Business Context
A manufacturing company deploys ONNX models to edge devices for detecting machine anomalies. Before deployment, the MLOps pipeline validates the structural integrity of the 'v2.0' ONNX model to ensure no corruption occurred during the training export process.
Data Preparation
Simulation of a valid ONNX binary file loaded into a CAS table.
Copied!
| 1 | DATA casuser.onnx_iot_valid; |
| 2 | LENGTH model_blob varbinary; |
| 3 | /* Simulating ONNX header signature */ |
| 4 | model_blob = '080012204...'x; |
| 5 | OUTPUT; |
| 6 | RUN; |
Étapes de réalisation
1
Verify the presence of the table and the binary column.
Copied!
| 1 | |
| 2 | PROC CAS; |
| 3 | |
| 4 | TABLE.tableInfo / caslib='casuser' TABLE='onnx_iot_valid'; |
| 5 | |
| 6 | RUN; |
| 7 |
2
Execute astore.check to validate the model blob.
Copied!
| 1 | |
| 2 | PROC CAS; |
| 3 | |
| 4 | astore.check / onnx={TABLE={caslib='casuser', name='onnx_iot_valid'}}; |
| 5 | |
| 6 | RUN; |
| 7 |
Expected Result
The action completes without errors. The log confirms that the ONNX model is valid and ready for subsequent steps like 'describe' or 'score'.