Add a note for TPU issues for deberta_v3 (#758)

Author: mattdangerw

keras_nlp/models/deberta_v3/deberta_v3_backbone.py

@@ -47,6 +47,9 @@ class DebertaV3Backbone(Backbone):
     dimensions. To load preset architectures and weights, use the `from_preset`
     constructor.
 
+    Note: `DebertaV3Backbone` has a performance issue on TPUs, and we recommend
+    other models for TPU training and inference.
+
     Disclaimer: Pre-trained models are provided on an "as is" basis, without
     warranties or conditions of any kind. The underlying model is provided by a
     third party and subject to a separate license, available

keras_nlp/models/deberta_v3/deberta_v3_classifier.py

@@ -44,6 +44,9 @@ class DebertaV3Classifier(Task):
     `fit()`, `predict()`, and `evaluate()`. This is done by default when
     creating the model with `from_preset()`.
 
+    Note: `DebertaV3Backbone` has a performance issue on TPUs, and we recommend
+    other models for TPU training and inference.
+
     Disclaimer: Pre-trained models are provided on an "as is" basis, without
     warranties or conditions of any kind. The underlying model is provided by a
     third party and subject to a separate license, available