aironsuit 0.1.9

A model wrapper for automatic model design and visualization purposes.

AIronSuit

AIronSuit (Beta) is a Python library for automatic model design/selection and visualization purposes built to work with tensorflow (or pytorch in the future) as a backend. It aims to accelerate the development of deep learning approaches for research/development purposes by providing components relying on cutting edge approaches. It is flexible and its components can be replaced by customized ones from the user. The user mostly focuses on defining the input and output, and AIronSuit takes care of its optimal mapping.

Key features:

1. Automatic model design/selection with hyperopt.
2. Parallel computing for multiple models across multiple GPUs when using a k-fold approach.
3. Built-in model trainer that saves training progression to be visualized with TensorBoard.
4. Machine learning tools from AIronTools: model_constructor, custom_block, layer_constructor, preprocessing utils, etc.
5. Flexibility: the user can replace AIronSuit components by a user customized one. For instance, the model constructor can be easily replaced by a user customized one.

Installation

pip install aironsuit

Example

# Databricks notebook source
import numpy as np
from tensorflow.keras.datasets import mnist
from tensorflow.keras.models import Model
from tensorflow.keras.utils import to_categorical
from tensorflow.keras.layers import Input
import os
os.environ['AIRONSUIT_BACKEND'] = 'tensorflow'
from aironsuit.suit import AIronSuit
from airontools.model_constructors import layer_constructor
from airontools.tools import path_management
HOME = os.path.expanduser("~")

# COMMAND ----------

# Example Set-Up #

project_name = 'simple_mnist'
working_path = os.path.join(HOME, project_name)
num_classes = 10
batch_size = 128
epochs = 20

# COMMAND ----------

# Load data
(x_train, y_train), (x_test, y_test) = mnist.load_data()

# Preprocess data
x_train = np.expand_dims(x_train.astype('float32') / 255, -1)
x_test = np.expand_dims(x_test.astype('float32') / 255, -1)
y_train = to_categorical(y_train, num_classes)
y_test = to_categorical(y_test, num_classes)

# COMMAND ----------

# Create model
input_shape = (28, 28, 1)
inputs = Input(shape=input_shape)
outputs = layer_constructor(x=inputs, input_shape=input_shape, units=10, activation='softmax', filters=5,
                            kernel_size=15)
model = Model(inputs=inputs, outputs=outputs)
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

# COMMAND ----------

# Invoke AIronSuit
aironsuit = AIronSuit(model=model)

# COMMAND ----------

# Training
path_management(working_path, modes=['rm', 'make'])
aironsuit.train(
    epochs=epochs,
    x_train=x_train,
    y_train=y_train,
    path=working_path)
aironsuit.summary()

# COMMAND ----------

# Evaluate
score = aironsuit.evaluate(x_test, y_test)
print('Test loss:', score[0])
print('Test accuracy:', score[1])

# COMMAND ----------

# Save Model
aironsuit.save_model(os.path.join(working_path, project_name + '_model'))
del aironsuit, model

# COMMAND ----------

# Re-Invoke AIronSuit and load model
aironsuit = AIronSuit()
aironsuit.load_model(os.path.join(working_path, project_name + '_model'))
aironsuit.model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

# Further Training
aironsuit.train(
    epochs=epochs,
    x_train=x_train,
    y_train=y_train)

# COMMAND ----------

# Evaluate
score = aironsuit.evaluate(x_test, y_test)
print('Test loss:', score[0])
print('Test accuracy:', score[1])

More Examples

see usage examples in aironsuit/examples