feat: implement DeepSeek Janus chat interface

- Add streaming response generation
- Implement chat UI with real-time updates
- Add multi-page structure
- Update dependencies
- Improve error handling

.gitignore

@@ -1,5 +1,7 @@
 __pycache__/
 *.py[cod]
+.env*
+.venv
 *.so
 .Python
 env/
@@ -7,5 +9,5 @@ build/
 dist/
 eggs/
 *.egg-info/
-.env
+.streamlit/secrets.toml
 generated_image.png

app.py

@@ -24,11 +24,23 @@ st.markdown("""
 """, unsafe_allow_html=True)
 
 st.title("🧠 DeepSeek AI Assistant")
+
+if "model_loaded" not in st.session_state:
+    st.session_state.model_loaded = False
+
 st.markdown("""
-Welcome! Choose a feature from the sidebar to get started.
+👈 Select 'Chat' from the sidebar to start chatting!
+
+### Features:
+- Real-time response generation
+- Context-aware conversations
+- Professional responses
+- Memory efficient
 
-- 💭 Chat Interface: Have a conversation with the AI
-- More features coming soon...
+### Tips:
+- Be specific in your questions
+- Use clear language
+- Start with simple queries
 """)
 
 @st.cache_resource

pages/01_chat.py

@@ -3,77 +3,91 @@ from utils import load_model, generate_stream
 import time
 
 def init_chat():
-    if "messages" not in st.session_state:
-        st.session_state.messages = []
-        st.session_state.model, st.session_state.tokenizer = load_model()
-        st.session_state.generating = False
+    try:
+        # Initialize session state
+        if "messages" not in st.session_state:
+            st.session_state.messages = []
+        if "generating" not in st.session_state:
+            st.session_state.generating = False
+            
+        # Load model if needed
+        if "model" not in st.session_state or "tokenizer" not in st.session_state:
+            with st.spinner("📚 Initializing AI model..."):
+                st.session_state.model, st.session_state.tokenizer = load_model()
+                
+    except Exception as e:
+        st.error(f"Initialization error: {str(e)}")
+        st.button("🔄 Retry Loading")
+        st.stop()
 
-# Chat interface with error handling
-try:
-    st.title("💭 Chat Interface")
-    init_chat()
-    
-    # Sidebar controls with session management
-    with st.sidebar:
-        st.markdown("### Chat Controls")
-        cols = st.columns(2)
-        with cols[0]:
-            if st.button("🗑️ Clear Chat", use_container_width=True):
-                st.session_state.messages = []
-                st.rerun()
-        with cols[1]:
-            if st.button("🔄 Reset Model", use_container_width=True):
-                st.cache_resource.clear()
-                st.rerun()
+st.title("💭 Chat Interface")
+
+# Initialize
+init_chat()
+
+# Verify model loaded
+if not st.session_state.get("model_loaded", False):
+    st.warning("⚠️ Model not fully loaded. Please wait...")
+    st.stop()
+
+# Sidebar controls
+with st.sidebar:
+    st.markdown("### Chat Controls")
+    cols = st.columns(2)
+    with cols[0]:
+        if st.button("🗑️ Clear Chat", use_container_width=True):
+            st.session_state.messages = []
+            st.session_state.generating = False
+            st.rerun()
+    with cols[1]:
+        if st.button("🔄 Reset Model", use_container_width=True):
+            st.session_state.clear()
+            st.cache_resource.clear()
+            st.rerun()
+
+# Chat history
+chat_container = st.container()
+with chat_container:
+    for msg in st.session_state.messages:
+        with st.chat_message(msg["role"]):
+            st.markdown(msg["content"])
+
+# Input handling
+if prompt := st.chat_input(
+    "Ask me anything...",
+    disabled=st.session_state.generating
+):
+    # Update generating state
+    st.session_state.generating = True
     
-    # Chat history with proper formatting
-    chat_container = st.container()
-    with chat_container:
-        for msg in st.session_state.messages:
-            with st.chat_message(msg["role"]):
-                st.markdown(msg["content"])
+    # Show user message
+    st.session_state.messages.append({"role": "user", "content": prompt})
+    with st.chat_message("user"):
+        st.markdown(prompt)
     
-    # Input handling with safeguards
-    if prompt := st.chat_input(
-        "Ask me anything...",
-        disabled=st.session_state.get("generating", False)
-    ):
-        if not st.session_state.generating:
-            st.session_state.generating = True
+    # Generate and show response
+    with st.chat_message("assistant"):
+        try:
+            context = "\n".join([
+                f"{m['role']}: {m['content']}"
+                for m in st.session_state.messages[-3:]
+            ])
             
-            # Show user message
-            st.session_state.messages.append({"role": "user", "content": prompt})
-            with st.chat_message("user"):
-                st.markdown(prompt)
+            response = generate_stream(
+                context,
+                st.session_state.model,
+                st.session_state.tokenizer
+            )
             
-            # Generate and show response
-            with st.chat_message("assistant"):
-                try:
-                    # Get recent context
-                    context = "\n".join([
-                        f"{m['role']}: {m['content']}"
-                        for m in st.session_state.messages[-3:]
-                    ])
-                    
-                    response = generate_stream(
-                        context,
-                        st.session_state.model,
-                        st.session_state.tokenizer
-                    )
-                    
-                    if response:
-                        st.session_state.messages.append({
-                            "role": "assistant",
-                            "content": response
-                        })
-                    
-                except Exception as e:
-                    st.error("Failed to generate response. Please try again.")
-                    st.error(f"Error details: {str(e)}")
-                
-                finally:
-                    st.session_state.generating = False
-
-except Exception as e:
-    st.error(f"Application error: {str(e)}")
-    st.button("🔄 Restart App")
+            if response:
+                st.session_state.messages.append({
+                    "role": "assistant",
+                    "content": response
+                })
+            
+        except Exception as e:
+            st.error("Failed to generate response. Please try again.")
+            st.error(f"Error details: {str(e)}")
+        
+        finally:
+            st.session_state.generating = False

utils.py

@@ -9,30 +9,37 @@ logging.basicConfig(level=logging.INFO)
 
 @st.cache_resource
 def load_model():
+    if "model_loaded" not in st.session_state:
+        st.session_state.model_loaded = False
+        
     model_name = "deepseek-ai/Janus-Pro-7B"
     
     try:
-        tokenizer = AutoTokenizer.from_pretrained(
-            model_name,
-            trust_remote_code=True,
-            padding_side='left'
-        )
-        tokenizer.pad_token = tokenizer.eos_token
-        
-        model = AutoModelForCausalLM.from_pretrained(
-            model_name,
-            torch_dtype=torch.float32,
-            low_cpu_mem_usage=True,
-            trust_remote_code=True,
-            device_map='cpu'
-        )
-        
-        model.eval()
-        torch.set_num_threads(8)
-        return model, tokenizer
-        
+        with st.spinner("🔄 Loading model (first run only)..."):
+            tokenizer = AutoTokenizer.from_pretrained(
+                model_name,
+                trust_remote_code=True,
+                padding_side='left'
+            )
+            tokenizer.pad_token = tokenizer.eos_token
+            
+            model = AutoModelForCausalLM.from_pretrained(
+                model_name,
+                torch_dtype=torch.float32,
+                low_cpu_mem_usage=True,
+                trust_remote_code=True,
+                device_map='cpu'
+            )
+            
+            model.eval()
+            torch.set_num_threads(8)
+            st.session_state.model_loaded = True
+            
+            return model, tokenizer
+            
     except Exception as e:
-        st.error(f"Error loading model: {str(e)}")
+        st.error(f"❌ Error loading model: {str(e)}")
+        st.info("Try refreshing the page or clearing the cache.")
         st.stop()
 
 def stream_tokens(response: str, delay: float = 0.01) -> Generator[str, None, None]:
@@ -49,6 +56,10 @@ def stream_tokens(response: str, delay: float = 0.01) -> Generator[str, None, No
 
 def generate_stream(prompt: str, model: AutoModelForCausalLM, tokenizer: AutoTokenizer) -> Optional[str]:
     try:
+        # Safety checks
+        if not model or not tokenizer:
+            raise ValueError("Model or tokenizer not initialized")
+            
         # Format prompt with safety checks
         safe_prompt = prompt.strip().replace("<", "&lt;").replace(">", "&gt;")
         chat_prompt = f"""### Human: {safe_prompt}
@@ -124,7 +135,6 @@ def generate_stream(prompt: str, model: AutoModelForCausalLM, tokenizer: AutoTok
             return response
             
     except Exception as e:
-        logger = logging.getLogger(__name__)
-        logger.error(f"Generation error: {str(e)}")
-        st.error("Something went wrong. Please try again with a different prompt.")
-        return None
+        st.error(f"Generation error: {str(e)}")
+        return "I apologize, but I couldn't generate a response. Please try again."
+```