# meta-pytorch/torchrec

Pytorch domain library for recommendation systems

Repository: https://github.com/meta-pytorch/torchrec
Canonical: https://ross.abutalabs.com/products/torchrec
Homepage: https://pytorch.org/torchrec/
Language: Python
License: BSD-3-Clause
License Family: permissive
Topics: pytorch, gpu, deep-learning, cuda, recommender-system, recommendation-system, sharding
Last push: 2026-08-26T21:05:18+00:00

## Health v2 (maintenance only)
Score: 90/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 73, longevity 100
- inputs: {"age_days": 1878, "days_push": 7, "days_rel": 21, "gap_med": 91.5, "n_releases_24m": 9}
- flags: no_readme
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2599, forks 684 (observed 2026-08-28T04:07:03.289020+00:00)

## What it is
TorchRec is a PyTorch domain library for building recommendation systems at scale. It provides distributed sharding of large embedding tables and GPU-accelerated training and inference for recommender models.

## Use cases
- train large recommendation models on multiple gpus
- shard huge embedding tables across a cluster
- build a recommender system with pytorch
- run distributed training for deep learning recommendation models
- serve recommendation model inference efficiently on gpu

## When to choose
- you need to scale embedding tables beyond a single gpu's memory
- you are building recommender models in the pytorch ecosystem
- you want optimized gpu kernels for sparse embeddings

## When to avoid
- you need a simple non-distributed recommender prototype
- your stack is tensorflow-based rather than pytorch
- you need a turnkey recommendation service rather than a training library

## Facets
- artifact type: library
- maturity: active
- function: deep-learning, machine-learning, gpu-computing
- domain: machine-learning, deep-learning, gpu-computing
- platform: python
- tags: pytorch, recommender-systems, sharding, distributed-training, embedding-tables, recommendation-systems, gpu, linux, cuda

## Member repositories
- meta-pytorch/torchrec (main) score 90

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:03.289020+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:21:17.862019+00:00, confidence not recorded.
  - registry_pypi: https://pypi.org/pypi/torchrec/json (fetched 2026-08-29T10:04:14.371188+00:00, sha 296a4d1e9843)
- Data as of 2026-08-30T08:39:29.467469+00:00.