Train, convert and predict with ONNX Runtime

This example demonstrates an end to end scenario starting with the training of a machine learned model to its use in its converted from.

Train a logistic regression

The first step consists in retrieving the iris datset.

from sklearn.datasets import load_iris
iris = load_iris()
X, y = iris.data, iris.target

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y)

Then we fit a model.

from sklearn.linear_model import LogisticRegression
clr = LogisticRegression()
clr.fit(X_train, y_train)

We compute the prediction on the test set and we show the confusion matrix.

from sklearn.metrics import confusion_matrix

pred = clr.predict(X_test)
print(confusion_matrix(y_test, pred))

Out:

[[11  0  0]
 [ 0 12  4]
 [ 0  0 11]]

Conversion to ONNX format

We use module sklearn-onnx to convert the model into ONNX format.

from skl2onnx import convert_sklearn
from skl2onnx.common.data_types import FloatTensorType

initial_type = [('float_input', FloatTensorType([None, 4]))]
onx = convert_sklearn(clr, initial_types=initial_type)
with open("logreg_iris.onnx", "wb") as f:
    f.write(onx.SerializeToString())

We load the model with ONNX Runtime and look at its input and output.

import onnxruntime as rt
sess = rt.InferenceSession("logreg_iris.onnx")

print("input name='{}' and shape={}".format(
    sess.get_inputs()[0].name, sess.get_inputs()[0].shape))
print("output name='{}' and shape={}".format(
    sess.get_outputs()[0].name, sess.get_outputs()[0].shape))

Out:

input name='float_input' and shape=[None, 4]
output name='output_label' and shape=[1]

We compute the predictions.

input_name = sess.get_inputs()[0].name
label_name = sess.get_outputs()[0].name

import numpy
pred_onx = sess.run([label_name], {input_name: X_test.astype(numpy.float32)})[0]
print(confusion_matrix(pred, pred_onx))

Out:

[[11  0  0]
 [ 0 12  0]
 [ 0  0 15]]

The prediction are perfectly identical.

Probabilities

Probabilities are needed to compute other relevant metrics such as the ROC Curve. Let’s see how to get them first with scikit-learn.

prob_sklearn = clr.predict_proba(X_test)
print(prob_sklearn[:3])

Out:

[[1.04597336e-03 3.26972202e-01 6.71981824e-01]
 [8.07529571e-01 1.92267362e-01 2.03067523e-04]
 [3.75046145e-02 6.77776609e-01 2.84718777e-01]]

And then with ONNX Runtime. The probabilies appear to be

prob_name = sess.get_outputs()[1].name
prob_rt = sess.run([prob_name], {input_name: X_test.astype(numpy.float32)})[0]

import pprint
pprint.pprint(prob_rt[0:3])

Out:

[{0: 0.0010459469631314278, 1: 0.32697227597236633, 2: 0.6719817519187927},
 {0: 0.807529628276825, 1: 0.19226738810539246, 2: 0.00020308367675170302},
 {0: 0.037504520267248154, 1: 0.6777766942977905, 2: 0.2847188115119934}]

Let’s benchmark.

from timeit import Timer

def speed(inst, number=10, repeat=20):
    timer = Timer(inst, globals=globals())
    raw = numpy.array(timer.repeat(repeat, number=number))
    ave = raw.sum() / len(raw) / number
    mi, ma = raw.min() / number, raw.max() / number
    print("Average %1.3g min=%1.3g max=%1.3g" % (ave, mi, ma))
    return ave

print("Execution time for clr.predict")
speed("clr.predict(X_test)")

print("Execution time for ONNX Runtime")
speed("sess.run([label_name], {input_name: X_test.astype(numpy.float32)})[0]")

Out:

Execution time for clr.predict
Average 2.35e-05 min=1.95e-05 max=4.43e-05
Execution time for ONNX Runtime
Average 2.9e-05 min=2.69e-05 max=4.27e-05

Let’s benchmark a scenario similar to what a webservice experiences: the model has to do one prediction at a time as opposed to a batch of prediction.

def loop(X_test, fct, n=None):
    nrow = X_test.shape[0]
    if n is None:
        n = nrow
    for i in range(0, n):
        im = i % nrow
        fct(X_test[im: im+1])

print("Execution time for clr.predict")
speed("loop(X_test, clr.predict, 100)")

def sess_predict(x):
    return sess.run([label_name], {input_name: x.astype(numpy.float32)})[0]

print("Execution time for sess_predict")
speed("loop(X_test, sess_predict, 100)")

Out:

Execution time for clr.predict
Average 0.00202 min=0.00181 max=0.00239
Execution time for sess_predict
Average 0.00155 min=0.00128 max=0.00247

Let’s do the same for the probabilities.

print("Execution time for predict_proba")
speed("loop(X_test, clr.predict_proba, 100)")

def sess_predict_proba(x):
    return sess.run([prob_name], {input_name: x.astype(numpy.float32)})[0]

print("Execution time for sess_predict_proba")
speed("loop(X_test, sess_predict_proba, 100)")

Out:

Execution time for predict_proba
Average 0.00396 min=0.0034 max=0.00516
Execution time for sess_predict_proba
Average 0.00171 min=0.0014 max=0.00269

This second comparison is better as ONNX Runtime, in this experience, computes the label and the probabilities in every case.

Benchmark with RandomForest

We first train and save a model in ONNX format.

from sklearn.ensemble import RandomForestClassifier
rf = RandomForestClassifier()
rf.fit(X_train, y_train)

initial_type = [('float_input', FloatTensorType([1, 4]))]
onx = convert_sklearn(rf, initial_types=initial_type)
with open("rf_iris.onnx", "wb") as f:
    f.write(onx.SerializeToString())

We compare.

sess = rt.InferenceSession("rf_iris.onnx")

def sess_predict_proba_rf(x):
    return sess.run([prob_name], {input_name: x.astype(numpy.float32)})[0]

print("Execution time for predict_proba")
speed("loop(X_test, rf.predict_proba, 100)")

print("Execution time for sess_predict_proba")
speed("loop(X_test, sess_predict_proba_rf, 100)")

Out:

Execution time for predict_proba
Average 0.0578 min=0.0551 max=0.0601
Execution time for sess_predict_proba
Average 0.00177 min=0.00145 max=0.0029

Let’s see with different number of trees.

measures = []

for n_trees in range(5, 51, 5):
    print(n_trees)
    rf = RandomForestClassifier(n_estimators=n_trees)
    rf.fit(X_train, y_train)
    initial_type = [('float_input', FloatTensorType([1, 4]))]
    onx = convert_sklearn(rf, initial_types=initial_type)
    with open("rf_iris_%d.onnx" % n_trees, "wb") as f:
        f.write(onx.SerializeToString())
    sess = rt.InferenceSession("rf_iris_%d.onnx" % n_trees)
    def sess_predict_proba_loop(x):
        return sess.run([prob_name], {input_name: x.astype(numpy.float32)})[0]
    tsk = speed("loop(X_test, rf.predict_proba, 100)", number=5, repeat=5)
    trt = speed("loop(X_test, sess_predict_proba_loop, 100)", number=5, repeat=5)
    measures.append({'n_trees': n_trees, 'sklearn': tsk, 'rt': trt})

from pandas import DataFrame
df = DataFrame(measures)
ax = df.plot(x="n_trees", y="sklearn", label="scikit-learn", c="blue", logy=True)
df.plot(x="n_trees", y="rt", label="onnxruntime",
                ax=ax, c="green", logy=True)
ax.set_xlabel("Number of trees")
ax.set_ylabel("Prediction time (s)")
ax.set_title("Speed comparison between scikit-learn and ONNX Runtime\nFor a random forest on Iris dataset")
ax.legend()

Out:

5
Average 0.0353 min=0.0324 max=0.0386
Average 0.00156 min=0.00135 max=0.00173
10
Average 0.0541 min=0.0519 max=0.0567
Average 0.00157 min=0.00145 max=0.00167
15
Average 0.079 min=0.0768 max=0.0806
Average 0.00187 min=0.00155 max=0.00216
20
Average 0.103 min=0.0994 max=0.109
Average 0.0017 min=0.00168 max=0.00176
25
Average 0.121 min=0.117 max=0.125
Average 0.00195 min=0.00159 max=0.00231
30
Average 0.149 min=0.145 max=0.153
Average 0.00239 min=0.0016 max=0.00324
35
Average 0.17 min=0.163 max=0.178
Average 0.00199 min=0.00162 max=0.00269
40
Average 0.191 min=0.189 max=0.192
Average 0.00183 min=0.00171 max=0.00197
45
Average 0.214 min=0.212 max=0.216
Average 0.00252 min=0.00187 max=0.00345
50
Average 0.236 min=0.226 max=0.243
Average 0.00237 min=0.00182 max=0.00326

Total running time of the script: ( 0 minutes 49.102 seconds)