diff --git a/examples/seq2seq/finetune_trainer.py b/examples/seq2seq/finetune_trainer.py
index 5159b82ef3..30cc30353a 100755
--- a/examples/seq2seq/finetune_trainer.py
+++ b/examples/seq2seq/finetune_trainer.py
@@ -98,6 +98,7 @@ class DataTrainingArguments:
         metadata={
             "help": "The maximum total sequence length for validation target text after tokenization. Sequences longer "
             "than this will be truncated, sequences shorter will be padded."
+            " This argument is also used to override the ``max_length`` param of ``model.generate``, which is used during ``evaluate`` and ``predict``"
         },
     )
     test_max_target_length: Optional[int] = field(
diff --git a/examples/seq2seq/utils.py b/examples/seq2seq/utils.py
index 8014463122..437cdf2e63 100644
--- a/examples/seq2seq/utils.py
+++ b/examples/seq2seq/utils.py
@@ -434,7 +434,8 @@ def use_task_specific_params(model, task):
 
     if task_specific_params is not None:
         pars = task_specific_params.get(task, {})
-        logger.info(f"using task specific params for {task}: {pars}")
+        logger.info(f"setting model.config to task specific params for {task}:\n {pars}")
+        logger.info("note: command line args may override some of these")
         model.config.update(pars)