From cdeaded2516509b10e9494c2884dcef8b06a4fe2 Mon Sep 17 00:00:00 2001
From: Andrei Betlen <abetlen@gmail.com>
Date: Wed, 10 May 2023 16:12:17 -0400
Subject: [PATCH] Bugfix: Ensure logs are printed when streaming

---
 llama_cpp/llama.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/llama_cpp/llama.py b/llama_cpp/llama.py
index 7be51e1..bbce4b2 100644
--- a/llama_cpp/llama.py
+++ b/llama_cpp/llama.py
@@ -709,6 +709,9 @@ class Llama:
                 print("Llama._create_completion: cache save", file=sys.stderr)
             self.cache[prompt_tokens + completion_tokens] = self.save_state()
 
+        if self.verbose:
+            llama_cpp.llama_print_timings(self.ctx)
+
         if stream:
             yield {
                 "id": completion_id,
@@ -780,9 +783,6 @@ class Llama:
                 "top_logprobs": top_logprobs,
             }
 
-        if self.verbose:
-            llama_cpp.llama_print_timings(self.ctx)
-
         yield {
             "id": completion_id,
             "object": "text_completion",