[`generate`] Only warn users if the `generation_config`'s `max_length` is set...

[`generate`]  Only warn users if the `generation_config`'s `max_length` is set to the default value (#25030)

* check max length is default

* nit

* update warning: no-longer deprecate

* comment in the configuration_utils in case max length's default gets changed in the futur

src/transformers/generation/configuration_utils.py

@@ -234,6 +234,7 @@ class GenerationConfig(PushToHubMixin):
 
     def __init__(self, **kwargs):
         # Parameters that control the length of the output
+        # if the default `max_length` is updated here, make sure to update the `generate` tests following https://github.com/huggingface/transformers/pull/25030
         self.max_length = kwargs.pop("max_length", 20)
         self.max_new_tokens = kwargs.pop("max_new_tokens", None)
         self.min_length = kwargs.pop("min_length", 0)

src/transformers/generation/flax_utils.py

@@ -377,11 +377,11 @@ class FlaxGenerationMixin:
         # Prepare `max_length` depending on other stopping criteria.
         input_ids_seq_length = input_ids.shape[-1]
         has_default_max_length = kwargs.get("max_length") is None and generation_config.max_length is not None
-        if has_default_max_length and generation_config.max_new_tokens is None:
+        if has_default_max_length and generation_config.max_new_tokens is None and generation_config.max_length != 20:
+            # 20 is the default max_length of the generation config
             warnings.warn(
-                f"Using `max_length`'s default ({generation_config.max_length}) to control the generation length. "
-                "This behaviour is deprecated and will be removed from the config in v5 of Transformers -- we"
-                " recommend using `max_new_tokens` to control the maximum length of the generation.",
+                f"Using the model-agnostic default `max_length` (={generation_config.max_length}) "
+                "to control the generation length.  recommend setting `max_new_tokens` to control the maximum length of the generation.",
                 UserWarning,
             )
         elif generation_config.max_new_tokens is not None:

src/transformers/generation/tf_utils.py

@@ -850,11 +850,11 @@ class TFGenerationMixin:
         # 7. Prepare `max_length` depending on other stopping criteria.
         input_ids_seq_length = shape_list(input_ids)[-1]
         has_default_max_length = kwargs.get("max_length") is None and generation_config.max_length is not None
-        if has_default_max_length and generation_config.max_new_tokens is None:
+        if has_default_max_length and generation_config.max_new_tokens is None and generation_config.max_length != 20:
+            # 20 is the default max_length of the generation config
             warnings.warn(
-                f"Using `max_length`'s default ({generation_config.max_length}) to control the generation length. "
-                "This behaviour is deprecated and will be removed from the config in v5 of Transformers -- we"
-                " recommend using `max_new_tokens` to control the maximum length of the generation.",
+                f"Using the model-agnostic default `max_length` (={generation_config.max_length}) "
+                "to control the generation length.  recommend setting `max_new_tokens` to control the maximum length of the generation.",
                 UserWarning,
             )
         elif generation_config.max_new_tokens is not None:

src/transformers/generation/utils.py

@@ -1365,11 +1365,11 @@ class GenerationMixin:
         # 6. Prepare `max_length` depending on other stopping criteria.
         input_ids_seq_length = input_ids.shape[-1]
         has_default_max_length = kwargs.get("max_length") is None and generation_config.max_length is not None
-        if has_default_max_length and generation_config.max_new_tokens is None:
+        if has_default_max_length and generation_config.max_new_tokens is None and generation_config.max_length != 20:
+            # 20 is the default max_length of the generation config
             warnings.warn(
-                f"Using `max_length`'s default ({generation_config.max_length}) to control the generation length. "
-                "This behaviour is deprecated and will be removed from the config in v5 of Transformers -- we"
-                " recommend using `max_new_tokens` to control the maximum length of the generation.",
+                f"Using the model-agnostic default `max_length` (={generation_config.max_length}) "
+                "to control the generation length.  recommend setting `max_new_tokens` to control the maximum length of the generation.",
                 UserWarning,
             )
         elif generation_config.max_new_tokens is not None:

src/transformers/models/musicgen/modeling_musicgen.py

@@ -1303,11 +1303,10 @@ class MusicgenForCausalLM(MusicgenPreTrainedModel):
         # 5. Prepare `max_length` depending on other stopping criteria.
         input_ids_seq_length = input_ids.shape[-1]
         has_default_max_length = kwargs.get("max_length") is None and generation_config.max_length is not None
-        if has_default_max_length and generation_config.max_new_tokens is None:
+        if has_default_max_length and generation_config.max_new_tokens is None and generation_config.max_length != 20:
             logger.warning(
-                f"Using `max_length`'s default ({generation_config.max_length}) to control the generation length. "
-                "This behaviour is deprecated and will be removed from the config in v5 of Transformers -- we"
-                " recommend using `max_new_tokens` to control the maximum length of the generation.",
+                f"Using the model-agnostic default `max_length` (={generation_config.max_length}) "
+                "to control the generation length.  recommend setting `max_new_tokens` to control the maximum length of the generation.",
                 UserWarning,
             )
         elif generation_config.max_new_tokens is not None:
@@ -2332,9 +2331,8 @@ class MusicgenForConditionalGeneration(PreTrainedModel):
         has_default_max_length = kwargs.get("max_length") is None and generation_config.max_length is not None
         if has_default_max_length and generation_config.max_new_tokens is None:
             logger.warning(
-                f"Using `max_length`'s default ({generation_config.max_length}) to control the generation length. "
-                "This behaviour is deprecated and will be removed from the config in v5 of Transformers -- we"
-                " recommend using `max_new_tokens` to control the maximum length of the generation.",
+                f"Using the model-agnostic default `max_length` (={generation_config.max_length}) "
+                "to control the generation length.  recommend setting `max_new_tokens` to control the maximum length of the generation.",
                 UserWarning,
             )
         elif generation_config.max_new_tokens is not None: