AWhat changed
Integrated data, operator-model, pipeline and optimizer parallelism with rematerialization in MindSpore Auto-parallel for large Chinese autoregressive model training on Ascend accelerators.
Scientific scope
Integrated data, operator-model, pipeline and optimizer parallelism with rematerialization in MindSpore Auto-parallel for large Chinese autoregressive model training on Ascend accelerators. Transformer/GPT pretraining, parallelization and rematerialization are inherited. Acquired delta is integrated training system and Chinese model resource, not each parallel dimension. Processor/model scales and language evaluations are historical reported experiments, not current performance guarantees.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Transformer/GPT pretraining, parallelization and rematerialization are inherited. Acquired delta is integrated training system and Chinese model resource, not each parallel dimension. Processor/model scales and language evaluations are historical reported experiments, not current performance guarantees.
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.
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