apache / mxnet UNCLAIMED

Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more

0 0 0 C++

[FEATURE] Add tanh approximation for GeLU activation (#21034)

* Add gelu tanh activation

* fix sanity

* change tolerance

* apply review

* change backend function name to gelu_erf

* define const

bgawrych committed 3y ago

cdffaf09945f9cf3afd85ff91c78d66673247b15

Parent: c486a0e

Committed by GitHub <noreply@github.com> on 6/28/2022, 12:53:16 PM