This is an FP16 model
#19
by
daryltucker
- opened
This model is primarily FP16, with just a few FP32 normalization layers.
When attempting to quantize this to FP16, the resulting model is identical.
This model is primarily FP16, with just a few FP32 normalization layers.
When attempting to quantize this to FP16, the resulting model is identical.