BWhat changed
Balanced packed-document training using profiled compute/communication costs, variable-length micro-batches and delayed outlier documents, alongside per-document context sharding to address pipeline and context imbalance.
Scientific scope
Balanced packed-document training using profiled compute/communication costs, variable-length micro-batches and delayed outlier documents, alongside per-document context sharding to address pipeline and context imbalance. Document attention work is not determined by token count alone. Packing is heuristic under memory and delay constraints; delaying long documents changes sample timing and may affect data randomness. Reported balance and speedups are measured for selected configurations, not an exact globally optimal schedule for every model.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Document attention work is not determined by token count alone. Packing is heuristic under memory and delay constraints; delaying long documents changes sample timing and may affect data randomness. Reported balance and speedups are measured for selected configurations, not an exact globally optimal schedule for every model.
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