vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-15 11:29:43 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fixed inccorect padding for flash attn with swa

Browse source
This commit is contained in:
Concedo 2025-06-30 20:32:14 +08:00
parent 186227fc26
commit 989f9e6b98
1 changed files with 1 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

1
src/llama-kv-cache-unified-iswa.cpp
View file

@ -31,6 +31,7 @@ llama_kv_cache_unified_iswa::llama_kv_cache_unified_iswa(
//kcpp: pad the swa kv cache as well, similar to extra_context_handle_fragmentation
size_swa += 32;
size_swa = GGML_PAD(size_swa, n_pad);
// when using full-size SWA cache, we set the SWA cache size to be equal to the base cache size
if (swa_full) {