Fix provider availability check on ORT 1.16.0 release (#1403)

fix provider availablity on ORT 1.16.0 release

optimum/onnxruntime/utils.py

@@ -22,6 +22,7 @@
 
 import numpy as np
 import torch
+from packaging import version
 from transformers.utils import logging
 
 import onnxruntime as ort
@@ -218,8 +219,13 @@ def validate_provider_availability(provider: str):
     Args:
         provider (str): Name of an ONNX Runtime execution provider.
     """
-    # disable on Windows as reported in https://github.com/huggingface/optimum/issues/769
-    if os.name != "nt" and provider in ["CUDAExecutionProvider", "TensorrtExecutionProvider"]:
+    # Disable on Windows as reported in https://github.com/huggingface/optimum/issues/769.
+    # Disable as well for ORT 1.16.0 that has changed changed the way _ld_preload.py is filled: https://github.com/huggingface/optimum/issues/1402.
+    if (
+        version.parse(ort.__version__) < version.parse("1.16.0")
+        and os.name != "nt"
+        and provider in ["CUDAExecutionProvider", "TensorrtExecutionProvider"]
+    ):
         path_cuda_lib = os.path.join(ort.__path__[0], "capi", "libonnxruntime_providers_cuda.so")
         path_trt_lib = os.path.join(ort.__path__[0], "capi", "libonnxruntime_providers_tensorrt.so")
         path_dependecy_loading = os.path.join(ort.__path__[0], "capi", "_ld_preload.py")