Update app.py

app.py

@@ -17,8 +17,8 @@ all_questions = questions_english + questions_arabic
 vectorizer = TfidfVectorizer().fit(all_questions)
 questions_tfidf = vectorizer.transform(all_questions)
 
-# Initialize Hugging Face Inference Client with the specified model
-client = InferenceClient("meta-llama/Llama-3.2-11B-Vision-Instruct")
+# Initialize Hugging Face Inference Client
+client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
 def find_similar_question(user_message: str, threshold: float = 0.7) -> str:
     """
@@ -27,21 +27,21 @@ def find_similar_question(user_message: str, threshold: float = 0.7) -> str:
     """
     # Transform the user's message using the same vectorizer
     user_tfidf = vectorizer.transform([user_message])
-
+    
     # Compute cosine similarity with all dataset questions
     similarities = cosine_similarity(user_tfidf, questions_tfidf).flatten()
-
+    
     # Find the index of the most similar question
     max_similarity_index = similarities.argmax()
     max_similarity_score = similarities[max_similarity_index]
-
+    
     # If the similarity score exceeds the threshold, return the corresponding answer
     if max_similarity_score >= threshold:
         if max_similarity_index < len(questions_english):
             return df['Answer'].iloc[max_similarity_index]
         else:
             return df['إجابة'].iloc[max_similarity_index - len(questions_english)]
-
+    
     # Return None if no similar question is found
     return None
 
@@ -62,17 +62,25 @@ def respond(
         return
 
     # If no similar question is found, use the Hugging Face model
-    prompt = f"{system_message}\nUser: {message}\nAssistant:"
+    messages = [{"role": "system", "content": system_message}]
+
+    for val in history:
+        if val[0]:
+            messages.append({"role": "user", "content": val[0]})
+        if val[1]:
+            messages.append({"role": "assistant", "content": val[1]})
+
+    messages.append({"role": "user", "content": message})
 
     response = ""
-    for msg in client.text_generation(
-        prompt,
-        max_new_tokens=max_tokens,
+    for msg in client.chat_completion(
+        messages,
+        max_tokens=max_tokens,
         stream=True,
         temperature=temperature,
         top_p=top_p,
     ):
-        token = msg.get("generated_text", "")
+        token = msg.choices[0].delta.content
         response += token
         yield response