Skip to main content

DLTKDL is a wrapper for TensorFlow and Keras that makes deep learning and AI more accessible and easier to apply

Project description

DLTK Deep Learning API or DLTKDL is an open source lightweight wrapper for the deep learning library TensorFlow and Keras (and other libraries) to help build, train, and deploy neural networks and other machine learning models. Inspired by ML framework extensions like fastai and ludwig, it is designed to make deep learning and AI more accessible and easier to apply for both newcomers and experienced practitioners. With only a few lines of code, dltkdl allows you to easily and quickly:

Installation Instructions

Make sure pip is up-to-date with:

Install TensorFlow 2 if it is not already installed (e.g., pip install tensorflow==2.3.0)

Install dltkdl:

git clone

cd Deep-Learning

python install or pip install .

The above should be all you need on Linux systems and cloud computing environments like Google Colab and AWS EC2. If you are using dltkdl on a Windows computer, you can follow these more detailed instructions that include some extra steps.

Some important things to note about installation:

If using dltkdl on a local machine with a GPU (versus Google Colab, for example), you'll need to install GPU support for TensorFlow 2.

On Google Colab, TensorFlow 2 should be already installed. You should be able to use dltkdl with any version of TensorFlow 2. However, we currently recommend using TensorFlow 2.3.0 (if possible) due to a TensorFlow bug that will not be fixed until TensorFlow 2.4 that affects the Learning-Rate-Finder.

Since some dltkdl dependencies have not yet been migrated to tf.keras in TensorFlow 2 (or may have other issues), dltkdl is temporarily using forked versions of some libraries. If not installed, dltkdl will complain when a method or function needing either of these libraries is invoked.

Available Deep Learning algorithms:

Artificial Neural Network(ANN) :

  1. Multilayer perceptron or mlp: a configurable multilayer perceptron with categorical variable embeddings.

Convolutional Neural Network(CNN) :

  1. pretrained_resnet50: 50-layer Residual Network (pretrained on ImageNet)
  2. resnet50: 50-layer Residual Network (randomly initialized)
  3. pretrained_mobilenet: MobileNet Neural Network (pretrained on ImageNet)
  4. mobilenet: MobileNet Neural Network (randomly initialized)
  5. pretrained_inception: Inception Version 3 (pretrained on ImageNet)
  6. inception: Inception Version 3 (randomly initialized)
  7. wrn22: 22-layer Wide Residual Network (randomly initialized)
  8. default_cnn: a default Convolutional Neural Network

Recurrent neural Network(RNN) and Long short-term memory(LSTM) :

  1. fasttext: a fastText-like model []
  2. logreg: logistic regression using a trainable Embedding layer
  3. nbsvm: NBSVM model []
  4. bigru: Bidirectional GRU with pretrained word vectors []
  5. standard_gru: simple 2-layer GRU with randomly initialized embeddings
  6. bert: Bidirectional Encoder Representations from Transformers (BERT) []
  7. distilbert: distilled, smaller, and faster BERT from Hugging Face []


You can also chckout step-by step implementation by following python notebooks under tutorial folder.

Example Hourse price prediction using Multilayer perceptron Neural Network

import urllib.request
import pandas as pd
import numpy as np
pd.set_option('display.max_columns', None)
import dltkdl
from dltkdl import tabledata

train_df = pd.read_csv('train.csv', index_col=0)
# print(train_df)
train_df.drop(['Alley', 'PoolQC', 'MiscFeature', 'Fence', 'FireplaceQu', 'Utilities'], 1, inplace=True)
trn, val, preproc = tabledata.load_from_dataframe(train_df, is_regression=True, label_columns='SalePrice',
# Invoking multilayer perceptron NN
model = tabledata.tabular_regression_model('mlp', trn)
learner = dltkdl.get_learner(model, train_data=trn, val_data=val, batch_size=128)
learner.lr_find(show_plot=True, max_epochs=16)

Example: Classifying Images of Dogs and Cats Using a Pretrained ResNet50 model

import dltkdl
from dltkdl import vision as vis

# load data
(train_data, val_data, preproc) = vis.images_from_folder(
                                              data_aug = vis.get_data_aug(horizontal_flip=True),
                                              train_test_names=['train', 'valid'], 
                                              target_size=(224,224), color_mode='rgb')

# load model
model = vis.image_classifier('pretrained_resnet50', train_data, val_data, freeze_layers=80)

# wrap model and data in dltkdl.Learner object
learner = dltkdl.get_learner(model=model, train_data=train_data, val_data=val_data, 
                             workers=8, use_multiprocessing=False, batch_size=64)

# find good learning rate
learner.lr_find()             # briefly simulate training to find good learning rate
learner.lr_plot()             # visually identify best learning rate

# train using triangular policy with ModelCheckpoint and implicit ReduceLROnPlateau and EarlyStopping
learner.autofit(1e-4, checkpoint_folder='/tmp/saved_weights')

Example: Language translation(English to Dutch)

from dltkdl import text 
translator = text.Translator(model_name='Helsinki-NLP/opus-mt-en-nl')
src_text = '''My name is Sarah and I live in London.'''

Output: Mijn naam is Sarah en ik woon in Londen.

Example: Text classification

# load text data
categories = ['alt.atheism', 'soc.religion.christian', '', '']
from sklearn.datasets import fetch_20newsgroups

train_b = fetch_20newsgroups(subset='train', categories=categories, shuffle=True)
test_b = fetch_20newsgroups(subset='test', categories=categories, shuffle=True)
(x_train, y_train) = (,
(x_test, y_test) = (,

# build, train, and validate model (Transformer is wrapper around transformers library)
import dltkdl
from dltkdl import text

MODEL_NAME = 'distilbert-base-uncased'
t = text.Transformer(MODEL_NAME, maxlen=500, class_names=train_b.target_names)
trn = t.preprocess_train(x_train, y_train)
val = t.preprocess_test(x_test, y_test)
model = t.get_classifier()
learner = dltkdl.get_learner(model, train_data=trn, val_data=val, batch_size=6)
learner.fit_onecycle(5e-5, 4)
learner.validate(class_names=t.get_classes())  # class_names must be string values

                        precision    recall  f1-score   support

           alt.atheism       0.91      0.93      0.92       319       0.98      0.96      0.97       389
            0.97      0.96      0.96       396
soc.religion.christian       0.95      0.97      0.96       398

              accuracy                           0.96      1502
             macro avg       0.95      0.95      0.95      1502
          weighted avg       0.96      0.96      0.96      1502

array([[296,   1,   6,  16],
       [ 12, 372,   5,   0],
       [  8,   6, 379,   3],
       [  8,   2,   0, 388]])

Founding Member

Lead Maintainer

Core Maintainer

Core Contributers


The content of this project itself is licensed under GNU LGPL, Version 3 (LGPL-3)

Project details

Download files

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

Files for DLTKDL, version 51.0.8
Filename, size File type Python version Upload date Hashes
Filename, size DLTKDL-51.0.8.tar.gz (145.9 kB) File type Source Python version None Upload date Hashes View
Filename, size DLTKDL-51.0.8-py3-none-any.whl (185.7 kB) File type Wheel Python version py3 Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page