Comment/message updates: b/181866850 tracks next steps after b/149576200.

PiperOrigin-RevId: 360932205

official/nlp/modeling/layers/text_layers.py

@@ -330,8 +330,7 @@ class SentencepieceTokenizer(tf.keras.layers.Layer):
     if bool(model_file_path) == bool(model_serialized_proto):
       raise ValueError("Exact one of `model_file_path` and "
                        "`model_serialized_proto` can be specified.")
-    # TODO(chendouble): After b/149576200 is resolved, support
-    # tokenize_with_offsets when strip_diacritics is True,
+    # TODO(b/181866850): Support tokenize_with_offsets for strip_diacritics=True
     if tokenize_with_offsets and strip_diacritics:
       raise ValueError("`tokenize_with_offsets` is not supported when "
                        "`strip_diacritics` is set to True.")
@@ -378,8 +377,8 @@ class SentencepieceTokenizer(tf.keras.layers.Layer):
     """
     if self._strip_diacritics:
       if self.tokenize_with_offsets:
-        raise ValueError("`tokenize_with_offsets` is not supported yet due to "
-                         "b/149576200, when `strip_diacritics` is set to True.")
+        raise ValueError("`tokenize_with_offsets` is not supported yet when "
+                         "`strip_diacritics` is set to True (b/181866850).")
       inputs = text.normalize_utf8(inputs, "NFD")
       inputs = tf.strings.regex_replace(inputs, r"\p{Mn}", "")
 

official/nlp/tools/export_tfhub.py

@@ -145,7 +145,7 @@ flags.DEFINE_integer(
     "sequence length for the bert_pack_inputs subobject."
     "Needed for --export_type preprocessing.")
 flags.DEFINE_bool(
-    "tokenize_with_offsets", False,  # Broken by b/149576200.
+    "tokenize_with_offsets", False,  # TODO(b/181866850)
     "Whether to export a .tokenize_with_offsets subobject for "
     "--export_type preprocessing.")
 flags.DEFINE_multi_string(

official/nlp/tools/export_tfhub_lib_test.py

@@ -564,7 +564,7 @@ class ExportPreprocessingTest(tf.test.TestCase, parameterized.TestCase):
   def test_exported_callables(self, use_sp_model):
     preprocess = tf.saved_model.load(self._do_export(
         ["d", "ef", "abc", "xy"], do_lower_case=True,
-        tokenize_with_offsets=not use_sp_model,  # TODO(b/149576200): drop this.
+        tokenize_with_offsets=not use_sp_model,  # TODO(b/181866850): drop this.
         experimental_disable_assert=True,  # TODO(b/175369555): drop this.
         use_sp_model=use_sp_model))
 
@@ -590,7 +590,7 @@ class ExportPreprocessingTest(tf.test.TestCase, parameterized.TestCase):
 
     # .tokenize_with_offsets()
     if use_sp_model:
-      # TODO(b/149576200): Enable tokenize_with_offsets when it works and test.
+      # TODO(b/181866850): Enable tokenize_with_offsets when it works and test.
       self.assertFalse(hasattr(preprocess, "tokenize_with_offsets"))
     else:
       token_ids, start_offsets, limit_offsets = (
@@ -691,7 +691,7 @@ class ExportPreprocessingTest(tf.test.TestCase, parameterized.TestCase):
   def test_shapes(self, use_sp_model):
     preprocess = tf.saved_model.load(self._do_export(
         ["abc", "def"], do_lower_case=True,
-        tokenize_with_offsets=not use_sp_model,  # TODO(b/149576200): drop this.
+        tokenize_with_offsets=not use_sp_model,  # TODO(b/181866850): drop this.
         experimental_disable_assert=True,  # TODO(b/175369555): drop this.
         use_sp_model=use_sp_model))
 
@@ -711,7 +711,7 @@ class ExportPreprocessingTest(tf.test.TestCase, parameterized.TestCase):
               tf.TensorSpec([batch_size], tf.string)),
           token_out_shape,
           "with batch_size=%s" % batch_size)
-      # TODO(b/149576200): Enable tokenize_with_offsets when it works and test.
+      # TODO(b/181866850): Enable tokenize_with_offsets when it works and test.
       if use_sp_model:
         self.assertFalse(hasattr(preprocess, "tokenize_with_offsets"))
       else:
@@ -771,7 +771,7 @@ class ExportPreprocessingTest(tf.test.TestCase, parameterized.TestCase):
                           "quick", "fox", "lazy", "dog"]
     preprocess = tf.saved_model.load(self._do_export(
         non_special_tokens, do_lower_case=True,
-        tokenize_with_offsets=use_bert,  # TODO(b/149576200): drop this.
+        tokenize_with_offsets=use_bert,  # TODO(b/181866850): drop this.
         experimental_disable_assert=True,  # TODO(b/175369555): drop this.
         add_mask_token=True, use_sp_model=not use_bert))
     vocab_size = len(non_special_tokens) + (5 if use_bert else 7)