Efficient Training of Large Language Models on Legacy GPUs with HetSeq and PyTorch.

Bruno Leite Franco, Edson Emlio Scalabrin

Browse the full CSCWD paper archive.