CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-04 09:32:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : update llama_kv_self API

Browse Source 
ggml-ci
This commit is contained in:
Georgi Gerganov
2025-01-14 16:47:34 +02:00
parent fd05ab87aa
commit 17b363afd3
30 changed files with 387 additions and 205 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
examples/save-load-state/save-load-state.cpp
View File

@@ -156,8 +156,6 @@ int main(int argc, char ** argv) {
// make new context
llama_context * ctx3 = llama_init_from_model(model, common_context_params_to_llama(params));
llama_kv_cache * kv3 = llama_get_kv_cache(ctx3);
llama_sampler * smpl3 = llama_sampler_chain_init(sparams);
llama_sampler_chain_add(smpl3, llama_sampler_init_dist(params.sampling.seed));
@@ -198,7 +196,7 @@ int main(int argc, char ** argv) {
fprintf(stderr, "%s : seq 0 copied, %zd bytes\n", __func__, ncopy);
// erase whole kv
llama_kv_cache_clear(kv3);
llama_kv_self_clear(ctx3);
fprintf(stderr, "%s : kv cache cleared\n", __func__);
// restore kv into seq 1