Update app.py

app.py

@@ -5,7 +5,6 @@ and provides retrieval-augmented chat through Hugging Face router.
 """
 
 import os
-import re
 import json
 import time
 import textwrap
@@ -31,7 +30,6 @@ EMBED_MODEL = "text-embedding-3-small"        # Embedding model
 HF_REPO = "rahul7star/OhamLab-LLM"            # Knowledge repo
 CACHE_PATH = "/tmp/ohamlab_emb_cache.json"    # Cache file
 
-# Client
 client = OpenAI(base_url="https://router.huggingface.co/v1", api_key=HF_TOKEN)
 api = HfApi(token=HF_TOKEN)
 
@@ -39,7 +37,7 @@ api = HfApi(token=HF_TOKEN)
 # 2. Load and Chunk Markdown Files
 # ---------------------------
 def load_ohamlab_knowledge():
-    """Loads all .md files from Hugging Face repo and splits into ~500-char chunks."""
+    """Load all .md files from Hugging Face repo and split into ~500-char chunks."""
     files = list_repo_files(HF_REPO, repo_type="model", token=HF_TOKEN)
     md_files = [f for f in files if f.endswith(".md")]
     chunks = []
@@ -64,7 +62,7 @@ def load_ohamlab_knowledge():
 # 3. Generate or Load Embeddings (with Cache)
 # ---------------------------
 def get_embeddings_with_cache():
-    """Generate or load cached embeddings for OhamLab context."""
+    """Generate or load cached embeddings for OhamLab knowledge."""
     if os.path.exists(CACHE_PATH):
         try:
             with open(CACHE_PATH, "r") as f:
@@ -103,7 +101,7 @@ OHAMLAB_TEXTS, OHAMLAB_EMBS = get_embeddings_with_cache()
 # 4. Semantic Retrieval
 # ---------------------------
 def retrieve_knowledge(query, top_k=3):
-    """Retrieve top-k most relevant text snippets."""
+    """Retrieve top-k most relevant text snippets from markdown knowledge bank."""
     try:
         q_emb = client.embeddings.create(model=EMBED_MODEL, input=[query]).data[0].embedding
         sims = np.dot(OHAMLAB_EMBS, q_emb) / (
@@ -120,13 +118,13 @@ def retrieve_knowledge(query, top_k=3):
 # ---------------------------
 def build_system_prompt(context: str, mode: str = "chat") -> str:
     return textwrap.dedent(f"""
-    You are OhamLab — AI Intelligence Software
+    You are OhamLab — AI Intelligence Software.
 
     Guidelines:
+    - Only answer using information retrieved from the OhamLab knowledge bank (.md files).
+    - Do not answer anything outside of this knowledge; if unknown, respond: "I could not find an answer in my knowledge base."
     - Always answer with clarity, scientific accuracy, and concise insight.
-    - Incorporate OhamLab research knowledge when relevant.
     - Avoid code unless explicitly requested.
-    - Be confident but label speculation clearly.
     - Mode: {mode.upper()}
 
     --- OhamLab Context (Retrieved Snippets) ---
@@ -147,45 +145,31 @@ def generate_response(user_input, history, mode="chat"):
         resp = client.chat.completions.create(
             model=MODEL_ID,
             messages=messages,
-            temperature=0.7,
+            temperature=0.0,  # deterministic for knowledge-only responses
             max_tokens=1200,
         )
         return resp.choices[0].message.content.strip()
     except Exception as e:
         print(f"⚠️ Model call failed: {e}")
-        return "⚠️ OahmLab encountered a temporary issue generating your response."
+        return "⚠️ OhamLab encountered a temporary issue generating your response."
 
 # ---------------------------
-# 7. Gradio Chat UI
+# 7. Chat Logic
 # ---------------------------
-import traceback
-import gradio as gr
-
-# ---------------------------
-# Chat Logic
-# ---------------------------
-
-
-
 def chat_with_model(user_message, chat_history):
-    """
-    Maintains full conversational context and returns updated chat history.
-    The assistant speaks as 'OhamLab'.
-    """
+    """Maintains full conversational context and returns updated chat history."""
     if not user_message:
         return chat_history, ""
 
     if chat_history is None:
         chat_history = []
 
-    # Convert Gradio message list (dict-based) to usable context
     history = [
         {"role": m["role"], "content": m["content"]}
         for m in chat_history
         if isinstance(m, dict) and "role" in m
     ]
 
-    # Append current user message
     history.append({"role": "user", "content": user_message})
 
     try:
@@ -194,99 +178,34 @@ def chat_with_model(user_message, chat_history):
         tb = traceback.format_exc()
         bot_reply = f"⚠️ OhamLab encountered an error:\n\n{e}\n\n{tb}"
 
-    # Add OhamLab's response as assistant role
     history.append({"role": "assistant", "content": bot_reply})
-
     return history, ""
 
-
 def reset_chat():
-    """Resets the chat session."""
     return []
 
-
 # ---------------------------
-# Gradio Chat UI
+# 8. Gradio Chat UI
 # ---------------------------
-
 def build_ui():
     with gr.Blocks(
         theme=gr.themes.Soft(primary_hue="indigo"),
         css="""
-            /* --- Hide share/delete icons --- */
-            #ohamlab .wrap.svelte-1lcyrj3 > div > div > button {
-                display: none !important;
-            }
             [data-testid="share-btn"],
             [data-testid="delete-btn"],
             .message-controls,
-            .message-actions {
-                display: none !important;
-                visibility: hidden !important;
-            }
-
-            /* --- User (Right) Message Bubble --- */
-            #ohamlab .message.user {
-                background-color: #4f46e5 !important;
-                color: white !important;
-                border-radius: 14px !important;
-                align-self: flex-end !important;
-                text-align: right !important;
-                margin-left: 25%;
-            }
-
-            /* --- OhamLab (Left) Message Bubble --- */
-            #ohamlab .message.assistant {
-                background-color: #f8f9fa !important;
-                color: #111 !important;
-                border-radius: 14px !important;
-                align-self: flex-start !important;
-                text-align: left !important;
-                margin-right: 25%;
-            }
-
-            
-            #ohamlab .chatbot .wrap.svelte-1lcyrj3 > div > div > button {
-                   display: none !important; /* hide share/delete icons */
-               }
-         
-            /* --- Overall Container --- */
-            .gradio-container {
-                max-width: 900px !important;
-                margin: auto;
-                padding-top: .5rem;
-            }
-            textarea {
-                resize: none !important;
-                border-radius: 12px !important;
-                border: 1px solid #d1d5db !important;
-                box-shadow: 0 1px 3px rgba(0,0,0,0.08);
-            }
-            button.primary {
-                background-color: #4f46e5 !important;
-                color: white !important;
-                border-radius: 10px !important;
-                padding: 0.6rem 1.4rem !important;
-                font-weight: 600;
-                transition: all 0.2s ease-in-out;
-            }
-            button.primary:hover {
-                background-color: #4338ca !important;
-            }
-            button.secondary {
-                background-color: #f3f4f6 !important;
-                border-radius: 10px !important;
-                color: #374151 !important;
-                font-weight: 500;
-                transition: all 0.2s ease-in-out;
-            }
-            button.secondary:hover {
-                background-color: #e5e7eb !important;
-            }
-        """,
+            .message-actions { display: none !important; visibility: hidden !important; }
+            #ohamlab .message.user { background-color: #4f46e5 !important; color: white !important; border-radius: 14px !important; align-self: flex-end !important; text-align: right !important; margin-left: 25%; }
+            #ohamlab .message.assistant { background-color: #f8f9fa !important; color: #111 !important; border-radius: 14px !important; align-self: flex-start !important; text-align: left !important; margin-right: 25%; }
+            .gradio-container { max-width: 900px !important; margin: auto; padding-top: .5rem; }
+            textarea { resize: none !important; border-radius: 12px !important; border: 1px solid #d1d5db !important; box-shadow: 0 1px 3px rgba(0,0,0,0.08); }
+            button.primary { background-color: #4f46e5 !important; color: white !important; border-radius: 10px !important; padding: 0.6rem 1.4rem !important; font-weight: 600; transition: all 0.2s ease-in-out; }
+            button.primary:hover { background-color: #4338ca !important; }
+            button.secondary { background-color: #f3f4f6 !important; border-radius: 10px !important; color: #374151 !important; font-weight: 500; transition: all 0.2s ease-in-out; }
+            button.secondary:hover { background-color: #e5e7eb !important; }
+        """
     ) as demo:
-        
-        # Chatbot area
+
         chatbot = gr.Chatbot(
             label="💠 OhamLab Conversation",
             height=520,
@@ -295,7 +214,6 @@ def build_ui():
             avatar_images=[None, None],
         )
 
-        # Input box (full width)
         with gr.Row():
             msg = gr.Textbox(
                 placeholder="Ask OhamLab anything ..",
@@ -305,19 +223,16 @@ def build_ui():
                 container=False,
             )
 
-        # Buttons (Send + Clear)
         with gr.Row(equal_height=True, variant="compact"):
-            send = gr.Button("Send", variant="primary", elem_classes=["primary"])
-            clear = gr.Button("Clear", variant="secondary", elem_classes=["secondary"])
+            send = gr.Button("Send", variant="primary")
+            clear = gr.Button("Clear", variant="secondary")
 
-        # Wiring
         send.click(chat_with_model, inputs=[msg, chatbot], outputs=[chatbot, msg])
         msg.submit(chat_with_model, inputs=[msg, chatbot], outputs=[chatbot, msg])
         clear.click(reset_chat, outputs=chatbot)
 
     return demo
 
-
 # ---------------------------
 # Entrypoint
 # ---------------------------
@@ -325,5 +240,3 @@ if __name__ == "__main__":
     print("🚀 Starting OhamLab Assistant...")
     demo = build_ui()
     demo.launch(server_name="0.0.0.0", server_port=7860)
-
-