The latest #MLPerf inference results are in and they show #DeepSparse providing ~50x improvements over baseline BERT-Large reference implementation on both AWS ARM and GCP x86 instances. See how and replicate our results today: