概率或原始分数

分类器通常会返回一个概率矩阵。默认情况下，sklearn-onnx 会创建一个返回概率的 ONNX 图，但如果模型实现了 decision_function 方法，它可能会跳过此步骤并返回原始分数。选项 'raw_scores' 用于更改默认行为。让我们通过一个简单的例子来看看。

训练模型并进行转换

import numpy
import sklearn
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
import onnxruntime as rt
import onnx
import skl2onnx
from skl2onnx.common.data_types import FloatTensorType
from skl2onnx import convert_sklearn
from sklearn.linear_model import LogisticRegression

iris = load_iris()
X, y = iris.data, iris.target
X_train, X_test, y_train, y_test = train_test_split(X, y)
clr = LogisticRegression(max_iter=500)
clr.fit(X_train, y_train)
print(clr)

initial_type = [("float_input", FloatTensorType([None, 4]))]
onx = convert_sklearn(clr, initial_types=initial_type, target_opset=12)

LogisticRegression(max_iter=500)

输出类型

让我们用 onnxruntime 确认概率的输出类型是字典列表。

sess = rt.InferenceSession(onx.SerializeToString(), providers=["CPUExecutionProvider"])
res = sess.run(None, {"float_input": X_test.astype(numpy.float32)})
print("skl", clr.predict_proba(X_test[:1]))
print("onnx", res[1][:2])

skl [[9.58822528e-01 4.11771279e-02 3.43780611e-07]]
onnx [{0: 0.9588225483894348, 1: 0.041177116334438324, 2: 3.437804991790472e-07}, {0: 1.1261729923717212e-05, 1: 0.0612429603934288, 2: 0.9387457966804504}]

原始分数和 decision_function

initial_type = [("float_input", FloatTensorType([None, 4]))]
options = {id(clr): {"raw_scores": True}}
onx2 = convert_sklearn(
    clr, initial_types=initial_type, options=options, target_opset=12
)

sess2 = rt.InferenceSession(
    onx2.SerializeToString(), providers=["CPUExecutionProvider"]
)
res2 = sess2.run(None, {"float_input": X_test.astype(numpy.float32)})
print("skl", clr.decision_function(X_test[:1]))
print("onnx", res2[1][:2])

skl [[ 5.9963453   2.84852226 -8.84486756]]
onnx [{0: 5.996345520019531, 1: 2.848522186279297, 2: -8.844867706298828}, {0: -6.6440277099609375, 1: 1.957166075706482, 2: 4.686861991882324}]

此示例使用的版本

print("numpy:", numpy.__version__)
print("scikit-learn:", sklearn.__version__)
print("onnx: ", onnx.__version__)
print("onnxruntime: ", rt.__version__)
print("skl2onnx: ", skl2onnx.__version__)

numpy: 2.3.1
scikit-learn: 1.6.1
onnx:  1.19.0
onnxruntime:  1.23.0
skl2onnx:  1.19.1