Erjia Guan f1a6f32b72 [DataLoader] Make distributed lazily initialized & share seed via PG (#85279) ... Fixes #84492 https://github.com/pytorch/data/issues/772 ## Changes - Move the logic of distributed sharding from the constructor of DataLoader to the constructor of DataLoaderIterator. This would prevent the Error caused by lazy distributed process initialization - Replace distributed store by process group (`gloo`) to share the random seed because `mpi` backend doesn't provide distributed store. Pull Request resolved: https://github.com/pytorch/pytorch/pull/85279 Approved by: https://github.com/NivekT, https://github.com/VitalyFedyunin		2022-09-23 18:52:52 +00:00
..
__init__.py	[DataLoader] Make distributed lazily initialized & share seed via PG (#85279)	2022-09-23 18:52:52 +00:00
collate.py	Integrate xdoctest - Rebased (#82797)	2022-08-12 02:08:01 +00:00
fetch.py	Fix fetch function which breaks user code (#85099)	2022-09-15 21:48:28 +00:00
pin_memory.py	[DataLoader] Move loop content into a function to ensure we don't preserve anything (#83595)	2022-08-18 20:54:47 +00:00
serialization.py	[DataPipe] Revamp serialization logic of DataPipes	2022-05-10 16:16:46 +00:00
signal_handling.py
worker.py	Set shuffle to DataPipes with set_shuffle API (#83741)	2022-09-13 13:38:58 +00:00