vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-12 09:59:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 3

Merge branch 'master' into concedo_experimental

Browse source 
# Conflicts:
#	Makefile
#	README.md
#	flake.lock
#	ggml-cuda.cu
#	llama.cpp
#	tests/test-backend-ops.cpp
#	tests/test-quantize-fns.cpp
This commit is contained in:
Concedo 2024-02-28 13:41:35 +08:00
commit ad638285de
26 changed files with 3393 additions and 589 deletions
Download patch file Download diff file Expand all files Collapse all files

2
otherarch/tools/unused/mainfn.txt
View file

@ -94,7 +94,7 @@ int mainfn() {
const int n_left = n_past - kcpp_params->n_keep - 1;
const int n_discard = n_left/2;
llama_kv_cache_seq_rm (llama_ctx_v4, 0, kcpp_params->n_keep + 1 , kcpp_params->n_keep + n_discard + 1);
llama_kv_cache_seq_shift(llama_ctx_v4, 0, kcpp_params->n_keep + 1 + n_discard, n_past, -n_discard);
llama_kv_cache_seq_add(llama_ctx_v4, 0, kcpp_params->n_keep + 1 + n_discard, n_past, -n_discard);
n_past -= n_discard;
}