# Lyken17/Efficient-PyTorch

My best practice of training large dataset using PyTorch.

Repository: https://github.com/Lyken17/Efficient-PyTorch
Canonical: https://ross.abutalabs.com/products/efficient-pytorch
Language: Python
License Family: other
Last push: 2024-05-09T16:37:32+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3079, "days_push": 846, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1103, forks 136 (observed 2026-08-28T04:03:36.064760+00:00)

## What it is
A collection of best practices and example code for efficiently training large datasets like ImageNet with PyTorch. It demonstrates techniques such as packing small JPEGs into LMDB to eliminate disk I/O bottlenecks, achieving ~730 images/second training ResNet-50.

## Use cases
- speed up pytorch training on large image datasets
- avoid disk io bottleneck when training imagenet
- convert image folders to lmdb for faster data loading
- benchmark pytorch resnet-50 training throughput
- learn best practices for efficient pytorch data pipelines

## When to choose
- you train large-scale image classification models in PyTorch and hit data loading bottlenecks
- you want reference code for LMDB-based data pipelines and mixed-precision training tips

## When to avoid
- you need a maintained library with an API rather than example code
- your project uses TensorFlow, JAX, or non-image modalities
- you need a permissively licensed dependency, since the repo has no license

## Facets
- artifact type: learning-resource
- maturity: maintenance
- function: machine-learning, benchmarking, etl
- domain: deep-learning, machine-learning, developer-tools
- platform: python
- tags: pytorch, training-pipeline, data-loading, lmdb, performance-optimization, imagenet, best-practices, data-engineering, gpu, linux

## Member repositories
- Lyken17/Efficient-PyTorch (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:36.064760+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 2026-08-30T06:45:30.995056+00:00, confidence not recorded.
  - readme: https://github.com/Lyken17/Efficient-PyTorch (fetched 2026-08-28T04:03:36.064760+00:00, sha 0eeb8d90cd7e)
- Data as of 2026-08-30T08:39:29.467469+00:00.