Use random_attention_mask for TF tests (#16517)

* use random_attention_mask for TF tests

* Fix for TFCLIP test (for now).
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

tests/openai/test_modeling_tf_openai.py

@@ -20,7 +20,7 @@ from transformers import OpenAIGPTConfig, is_tf_available
 from transformers.testing_utils import require_tf, slow
 
 from ..test_configuration_common import ConfigTester
-from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_tf_available():
@@ -70,7 +70,7 @@ class TFOpenAIGPTModelTester:
 
         input_mask = None
         if self.use_input_mask:
-            input_mask = ids_tensor([self.batch_size, self.seq_length], vocab_size=2)
+            input_mask = random_attention_mask([self.batch_size, self.seq_length])
 
         token_type_ids = None
         if self.use_token_type_ids:

tests/rembert/test_modeling_tf_rembert.py

@@ -20,7 +20,7 @@ from transformers import RemBertConfig, is_tf_available
 from transformers.testing_utils import require_tf, slow
 
 from ..test_configuration_common import ConfigTester
-from ..test_modeling_tf_common import TFModelTesterMixin, floats_tensor, ids_tensor
+from ..test_modeling_tf_common import TFModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
 
 
 if is_tf_available():
@@ -95,7 +95,7 @@ class TFRemBertModelTester:
 
         input_mask = None
         if self.use_input_mask:
-            input_mask = ids_tensor([self.batch_size, self.seq_length], vocab_size=2)
+            input_mask = random_attention_mask([self.batch_size, self.seq_length])
 
         token_type_ids = None
         if self.use_token_type_ids:

tests/roberta/test_modeling_tf_roberta.py

@@ -20,7 +20,7 @@ from transformers import RobertaConfig, is_tf_available
 from transformers.testing_utils import require_sentencepiece, require_tf, require_tokenizers, slow
 
 from ..test_configuration_common import ConfigTester
-from ..test_modeling_tf_common import TFModelTesterMixin, floats_tensor, ids_tensor
+from ..test_modeling_tf_common import TFModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
 
 
 if is_tf_available():
@@ -72,7 +72,7 @@ class TFRobertaModelTester:
 
         input_mask = None
         if self.use_input_mask:
-            input_mask = ids_tensor([self.batch_size, self.seq_length], vocab_size=2)
+            input_mask = random_attention_mask([self.batch_size, self.seq_length])
 
         token_type_ids = None
         if self.use_token_type_ids:

tests/roformer/test_modeling_tf_roformer.py

@@ -20,7 +20,7 @@ from transformers import RoFormerConfig, is_tf_available
 from transformers.testing_utils import require_tf, slow
 
 from ..test_configuration_common import ConfigTester
-from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_tf_available():
@@ -95,7 +95,7 @@ class TFRoFormerModelTester:
 
         input_mask = None
         if self.use_input_mask:
-            input_mask = ids_tensor([self.batch_size, self.seq_length], vocab_size=2)
+            input_mask = random_attention_mask([self.batch_size, self.seq_length])
 
         token_type_ids = None
         if self.use_token_type_ids:

tests/t5/test_modeling_tf_t5.py

@@ -20,7 +20,7 @@ from transformers.testing_utils import require_sentencepiece, require_tf, requir
 from transformers.utils import cached_property
 
 from ..test_configuration_common import ConfigTester
-from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_tf_available():
@@ -58,7 +58,7 @@ class TFT5ModelTester:
 
         input_mask = None
         if self.use_input_mask:
-            input_mask = ids_tensor([self.batch_size, self.seq_length], vocab_size=2)
+            input_mask = random_attention_mask([self.batch_size, self.seq_length])
 
         token_labels = None
         if self.use_labels:

tests/tapas/test_modeling_tf_tapas.py

@@ -38,7 +38,7 @@ from transformers.testing_utils import require_tensorflow_probability, require_t
 from transformers.utils import cached_property
 
 from ..test_configuration_common import ConfigTester
-from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_tf_available():
@@ -158,7 +158,7 @@ class TFTapasModelTester:
 
         input_mask = None
         if self.use_input_mask:
-            input_mask = ids_tensor([self.batch_size, self.seq_length], vocab_size=2)
+            input_mask = random_attention_mask([self.batch_size, self.seq_length])
 
         token_type_ids = []
         for type_vocab_size in self.type_vocab_sizes:

tests/test_modeling_tf_common.py

@@ -1440,7 +1440,7 @@ def ids_tensor(shape, vocab_size, rng=None, name=None, dtype=None):
 def random_attention_mask(shape, rng=None, name=None, dtype=None):
     attn_mask = ids_tensor(shape, vocab_size=2, rng=None, name=None, dtype=dtype)
     # make sure that at least one token is attended to for each batch
-    attn_mask = tf.concat([tf.constant(value=1, shape=(shape[0], 1), dtype=dtype), attn_mask[:, 1:]], axis=1)
+    attn_mask = tf.concat([attn_mask[:, :-1], tf.ones_like(attn_mask[:, -1:], dtype=dtype)], axis=-1)
     return attn_mask
 
 

tests/xlm/test_modeling_tf_xlm.py

@@ -20,7 +20,7 @@ from transformers import is_tf_available
 from transformers.testing_utils import require_tf, slow
 
 from ..test_configuration_common import ConfigTester
-from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_tf_available():
@@ -75,7 +75,7 @@ class TFXLMModelTester:
 
     def prepare_config_and_inputs(self):
         input_ids = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
-        input_mask = ids_tensor([self.batch_size, self.seq_length], 2, dtype=tf.float32)
+        input_mask = random_attention_mask([self.batch_size, self.seq_length], dtype=tf.float32)
 
         input_lengths = None
         if self.use_input_lengths:

tests/xlnet/test_modeling_tf_xlnet.py

@@ -22,7 +22,7 @@ from transformers import XLNetConfig, is_tf_available
 from transformers.testing_utils import require_tf, slow
 
 from ..test_configuration_common import ConfigTester
-from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_tf_available():
@@ -75,7 +75,7 @@ class TFXLNetModelTester:
         input_ids_1 = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
         input_ids_2 = ids_tensor([self.batch_size, self.seq_length], self.vocab_size)
         segment_ids = ids_tensor([self.batch_size, self.seq_length], self.type_vocab_size)
-        input_mask = ids_tensor([self.batch_size, self.seq_length], 2, dtype=tf.float32)
+        input_mask = random_attention_mask([self.batch_size, self.seq_length], dtype=tf.float32)
 
         input_ids_q = ids_tensor([self.batch_size, self.seq_length + 1], self.vocab_size)
         perm_mask = tf.zeros((self.batch_size, self.seq_length + 1, self.seq_length), dtype=tf.float32)