Workaround with L0_trt_reformat_free

Author: yinggeh

src/python/library/tritonclient/grpc/_infer_input.py

@@ -144,6 +144,7 @@ def set_shape(self, shape):
         """
         self._input.ClearField("shape")
         self._input.shape.extend(shape)
+        self.validate_data()
         return self
 
     def set_data_from_numpy(self, input_tensor):

src/python/library/tritonclient/grpc/_utils.py

@@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 
-# Copyright 2023-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# Copyright 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 #
 # Redistribution and use in source and binary forms, with or without
 # modification, are permitted provided that the following conditions
@@ -96,7 +96,6 @@ def _get_inference_request(
     if request_id != "":
         request.id = request_id
     for infer_input in inputs:
-        infer_input.validate_data()
         request.inputs.extend([infer_input._get_tensor()])
         if infer_input._get_content() is not None:
             request.raw_input_contents.extend([infer_input._get_content()])

src/python/library/tritonclient/http/_infer_input.py

@@ -145,6 +145,7 @@ def set_shape(self, shape):
             The updated input
         """
         self._shape = shape
+        self.validate_data()
         return self
 
     def set_data_from_numpy(self, input_tensor, binary_data=True):

src/python/library/tritonclient/http/_utils.py

@@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 
-# Copyright 2023-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# Copyright 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 #
 # Redistribution and use in source and binary forms, with or without
 # modification, are permitted provided that the following conditions
@@ -106,11 +106,7 @@ def _get_inference_request(
     if timeout is not None:
         parameters["timeout"] = timeout
 
-    infer_request["inputs"] = []
-    for infer_input in inputs:
-        infer_input.validate_data()
-        infer_request["inputs"].append(infer_input._get_tensor())
-
+    infer_request["inputs"] = [this_input._get_tensor() for this_input in inputs]
     if outputs:
         infer_request["outputs"] = [
             this_output._get_tensor() for this_output in outputs