Download Model¶
netspresso.np_qai.quantizer.NPQAIQuantizer.download_model(job, filename)
¶
Download a model from the QAI Hub and handle ONNX zip conversion.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
job
|
Union[CompileJob, QuantizeJob]
|
The job to download the model from (CompileJob or QuantizeJob). |
required |
filename
|
str
|
The filename to save the model to. |
required |
Returns:
| Name | Type | Description |
|---|---|---|
str |
str
|
The actual path of the downloaded and processed model. |
Note
Since QAI Hub July 2025 update, ONNX models are always downloaded as .zip files with external weights. Other formats (.tflite, .so, .bin) are downloaded directly. Reference: https://app.aihub.qualcomm.com/docs/hub/release_notes.html#released-july-14-2025
This function automatically extracts and converts ONNX zip files to single .onnx files with embedded weights. Non-ONNX models are returned as-is.