update

app.py

@@ -20,8 +20,8 @@ parser.add_argument("-t", "--threads", default=0, type=int, help="number of thre
 parser.add_argument("--plain", action="store_true", help="display in plain text without markdown support")
 args = parser.parse_args()
 
-from pyllamacpp.model import Model
-model = Model(model_path=model_file)
+from llama_cpp import Llama
+llm = Llama(model_path=model_file)
 
 
 
@@ -41,7 +41,10 @@ def predict(input, system_prompt, chatbot, max_length, ctx_length, top_p, temper
         num_threads=0,
         stream=True,
     )
-    for response_piece in model.generate(input):
+    output = llm(input)
+    response = output['choices'][0]['text']
+
+    for response_piece in response:
         response += response_piece
         chatbot[-1] = (chatbot[-1][0], response)
 

requirements.txt

@@ -1,2 +1,2 @@
 gradio
-git+https://github.com/zhangheli/pyllamacpp.git@main
+llama_cpp_python==0.2.14