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

llama-bench : use two tokens in the warmup run for prompt evals (#3059)

Browse source
This commit is contained in:
slaren 2023-09-07 15:52:34 +02:00 • committed by GitHub
parent be8c9c245b
commit 15b67a66c2
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 6 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

7
examples/llama-bench/llama-bench.cpp
View file

@ -986,7 +986,12 @@ int main(int argc, char ** argv) {
test t(inst, lmodel, ctx); test t(inst, lmodel, ctx);
// warmup run // warmup run
test_gen(ctx, 1, 0, t.n_threads); if (t.n_prompt > 0) {
test_prompt(ctx, std::min(2, t.n_batch), 0, t.n_batch, t.n_threads);
}
if (t.n_gen > 0) {
test_gen(ctx, 1, 0, t.n_threads);
}
for (int i = 0; i < params.reps; i++) { for (int i = 0; i < params.reps; i++) {
uint64_t t_start = get_time_ns(); uint64_t t_start = get_time_ns();