insightface
insightface copied to clipboard
torch2onnx file size
I recently trained a wf42m-vitB-pfc03 model using the wf42m_pfc03_40epoch_8gpu_vit_b config. It came out as 444MB. After converting to onnx using the provided torch2onnx script, it came out slightly larger in size. When using the simplify
argument it became slightly bigger than without.
Is this suppose to happen?