Skip to main content

Quick Starter Code with different specifications stitched together in tensorflow 2.0 .

Project description

tf-stitch

Starter code with best practices for different Deep Learning problems in one command.

This Package provides Tensorflow 2.0 boilerplate code , for deep learning problem domains.

To install, in command line enter -

pip install tf-stitch

View Pypi project on https://pypi.org/project/tf-stitch/

Why use this?

If you have been working in deep learning or have just started, you will find yourself writing the same imports, same training loop, same data fetching commands, same model declaration every time.

Personally, found myself copy-pasting MNIST models to get started with a project. And Even copy-pasting does not work as there are many parts of deep learning code that are needed in different combinations for different problems.

This Package is created just to solve this problem for our most beloved framework Tensorflow 2.0. Deep learning is an exciting field full of awesome ideas and new concepts popping every day. Custom stitched boiler code lets you focus on the uniqueness of your implementation and not the repetitive code.

Keras and Tensorflow 2.0, because of simple API, allows you to implement many ideas in less time. This Package just aims to double(2X) that efficiency.

USAGE

usage: tf-stitch output_file [-d DOMAIN]
[--dataset DATASET] [--model MODEL] [--training TRAINING]
[--testing TESTING]

Example -

tf-stitch output_script.py -d=vision

or

tf-stitch output_notebook.ipynb -d=nlp

Here output_script.py and output_notebook.ipynb are the generated output files with starter code. output_file is required for command to run. File can have extension .py or .ipynb for generating python script or jupyter notebook.

Arguments:

-d , --domain DOMAIN Domain problem type for selecting appropriate model and dataset. See tf_stitch/template.json for default datasets and model for each domain option. Current domain options are -

  • vision
  • nlp
  • structured

Either provide domain argument or dataset and model arguments

- dataset DATASET : Select any of Tensorflow datasets. See catalog.

--model MODEL : Select a Deep Learning Model Current available options -

  • conv - Convolution model
  • rnn - GRU sequence model
  • dnn - Fully connected network
  • custom - Custom layer network

--training TRAINING : Type of training loop. Options -

  • custom
  • built-in

--testing TESTING : If to include testing, True or False

Example -

tf-stitch output.ipynb --dataset = cifar100 --model = conv 
--training = custom --testing = True

All the templates are in their respective folder in tf_stitch/templates . Feel free to change templates according to your taste and also submit a pull request if it can be helpful for others as well.

Contributing

Hey There, this project requires your feedback and contribution to improve. You can contribute, but not restricted to, in following ways -

  • Providing Feedback and submitting feature request.
  • Apply best and latest TF2.0 practices.
  • Add more template and domain options.

Looking forward to hear from you : )

Project details


Download files

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

Source Distribution

tf-stitch-1.2.1.tar.gz (6.3 kB view details)

Uploaded Source

Built Distribution

tf_stitch-1.2.1-py3-none-any.whl (14.2 kB view details)

Uploaded Python 3

File details

Details for the file tf-stitch-1.2.1.tar.gz.

File metadata

  • Download URL: tf-stitch-1.2.1.tar.gz
  • Upload date:
  • Size: 6.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.5.0.1 requests/2.23.0 setuptools/47.3.1 requests-toolbelt/0.9.1 tqdm/4.40.0 CPython/3.6.6

File hashes

Hashes for tf-stitch-1.2.1.tar.gz
Algorithm Hash digest
SHA256 aa7b8e2d9eff9d100db1c2cfffda1077f1ef7afcfb290f0dffac93476ac90e9c
MD5 5d3f84ccab579704a2754ddf62d6cc31
BLAKE2b-256 db31b89e4f04ff1ea85fb8a2f5b119a5f0a87900a5bfee0c93e0af08783705c4

See more details on using hashes here.

File details

Details for the file tf_stitch-1.2.1-py3-none-any.whl.

File metadata

  • Download URL: tf_stitch-1.2.1-py3-none-any.whl
  • Upload date:
  • Size: 14.2 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.2.0 pkginfo/1.5.0.1 requests/2.23.0 setuptools/47.3.1 requests-toolbelt/0.9.1 tqdm/4.40.0 CPython/3.6.6

File hashes

Hashes for tf_stitch-1.2.1-py3-none-any.whl
Algorithm Hash digest
SHA256 c15b7a1747f0bd246805491fd92104bdff85fd0e3eba90082714d99524971212
MD5 a1b409a23d950846f8be7d975759f78f
BLAKE2b-256 380227f9271d04f2de2fa6108846a50208c59f1410f2aad9b684c31b44240a58

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