apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2026-04-23 00:17:25 +08:00
Code Issues Actions 19 Packages Projects Releases Wiki Activity

init (#6642)

Browse Source 
Co-authored-by: gongweibao <gognweibao@baidu.com>
This commit is contained in:
gongweibao
2026-03-04 21:55:31 +08:00
committed by GitHub
parent 5c8f5184d9
commit ddb06ff83f
306 changed files with 40627 additions and 34418 deletions
Download Patch File Download Diff File Expand all files Collapse all files
custom_ops/gpu_ops/fused_rotary_position_encoding.cu
-3
View File
@@ -15,7 +15,6 @@
#include "helper.h"
#include "paddle/extension.h"
template <typename T, bool IS_NEOX>
inline __device__ void apply_token_rotary_embedding_kernel(
T* __restrict__ arr,
@@ -43,7 +42,6 @@ inline __device__ void apply_token_rotary_embedding_kernel(
arr[y_index] = y * cos + x * sin;
}
template <typename T, bool IS_NEOX>
__global__ void apply_rotary_embedding_kernel(
T* __restrict__ query, // [num_tokens, num_heads, head_size]
@@ -84,7 +82,6 @@ __global__ void apply_rotary_embedding_kernel(
}
}
void FusedRotaryPositionEncoding(
paddle::Tensor& query, // [num_tokens, num_heads, head_size] or
// [num_tokens, num_heads * head_size]