server : fix division by zero when reporting stats (#16501)

diff --git a/tools/server/server.cpp b/tools/server/server.cpp
index 5293a98f034f1c5801e7b60d3e96efd1333a86c5..60326e8e50efe9dc6360c9499bdaf356a1e3e8b4 100644 (file)
--- a/tools/server/server.cpp
+++ b/tools/server/server.cpp
@@ -4226,7 +4226,7 @@ struct server_context {
                     metrics.on_prompt_eval(slot);
                 }
 
-                slot.t_token_generation = (t_current - slot.t_start_generation) / 1e3;
+                slot.t_token_generation = std::max<int64_t>(1, t_current - slot.t_start_generation) / 1e3;
 
                 completion_token_output result;
                 result.tok          = id;