summaryrefslogtreecommitdiff
path: root/examples/parallel/parallel.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'examples/parallel/parallel.cpp')
-rw-r--r--examples/parallel/parallel.cpp4
1 files changed, 2 insertions, 2 deletions
diff --git a/examples/parallel/parallel.cpp b/examples/parallel/parallel.cpp
index 69f9526a..eb64adef 100644
--- a/examples/parallel/parallel.cpp
+++ b/examples/parallel/parallel.cpp
@@ -157,7 +157,7 @@ int main(int argc, char ** argv) {
for (size_t i = 0; i < clients.size(); ++i) {
auto & client = clients[i];
client.id = i;
- client.ctx_sampling = llama_sampling_init(params);
+ client.ctx_sampling = llama_sampling_init(params.sparams);
}
std::vector<llama_token> tokens_system;
@@ -330,7 +330,7 @@ int main(int argc, char ** argv) {
const llama_token id = llama_sampling_sample(client.ctx_sampling, ctx, NULL, client.i_batch - i);
- llama_sampling_accept(client.ctx_sampling, ctx, id);
+ llama_sampling_accept(client.ctx_sampling, ctx, id, true);
if (client.n_decoded == 1) {
// start measuring generation time after the first token to make sure all concurrent clients