TorchRec文档

TorchRec是一个PyTorch域库，用于提供大规模推荐系统(RecSys)所需的通用稀疏和并行原语。它允许作者使用跨多个gpu分割的大型嵌入表来训练模型。有关安装说明，请访问https://github.com/pytorch/torchrec#readme

有关PyTorch的安装

Torchrec需要Python >= 3.7和CUDA >= 11.0(强烈推荐CUDA以提高性能，但不是必需的)。下面的示例展示了如何使用CUDA 11.6进行安装。这个设置假定您已经安装了conda。

在Linux上运行Python 3.7、3.8和3.9的实验性二进制文件可以通过pip轮子安装。

TO use the library without cuda, use the *-cpu fbgemm installations. However, this will be much slower than the CUDA variant.

Nightly

conda install pytorch pytorch-cuda=11.7 -c pytorch-nightly -c nvidia
pip install torchrec_nightly

Stable

conda install pytorch pytorch-cuda=11.7 -c pytorch -c nvidia
pip install torchrec

If you have no CUDA device:

Nightly

pip uninstall fbgemm-gpu-nightly -y
pip install fbgemm-gpu-nightly-cpu

Stable

pip uninstall fbgemm-gpu -y
pip install fbgemm-gpu-cpu

在本教程中，我们介绍了主要的torchRec API，称为DistributedModelParallel，或DMP。与pytorch的DistributedDataParallel一样，DMP包装了一个模型以支持分布式训练。

TorchRec API

Contents:

• torchrec.datasets

• torchrec.datasets.criteo

• torchrec.datasets.movielens

• torchrec.datasets.random

• torchrec.datasets.utils

• torchrec.datasets.scripts

• torchrec.datasets.scripts.contiguous_preproc_criteo

• torchrec.datasets.scripts.npy_preproc_criteo

• torchrec.distributed

• torchrec.distributed.collective_utils

• torchrec.distributed.comm

• torchrec.distributed.comm_ops

• torchrec.distributed.dist_data

• torchrec.distributed.embedding

• torchrec.distributed.embedding_lookup

• torchrec.distributed.embedding_sharding

• torchrec.distributed.embedding_types

• torchrec.distributed.embeddingbag

• torchrec.distributed.grouped_position_weighted

• torchrec.distributed.model_parallel

• torchrec.distributed.quant_embeddingbag

• torchrec.distributed.train_pipeline

• torchrec.distributed.types

• torchrec.distributed.utils

• torchrec.distributed.planner

• torchrec.distributed.planner.constants

• torchrec.distributed.planner.enumerators

• torchrec.distributed.planner.partitioners

• torchrec.distributed.planner.perf_models

• torchrec.distributed.planner.planners

• torchrec.distributed.planner.proposers

• torchrec.distributed.planner.shard_estimators

• torchrec.distributed.planner.stats

• torchrec.distributed.planner.storage_reservations

• torchrec.distributed.planner.types

• torchrec.distributed.planner.utils

• torchrec.distributed.sharding

• torchrec.distributed.sharding.cw_sharding

• torchrec.distributed.dist_data

• torchrec.distributed.sharding.dp_sharding

• torchrec.distributed.sharding.rw_sharding

• torchrec.distributed.sharding.tw_sharding

• torchrec.distributed.sharding.twcw_sharding

• torchrec.distributed.sharding.twrw_sharding

• torchrec.fx

• torchrec.fx.tracer

• Module contents

• torchrec.inference

• torchrec.inference.model_packager

• torchrec.inference.modules

• Module contents

• torchrec.models

• torchrec.models.deepfm

• torchrec.models.dlrm

• Module contents

• torchrec.modules

• torchrec.modules.activation

• torchrec.modules.crossnet

• torchrec.modules.deepfm

• torchrec.modules.embedding_configs

• torchrec.modules.embedding_modules

• torchrec.modules.feature_processor

• torchrec.modules.lazy_extension

• torchrec.modules.mlp

• torchrec.modules.utils

• Module contents

• torchrec.optim

• torchrec.optim.clipping

• torchrec.optim.fused

• torchrec.optim.keyed

• torchrec.optim.warmup

• Module contents

• torchrec.quant

• torchrec.quant.embedding_modules

• Module contents

• torchrec.sparse

• torchrec.sparse.jagged_tensor

• Module contents