ONNX Export¶

Export trained MLP surrogates for deployment anywhere.

MLP only

GP models store training data for inference and cannot be exported. Use method="mlp" if you need portable surrogates.

Basic export¶

surrogate.export("model.onnx")
# Creates: model__CL.onnx, model__CD.onnx  (one file per output)

Loading with onnxruntime¶

import onnxruntime as ort
import numpy as np

session = ort.InferenceSession("model__CL.onnx")
cl = session.run(None, {"X": x_new.astype(np.float32)})[0]

Deployment targets¶

Target	Description
Real-time control	C++ inference via ONNX Runtime C API
Edge devices	TensorRT-optimized ONNX on Jetson
Cloud serving	Triton Inference Server with ONNX backend
Python applications	`onnxruntime` package (pip install)