-
Myle Ott authored
Summary: * Add FusedLayerNorm and FusedAdam * Softmax and zero grad optimizations Pull Request resolved: https://github.com/pytorch/fairseq/pull/531 Differential Revision: D14218457 Pulled By: myleott fbshipit-source-id: 5656b2d0152cd85f77dc21ec0e1439ec04b9fa89
48d9afbe
