Update app.py

Delete the safetensors after converting it into fp16 to save space

app.py

@@ -11,6 +11,8 @@ from gradio_huggingfacehub_search import HuggingfaceHubSearch
 from apscheduler.schedulers.background import BackgroundScheduler
 from datetime import datetime
 import numpy as np
+import shutil
+
 
 HF_TOKEN = os.environ.get("HF_TOKEN")
 os.environ["GRADIO_ANALYTICS_ENABLED"] = "False"
@@ -237,7 +239,7 @@ def process_model(model_id, q_method, use_imatrix, imatrix_q_method, private_rep
               result = subprocess.run(["python", CONVERSION_SCRIPT, local_dir, "--outtype", "f16", "--outfile", fp16], shell=False, capture_output=True)
               if result.returncode != 0:
                   raise Exception(f"Error converting to fp16: {result.stderr.decode()}")
-
+          shutil.rmtree(downloads_dir)
           imatrix_path = Path(outdir)/"imatrix.dat"
           if use_imatrix:
               train_data_path = train_data_file.name if train_data_file else "llama.cpp/groups_merged.txt"