tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-17 13:40:55 -04:00
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

6
examples/infill/infill.cpp
View File

@@ -139,8 +139,6 @@ int main(int argc, char ** argv) {
return 1;
}
llama_kv_cache * kv = llama_get_kv_cache(ctx);
const llama_vocab * vocab = llama_model_get_vocab(model);
const int n_ctx_train = llama_model_n_ctx_train(model);
@@ -334,8 +332,8 @@ int main(int argc, char ** argv) {
LOG_DBG("context full, swapping: n_past = %d, n_left = %d, n_ctx = %d, n_keep = %d, n_discard = %d\n",
n_past, n_left, n_ctx, params.n_keep, n_discard);
llama_kv_cache_seq_rm (kv, 0, params.n_keep + 1 , params.n_keep + n_discard + 1);
llama_kv_cache_seq_add(kv, 0, params.n_keep + 1 + n_discard, n_past, -n_discard);
llama_kv_self_seq_rm (ctx, 0, params.n_keep + 1 , params.n_keep + n_discard + 1);
llama_kv_self_seq_add(ctx, 0, params.n_keep + 1 + n_discard, n_past, -n_discard);
n_past -= n_discard;