Not able to load QNN Context Binary Model

[albertsuchvi]

Describe the issue

Python API: When trying to load a QNN Context Binary model saved from a previous run on the QNNExecutionProvider, the program finishes unexpectedly. No exception is raised, so I do not have a way to know where it is exactly failing.

To reproduce

Step1: run a model using the QNNExecutionProvider and option ep.context_enable = 1, this generates an ONNX model with an EPContext node containing the binary context cache for the provider

Step2: try to load the generated model. The python program fails without throwing any exception

Attached (in zip file)

• sample code (run_esrgan_upscale_model.zip

Result of pip -list:
Package Version

---

coloredlogs 15.0.1
flatbuffers 24.12.23
humanfriendly 10.0
mpmath 1.3.0
numpy 2.2.2
onnx 1.17.0
onnxruntime-qnn 1.20.1
opencv-python 4.11.0.86
packaging 24.2
pip 24.2
protobuf 5.29.3
pyreadline3 3.5.4
setuptools 75.1.0
sympy 1.13.3
wheel 0.44.0

Urgency

No response

Platform

Windows

OS Version

Windows11

ONNX Runtime Installation

Released Package

ONNX Runtime Version or Commit ID

onnxruntime-qnn 1.20.1

ONNX Runtime API

Python

Architecture

ARM64

Execution Provider

Other / Unknown

Execution Provider Library Version

QNNExecutiopnProvider