Evaluate a quantize model
Psancs05 opened this issue · comments
I am trying to run an evaluation in a model that is quantized. I have to instantiate it using 'accelerate' to use the GPU, because otherwise it cannot fit in memory. The problem is that when using compute() in the metric I want, I get this error:
ValueError: The model has been loaded with
accelerateand therefore cannot be moved to a specific device. Please discard the
device argument when creating your pipeline object.
Is there any way to allow using the compute() method in a model that is in GPU?
I'm also having the same problem... @Psancs05 were you able to find a workaround for this?
yeah me too what's the solution for it