Adding a new return_full_text parameter to TextGenerationPipeline. (#9852)

* Adding a new `return_full_text` parameter to TextGenerationPipeline.

For text-generation, it's sometimes used as prompting text.
In that context, prefixing `generated_text` with the actual input
forces the caller to take an extra step to remove it.

The proposed change adds a new parameter (for backward compatibility).
`return_full_text` that enables the caller to prevent adding the prefix.

* Doc quality.

src/transformers/pipelines/text_generation.py

@@ -44,10 +44,11 @@ class TextGenerationPipeline(Pipeline):
         "TFCTRLLMHeadModel",
     ]
 
-    def __init__(self, *args, **kwargs):
+    def __init__(self, *args, return_full_text=True, **kwargs):
         super().__init__(*args, **kwargs)
 
         self.check_model_type(self.ALLOWED_MODELS)
+        self.return_full_text = return_full_text
 
     # overriding _parse_and_tokenize to allow for unusual language-modeling tokenizer arguments
     def _parse_and_tokenize(self, *args, **kwargs):
@@ -65,6 +66,7 @@ class TextGenerationPipeline(Pipeline):
         text_inputs,
         return_tensors=False,
         return_text=True,
+        return_full_text=None,
         clean_up_tokenization_spaces=False,
         prefix=None,
         **generate_kwargs
@@ -79,6 +81,9 @@ class TextGenerationPipeline(Pipeline):
                 Whether or not to include the tensors of predictions (as token indices) in the outputs.
             return_text (:obj:`bool`, `optional`, defaults to :obj:`True`):
                 Whether or not to include the decoded texts in the outputs.
+            return_full_text (:obj:`bool`, `optional`, defaults to :obj:`True`):
+                If set to :obj:`False` only added text is returned, otherwise the full text is returned Only meaningful
+                if `return_text` is set to True.
             clean_up_tokenization_spaces (:obj:`bool`, `optional`, defaults to :obj:`False`):
                 Whether or not to clean up the potential extra spaces in the text output.
             prefix (:obj:`str`, `optional`):
@@ -94,6 +99,8 @@ class TextGenerationPipeline(Pipeline):
             - **generated_token_ids** (:obj:`torch.Tensor` or :obj:`tf.Tensor`, present when ``return_tensors=True``)
               -- The token ids of the generated text.
         """
+        prefix = prefix if prefix is not None else self.model.config.prefix
+        return_full_text = return_full_text if return_full_text is not None else self.return_full_text
 
         if isinstance(text_inputs, str):
             text_inputs = [text_inputs]
@@ -101,7 +108,6 @@ class TextGenerationPipeline(Pipeline):
         for prompt_text in text_inputs:
             # Manage correct placement of the tensors
             with self.device_placement():
-                prefix = prefix if prefix is not None else self.model.config.prefix
                 if prefix is None and self.model.__class__.__name__ in [
                     "XLNetLMHeadModel",
                     "TransfoXLLMHeadModel",
@@ -168,7 +174,12 @@ class TextGenerationPipeline(Pipeline):
                             )
                         )
 
-                    record["generated_text"] = prompt_text + text[prompt_length:]
+                    if return_full_text:
+                        all_text = prompt_text + text[prompt_length:]
+                    else:
+                        all_text = text[prompt_length:]
+
+                    record["generated_text"] = all_text
 
                 result.append(record)
             results += [result]

tests/test_pipelines_text_generation.py

@@ -15,6 +15,7 @@
 import unittest
 
 from transformers import pipeline
+from transformers.testing_utils import require_torch
 
 from .test_pipelines_common import MonoInputPipelineCommonMixin
 
@@ -41,3 +42,21 @@ class TextGenerationPipelineTests(MonoInputPipelineCommonMixin, unittest.TestCas
         self.assertEqual(type(outputs[0][0]["generated_text"]), str)
         self.assertEqual(list(outputs[1][0].keys()), ["generated_text"])
         self.assertEqual(type(outputs[1][0]["generated_text"]), str)
+
+    @require_torch
+    def test_generation_output_style(self):
+        text_generator = pipeline(task="text-generation", model=self.small_models[0])
+        # text-generation is non-deterministic by nature, we can't fully test the output
+
+        outputs = text_generator("This is a test")
+        self.assertIn("This is a test", outputs[0]["generated_text"])
+
+        outputs = text_generator("This is a test", return_full_text=False)
+        self.assertNotIn("This is a test", outputs[0]["generated_text"])
+
+        text_generator = pipeline(task="text-generation", model=self.small_models[0], return_full_text=False)
+        outputs = text_generator("This is a test")
+        self.assertNotIn("This is a test", outputs[0]["generated_text"])
+
+        outputs = text_generator("This is a test", return_full_text=True)
+        self.assertIn("This is a test", outputs[0]["generated_text"])