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

fix-review

Browse source
This commit is contained in:
simonJJJ 2024-04-16 13:29:42 +08:00
parent 0010280042
commit 7355ca84b5
1 changed files with 0 additions and 6 deletions
Download patch file Download diff file Expand all files Collapse all files

6
llama.cpp
View file

@ -8537,12 +8537,6 @@ struct llm_build_context {
Vcur = ggml_add(ctx0, Vcur, model.layers[il].bv);
cb(Vcur, "Vcur", il);
// these nodes are added to the graph together so that they are not reordered
// by doing so, the number of splits in the graph is reduced
ggml_build_forward_expand(gf, Qcur);
ggml_build_forward_expand(gf, Kcur);
ggml_build_forward_expand(gf, Vcur);
Qcur = ggml_rope_custom(
ctx0, ggml_reshape_3d(ctx0, Qcur, n_embd_head, n_head, n_tokens), inp_pos,
n_rot, rope_type, 0, n_orig_ctx, freq_base, freq_scale,