Update note about torchxla2

docs/tutorials/pytorch-export.ipynb

@@ -139,7 +139,7 @@
         "\n",
         "_Dynamic batch dimensions can be specified as a part of the initial `torch.export` step._\n",
         "\n",
-        "_`torch_xla` doesn't yet support exporting models with dynamic shapes to StableHLO, so we recommend using the experimental library [`torch_xla2`](https://github.com/pytorch/xla/tree/master/experimental/torch_xla2) for this. It has a similar API for exporting, converts PyTorch models to JAX prior to exporting to StableHLO, and has high opset coverage along with great dynamic shape support._"
+        "_`torch_xla`'s support for exporting dynamic models is limited, for these cases we recommend using [`torch_xla2`](https://github.com/pytorch/xla/tree/master/experimental/torch_xla2) for this. This lowering path leverages JAX for lowering to StableHLO, and has high opset coverage with much broader support for exported programs with dynamic shapes._"
       ]
     },
     {