Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-25 13:58:46 +01:00
Code Issues Packages Projects Releases Wiki Activity

cuda : fix RoPE after #2268 (#3897)

Browse Source
This commit is contained in:
cebtenzzre 2023-11-02 01:49:44 -04:00 committed by GitHub
parent 0eb332a10f
commit 2fffa0d61f
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
ggml-cuda.cu
View File

@ -4539,7 +4539,7 @@ static __global__ void rope(
const int i2 = row/p_delta_rows; const int i2 = row/p_delta_rows;
const int p = has_pos ? pos[i2] : 0; const int p = has_pos ? pos[i2] : 0;
const float theta_base = p*powf(freq_base, -col/ncols); const float theta_base = p*powf(freq_base, -float(col)/ncols);
float cos_theta, sin_theta; float cos_theta, sin_theta;
rope_yarn(theta_base, freq_scale, corr_dims, col, ext_factor, attn_factor, &cos_theta, &sin_theta); rope_yarn(theta_base, freq_scale, corr_dims, col, ext_factor, attn_factor, &cos_theta, &sin_theta);
@ -4566,8 +4566,8 @@ static __global__ void rope_neox(
const int i = row*ncols + col/2; const int i = row*ncols + col/2;
const int i2 = row/p_delta_rows; const int i2 = row/p_delta_rows;
// simplified from `(row * ncols + col) * (-1 / ncols)` // simplified from `(ib * ncols + col) * (-1 / ncols)`, where ib is assumed to be zero
const float cur_rot = -col/ncols - row; const float cur_rot = -float(col)/ncols;
const int p = has_pos ? pos[i2] : 0; const int p = has_pos ? pos[i2] : 0;
const float theta_base = p*powf(freq_base, cur_rot); const float theta_base = p*powf(freq_base, cur_rot);