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

common : add -dkvc arg for enabling kv cache dumps

Browse source
This commit is contained in:
Georgi Gerganov 2023-11-23 18:47:56 +02:00
parent 5df7d06c42
commit f8e9f11428
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
4 changed files with 14 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

8
examples/parallel/parallel.cpp
View file

@ -113,6 +113,8 @@ int main(int argc, char ** argv) {
// insert new requests as soon as the previous one is done
const bool cont_batching = params.cont_batching;
const bool dump_kv_cache = params.dump_kv_cache;
#ifndef LOG_DISABLE_LOGS
log_set_target(log_filename_generator("parallel", "log"));
LOG_TEE("Log start\n");
@ -203,8 +205,10 @@ int main(int argc, char ** argv) {
LOG_TEE("Processing requests ...\n\n");
while (true) {
llama_kv_cache_view_update(ctx, &kvc_view);
dump_kv_cache_view_seqs(kvc_view, 40);
if (dump_kv_cache) {
llama_kv_cache_view_update(ctx, &kvc_view);
dump_kv_cache_view_seqs(kvc_view, 40);
}
llama_batch_clear(batch);