ikawrakow/ik_llama.cpp
1
0
Fork 0
mirror of https://github.com/ikawrakow/ik_llama.cpp.git synced 2026-03-05 11:30:09 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix bug: missing parentheses in logical expression (#275)

Browse Source 
This results in GGGGGGGGGGGGG when generating with
mla = 3, fa = 0.

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow
2025-03-21 13:23:01 +01:00
committed by GitHub
parent 022660f7ab
commit 3d6e25c82d
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/llama.cpp
View File

@@ -13869,7 +13869,7 @@ struct llm_build_context {
ggml_tensor * q = ggml_concat(ctx0, q_nope2, ggml_permute(ctx0, q_rope, 0, 2, 1, 3), 0); ggml_tensor * q = ggml_concat(ctx0, q_nope2, ggml_permute(ctx0, q_rope, 0, 2, 1, 3), 0);
cb(q, "q", il); cb(q, "q", il);
if (lctx.cparams.flash_attn && lctx.cparams.mla_attn == 1 || lctx.cparams.mla_attn == 3) { if (lctx.cparams.flash_attn && (lctx.cparams.mla_attn == 1 || lctx.cparams.mla_attn == 3)) {
ggml_tensor * kv_cache_lora = ggml_view_2d(ctx0, kv_self.kv_l[il], ggml_tensor * kv_cache_lora = ggml_view_2d(ctx0, kv_self.kv_l[il],
kv_lora_rank, n_kv, kv_lora_rank, n_kv,
ggml_row_size(kv_self.kv_l[il]->type, kv_lora_rank + n_embd_head_qk_rope), 0); ggml_row_size(kv_self.kv_l[il]->type, kv_lora_rank + n_embd_head_qk_rope), 0);