Will dataset object be replicated on each worker process for map-style dataset?

In the torch.utils.data — PyTorch 1.9.0 documentation about utils.data, it is clearly pointed out that each worker process gets a replica of the dataset object for iterable-style datasets, but it’s not clear whether that is also the case for map-style dataset.

So will dataset object be replicated on each worker process for map-style dataset😕?