torchrec
torchrec copied to clipboard
pytorch
Pytorch domain library for recommendation systems
Summary: For test_concat, the first values should be from [0:3] given representation ``` # f0 [] [1 2] [] [3] # f1 [4] [5] [] [6 7 8] # f2...
Differential Revision: D62699600
Summary: # Problem `EmbeddingTableType` is introduced recently for ZCH. There are two same definitions of `EmbeddingTableType` * one is defined in [torchrec/fb/distributed/sparsenn_configs.py](https://www.internalfb.com/code/fbsource/[3a83814d95f27eec3bf425d468a5234d6ca8a381]/fbcode/torchrec/fb/distributed/sparsenn_configs.py?lines=110) * The other is defined in [minimal_viable_ai/core/model_family_api/configs.py](https://www.internalfb.com/code/fbsource/[bbe727e13ae251889c5df57b58f2cbc05675662e]/fbcode/minimal_viable_ai/core/model_family_api/configs.py?lines=76) **Problem...
Summary: For DTensor based gradient clipping, in the 2D or nD parallelism case we don't know the specific dimension for FSDP, to avoid potential accuracy issue, we always do global...
Summary: X-link: https://github.com/pytorch/pytorch/pull/135653 We introduced the dispatchable backend for a ProcessGroup and collective in https://github.com/pytorch/pytorch/issues/86225. This PR is a follow-up cleanup to clean up the option of a ProcessGroup and...
This is an example of how we can hide content in .rst files that we don't want to expose yet. We can hide the APIs that we don't want to...
Summary: as per title Reviewed By: TroyGarden Differential Revision: D62321012
Summary: ^ Differential Revision: D62063705
