summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorGeorgi Gerganov <ggerganov@gmail.com>2023-09-03 13:42:56 +0300
committerGeorgi Gerganov <ggerganov@gmail.com>2023-09-03 13:43:17 +0300
commit8f429fa5111901f9646cf998643ac5310846d487 (patch)
treedb1ceb2a6696a70eff42487866d1986dc0c6902d
parent6519e9c99cffbad19b31bcba86df48c500628c09 (diff)
perplexity : fix ETA by warming up the model with an empty run
-rw-r--r--common/common.cpp8
-rw-r--r--examples/main/main.cpp8
2 files changed, 8 insertions, 8 deletions
diff --git a/common/common.cpp b/common/common.cpp
index 41fc59ce..a1c3dc78 100644
--- a/common/common.cpp
+++ b/common/common.cpp
@@ -752,6 +752,14 @@ std::tuple<struct llama_model *, struct llama_context *> llama_init_from_gpt_par
params.logit_bias[llama_token_eos(lctx)] = -INFINITY;
}
+ {
+ LOG("warming up the model with an empty run\n");
+
+ const std::vector<llama_token> tmp = { llama_token_bos(lctx), };
+ llama_eval(lctx, tmp.data(), tmp.size(), 0, params.n_threads);
+ llama_reset_timings(lctx);
+ }
+
return std::make_tuple(model, lctx);
}
diff --git a/examples/main/main.cpp b/examples/main/main.cpp
index 7117db4b..db98312c 100644
--- a/examples/main/main.cpp
+++ b/examples/main/main.cpp
@@ -492,14 +492,6 @@ int main(int argc, char ** argv) {
std::vector<llama_token> embd;
std::vector<llama_token> embd_guidance;
- {
- LOG("warming up the model with an empty run\n");
-
- const std::vector<llama_token> tmp = { llama_token_bos(ctx), };
- llama_eval(ctx, tmp.data(), tmp.size(), 0, params.n_threads);
- llama_reset_timings(ctx);
- }
-
while ((n_remain != 0 && !is_antiprompt) || params.interactive) {
// predict
if (embd.size() > 0) {