Evaluation of pre-training large language models on leadership-class supercomputers

Connections are not prepared for this record yet, or its source metadata has changed. The original record remains available while background snapshots are rebuilt.