You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I think that the tensor224\225 is repeatedly quantized, we need check if(weight_tensor->data_type != TENGINE_DT_UINT8) then quantize it. After i do this, it works.
The text was updated successfully, but these errors were encountered:
My model has two inputs, so I modify its quantization code as follows:
The weights are shared as follows:
![Screenshot from 2021-09-03 16-30-10](https://user-images.githubusercontent.com/25733341/131975588-a918e973-e580-4c7f-8792-6629aa9b197d.png)
![Screenshot from 2021-09-03 16-30-21](https://user-images.githubusercontent.com/25733341/131975602-b3ddf15f-12c5-44c0-ab89-f515f69870c5.png)
output:
![Screenshot from 2021-09-03 16-33-02](https://user-images.githubusercontent.com/25733341/131975879-3f1c2679-b29f-490e-bc03-791b48707b63.png)
I think that the tensor224\225 is repeatedly quantized, we need check if(weight_tensor->data_type != TENGINE_DT_UINT8) then quantize it. After i do this, it works.
The text was updated successfully, but these errors were encountered: