How is it quantized, and why does the configuration show float32?

#1
by Sohaibsoussi - opened

Hello πŸ‘‹
I'm just curious about how did you quantize your model and push it on the hub since I tried multiple times and it failed πŸ˜”
I also saw in the configuration that it was in float32 , does it mean that hf do not accept qint8 format?

Sign up or log in to comment