Skip to content

P
Paddle

PaddlePaddle / Paddle
大约 1 年 前同步成功

通知 2298
Star 20931
Fork 5422
P
Paddle
未验证 提交 bf0ef606 编写于 作者: Z zhaoyingli 提交者： GitHub
浏览文件
操作

disable fast_layer_norm (#56263)

上级 c27bd049
显示空白变更内容
内联 并排
Showing with 1 addition and 1 deletion
paddle/phi/kernels/gpu/layer_norm_kernel.cu
浏览文件 @ bf0ef606
...@@ -606,7 +606,7 @@ void LayerNormKernel(const Context &dev_ctx, ...@@ -606,7 +606,7 @@ void LayerNormKernel(const Context &dev_ctx,
if ((feature_size >= 768 && feature_size <= 2048 && feature_size % 256 == 0 || if ((feature_size >= 768 && feature_size <= 2048 && feature_size % 256 == 0 ||
feature_size == 4096) && feature_size == 4096) &&
scale != nullptr && bias != nullptr) { scale != nullptr && bias != nullptr) {
can_call_fast_kernel = true; can_call_fast_kernel = false;
} }
if (can_call_fast_kernel) { if (can_call_fast_kernel) {
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或