DOI: 10.1145/3839112 ISSN: 1542-7730

Finding the Real Bottleneck in Large-Scale ML Training

Ankit Sinha

Real efficiency requires fixing the underlying systems