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

kv_cache management

Browse source
This commit is contained in:
Leon Ericsson 2023-12-16 12:12:33 +01:00
parent 1665ad8bf1
commit 21431197a1
1 changed files with 4 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

4
examples/lookup/lookup.cpp
View file

@ -130,6 +130,10 @@ int main(int argc, char ** argv){
break; break;
} }
// KV cache management
// clean the cache of draft tokens that weren't accepted
llama_kv_cache_seq_rm(ctx, 0, n_past, -1);
llama_batch_clear(batch_tgt); llama_batch_clear(batch_tgt);
llama_batch_add(batch_tgt, draft[0], n_past, { 0 }, true); llama_batch_add(batch_tgt, draft[0], n_past, { 0 }, true);