Trainer: delegate default generation values to `generation_config` (#25987)

examples/pytorch/question-answering/trainer_seq2seq_qa.py

@@ -46,12 +46,13 @@ class QuestionAnsweringSeq2SeqTrainer(Seq2SeqTrainer):
         **gen_kwargs,
     ) -> Dict[str, float]:
         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
-        )
+
+        # Use legacy argument setting if a) the option is not explicitly passed; and b) the argument is set in the
+        # training args
+        if gen_kwargs.get("max_length") is None and self.args.generation_max_length is not None:
+            gen_kwargs["max_length"] = self.args.generation_max_length
+        if gen_kwargs.get("num_beams") is None and self.args.generation_num_beams is not None:
+            gen_kwargs["num_beams"] = self.args.generation_num_beams
         self._gen_kwargs = gen_kwargs
 
         eval_dataset = self.eval_dataset if eval_dataset is None else eval_dataset

src/transformers/trainer_seq2seq.py

@@ -149,11 +149,17 @@ class Seq2SeqTrainer(Trainer):
         """
 
         gen_kwargs = gen_kwargs.copy()
-        if gen_kwargs.get("max_length") is None and gen_kwargs.get("max_new_tokens") is None:
+
+        # Use legacy argument setting if a) the option is not explicitly passed; and b) the argument is set in the
+        # training args
+        if (
+            gen_kwargs.get("max_length") is None
+            and gen_kwargs.get("max_new_tokens") is None
+            and self.args.generation_max_length is not None
+        ):
             gen_kwargs["max_length"] = 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
-        )
+        if gen_kwargs.get("num_beams") is None and self.args.generation_num_beams is not None:
+            gen_kwargs["num_beams"] = 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)
@@ -206,11 +212,17 @@ class Seq2SeqTrainer(Trainer):
         """
 
         gen_kwargs = gen_kwargs.copy()
-        if gen_kwargs.get("max_length") is None and gen_kwargs.get("max_new_tokens") is None:
+
+        # Use legacy argument setting if a) the option is not explicitly passed; and b) the argument is set in the
+        # training args
+        if (
+            gen_kwargs.get("max_length") is None
+            and gen_kwargs.get("max_new_tokens") is None
+            and self.args.generation_max_length is not None
+        ):
             gen_kwargs["max_length"] = 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
-        )
+        if gen_kwargs.get("num_beams") is None and self.args.generation_num_beams is not None:
+            gen_kwargs["num_beams"] = 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)
@@ -256,16 +268,14 @@ class Seq2SeqTrainer(Trainer):
 
         # XXX: adapt synced_gpus for fairscale as well
         # Priority (handled in generate):
-        # gen_kwargs > model.generation_config > default GenerationConfig()
-
+        # non-`None` gen_kwargs > model.generation_config > default GenerationConfig()
         if len(gen_kwargs) == 0 and hasattr(self, "_gen_kwargs"):
             gen_kwargs = self._gen_kwargs.copy()
+        if "num_beams" in gen_kwargs and gen_kwargs["num_beams"] is None:
+            gen_kwargs.pop("num_beams")
+        if "max_length" in gen_kwargs and gen_kwargs["max_length"] is None:
+            gen_kwargs.pop("max_length")
 
-        if gen_kwargs.get("max_length") is None and gen_kwargs.get("max_new_tokens") is None:
-            gen_kwargs["max_length"] = 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