CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-06 09:46:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

cont : fix build

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-06-24 15:59:47 +03:00
parent 132143938f
commit 401c13e3c3
1 changed files with 5 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/llama-kv-cache-unified.cpp
View File

@@ -908,10 +908,12 @@ ggml_tensor * llama_kv_cache_unified::get_k(ggml_context * ctx, int32_t il, uint
auto * k = layers[ikv].k; auto * k = layers[ikv].k;
const auto ns = sinfo.s1 - sinfo.s0 + 1; assert(sinfo.s1 >= sinfo.s0);
const uint32_t ns = sinfo.s1 - sinfo.s0 + 1;
assert(ns > 0); assert(ns > 0);
assert(ns <= (int) n_seq_virt); assert(ns <= n_seq_virt);
const uint64_t size_virt = ggml_row_size(k->type, hparams.n_embd_k_gqa(il)*get_size()); const uint64_t size_virt = ggml_row_size(k->type, hparams.n_embd_k_gqa(il)*get_size());
@@ -928,7 +930,7 @@ ggml_tensor * llama_kv_cache_unified::get_v(ggml_context * ctx, int32_t il, uint
auto * v = layers[ikv].v; auto * v = layers[ikv].v;
const auto ns = sinfo.s1 - sinfo.s0 + 1; const uint32_t ns = sinfo.s1 - sinfo.s0 + 1;
assert(ns > 0); assert(ns > 0);
assert(ns <= n_seq_virt); assert(ns <= n_seq_virt);