Skip to content

Latest commit

 

History

History

keras_tf1

README.md

mlflow-examples - keras_tf1

Overview

  • Keras with TensorFlow 1.x train and predict.
  • Saves model as keras flavor.
  • MNIST dataset.
  • Option to autolog parameters and metrics.
  • Option to log and score model as ONNX.
  • Setup: conda.yaml.

Setup

Libraries

pip install keras==2.2.5
pip install tensorflow==1.15.0

Training

Source: train.py.

To run with user logging (no autologging).

python train.py --experiment_name keras_mnist --epochs 3 --batch_size 128

To log a model as ONNX flavor under the artifact path onnx-model.

python train.py --experiment_name keras_mnist --epochs 3 --batch_size 128 --log_as_onnx

Autologging

There are two autologging options:

  • keras_autolog - calls mlflow.keras.autolog()
  • tensorflow_autolog - calls mlflow.tensorflow.autolog()
python train.py --experiment_name keras_mnist --epochs 3 --batch_size 128 --keras_autolog

Autologging will create a model under the name model.

Autlogging Parameters:

acc
loss

Autlogging Metrics:

batch_size
class_weight
epochs
epsilon
initial_epoch
learning_rate
max_queue_size
num_layers
optimizer_name
sample_weight
shuffle
steps_per_epoch
use_multiprocessing
validation_freq
validation_split
validation_steps
workers

Scoring

Score as Keras flavor

Source: keras_predict.py.

python keras_predict.py --model_uri runs:/7e674524514846799310c41f10d6b99d/keras-model

predictions.type: <class 'numpy.ndarray'>
predictions.shape: (10000,)
predictions: [7 2 1 ... 4 5 6]

Score as Pyfunc flavor

Source: pyfunc_predict.py.

Score Keras model with Pyfunc

python pyfunc_predict.py --model_uri runs:/7e674524514846799310c41f10d6b99d/keras-model

predictions.type: <class 'pandas.core.frame.DataFrame'>
predictions.shape: (10000, 10)

predictions:                  0             1  ...             8             9
0     7.356894e-07  2.184515e-09  ...  2.648242e-07  1.557131e-05
1     3.679516e-08  5.211977e-06  ...  2.588275e-07  4.540044e-12
...            ...           ...  ...           ...           ...
9998  5.653655e-08  3.749759e-09  ...  1.073899e-04  1.215128e-09
9999  2.790610e-08  2.516971e-11  ...  6.860461e-10  2.355604e-10

Score ONNX model with Pyfunc

python pyfunc_predict.py --model_uri runs:/7e674524514846799310c41f10d6b99d/onnx-model

predictions.type: <class 'pandas.core.frame.DataFrame'>
predictions.shape: (100000, 1)
predictions:             dense_2
0      7.529760e-07
1      3.086328e-09
...             ...
99998  9.314070e-10
99999  2.785560e-10

Score as ONNX flavor

Source: onnx_predict.py and onnx_utils.py.

python onnx_predict.py --model_uri runs:/7e674524514846799310c41f10d6b99d/onnx-model

predictions.type: <class 'numpy.ndarray'>
predictions.shape: (10000, 10)
predictions: 
[[7.5297595e-07 3.0863279e-09 2.5705955e-04 ... 9.9953580e-01 3.5329055e-07 1.3658248e-05]
 [3.7254765e-08 8.8118195e-06 9.9951375e-01 ... 2.6982589e-12 9.4671401e-07 2.8832321e-12]
 ...
 [8.6566203e-08 6.8279524e-09 1.0189680e-08 ... 1.5083194e-09 2.0773137e-04 1.7879515e-09]
 [3.8302844e-08 1.6128991e-11 1.1180904e-05 ... 4.9651490e-12 9.3140695e-10 2.7855604e-10]]