Fix TF Roberta for mixed precision training (#11675)

src/transformers/models/roberta/modeling_tf_roberta.py

@@ -541,7 +541,9 @@ class TFRobertaMainLayer(tf.keras.layers.Layer):
         # Since we are adding it to the raw scores before the softmax, this is
         # effectively the same as removing these entirely.
         extended_attention_mask = tf.cast(extended_attention_mask, dtype=embedding_output.dtype)
-        extended_attention_mask = tf.multiply(tf.subtract(1.0, extended_attention_mask), -10000.0)
+        one_cst = tf.constant(1.0, dtype=embedding_output.dtype)
+        ten_thousand_cst = tf.constant(-10000.0, dtype=embedding_output.dtype)
+        extended_attention_mask = tf.multiply(tf.subtract(one_cst, extended_attention_mask), ten_thousand_cst)
 
         # Prepare head mask if needed
         # 1.0 in head_mask indicate we keep the head