Update app.py

app.py

@@ -18,13 +18,13 @@ def get_text_from_url(url):
     visible_texts = filter(tag_visible, texts)
     return "\n".join(t.strip() for t in visible_texts)
 
-# Pre-fetch and truncate homepage text to keep the prompt concise
+# Pre-fetch and truncate homepage text to reduce the prompt length.
 text_list = []
 homepage_url = "https://sites.google.com/view/abhilashnandy/home/"
 extensions = ["", "pmrf-profile-page"]
 for ext in extensions:
     full_text = get_text_from_url(homepage_url + ext)
-    truncated_text = full_text[:1000]  # use only the first 1000 characters per page
+    truncated_text = full_text[:1000]  # using only the first 1000 characters per extension
     text_list.append(truncated_text)
 
 SYSTEM_MESSAGE = (
@@ -32,8 +32,8 @@ SYSTEM_MESSAGE = (
     "Context: " + " ".join(text_list)
 )
 
-# Switch to a model that is known to have a valid config and is optimized for CPU inference.
-client = InferenceClient("nomic-ai/gpt4all-j")
+# Use a model that is both lightweight and includes a proper chat configuration.
+client = InferenceClient("TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ")
 
 def respond(message, history: list[tuple[str, str]], system_message=SYSTEM_MESSAGE,
             max_tokens=100, temperature=0.7, top_p=0.95):
@@ -43,7 +43,7 @@ def respond(message, history: list[tuple[str, str]], system_message=SYSTEM_MESSA
         messages.append({"role": "assistant", "content": "Answer: " + a})
     messages.append({"role": "user", "content": message})
     try:
-        # Use streaming mode to start receiving output as it is generated.
+        # Enable streaming mode to receive output faster.
         response_stream = client.chat_completion(
             messages,
             max_tokens=max_tokens,