evahteev
diff --git a/‎llama_cpp/llama.py
Copy file name to clipboardExpand all lines: llama_cpp/llama.py
+1Lines changed: 1 addition & 0 deletions b/‎llama_cpp/llama.py
Copy file name to clipboardExpand all lines: llama_cpp/llama.py
+1Lines changed: 1 addition & 0 deletions
@@ -280,6 +280,7 @@ def generate(
             if self.verbose:
                 print("generate cache hit", file=sys.stderr)
             reset = False
+            tokens = tokens[len(self.tokens) :]
         ###
         if reset:
             self.reset()