Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Enhancement] fix data type in fused-bias-leakyrelu for apex fp16 training #981

Merged
merged 1 commit into from
Apr 24, 2021
Merged

[Enhancement] fix data type in fused-bias-leakyrelu for apex fp16 training #981

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 5 additions & 5 deletions mmcv/ops/fused_bias_leakyrelu.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,10 +45,9 @@ def backward(ctx, gradgrad_input, gradgrad_bias):
# The second order deviation, in fact, contains two parts, while the
# the first part is zero. Thus, we direct consider the second part
# which is similar with the first order deviation in implementation.
gradgrad_out = ext_module.fused_bias_leakyrelu(gradgrad_input,
gradgrad_bias, out, 3,
1, ctx.negative_slope,
ctx.scale)
gradgrad_out = ext_module.fused_bias_leakyrelu(
gradgrad_input, gradgrad_bias.to(out.dtype), out, 3, 1,
ctx.negative_slope, ctx.scale)

return gradgrad_out, None, None, None

Expand Down Expand Up @@ -139,7 +138,8 @@ def fused_bias_leakyrelu(input, bias, negative_slope=0.2, scale=2**0.5):
if not input.is_cuda:
return bias_leakyrelu_ref(input, bias, negative_slope, scale)

return FusedBiasLeakyReLUFunction.apply(input, bias, negative_slope, scale)
return FusedBiasLeakyReLUFunction.apply(input, bias.to(input.dtype),
negative_slope, scale)


def bias_leakyrelu_ref(x, bias, negative_slope=0.2, scale=2**0.5):
Expand Down