Closed Amin456789 closed 6 months ago
Hi! I tested it over ONNX with cpu. It seems that the memory is not the main constraint in this case. Quantization can reduce the RAM consumption for ~1GB. However, the speed can be slower.
thank you for the answer
Hello!
thank you for this, this is pretty cool!
i have two questions
does it work with cpu too? i used onnx int8 version in the past and it worked very well with cpu, how about tiny sd?
can u please work on AnimateDiff support too? it will be very handy if we have a tiny animatediff with cpu support for us laptop cpu users
kinds regards