fix: dataset

app.py

@@ -21,6 +21,16 @@ def load_models():
     model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
     return tokenizer, model
 
+@st.cache_data
+def load_dataset():
+    # Create initial dataset if it doesn't exist
+    if not os.path.exists(DATASET_PATH):
+        with st.spinner("Building initial dataset from autism research papers..."):
+            import faiss_index.index as idx
+            papers = idx.fetch_arxiv_papers("autism research", max_results=100)
+            idx.build_faiss_index(papers, dataset_dir=DATASET_DIR)
+    return load_from_disk(DATASET_PATH)
+
 def generate_answer(question, context, max_length=200):
     tokenizer, model = load_models()