diff --git a/modules/llamacpp_model_alternative.py b/modules/llamacpp_model_alternative.py
index 8fea2ab4..6bdf9bc3 100644
--- a/modules/llamacpp_model_alternative.py
+++ b/modules/llamacpp_model_alternative.py
@@ -6,7 +6,7 @@ Documentation:
 https://abetlen.github.io/llama-cpp-python/
 '''
 
-from llama_cpp import Llama
+from llama_cpp import Llama, LlamaCache
 
 from modules import shared
 from modules.callbacks import Iteratorize
@@ -27,6 +27,7 @@ class LlamaCppModel:
             'n_threads': shared.args.threads or None
         }
         self.model = Llama(**params)
+        self.model.set_cache(LlamaCache)
 
         # This is ugly, but the model and the tokenizer are the same object in this library.
         return result, result
diff --git a/requirements.txt b/requirements.txt
index 996afe28..34eee23d 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -14,5 +14,5 @@ tqdm
 git+https://github.com/huggingface/peft
 transformers==4.28.0
 bitsandbytes==0.38.1; platform_system != "Windows"
-llama-cpp-python==0.1.33; platform_system != "Windows"
-https://github.com/abetlen/llama-cpp-python/releases/download/v0.1.33/llama_cpp_python-0.1.33-cp310-cp310-win_amd64.whl; platform_system == "Windows"
+llama-cpp-python==0.1.34; platform_system != "Windows"
+https://github.com/abetlen/llama-cpp-python/releases/download/v0.1.34/llama_cpp_python-0.1.34-cp310-cp310-win_amd64.whl; platform_system == "Windows"