Open Models Almanac
Methodsoptimizationtraining parallelism

single mention · optimization

Batch-size scaling with GPU count

The batch size is scaled with the number of GPUs as a remedy for rollout staleness.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See training parallelism for the methods filed beside it.

Also called scale the batch size with the GPU count.

Filed under optimization › training parallelism