Skip to main content
Models deployed to MemryX accelerators can be verified before initializing the async accelerator with the secure loader: 
from memryx import AsyncAccl
from thistle_secure_loader import secure_load

def memryx_loader(path: str):
    return AsyncAccl(path)

accl = secure_load("model.onnx", memryx_loader)
print("MemryX model verified securely.")
The secure_load call verifies the model file’s signature using tuc before passing it to AsyncAccl. If the signature check fails, a ModelVerificationError is raised and the model is never loaded.

​
Requirements

The memryx package is Linux only (x86_64 / ARM64) and is not available for macOS or Windows.
--extra-index-url https://developer.memryx.com/pip
memryx