The Quantization Release Blog Post and various demos in the docs make heavy use of the torch.quantization.get_default_qconfig
method for configuring quantization algorithm behavior. Oddly, the torch.quantization API Reference makes no reference to this function.
This is unfortunate, because it means there isn’t anywhere to go to get a list of valid parameters to this method. If others agree, and this isn’t intentional, I’d be happy to fine a ticket for this in the GH tracker.