Commit 8d3a25ab authored by Nathan Luehr's avatar Nathan Luehr Committed by Gunhan Gulsoy
Browse files

Update Eigen hash for fix of predux bug (#14770) 

For Maxwell and earlier GPUs, Eigen was incorectly casting fp16 values to
unsigned int during some reductions. This results in incorrect results in
Tensorflow's xent and sparse_xent ops when applied to fp16 data.
parent 04807b62
Show whitespace changes
Inline Side-by-side
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please to comment