From 6e298d8fca1ee5f25239e54aa5f3eed2eee4651e Mon Sep 17 00:00:00 2001
From: Andrei Betlen <abetlen@gmail.com>
Date: Fri, 14 Apr 2023 22:21:19 -0400
Subject: [PATCH] Set kv cache size to f16 by default

---
 llama_cpp/llama.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/llama_cpp/llama.py b/llama_cpp/llama.py
index ecfd2f4..cd737c5 100644
--- a/llama_cpp/llama.py
+++ b/llama_cpp/llama.py
@@ -21,7 +21,7 @@ class Llama:
         n_ctx: int = 512,
         n_parts: int = -1,
         seed: int = 1337,
-        f16_kv: bool = False,
+        f16_kv: bool = True,
         logits_all: bool = False,
         vocab_only: bool = False,
         use_mmap: bool = True,