docs: Update README to reflect torch 2 support

Update the README file to remove the "experimental" tag from the documentaion.

The existance of the tag was an oversight as Torch 2.x has been supported for 18+ months at this point.

Signed-off-by: J Wyman <[email protected]>

Author: whoisj

README.md

@@ -1,5 +1,5 @@
 <!--
-# Copyright 2020-2024, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# Copyright 2020-2025, 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
@@ -81,8 +81,8 @@ Currently, Triton requires that a specially patched version of
 PyTorch be used with the PyTorch backend. The full source for
 these PyTorch versions are available as Docker images from
 [NGC](https://ngc.nvidia.com). For example, the PyTorch version
-compatible with the 22.12 release of Triton is available as
-nvcr.io/nvidia/pytorch:22.12-py3.
+compatible with the 25.09 release of Triton is available as
+nvcr.io/nvidia/pytorch:25.09-py3.
 
 Copy over the LibTorch and Torchvision headers and libraries from the
 [PyTorch NGC container](https://ngc.nvidia.com/catalog/containers/nvidia:pytorch)
@@ -306,50 +306,7 @@ instance in the
 to ensure that the model instance and the tensors used for inference are
 assigned to the same GPU device as on which the model was traced.
 
-# PyTorch 2.0 Backend \[Experimental\]
-
-> [!WARNING]
-> *This feature is subject to change and removal.*
-
-Starting from 24.01, PyTorch models can be served directly via
-[Python runtime](src/model.py). By default, Triton will use the
-[LibTorch runtime](#pytorch-libtorch-backend) for PyTorch models. To use Python
-runtime, provide the following
-[runtime setting](https://github.com/triton-inference-server/backend/blob/main/README.md#backend-shared-library)
-in the model configuration:
-
-```
-runtime: "model.py"
-```
-
-## Dependencies
-
-### Python backend dependency
-
-This feature depends on
-[Python backend](https://github.com/triton-inference-server/python_backend),
-see
-[Python-based Backends](https://github.com/triton-inference-server/backend/blob/main/docs/python_based_backends.md)
-for more details.
-
-### PyTorch dependency
-
-This feature will take advantage of the
-[`torch.compile`](https://pytorch.org/docs/stable/generated/torch.compile.html#torch-compile)
-optimization, make sure the
-[PyTorch 2.0+ pip package](https://pypi.org/project/torch) is available in the
-same Python environment.
-
-Alternatively, a [Python Execution Environment](#using-custom-python-execution-environments)
-with the PyTorch dependency may be used. It can be created with the
-[provided script](tools/gen_pb_exec_env.sh). The resulting
-`pb_exec_env_model.py.tar.gz` file should be placed at the same
-[backend shared library](https://github.com/triton-inference-server/backend/blob/main/README.md#backend-shared-library)
-directory as the [Python runtime](src/model.py).
-
-## Model Layout
-
-### PyTorch 2.0 models
+## PyTorch 2.0 models
 
 The model repository should look like:
 
@@ -369,7 +326,7 @@ The `model.pt` may be optionally provided which contains the saved
 [`state_dict`](https://pytorch.org/tutorials/beginner/saving_loading_models.html#saving-loading-model-for-inference)
 of the model.
 
-### TorchScript models
+## TorchScript models
 
 The model repository should look like: