Vectorization is now wrapped in a sklearn compatible transformer for pipelining. Next step ahead would be paralleization over multiple CPUs to speed up the processing. Right now single core is a major bottleneck before training and inference while working with large datasets.
Vectorization is now wrapped in a sklearn compatible transformer for pipelining. Next step ahead would be paralleization over multiple CPUs to speed up the processing. Right now single core is a major bottleneck before training and inference while working with large datasets.