asd

app.py

@@ -1,28 +1,32 @@
 from fastapi import FastAPI
 from pydantic import BaseModel
-from huggingface_hub import hf_hub_download
 from llama_cpp import Llama
+from huggingface_hub import hf_hub_download
 import os
 
 REPO_ID = "google/gemma-2b-it-GGUF"
 FILENAME = "gemma-2b-it.gguf"
-HF_TOKEN = os.environ.get("HF_TOKEN")  # must be set in HF Spaces Secrets
+HF_TOKEN = os.environ.get("HF_TOKEN")
 MODEL_DIR = "./models"
+CACHE_DIR = "./models/.hf_cache"
 MODEL_PATH = os.path.join(MODEL_DIR, FILENAME)
 
-# Step 1: Auto-download model if not exists
+# Make sure directories exist
+os.makedirs(MODEL_DIR, exist_ok=True)
+os.makedirs(CACHE_DIR, exist_ok=True)
+
 if not os.path.exists(MODEL_PATH):
-    os.makedirs(MODEL_DIR, exist_ok=True)
     try:
         print("📦 Downloading model from Hugging Face Hub...")
         hf_hub_download(
             repo_id=REPO_ID,
             filename=FILENAME,
             token=HF_TOKEN,
+            cache_dir=CACHE_DIR,
             local_dir=MODEL_DIR,
-            local_dir_use_symlinks=False
+            local_dir_use_symlinks=False  # even though deprecated, keep for compatibility
         )
-        print("✅ Model downloaded.")
+        print(f"✅ Model downloaded to {MODEL_PATH}")
     except Exception as e:
         print(f"❌ Download failed: {e}")
         raise