Update app.py

app.py

@@ -3,7 +3,25 @@ import streamlit as st
 from phi.assistant import Assistant
 from phi.tools.arxiv_toolkit import ArxivToolkit
 from huggingface_hub import InferenceClient
-from phi.llm.huggingface import HuggingFaceLLM  # Added for proper LLM wrapping
+
+# Define a wrapper for Hugging Face LLM
+class HuggingFaceLLM:
+    def __init__(self, client):
+        self.client = client
+
+    def chat_completion(self, messages, max_tokens=512, stream=False, temperature=0.7, top_p=0.95):
+        response = self.client.post(
+            payload={
+                "inputs": messages,
+                "parameters": {
+                    "max_tokens": max_tokens,
+                    "temperature": temperature,
+                    "top_p": top_p,
+                },
+                "stream": stream,
+            }
+        )
+        return response
 
 # Initialize the Hugging Face Inference Client
 raw_client = InferenceClient(model="HuggingFaceH4/zephyr-7b-beta")
@@ -37,7 +55,7 @@ if st.button("Search") and query:
         # Generate response using Zephyr
         response = ""
         for message in client.chat_completion(messages, max_tokens=max_tokens, stream=True, temperature=temperature, top_p=top_p):
-            token = message.choices[0].delta.content
+            token = message["choices"][0]["delta"]["content"]
             response += token
 
         # Search arXiv and parse results