From f0de4953ae3c4ce09fe2631d29474dfce0db7256 Mon Sep 17 00:00:00 2001
From: Concedo <39025047+LostRuins@users.noreply.github.com>
Date: Thu, 14 Dec 2023 16:58:41 +0800
Subject: [PATCH] fixed length exceeding max ctx

---
 koboldcpp.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/koboldcpp.py b/koboldcpp.py
index e8d7e029a..af6a43a4f 100755
--- a/koboldcpp.py
+++ b/koboldcpp.py
@@ -310,13 +310,13 @@ def generate(prompt, memory="", max_length=32, max_context_length=512, temperatu
     inputs.memory = memory.encode("UTF-8")
     if max_length >= max_context_length:
         max_length = max_context_length-1
-    inputs.max_context_length = max_context_length   # this will resize the context buffer if changed
     global showmaxctxwarning
     if max_context_length > maxctx:
         if showmaxctxwarning:
             print(f"\n(Warning! Request max_context_length={max_context_length} exceeds allocated context size of {maxctx}. It will be reduced to fit. Consider launching with increased --contextsize to avoid errors. This message will only show once per session.)")
             showmaxctxwarning = False
         max_context_length = maxctx
+    inputs.max_context_length = max_context_length   # this will resize the context buffer if changed
     inputs.max_length = max_length
     inputs.temperature = temperature
     inputs.top_k = top_k