Update src/streamlit_app.py

src/streamlit_app.py

@@ -1,14 +1,14 @@
 from transformers import AutoModelForQuestionAnswering, AutoTokenizer
-from arabert.preprocess import ArabertPreprocessor
+from farasa.segmenter import FarasaSegmenter
 import streamlit as st
 import torch
 import os
 
 # Load model
-model = AutoModelForQuestionAnswering.from_pretrained("MarioMamdouh121/arabic-qa-model")
-tokenizer = AutoTokenizer.from_pretrained("MarioMamdouh121/arabic-qa-model")
+model = AutoModelForQuestionAnswering.from_pretrained("checkpoint-2817")
+tokenizer = AutoTokenizer.from_pretrained("checkpoint-2817")
 
-arabert_prep = ArabertPreprocessor(model_name="aubmindlab/bert-base-arabertv2")
+segmenter = FarasaSegmenter(interactive=False)
 
 # Streamlit interface
 st.title("Arabic Question Answering")
@@ -19,8 +19,8 @@ question = st.text_input("السؤال")
 
 if st.button("احصل على الجواب") and context and question:
     # Preprocess
-    context_proc = arabert_prep.preprocess(context)
-    question_proc = arabert_prep.preprocess(question)
+    context_proc = segmenter.segment(context)
+    question_proc = segmenter.segment(question)
 
     # Tokenize
     inputs = tokenizer(