Skip to main content

Set of utilities to work easier with Bender.

Project description

Set of utilities to work easier with Bender.

Currently there’s support for TensorFlow and Caffe, but we are working on more stuff!

Works on Python 2.7.+ and 3.+, with Tensorflow 1.2+.

To install:

pip install benderthon

TensorFlow is required too. The simplest way to install it is:

pip install tensorflow

There are other ways, see Installing Tensorflow. Benderthon does not install it by default to let the usage of a custom installation.

tf-freeze

Utility to convert TensorFlow checkpoints into minimal frozen graphs.

Usage

To take the checkpoint in checkpoint_path.ckpt, whose output is yielded by the node named Tanh, and save it to graph_with_weights.pb:

benderthon tf-freeze checkpoint_path.ckpt graph_with_weights.pb Tanh

Sample

The file sample.py contains a network example for MNIST dataset with 2 convolutional layers and 2 dens layers. If you run it, it will generate checkpoints files with prefix checkpoints/mnist.ckpt:

./sample.py

Then you can get a minimal protobuf version with the weights frozen:

benderthon tf-freeze checkpoints/mnist.ckpt output/mnist.pb Prediction

The generated file occupies half the original checkpoints (26MB to 13MB).

You can also get only the graph, which occupies just 13kB:

benderthon tf-freeze --no-weights checkpoints/mnist.ckpt output/mnist_only_graph.pb Prediction

To save the weights in a separate path for later processing:

benderthon tf-freeze --only-weights checkpoints/mnist.ckpt weights/ Prediction

caffe-freeze

This module cannot be accessed from the command line utility, it should be used from Python code, importing benderthon.caffe_freeze.

You need caffeflow package installed first:

pip install -e git://github.com/xmartlabs/caffeflow.git@4618f89#egg=caffeflow

Development

This utility is under development and the API is not stable. So, do not heavily rely on it.

To install locally you should do ./setup.py install, but first have pandoc and pypandoc installed.

License

Copyright 2017 Xmartlabs SRL.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

   http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distribution

benderthon-0.3.0-py2.py3-none-any.whl (14.0 kB view details)

Uploaded Python 2 Python 3

File details

Details for the file benderthon-0.3.0-py2.py3-none-any.whl.

File metadata

File hashes

Hashes for benderthon-0.3.0-py2.py3-none-any.whl
Algorithm Hash digest
SHA256 bd02b42bccb5ba2e00e6518036449e2c17c784518a6dd358dd4b69f83a8f24f0
MD5 8941813c9912ff4571ca535889be92a4
BLAKE2b-256 d6410388ce95d9e8ec6c452c4af2d259f18a1fec9f4318250379fd28164b34f3

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page