apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2026-04-23 17:11:21 +08:00
Code Issues Actions 6 Packages Projects Releases Wiki Activity

[Feature] DeepseekV3 use pd_build_static_op (#2948)

Browse Source 
Co-authored-by: K11OntheBoat <“ruianmaidanglao@163.com”>
This commit is contained in:
K11OntheBoat
2025-07-22 15:03:41 +08:00
committed by GitHub
parent 2a8a2c06de
commit e991777757
6 changed files with 13 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files
custom_ops/gpu_ops/multi_head_latent_attention.cu
+2 -1
View File
@@ -13,6 +13,7 @@
// limitations under the License.
#include "append_attn/multi_head_latent_attention_kernel.h"
#include "helper.h"
#include "mla_attn/batch_mla_with_paged_kv_cache.h"
template <paddle::DataType D>
@@ -410,7 +411,7 @@ std::vector<paddle::DataType> MultiHeadLatentAttentionInferDtype(
}
}
PD_BUILD_OP(multi_head_latent_attention)
PD_BUILD_STATIC_OP(multi_head_latent_attention)
.Inputs({"query",
"key_cache",
"value_cache",