Add logits_processor parameter, used by `generate`, to `Seq2SeqTrainer`...

Add logits_processor parameter, used by `generate`, to `Seq2SeqTrainer` methods `evaluate` and `predict` (#17805)

* Add logits_processor parameter, used by `generate`, to `Seq2SeqTrainer` methods `evaluate` and `predict`

* Add all generate parameters to `Seq2SeqTrainer`, and also to `QuestionAnsweringSeq2SeqTrainer` which overrides it

* Remove `self._num_beams` from trainer classes

* - Run fixup
- Fix "Constraint" not exposed
- Fix synced_gpus to actually read from param

* Use kwargs

* Copy kwargs before making changes to it

* Fix style issues unused imports

examples/pytorch/question-answering/trainer_seq2seq_qa.py

@@ -41,11 +41,16 @@ class QuestionAnsweringSeq2SeqTrainer(Seq2SeqTrainer):
         eval_examples=None,
         ignore_keys: Optional[List[str]] = None,
         metric_key_prefix: str = "eval",
-        max_length: Optional[int] = None,
-        num_beams: Optional[int] = None,
+        **gen_kwargs,
     ) -> Dict[str, float]:
-        self._max_length = max_length if max_length is not None else self.args.generation_max_length
-        self._num_beams = num_beams if num_beams is not None else self.args.generation_num_beams
+        gen_kwargs = gen_kwargs.copy()
+        gen_kwargs["max_length"] = (
+            gen_kwargs["max_length"] if gen_kwargs.get("max_length") is not None else self.args.generation_max_length
+        )
+        gen_kwargs["num_beams"] = (
+            gen_kwargs["num_beams"] if gen_kwargs.get("num_beams") is not None else self.args.generation_num_beams
+        )
+        self._gen_kwargs = gen_kwargs
 
         eval_dataset = self.eval_dataset if eval_dataset is None else eval_dataset
         eval_dataloader = self.get_eval_dataloader(eval_dataset)
@@ -87,7 +92,11 @@ class QuestionAnsweringSeq2SeqTrainer(Seq2SeqTrainer):
         self.control = self.callback_handler.on_evaluate(self.args, self.state, self.control, metrics)
         return metrics
 
-    def predict(self, predict_dataset, predict_examples, ignore_keys=None, metric_key_prefix: str = "test"):
+    def predict(
+        self, predict_dataset, predict_examples, ignore_keys=None, metric_key_prefix: str = "test", **gen_kwargs
+    ):
+        self._gen_kwargs = gen_kwargs.copy()
+
         predict_dataloader = self.get_test_dataloader(predict_dataset)
 
         # Temporarily disable metric computation, we will do it in the loop here.

src/transformers/trainer_seq2seq.py

@@ -33,8 +33,7 @@ class Seq2SeqTrainer(Trainer):
         eval_dataset: Optional[Dataset] = None,
         ignore_keys: Optional[List[str]] = None,
         metric_key_prefix: str = "eval",
-        max_length: Optional[int] = None,
-        num_beams: Optional[int] = None,
+        **gen_kwargs
     ) -> Dict[str, float]:
         """
         Run evaluation and returns metrics.
@@ -60,13 +59,23 @@ class Seq2SeqTrainer(Trainer):
             num_beams (`int`, *optional*):
                 Number of beams for beam search that will be used when predicting with the generate method. 1 means no
                 beam search.
+            gen_kwargs:
+                Additional `generate` specific kwargs.
 
         Returns:
             A dictionary containing the evaluation loss and the potential metrics computed from the predictions. The
             dictionary also contains the epoch number which comes from the training state.
         """
-        self._max_length = max_length if max_length is not None else self.args.generation_max_length
-        self._num_beams = num_beams if num_beams is not None else self.args.generation_num_beams
+
+        gen_kwargs = gen_kwargs.copy()
+        gen_kwargs["max_length"] = (
+            gen_kwargs["max_length"] if gen_kwargs.get("max_length") is not None else self.args.generation_max_length
+        )
+        gen_kwargs["num_beams"] = (
+            gen_kwargs["num_beams"] if gen_kwargs.get("num_beams") is not None else self.args.generation_num_beams
+        )
+        self._gen_kwargs = gen_kwargs
+
         return super().evaluate(eval_dataset, ignore_keys=ignore_keys, metric_key_prefix=metric_key_prefix)
 
     def predict(
@@ -74,8 +83,7 @@ class Seq2SeqTrainer(Trainer):
         test_dataset: Dataset,
         ignore_keys: Optional[List[str]] = None,
         metric_key_prefix: str = "test",
-        max_length: Optional[int] = None,
-        num_beams: Optional[int] = None,
+        **gen_kwargs
     ) -> PredictionOutput:
         """
         Run prediction and returns predictions and potential metrics.
@@ -98,6 +106,8 @@ class Seq2SeqTrainer(Trainer):
             num_beams (`int`, *optional*):
                 Number of beams for beam search that will be used when predicting with the generate method. 1 means no
                 beam search.
+            gen_kwargs:
+                Additional `generate` specific kwargs.
 
         <Tip>
 
@@ -114,8 +124,16 @@ class Seq2SeqTrainer(Trainer):
             - metrics (`Dict[str, float]`, *optional*): The potential dictionary of metrics (if the dataset contained
               labels).
         """
-        self._max_length = max_length if max_length is not None else self.args.generation_max_length
-        self._num_beams = num_beams if num_beams is not None else self.args.generation_num_beams
+
+        gen_kwargs = gen_kwargs.copy()
+        gen_kwargs["max_length"] = (
+            gen_kwargs["max_length"] if gen_kwargs.get("max_length") is not None else self.args.generation_max_length
+        )
+        gen_kwargs["num_beams"] = (
+            gen_kwargs["num_beams"] if gen_kwargs.get("num_beams") is not None else self.args.generation_num_beams
+        )
+        self._gen_kwargs = gen_kwargs
+
         return super().predict(test_dataset, ignore_keys=ignore_keys, metric_key_prefix=metric_key_prefix)
 
     def prediction_step(
@@ -155,11 +173,17 @@ class Seq2SeqTrainer(Trainer):
         inputs = self._prepare_inputs(inputs)
 
         # XXX: adapt synced_gpus for fairscale as well
-        gen_kwargs = {
-            "max_length": self._max_length if self._max_length is not None else self.model.config.max_length,
-            "num_beams": self._num_beams if self._num_beams is not None else self.model.config.num_beams,
-            "synced_gpus": True if is_deepspeed_zero3_enabled() else False,
-        }
+        gen_kwargs = self._gen_kwargs.copy()
+        gen_kwargs["max_length"] = (
+            gen_kwargs["max_length"] if gen_kwargs.get("max_length") is not None else self.model.config.max_length
+        )
+        gen_kwargs["num_beams"] = (
+            gen_kwargs["num_beams"] if gen_kwargs.get("num_beams") is not None else self.model.config.num_beams
+        )
+        default_synced_gpus = True if is_deepspeed_zero3_enabled() else False
+        gen_kwargs["synced_gpus"] = (
+            gen_kwargs["synced_gpus"] if gen_kwargs.get("synced_gpus") is not None else default_synced_gpus
+        )
 
         if "attention_mask" in inputs:
             gen_kwargs["attention_mask"] = inputs.get("attention_mask", None)