Update app.py

app.py

@@ -12,7 +12,11 @@ from itertools import islice
 
 FILEPATH_PATTERN = "structured_data_doc.parquet"
 NUM_PROC = os.cpu_count()
+parent_path  = dirname(getcwd())
 
+cache_path = join(parent_path,'.cache')
+if not exists(cache_path):
+    makedirs(cache_path)
 
 # Determine device based on GPU availability
 device = "cuda" if torch.cuda.is_available() else "cpu"
@@ -69,6 +73,7 @@ async def create_upload_file(file: UploadFile = File(...)):
     full_dataset = load_dataset("parquet",
                 data_files=FILEPATH_PATTERN,
                 split="train",
+                cache_path=cache_path,
                 keep_in_memory=True,
                 num_proc=NUM_PROC*2)
     # Generate and append embeddings to the train split