Quick and Distributed TensorFlow command framework
Project description
# tensorflow-qnd
[](https://badge.fury.io/py/tensorflow-qnd)
[]()
[](https://travis-ci.org/raviqqe/tensorflow-qnd)
[](https://unlicense.org)
Quick and Distributed TensorFlow command framework
tensorflow-qnd is a TensorFlow framework to create commands to train and
evaluate models on multiple computers.
While made to be used on multiple computers in a cluster, this library is also
useful to exploit multiple GPUs on a single machine.
## Installation
Python 3.5+ and TensorFlow 0.12+ are required.
```
pip3 install --user --upgrade tensorflow-qnd
```
## Usage
See [documentation](https://raviqqe.github.io/tensorflow-qnd/qnd).
## Examples
```python
import logging
import qnd
import tensorflow as tf
logging.getLogger().setLevel(logging.INFO)
def read_file(filename_queue):
_, serialized = tf.TFRecordReader().read(filename_queue)
scalar_feature = lambda dtype: tf.FixedLenFeature([], dtype)
features = tf.parse_single_example(serialized, {
"image_raw": scalar_feature(tf.string),
"label": scalar_feature(tf.int64),
})
image = tf.decode_raw(features["image_raw"], tf.uint8)
image.set_shape([28**2])
return tf.to_float(image) / 255 - 0.5, features["label"]
def minimize(loss):
return tf.contrib.layers.optimize_loss(
loss,
tf.contrib.framework.get_global_step(),
0.01,
"Adam")
def model(image, number):
h = tf.contrib.layers.fully_connected(image, 64)
h = tf.contrib.layers.fully_connected(h, 10, activation_fn=None)
loss = tf.reduce_mean(
tf.nn.sparse_softmax_cross_entropy_with_logits(h, number))
predictions = tf.argmax(h, axis=1)
return predictions, loss, minimize(loss), {
"accuracy": tf.reduce_mean(tf.to_float(tf.equal(predictions, number)))
}
run = qnd.def_run()
def main():
run(model, read_file)
if __name__ == "__main__":
main()
```
See also [examples](examples) directory.
## License
[The Unlicense](https://unlicense.org)
[](https://badge.fury.io/py/tensorflow-qnd)
[]()
[](https://travis-ci.org/raviqqe/tensorflow-qnd)
[](https://unlicense.org)
Quick and Distributed TensorFlow command framework
tensorflow-qnd is a TensorFlow framework to create commands to train and
evaluate models on multiple computers.
While made to be used on multiple computers in a cluster, this library is also
useful to exploit multiple GPUs on a single machine.
## Installation
Python 3.5+ and TensorFlow 0.12+ are required.
```
pip3 install --user --upgrade tensorflow-qnd
```
## Usage
See [documentation](https://raviqqe.github.io/tensorflow-qnd/qnd).
## Examples
```python
import logging
import qnd
import tensorflow as tf
logging.getLogger().setLevel(logging.INFO)
def read_file(filename_queue):
_, serialized = tf.TFRecordReader().read(filename_queue)
scalar_feature = lambda dtype: tf.FixedLenFeature([], dtype)
features = tf.parse_single_example(serialized, {
"image_raw": scalar_feature(tf.string),
"label": scalar_feature(tf.int64),
})
image = tf.decode_raw(features["image_raw"], tf.uint8)
image.set_shape([28**2])
return tf.to_float(image) / 255 - 0.5, features["label"]
def minimize(loss):
return tf.contrib.layers.optimize_loss(
loss,
tf.contrib.framework.get_global_step(),
0.01,
"Adam")
def model(image, number):
h = tf.contrib.layers.fully_connected(image, 64)
h = tf.contrib.layers.fully_connected(h, 10, activation_fn=None)
loss = tf.reduce_mean(
tf.nn.sparse_softmax_cross_entropy_with_logits(h, number))
predictions = tf.argmax(h, axis=1)
return predictions, loss, minimize(loss), {
"accuracy": tf.reduce_mean(tf.to_float(tf.equal(predictions, number)))
}
run = qnd.def_run()
def main():
run(model, read_file)
if __name__ == "__main__":
main()
```
See also [examples](examples) directory.
## License
[The Unlicense](https://unlicense.org)
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
tensorflow-qnd-0.0.6.tar.gz
(9.5 kB
view hashes)
Built Distributions
tensorflow_qnd-0.0.6-py3.5.egg
(25.2 kB
view hashes)
Close
Hashes for tensorflow_qnd-0.0.6-py3-none-any.whl
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 | 00f28100ff4657ced8f3fb59d8421dbfd64bef012ee271c4dcba697a89240f47 |
|
| MD5 | 0a84c0e773042ee542a08fd17d218bba |
|
| BLAKE2b-256 | 82569fc3c6325aa4a6aa5b906c9113f1e4c76faa3157d49def9dd28e1c71f0c1 |
