change path

app.py

@@ -17,11 +17,14 @@ app = FastAPI()
 
 
 model_name = "numind/NuExtract-1.5-tiny"
-# MODEL_CACHE_DIR = "/app/model_cache"
-# model_cache_path = snapshot_download(
-#     repo_id="numind/NuExtract-1.5-tiny",
-#     cache_dir=MODEL_CACHE_DIR
-# )
+MODEL_CACHE_DIR = "/app/model_cache"
+model_cache_path = snapshot_download(
+    repo_id="numind/NuExtract-1.5-tiny",
+    cache_dir=MODEL_CACHE_DIR
+)
+
+print(">>> MODEL CACHE PATH:", model_cache_path, os.listdir(model_cache_path))
+
 device     = "gpu" if torch.cuda.is_available() else "mps" if torch.backends.mps.is_available() else "cpu"
 dtype = torch.float16 if device in ("mps", "gpu") else torch.float32
 
@@ -41,14 +44,14 @@ def load_model():
     # model_name, torch_dtype=dtype, trust_remote_code=True
     # )
     model = AutoModelForCausalLM.from_pretrained(
-        "/app/model_cache/models--numind--NuExtract-1.5-tiny",  # include the full folder name
+        model_cache_path,
         local_files_only=True,
         torch_dtype=dtype,
         trust_remote_code=True
     ).to(device).eval()
     # tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
     tokenizer = AutoTokenizer.from_pretrained(
-        "/app/model_cache/models--numind--NuExtract-1.5-tiny",  # include the full folder name
+        model_cache_path,
         local_files_only=True,
         trust_remote_code=True
     )