Skip to content
OpenTrain AIFor AI Companies

OSDP: Optimal Sharded Data Parallel for Distributed Deep Learning

Youhe Jiang, Fangcheng Fu, Xupeng Miao, Xiaonan Nie, Bin CuiPublished Aug 1, 2023
DOI Publisher
Researcher verdict
Context only
Use as context only
Benchmark evidence
Missing
Not verified yet
Time to first repro
A few days
Plan setup time
Risk flags
2
Review before use

Abstract

Domain fit: AI-adjacent · Paper appears method- or tooling-adjacent to AI workflows with partial ecosystem coverage.

Large-scale deep learning models contribute to significant performance improvements on varieties of downstream tasks. Current data and model parallelism approaches utilize model replication and partition techniques to support the distributed training of ultra-large models. However, directly deploying these systems often leads to sub-optimal training efficiency due to the complex model architectures and the strict device memory constraints. In this paper, we propose Optimal Sharded Data Parallel (OSDP), an automated parallel training system that combines the advantages from both data and model parallelism. Given the model description and the device information, OSDP makes trade-offs between the memory consumption and the hardware utilization, thus automatically generates the distributed computation graph and maximizes the overall system throughput. In addition, OSDP introduces operator splitting to further alleviate peak memory footprints during training with negligible overheads, which enables the trainability of larger models as well as the higher throughput. Extensive experimental results of OSDP on multiple different kinds of large-scale models demonstrate that the proposed strategy outperforms the state-of-the-art in multiple regards.

Results and benchmarks

Freshness tier: cold
Large-scale deep learning models contribute to significant performance improvements on varieties of downstream tasks.

Implementation

No direct implementation yet

Maintained implementation evidence is not confirmed for this paper yet.

Use the implementation status and reproduction sections for the current action plan.

Implementation evidence summary
Confidence: low

Recommendation evidence is currently too limited for a maintained-repo choice. Use Implementation Status and Reproduction Path for a practical baseline plan.

Reproduction risks
  • Estimate is based on paper-only reproduction flow

Reproduction readiness

Time to first repro: days
Last checked: Aug 23, 2026

No repo

No verified implementation available

  • No maintained repository has been identified for this paper. Check adjacent implementations or HF artifacts below.

Hardware requirements

  • Expect multi-day setup/compute for meaningful reproduction based on current guidance.

Repositories and ecosystem

No additional verified repositories beyond the primary recommendation.

Hugging Face artifacts

No trustworthy direct or curated related Hugging Face artifacts were found yet. Use targeted searches to quickly locate candidate models, datasets, and demos.

Tip: start with models, then check datasets and spaces if you need evaluation data or demos.

Research context

7

Citations

33

References

Tasks

Computer science, Data parallelism, Computation, Throughput, Partition (number theory), Replication (statistics), Distributed memory, Distributed computing

Methods

Data modeling

Domains

Artificial intelligence, Computer Vision and Pattern Recognition

Evaluation and human feedback data

Open this paper in HFEPX to review benchmark signals, evaluation modes, and human-feedback protocol context.

Open in HFEPX
Explore similar papers