root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-25 02:44:36 +00:00
Code Issues Actions 9 Packages Projects Releases Wiki Activity

llava : replace ggml_cpy with ggml_cont

Browse Source
This commit is contained in:
slaren 2024-02-19 14:02:36 +01:00 committed by Georgi Gerganov
parent 337c9cbd52
commit 6fd413791a
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
examples/llava/clip.cpp
View File

@ -618,7 +618,7 @@ static ggml_cgraph * clip_image_build_graph(clip_ctx * ctx, const clip_image_f32
KQV = ggml_reshape_4d(ctx0, KQV, d_head, num_positions, n_head, batch_size); KQV = ggml_reshape_4d(ctx0, KQV, d_head, num_positions, n_head, batch_size);
KQV = ggml_cont(ctx0, ggml_permute(ctx0, KQV, 0, 2, 1, 3)); KQV = ggml_cont(ctx0, ggml_permute(ctx0, KQV, 0, 2, 1, 3));
cur = ggml_cpy(ctx0, KQV, ggml_new_tensor_3d(ctx0, GGML_TYPE_F32, hidden_size, num_positions, batch_size)); cur = ggml_cont_3d(ctx0, KQV, hidden_size, num_positions, batch_size);
} }
// attention output // attention output