AWhat changed
Combined table/row/column/data-parallel embedding placement, fused sparse operators, hierarchical memory and a dedicated scale-out network in a PyTorch-based training stack for very large recommendation models.
Scientific scope
Combined table/row/column/data-parallel embedding placement, fused sparse operators, hierarchical memory and a dedicated scale-out network in a PyTorch-based training stack for very large recommendation models. One2021 preprint/2022 publication family, with the expanded2023-stamped v7 byline retained separately. Four-dimensional embedding parallelism is not WLB-LLM's DP/TP/PP/CP scheme. Placement uses cost models and heuristics rather than globally optimal observed runtime; column sharding can affect row-wise optimizer semantics. Reported speed and model capacity are configuration-specific.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. One2021 preprint/2022 publication family, with the expanded2023-stamped v7 byline retained separately. Four-dimensional embedding parallelism is not WLB-LLM's DP/TP/PP/CP scheme. Placement uses cost models and heuristics rather than globally optimal observed runtime; column sharding can affect row-wise optimizer semantics. Reported speed and model capacity are configuration-specific.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
- High-performance, Distributed Training of Large-scale Deep Learning Recommendation Models
- Software-Hardware Co-design for Fast and Scalable Training of Deep Learning Recommendation Models
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction