Importing a custom model #1288
-
I am following the cnv_end2end example to apply it on a custom YOLO-based model. However, in the step of exporting the model into QONNX format, I can't find what specifically I should import in the first argument which asks for the model export_qonnx(cnv, torch.randn(1, 3, 32, 32), export_onnx_path). I mean that in YOLO there is yolo.py which has the detect class as well as the .yaml file so what should I import? |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
Hi, |
Beta Was this translation helpful? Give feedback.
Hi,
you should export a Brevitas PyTorch model, like this CNV example model: https://github.com/Xilinx/brevitas/blob/master/src/brevitas_examples/bnn_pynq/models/CNV.py