1832 [diff] |
[JAX] Refactor jax_jit to avoid DevicePut on pruned args. name old cpu/op new cpu/op delta eager_unary_dispatch 35.7µs ± 2% 35.9µs ± 3% ~ (p=0.841 n=5+5) eager_unary 36.4µs ± 2% 36.6µs ± 3% ~ (p=0.421 n=5+5) eager_binary_dispatch 45.6µs ± 1% 46.1µs ± 2% ~ (p=0.421 n=5+5) eager_binary 46.6µs ± 2% 47.0µs ± 5% ~ (p=1.000 n=5+5) jit_trivial_dispatch 41.4µs ± 1% 41.4µs ± 0% ~ (p=0.690 n=5+5) jit_trivial 42.4µs ± 1% 42.3µs ± 1% ~ (p=0.841 n=5+5) jit_simple_dispatch 8.85µs ± 3% 9.15µs ± 3% ~ (p=0.095 n=5+5) jit_simple 9.77µs ± 1% 9.82µs ± 2% ~ (p=0.548 n=5+5) jit_simple_many_args_dispatch_10 13.4µs ± 1% 13.6µs ± 3% ~ (p=0.222 n=5+5) jit_simple_many_args_10 14.0µs ± 2% 14.1µs ± 1% ~ (p=0.421 n=5+5) jit_simple_pruned_args_dispatch_10 8.05µs ± 3% 8.07µs ± 4% ~ (p=0.841 n=5+5) jit_simple_pruned_args_10 9.53µs ± 2% 9.43µs ± 2% ~ (p=0.222 n=5+5) jit_simple_many_args_dispatch_100 55.2µs ± 1% 54.8µs ± 2% ~ (p=0.310 n=5+5) jit_simple_many_args_100 55.8µs ± 1% 55.8µs ± 1% ~ (p=0.841 n=5+5) jit_simple_pruned_args_dispatch_100 14.3µs ± 4% 12.6µs ± 1% -11.41% (p=0.016 n=5+4) jit_simple_pruned_args_100 14.8µs ± 1% 13.3µs ± 2% -10.06% (p=0.008 n=5+5) jit_simple_many_args_dispatch_1000 489µs ± 1% 477µs ± 3% ~ (p=0.056 n=5+5) jit_simple_many_args_1000 495µs ± 3% 493µs ± 3% ~ (p=0.841 n=5+5) jit_simple_pruned_args_dispatch_1000 85.0µs ± 3% 65.3µs ± 3% -23.13% (p=0.008 n=5+5) jit_simple_pruned_args_1000 86.0µs ± 3% 66.4µs ± 3% -22.78% (p=0.008 n=5+5) jit_simple_many_args_dispatch_2000 1.09ms ± 4% 1.03ms ± 3% -5.97% (p=0.016 n=5+5) jit_simple_many_args_2000 1.07ms ± 3% 1.04ms ± 5% ~ (p=0.095 n=5+5) jit_simple_pruned_args_dispatch_2000 190µs ± 3% 144µs ± 3% -23.96% (p=0.008 n=5+5) jit_simple_pruned_args_2000 195µs ± 4% 147µs ± 3% -24.29% (p=0.008 n=5+5) jit_dispatch_without_transfer 76.0µs ± 1% 77.2µs ± 6% ~ (p=0.310 n=5+5) jit_dispatch_with_transfer 82.1µs ± 5% 81.3µs ± 2% ~ (p=0.421 n=5+5) sda_index_1 8.83µs ± 1% 8.73µs ± 2% ~ (p=0.222 n=5+5) name old time/op new time/op delta eager_unary_dispatch 35.7µs ± 2% 35.9µs ± 3% ~ (p=0.841 n=5+5) eager_unary 36.5µs ± 2% 37.1µs ± 4% ~ (p=0.222 n=5+5) eager_binary_dispatch 45.6µs ± 1% 46.1µs ± 2% ~ (p=0.421 n=5+5) eager_binary 46.8µs ± 3% 47.1µs ± 5% ~ (p=1.000 n=5+5) jit_trivial_dispatch 41.4µs ± 1% 41.4µs ± 0% ~ (p=0.690 n=5+5) jit_trivial 42.4µs ± 1% 42.3µs ± 1% ~ (p=0.841 n=5+5) jit_simple_dispatch 8.86µs ± 3% 9.15µs ± 3% ~ (p=0.095 n=5+5) jit_simple 9.82µs ± 1% 9.91µs ± 0% ~ (p=0.190 n=5+4) jit_simple_many_args_dispatch_10 13.4µs ± 1% 13.6µs ± 4% ~ (p=0.310 n=5+5) jit_simple_many_args_10 14.1µs ± 2% 14.2µs ± 1% ~ (p=0.421 n=5+5) jit_simple_pruned_args_dispatch_10 8.07µs ± 4% 8.07µs ± 4% ~ (p=0.841 n=5+5) jit_simple_pruned_args_10 9.59µs ± 2% 9.48µs ± 2% ~ (p=0.222 n=5+5) jit_simple_many_args_dispatch_100 55.2µs ± 1% 54.8µs ± 2% ~ (p=0.310 n=5+5) jit_simple_many_args_100 55.9µs ± 1% 55.9µs ± 1% ~ (p=0.841 n=5+5) jit_simple_pruned_args_dispatch_100 14.3µs ± 5% 12.6µs ± 1% -11.75% (p=0.016 n=5+4) jit_simple_pruned_args_100 14.8µs ± 2% 13.3µs ± 2% -10.19% (p=0.008 n=5+5) jit_simple_many_args_dispatch_1000 489µs ± 1% 477µs ± 3% ~ (p=0.056 n=5+5) jit_simple_many_args_1000 495µs ± 3% 493µs ± 3% ~ (p=0.841 n=5+5) jit_simple_pruned_args_dispatch_1000 85.0µs ± 3% 65.3µs ± 3% -23.13% (p=0.008 n=5+5) jit_simple_pruned_args_1000 86.1µs ± 3% 66.5µs ± 2% -22.72% (p=0.008 n=5+5) jit_simple_many_args_dispatch_2000 1.09ms ± 4% 1.03ms ± 3% -5.96% (p=0.016 n=5+5) jit_simple_many_args_2000 1.07ms ± 3% 1.04ms ± 5% ~ (p=0.095 n=5+5) jit_simple_pruned_args_dispatch_2000 190µs ± 3% 144µs ± 3% -23.97% (p=0.008 n=5+5) jit_simple_pruned_args_2000 195µs ± 4% 147µs ± 3% -24.31% (p=0.008 n=5+5) jit_dispatch_without_transfer 1.41ms ± 1% 1.40ms ± 1% ~ (p=0.095 n=5+5) jit_dispatch_with_transfer 1.40ms ± 2% 1.40ms ± 2% ~ (p=0.841 n=5+5) sda_index_1 8.83µs ± 1% 8.73µs ± 2% ~ (p=0.222 n=5+5) PiperOrigin-RevId: 374468578 Change-Id: I0a45af35b936a72f8271bd3e3a66e0d778619132
Documentation |
|---|
 |
TensorFlow is an end-to-end open source platform for machine learning. It has a comprehensive, flexible ecosystem of tools, libraries, and community resources that lets researchers push the state-of-the-art in ML and developers easily build and deploy ML-powered applications.
TensorFlow was originally developed by researchers and engineers working on the Google Brain team within Google's Machine Intelligence Research organization to conduct machine learning and deep neural networks research. The system is general enough to be applicable in a wide variety of other domains, as well.
TensorFlow provides stable Python and C++ APIs, as well as non-guaranteed backward compatible API for other languages.
Keep up-to-date with release announcements and security updates by subscribing to announce@tensorflow.org. See all the mailing lists.
See the TensorFlow install guide for the pip package, to enable GPU support, use a Docker container, and build from source.
To install the current release, which includes support for CUDA-enabled GPU cards (Ubuntu and Windows):
$ pip install tensorflow
A smaller CPU-only package is also available:
$ pip install tensorflow-cpu
To update TensorFlow to the latest version, add --upgrade flag to the above commands.
Nightly binaries are available for testing using the tf-nightly and tf-nightly-cpu packages on PyPi.
$ python
>>> import tensorflow as tf >>> tf.add(1, 2).numpy() 3 >>> hello = tf.constant('Hello, TensorFlow!') >>> hello.numpy() b'Hello, TensorFlow!'
For more examples, see the TensorFlow tutorials.
If you want to contribute to TensorFlow, be sure to review the contribution guidelines. This project adheres to TensorFlow's code of conduct. By participating, you are expected to uphold this code.
We use GitHub issues for tracking requests and bugs, please see TensorFlow Discuss for general questions and discussion, and please direct specific questions to Stack Overflow.
The TensorFlow project strives to abide by generally accepted best practices in open-source software development:
| Build Type | Status | Artifacts |
|---|---|---|
| Linux CPU |  | PyPI |
| Linux GPU |  | PyPI |
| Linux XLA |  | TBA |
| macOS |  | PyPI |
| Windows CPU |  | PyPI |
| Windows GPU |  | PyPI |
| Android |  |  |
| Raspberry Pi 0 and 1 |  | Py3 |
| Raspberry Pi 2 and 3 |  | Py3 |
| Libtensorflow MacOS CPU | Status Temporarily Unavailable | Nightly Binary Official GCS |
| Libtensorflow Linux CPU | Status Temporarily Unavailable | Nightly Binary Official GCS |
| Libtensorflow Linux GPU | Status Temporarily Unavailable | Nightly Binary Official GCS |
| Libtensorflow Windows CPU | Status Temporarily Unavailable | Nightly Binary Official GCS |
| Libtensorflow Windows GPU | Status Temporarily Unavailable | Nightly Binary Official GCS |
See TensorFlow SIG Build to find our list of community-supported TensorFlow builds.
Learn more about the TensorFlow community and how to contribute.