+ jD^RIt^RIHt!RR]P4tR#)NcTa]tRt^toRt]P V3RlRl4tRtVt R#)SharderaT This is an interface which allows user to create more advanced sharding strategies that are not easily be composed by the `ShardingSpec`. :class:`torch.distributed._shard.sharding_plan.ShardingPlan` could take an object of the `Sharder` and call `shard` to shard the module, then replace the original module with sharded module returned. cN<V^8dQhRS[PRS[P/#)modulereturn)nnModule)format __classdict__s"a/home/luca/screen-checking/.venv/lib/python3.14/site-packages/torch/distributed/_shard/sharder.py __annotate__Sharder.__annotate__s#   BII  "))  c R#)a) Shard a module base on the implementation of this method, and return the sharded version of the module. Args: module (:class:`torch.nn.Module`): The module to apply sharding to. Returns: A :class:`torch.nn.Module` object that represents a module that's already been sharded. N)selfrs&&r shard Sharder.shardsrrN) __name__ __module__ __qualname____firstlineno____doc__abcabstractmethodr__static_attributes____classdictcell__)r s@r rrs$     rr)rtorch.nnr ABCrrrr r!s  cgg r