[generate] model defaults being inherited only happens for newer models (#36881)

src/transformers/generation/utils.py

@@ -23,6 +23,7 @@ from typing import TYPE_CHECKING, Any, Callable, Dict, List, Optional, Tuple, Un
 import numpy as np
 import torch
 import torch.distributed as dist
+from packaging import version
 from torch import nn
 from torch.nn import functional as F
 
@@ -1552,7 +1553,7 @@ class GenerationMixin:
         return generation_config
 
     def _prepare_generation_config(
-        self, generation_config: Optional[GenerationConfig], **kwargs: Dict
+        self, generation_config: Optional[GenerationConfig], use_model_defaults: Optional[bool] = None, **kwargs: Dict
     ) -> Tuple[GenerationConfig, Dict]:
         """
         Prepares the base generation config, then applies any generation configuration options from kwargs. This
@@ -1591,12 +1592,18 @@ class GenerationMixin:
 
         generation_config = copy.deepcopy(generation_config)
 
-        # If `generation_config` is provided, let's fallback ALL default values to the model's generation config
         if not using_model_generation_config:
+            # If `generation_config` is provided:
+            # - `use_model_defaults`: let's fallback ALL default values to the model's generation config
+            # - otherwise: legacy behavior, let's just make sure we have the tokens defined
+            model_base_version = version.parse(version.parse(self.generation_config.transformers_version).base_version)
+            if use_model_defaults is True or (
+                use_model_defaults is None and model_base_version >= version.parse("4.50.0")
+            ):
                 modified_values = {}
                 default_generation_config = GenerationConfig()
                 for key, default_value in default_generation_config.__dict__.items():
-                if key.startswith("_"):  # metadata
+                    if key.startswith("_") or key == "transformers_version":  # metadata
                         continue
                     custom_gen_config_value = getattr(generation_config, key)
                     model_gen_config_value = getattr(self.generation_config, key)
@@ -1608,6 +1615,15 @@ class GenerationMixin:
                         f"`generation_config` default values have been modified to match model-specific defaults: "
                         f"{modified_values}. If this is not desired, please set these values explicitly."
                     )
+            else:
+                if generation_config.bos_token_id is None:
+                    generation_config.bos_token_id = self.generation_config.bos_token_id
+                if generation_config.eos_token_id is None:
+                    generation_config.eos_token_id = self.generation_config.eos_token_id
+                if generation_config.pad_token_id is None:
+                    generation_config.pad_token_id = self.generation_config.pad_token_id
+                if generation_config.decoder_start_token_id is None:
+                    generation_config.decoder_start_token_id = self.generation_config.decoder_start_token_id
 
         # Finally, apply any passed kwargs
         model_kwargs = generation_config.update(**kwargs)
@@ -1967,6 +1983,7 @@ class GenerationMixin:
         streamer: Optional["BaseStreamer"] = None,
         negative_prompt_ids: Optional[torch.Tensor] = None,
         negative_prompt_attention_mask: Optional[torch.Tensor] = None,
+        use_model_defaults: Optional[bool] = None,
         **kwargs,
     ) -> Union[GenerateOutput, torch.LongTensor]:
         r"""
@@ -2031,6 +2048,11 @@ class GenerationMixin:
                 size. This is an experimental feature, subject to breaking API changes in future versions.
             negative_prompt_attention_mask (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
                 Attention_mask for `negative_prompt_ids`.
+            use_model_defaults (`bool`, *optional*):
+                When it is `True`, unset parameters in `generation_config` will be set to the model-specific default
+                generation configuration (`model.generation_config`), as opposed to the global defaults
+                (`GenerationConfig()`). If unset, models saved starting from `v4.50` will consider this flag to be
+                `True`.
             kwargs (`Dict[str, Any]`, *optional*):
                 Ad hoc parametrization of `generation_config` and/or additional model-specific kwargs that will be
                 forwarded to the `forward` function of the model. If the model is an encoder-decoder model, encoder
@@ -2058,7 +2080,9 @@ class GenerationMixin:
         tokenizer = kwargs.pop("tokenizer", None)  # Pull this out first, we only use it for stopping criteria
         assistant_tokenizer = kwargs.pop("assistant_tokenizer", None)  # only used for assisted generation
 
-        generation_config, model_kwargs = self._prepare_generation_config(generation_config, **kwargs)
+        generation_config, model_kwargs = self._prepare_generation_config(
+            generation_config, use_model_defaults, **kwargs
+        )
         self._validate_model_kwargs(model_kwargs.copy())
         self._validate_assistant(assistant_model, tokenizer, assistant_tokenizer)
 

tests/models/gemma3/test_modeling_gemma3.py

@@ -575,8 +575,8 @@ class Gemma3IntegrationTest(unittest.TestCase):
 
     def test_generation_beyond_sliding_window_with_generation_config(self):
         """
-        Same as `test_generation_beyond_sliding_window`, but passing a GenerationConfig. Regression test for #36684 --
+        Similar to `test_generation_beyond_sliding_window`, but passing a GenerationConfig. Regression test for #36684
-        ensures `cache_implementation='hybrid'` is correctly inherited from the base `model.generation_config`.
+        -- ensures `cache_implementation='hybrid'` is correctly inherited from the base `model.generation_config`.
         """
         model_id = "google/gemma-3-1b-it"
         attn_implementation = "sdpa"
@@ -594,12 +594,16 @@ class Gemma3IntegrationTest(unittest.TestCase):
 
         # Make sure prefill is larger than sliding window
         input_size = inputs.input_ids.shape[-1]
-        self.assertTrue(input_size > model.config.sliding_window)
+        self.assertGreater(input_size, model.config.sliding_window)
 
-        generation_config = GenerationConfig(max_new_tokens=20)
+        generation_config = GenerationConfig(max_new_tokens=5, min_new_tokens=5)
+        out = model.generate(**inputs, generation_config=generation_config)
 
-        out = model.generate(**inputs, generation_config=generation_config)[:, input_size:]
+        # Generation works beyond sliding window
-        output_text = tokenizer.batch_decode(out)
+        self.assertGreater(out.shape[1], model.config.sliding_window)
+        self.assertEqual(out.shape[1], input_size + 5)
 
-        EXPECTED_COMPLETIONS = [" and I'm going to take a walk.\n\nI really enjoy the scenery, and I'", ", green, yellow, orange, purple, brown, black, white, gray.\n\nI'"]  # fmt: skip
+        # Note: Auto-inheritance only works for models saved starting from 4.50.0
-        self.assertEqual(output_text, EXPECTED_COMPLETIONS)
+        model.generation_config.transformers_version = "4.49.0"
+        with self.assertRaises(RuntimeError):  # errors out because it is not using hybrid cache
+            out = model.generate(**inputs, generation_config=generation_config)