Update app.py

app.py

@@ -1,65 +1,85 @@
 import os
+import requests
 import gradio as gr
 from huggingface_hub import InferenceClient
 
-"""
-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
-"""
-hf_token = os.getenv("apikey")
-client = InferenceClient(token=hf_token, model="HuggingFaceH4/zephyr-7b-beta")
+# API keys from environment
+hf_token = os.getenv("apikey")  # Hugging Face token
+tavily_token = os.getenv("tavily")  # Tavily key
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+# 30B model (chat-tuned)
+client = InferenceClient(token=hf_token, model="NousResearch/Nous-Hermes-2-Yi-34B")
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+# Web search fallback
+def search_tavily(query):
+    try:
+        res = requests.post("https://api.tavily.com/search", json={
+            "api_key": tavily_token,
+            "query": query,
+            "include_answer": True,
+            "search_depth": "basic"
+        })
+        return res.json().get("answer", "")
+    except:
+        return ""
 
-    messages.append({"role": "user", "content": message})
+# Main chat logic
+def respond(message, history: list[tuple[str, str]], max_tokens, temperature, top_p):
+    messages = []
+
+    for user, assistant in history:
+        if user:
+            messages.append({"role": "user", "content": user})
+        if assistant:
+            messages.append({"role": "assistant", "content": assistant})
 
+    messages.append({"role": "user", "content": message})
     response = ""
+    fallback_needed = False
 
-    for message in client.chat_completion(
+    # Attempt answer
+    for msg in client.chat_completion(
         messages,
         max_tokens=max_tokens,
         stream=True,
         temperature=temperature,
         top_p=top_p,
     ):
-        token = message.choices[0].delta.content
-
+        token = msg.choices[0].delta.content
         response += token
         yield response
 
+    # Check if it's unsure
+    triggers = ["i don't know", "no information", "i cannot", "as an ai", "unsure"]
+    if any(trigger in response.lower() for trigger in triggers):
+        fallback_needed = True
+
+    # Retry with web search if needed
+    if fallback_needed:
+        web_snippet = search_tavily(message)
+        if web_snippet:
+            prompt = f"User asked: {message}\nHere’s info from the web: {web_snippet}\nAnswer:"
+            response = ""
+            for msg in client.chat_completion(
+                [{"role": "user", "content": prompt}],
+                max_tokens=max_tokens,
+                stream=True,
+                temperature=temperature,
+                top_p=top_p,
+            ):
+                token = msg.choices[0].delta.content
+                response += token
+                yield response
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
+# Gradio interface without system message
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
         gr.Slider(minimum=1, maximum=2048, 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)",
-        ),
+        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p"),
     ],
 )
 
-
 if __name__ == "__main__":
-    demo.launch(share=True)
+    demo.launch()