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

Fixes "Not enough space in the context's memory pool" encountered on certain models, which seems to be caused by some imprecision related to the automatic casting of floating point values

Browse source 
(cherry picked from commit 1ad8f0d80e)
This commit is contained in:
Concedo 2023-12-14 16:00:44 +08:00
parent 0e31f53422
commit 34b3dac66d
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
llama.cpp
View file

@ -1565,7 +1565,7 @@ static bool llama_kv_cache_init(
cache.cells.clear();
cache.cells.resize(n_ctx);
cache.buf.resize(n_elements*(ggml_type_sizef(ktype) + ggml_type_sizef(vtype)) + 2u*n_layer*ggml_tensor_overhead());
cache.buf.resize(n_elements*((size_t)(ggml_type_sizef(ktype) + ggml_type_sizef(vtype))) + 2u*n_layer*ggml_tensor_overhead());
memset(cache.buf.data, 0, cache.buf.size);
struct ggml_init_params params;