vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fix rope_cuda parameter order

Browse source
This commit is contained in:
Cebtenzzre 2023-10-08 14:31:35 -04:00
parent 9bd050f133
commit babf0e0c8f
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
ggml-cuda.cu
View file

@ -5469,8 +5469,8 @@ static void scale_f32_cuda(const float * x, float * dst, const float scale, cons
template<typename T>
static void rope_cuda(
const T * x, T * dst, int ncols, int nrows, float freq_scale, float ext_factor, float attn_factor,
float theta_scale, const int32_t * pos, int p_delta_rows, rope_corr_dims corr_dims, cudaStream_t stream
const T * x, T * dst, int ncols, int nrows, const int32_t * pos, float freq_scale, int p_delta_rows,
float theta_scale, float ext_factor, float attn_factor, rope_corr_dims corr_dims, cudaStream_t stream
) {
GGML_ASSERT(ncols % 2 == 0);
const dim3 block_dims(1, CUDA_ROPE_BLOCK_SIZE, 1);