Fetching the paper…

Stochastic Weight Averaging in Parallel: Large-Batch Training that Generalizes Well · Around