Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-28 15:18:26 +01:00
Code Issues Packages Projects Releases Wiki Activity

cuda : fix soft_max to use correct mask size

Browse Source
This commit is contained in:
Georgi Gerganov 2024-02-01 16:47:20 +02:00
parent 910b15bb40
commit 2e46013749
No known key found for this signature in database
GPG Key ID: BF970631944C16B7
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
ggml-cuda.cu
View File

@ -9064,7 +9064,7 @@ static void ggml_cuda_op_soft_max(
const int64_t ne00 = src0->ne[0];
const int64_t nrows_x = ggml_nrows(src0);
const int64_t nrows_y = src1 ? ggml_nrows(src1) : 1;
const int64_t nrows_y = src1 ? src0->ne[1] : 1; // note: using number of queries since mask can be padded!
float scale = 1.0f;
memcpy(&scale, dst->op_params, sizeof(float));