vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-10 17:14:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix attn weight scaling after rebase

Browse source
This commit is contained in:
HimariO 2025-04-07 22:07:56 +08:00
parent 223edef897
commit b28ad7ecca
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
examples/llava/clip.cpp
View file

@ -778,7 +778,7 @@ static ggml_cgraph * clip_image_build_graph_legacy(clip_ctx * ctx, const clip_im
if (full_attn) {
KQ = ggml_soft_max_ext(ctx0, KQ, nullptr, 1.0f / sqrtf((float)d_head), 0.0f);
} else {
KQ = ggml_soft_max_ext(ctx0, KQ, window_mask, 1.0f, 0.0f);
KQ = ggml_soft_max_ext(ctx0, KQ, window_mask, 1.0f / sqrtf((float)d_head), 0.0f);
}
struct ggml_tensor * KQV = ggml_mul_mat(ctx0, V, KQ);