Update app.py

app.py

@@ -1,10 +1,9 @@
-
-import gradio as gr
+import streamlit as st
 from llama_cpp import Llama
 import os
 import requests
 
-MODEL_PATH = "irixium-12b-model_stock-q4_k_m.gguf"
+MODEL_PATH = "model.gguf"
 DEFAULT_SYSTEM_PROMPT = "You are a helpful assistant."
 
 def download_model(url, save_path):
@@ -60,11 +59,13 @@ def inference(message, history, model_url, system_prompt):
     if model_url and not os.path.exists(MODEL_PATH):
         download_result = download_model(model_url, MODEL_PATH)
         if "Error" in download_result:
-            return history + [{"role": "assistant", "content": download_result}], history
+            st.session_state.messages.append({"role": "assistant", "content": download_result})
+            return
 
     llm = load_model(MODEL_PATH)
     if isinstance(llm, str):
-        return history + [{"role": "assistant", "content": llm}], history
+        st.session_state.messages.append({"role": "assistant", "content": llm})
+        return
 
     messages = [{"role": "system", "content": system_prompt}]
     for item in history:
@@ -73,18 +74,25 @@ def inference(message, history, model_url, system_prompt):
 
     prompt = apply_chat_template(llm.model_path, messages, system_prompt)
     response = generate_response(prompt, llm)
-    history.append({"role": "assistant", "content": response})
-    return history, history
+    st.session_state.messages.append({"role": "assistant", "content": response})
+
+st.title("llama.cpp Chat")
+
+if "messages" not in st.session_state:
+    st.session_state.messages = []
 
-with gr.Blocks() as iface:
-    model_url_input = gr.Textbox(label="Model URL (GGUF)", placeholder="Enter GGUF model URL...")
-    system_prompt_input = gr.Textbox(label="System Prompt", value=DEFAULT_SYSTEM_PROMPT, lines=3)
-    chatbot = gr.Chatbot(type="messages")
-    message = gr.Textbox(label="Message")
-    send_button = gr.Button("Send")
-    state = gr.State([]) #This line is very important.
+model_url = st.text_input("Model URL (GGUF)", placeholder="Enter GGUF model URL...")
+system_prompt = st.text_area("System Prompt", value=DEFAULT_SYSTEM_PROMPT)
 
-    send_button.click(inference, inputs=[message, state, model_url_input, system_prompt_input], outputs=[chatbot, state])
-    message.submit(inference, inputs=[message, state, model_url_input, system_prompt_input], outputs=[chatbot, state])
+for message in st.session_state.messages:
+    with st.chat_message(message["role"]):
+        st.markdown(message["content"])
 
-iface.launch()
+if prompt := st.chat_input("Message"):
+    st.session_state.messages.append({"role": "user", "content": prompt})
+    with st.chat_message("user"):
+        st.markdown(prompt)
+    history = [m for m in st.session_state.messages if m["role"] != "system"]
+    inference(prompt, history, model_url, system_prompt)
+    with st.chat_message("assistant"):
+        st.markdown(st.session_state.messages[-1]["content"])