coreai_opt.quantization.spec.QuantizationGranularity¶ class coreai_opt.quantization.spec.QuantizationGranularity[source]¶ Bases: BaseModel, ConfigRegistryMixin Base class for quantization granularity specifications. get_block_size(tensor_shape)[source]¶ Get a list of block sizes based on the granularity. Parameters: tensor_shape (Size) Return type: tuple[int, …]
