Adding `num_return_sequences` support for text2text generation. (#14988)

* Adding `num_return_sequences` support for text2text generation.
Co-Authored-By: Enze <pu.miao@foxmail.com>

* Update tests/test_pipelines_text2text_generation.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update tests/test_pipelines_text2text_generation.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Enze <pu.miao@foxmail.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

src/transformers/pipelines/text2text_generation.py

@@ -157,18 +157,20 @@ class Text2TextGenerationPipeline(Pipeline):
         return {"output_ids": output_ids}
 
     def postprocess(self, model_outputs, return_type=ReturnType.TEXT, clean_up_tokenization_spaces=False):
-        record = {}
+        records = []
+        for output_ids in model_outputs["output_ids"]:
             if return_type == ReturnType.TENSORS:
                 record = {f"{self.return_name}_token_ids": model_outputs}
             elif return_type == ReturnType.TEXT:
                 record = {
                     f"{self.return_name}_text": self.tokenizer.decode(
-                    model_outputs["output_ids"][0],
+                        output_ids,
                         skip_special_tokens=True,
                         clean_up_tokenization_spaces=clean_up_tokenization_spaces,
                     )
                 }
-        return record
+            records.append(record)
+        return records
 
 
 @add_end_docstrings(PIPELINE_INIT_ARGS)

tests/test_pipelines_text2text_generation.py

@@ -50,6 +50,19 @@ class Text2TextGenerationPipelineTests(unittest.TestCase, metaclass=PipelineTest
         outputs = generator("Something there", do_sample=False)
         self.assertEqual(outputs, [{"generated_text": ""}])
 
+        num_return_sequences = 3
+        outputs = generator(
+            "Something there",
+            num_return_sequences=num_return_sequences,
+            num_beams=num_return_sequences,
+        )
+        target_outputs = [
+            {"generated_text": "Beide Beide Beide Beide Beide Beide Beide Beide Beide"},
+            {"generated_text": "Beide Beide Beide Beide Beide Beide Beide Beide"},
+            {"generated_text": ""},
+        ]
+        self.assertEqual(outputs, target_outputs)
+
     @require_tf
     def test_small_model_tf(self):
         generator = pipeline("text2text-generation", model="patrickvonplaten/t5-tiny-random", framework="tf")