From 971fe9f007aab94ac385373d011ef21f114243c2 Mon Sep 17 00:00:00 2001
From: YellowRoseCx <80486540+YellowRoseCx@users.noreply.github.com>
Date: Sat, 17 Jun 2023 06:54:29 -0500
Subject: [PATCH] add tokens per second output (#246)

* add tokens per second output

* Update gpttype_adapter.cpp

simplify

---------

Co-authored-by: LostRuins <39025047+LostRuins@users.noreply.github.com>
---
 gpttype_adapter.cpp | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/gpttype_adapter.cpp b/gpttype_adapter.cpp
index 775f29bb6..98cf3a2b9 100644
--- a/gpttype_adapter.cpp
+++ b/gpttype_adapter.cpp
@@ -1280,7 +1280,8 @@ generation_outputs gpttype_generate(const generation_inputs inputs, generation_o
     float pt1 = (time1*1000.0/(embd_inp.size()==0?1:embd_inp.size()));
     int realnpredict = params.n_predict-stopper_unused_tokens;
     float pt2 = (time2*1000.0/(realnpredict==0?1:realnpredict));
-    printf("\nTime Taken - Processing:%.1fs (%.0fms/T), Generation:%.1fs (%.0fms/T), Total:%.1fs", time1, pt1, time2, pt2, (time1 + time2));
+    float tokens_per_second = (realnpredict == 0 ? 0 : realnpredict / (time1 + time2));
+    printf("\nTime Taken - Processing:%.1fs (%.0fms/T), Generation:%.1fs (%.0fms/T), Total:%.1fs (%.1fT/s)", time1, pt1, time2, pt2, (time1 + time2), tokens_per_second);
     fflush(stdout);
     output.status = 1;
     generation_finished = true;