apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-12-24 13:28:13 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[Feature] Optim PaddleOCR-VL (#4873)

Browse Source 
* [Feature] Optim PaddleOCR-VL

* fix bug
This commit is contained in:
ming1753
2025-11-07 14:56:44 +08:00
committed by GitHub
parent bbe0820555
commit cba185f1fe
12 changed files with 535 additions and 112 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
custom_ops/gpu_ops/cpp_extensions.cc
View File

@@ -1059,6 +1059,15 @@ std::vector<paddle::Tensor> UpdateAttnMaskOffsets(
const paddle::Tensor& decode_states,
const paddle::Tensor& mask_rollback);
std::vector<paddle::Tensor> FusedNeoxRopeEmbedding(
const paddle::Tensor& qkv,
const paddle::Tensor& cos_emb,
const paddle::Tensor& sin_emb,
const int num_heads,
const int head_dim);
std::vector<paddle::Tensor> GeluTanh(paddle::Tensor& input);
PYBIND11_MODULE(fastdeploy_ops, m) {
m.def("get_expert_token_num",
&GetExpertTokenNum,
@@ -1648,4 +1657,10 @@ PYBIND11_MODULE(fastdeploy_ops, m) {
m.def("update_attn_mask_offsets",
&UpdateAttnMaskOffsets,
"update attention mask");
m.def("fused_neox_rope_embedding",
&FusedNeoxRopeEmbedding,
"fused_neox_rope_embedding function");
m.def("gelu_tanh", &GeluTanh, "gelu_tanh function");
}