Update src/qa.py

src/qa.py

@@ -1,16 +1,24 @@
 import os
-from sentence_transformers import SentenceTransformer
-from transformers import pipeline
-from vectorstore import search_faiss
-
-print("✅ qa.py loaded from:", __file__)
 
+# ----------------------------
 # Force Hugging Face to use /tmp for cache
+# ----------------------------
 CACHE_DIR = "/tmp/huggingface"
+os.makedirs(CACHE_DIR, exist_ok=True)
+
 os.environ["HF_HOME"] = CACHE_DIR
 os.environ["TRANSFORMERS_CACHE"] = CACHE_DIR
 os.environ["HF_DATASETS_CACHE"] = CACHE_DIR
 
+# ----------------------------
+# Now import libraries (they'll respect env vars)
+# ----------------------------
+from sentence_transformers import SentenceTransformer
+from transformers import pipeline
+from vectorstore import search_faiss
+
+print("✅ qa.py loaded from:", __file__)
+
 # ----------------------------
 # Query embedding model
 # ----------------------------
@@ -24,7 +32,6 @@ _query_model = SentenceTransformer(
 # ----------------------------
 MODEL_NAME = "google/flan-t5-small"
 
-# Make sure model downloads into /tmp
 _answer_model = pipeline(
     "text2text-generation",
     model=MODEL_NAME,