Update app.py

app.py

@@ -1,17 +1,9 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
-
 import requests
 from bs4 import BeautifulSoup
 from bs4.element import Comment
 
-def get_text_from_url(url):
-    response = requests.get(url)
-    soup = BeautifulSoup(response.text, 'html.parser')
-    texts = soup.find_all(text=True)
-    visible_texts = filter(tag_visible, texts)
-    return u"\n".join(t.strip() for t in visible_texts)
-
 def tag_visible(element):
     if element.parent.name in ['style', 'script', 'head', 'title', 'meta', '[document]']:
         return False
@@ -19,60 +11,56 @@ def tag_visible(element):
         return False
     return True
 
+def get_text_from_url(url):
+    response = requests.get(url)
+    soup = BeautifulSoup(response.text, 'html.parser')
+    texts = soup.find_all(text=True)
+    visible_texts = filter(tag_visible, texts)
+    # Only join non-empty lines
+    return "\n".join(t.strip() for t in visible_texts if t.strip())
+
+# Collect homepage text from various URL extensions
 text_list = []
 homepage_url = "https://sites.google.com/view/abhilashnandy/home/"
 extensions = ["", "pmrf-profile-page"]
 for ext in extensions:
-    url_text = get_text_from_url(homepage_url+ext)
+    url_text = get_text_from_url(homepage_url + ext)
     text_list.append(url_text)
-# Repeat for sub-links if necessary
-
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("stabilityai/stablelm-2-1_6b-chat")#("stabilityai/stablelm-2-1_6b-chat")#("TheBloke/TinyLlama-1.1B-Chat-v1.0-GPTQ")#("TheBloke/TinyLlama-1.1B-Chat-v1.0-GGUF")#("QuantFactory/Meta-Llama-3-8B-Instruct-GGUF")#("HuggingFaceH4/zephyr-7b-beta")
-
-SYSTEM_MESSAGE = "You are a QA chatbot to answer queries (in less than 30 words) on my homepage that has the following information -\n\n" + "\n\n".join(text_list) + "\n\n"
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message=SYSTEM_MESSAGE,
-    max_tokens=140,
-    temperature=0.7,
-    top_p=0.95,
-):
-    messages = [{"role": "system", "content": system_message}]
+# Build the system message containing homepage info
+SYSTEM_MESSAGE = (
+    "You are a QA chatbot to answer queries (in less than 30 words) on my homepage that has the following information -\n\n"
+    + "\n\n".join(text_list)
+    + "\n\n"
+)
 
-    for val in history:
-        if len(val) >= 1:
-            messages.append({"role": "user", "content": "Question: " + val[0]})
-        if len(val) >= 2:
-            messages.append({"role": "assistant", "content": "Answer: " + val[1]})
+# Use a lightweight model for low-latency CPU inference
+client = InferenceClient("microsoft/DialoGPT-small")
 
-    messages.append({"role": "user", "content": message})
+def respond(message, history: list[tuple[str, str]], system_message=SYSTEM_MESSAGE,
+            max_tokens=140, temperature=0.7, top_p=0.95):
+    # Construct the prompt including the system message and conversation history.
+    prompt = system_message
+    for user_q, bot_a in history:
+        prompt += f"Question: {user_q}\n"
+        prompt += f"Answer: {bot_a}\n"
+    prompt += f"Question: {message}\nAnswer:"
 
     try:
-        response = client.chat_completion(
-            messages,
-            max_tokens=max_tokens,
+        response = client.text_generation(
+            prompt,
+            max_new_tokens=max_tokens,
             temperature=temperature,
             top_p=top_p,
-            # stream=True,  # Disable streaming for debugging
         )
-        return response.choices[0].message["content"]
-
+        # Expecting a list of dict(s) with key "generated_text"
+        generated_text = response[0]["generated_text"]
+        # Attempt to extract the answer by splitting at "Answer:"
+        answer = generated_text.split("Answer:")[-1].strip().split("\n")[0].strip()
+        return answer
     except Exception as e:
         print(f"An error occurred: {e}")
-        return str(e) #"An error occurred while processing the response."
-
-
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-
-# initial_message = [("user", "Yo who dis Abhilash?")]
+        return str(e)
 
 markdown_note = "## Ask Anything About Me! (Might show a tad bit of hallucination!)"
 
@@ -81,24 +69,10 @@ demo = gr.Blocks()
 with demo:
     gr.Markdown(markdown_note)
     gr.ChatInterface(
-    respond,
-    examples = ["Yo who dis Abhilash?", "What is Abhilash's most recent publication?"],
-    # message=initial_message,
-    additional_inputs=[
-        # gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        # gr.Slider(minimum=1, maximum=8192, value=512, step=1, label="Max new tokens"),
-        # gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        # gr.Slider(
-        #     minimum=0.1,
-        #     maximum=1.0,
-        #     value=0.95,
-        #     step=0.05,
-        #     label="Top-p (nucleus sampling)",
-        # ),
-    ],
-    # value=initial_message
-)
-
+        respond,
+        examples=["Yo who dis Abhilash?", "What is Abhilash's most recent publication?"],
+        additional_inputs=[],
+    )
 
 if __name__ == "__main__":
     demo.launch()