BWhat changed
Documented output-head normalization, maximum-logit regularization, elastic distributed-training choices and intermediate checkpoints for bilingual language-model training, enabling training-dynamics research.
Scientific scope
Documented output-head normalization, maximum-logit regularization, elastic distributed-training choices and intermediate checkpoints for bilingual language-model training, enabling training-dynamics research. Normalization and logit regularization are adaptations with cited predecessors, not invention of normalization, z-loss, ZeRO or positional encodings. Acquire general language-model training only; medical/legal benchmark sections do not establish clinical outcomes and are outside new intake. Model sizes share one package.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Normalization and logit regularization are adaptations with cited predecessors, not invention of normalization, z-loss, ZeRO or positional encodings. Acquire general language-model training only; medical/legal benchmark sections do not establish clinical outcomes and are outside new intake. Model sizes share one package.
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.
- Baichuan 2: Open Large-scale Language Models
- Yang et al.: Baichuan2 training-stability and checkpoint resource — official research provenance
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