Generate: v4.42 deprecations 🧹🧹 (huggingface#31956)

v4_42 deprecations
MHRDYN7 · Jul 23, 2024 · 11aaad7 · 11aaad7
1 parent 48bbb2a
commit 11aaad7
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 71 deletions.
diff --git a/src/transformers/generation/utils.py b/src/transformers/generation/utils.py
@@ -2218,8 +2218,8 @@ def _dola_decoding(
         stopping_criteria: StoppingCriteriaList,
         generation_config: GenerationConfig,
         synced_gpus: bool,
-        streamer: Optional["BaseStreamer"] = None,
-        logits_warper: Optional[LogitsProcessorList] = None,
+        streamer: "BaseStreamer",
+        logits_warper: LogitsProcessorList,
         **model_kwargs,
     ) -> Union[GenerateNonBeamOutput, torch.LongTensor]:
         r"""
@@ -2818,34 +2818,6 @@ def _contrastive_search(
         else:
             return input_ids
 
-    def _greedy_search(
-        self,
-        input_ids: torch.LongTensor,
-        logits_processor: LogitsProcessorList,
-        stopping_criteria: StoppingCriteriaList,
-        generation_config: GenerationConfig,
-        synced_gpus: bool,
-        streamer: Optional["BaseStreamer"],
-        **model_kwargs,
-    ) -> Union[GenerateNonBeamOutput, torch.LongTensor]:
-        r"""
-        Deprecated. Use `._sample()` instead, passing the same arguments.
-        """
-
-        logger.warning_once(
-            "Calling `._greedy_search()` directly is deprecated and will be removed in v4.42. Use `._sample()` "
-            "instead, passing the same arguments."
-        )
-        return self._sample(
-            input_ids=input_ids,
-            logits_processor=logits_processor,
-            stopping_criteria=stopping_criteria,
-            generation_config=generation_config,
-            synced_gpus=synced_gpus,
-            streamer=streamer,
-            **model_kwargs,
-        )
-
     def _sample(
         self,
         input_ids: torch.LongTensor,
@@ -2854,7 +2826,7 @@ def _sample(
         generation_config: GenerationConfig,
         synced_gpus: bool,
         streamer: Optional["BaseStreamer"],
-        logits_warper: Optional[LogitsProcessorList] = None,
+        logits_warper: LogitsProcessorList,
         **model_kwargs,
     ) -> Union[GenerateNonBeamOutput, torch.LongTensor]:
         r"""
@@ -3053,7 +3025,6 @@ def _temporary_reorder_cache(self, past_key_values, beam_idx):
             past_key_values.reorder_cache(beam_idx)
         return past_key_values
 
-    # TODO (joao, v4.42): remove default for `logits_warper`
     def _beam_search(
         self,
         input_ids: torch.LongTensor,
@@ -3062,7 +3033,7 @@ def _beam_search(
         stopping_criteria: StoppingCriteriaList,
         generation_config: GenerationConfig,
         synced_gpus: bool,
-        logits_warper: Optional[LogitsProcessorList] = None,
+        logits_warper: LogitsProcessorList,
         **model_kwargs,
     ) -> Union[GenerateBeamOutput, torch.LongTensor]:
         r"""
@@ -3342,36 +3313,6 @@ def _beam_search(
         else:
             return sequence_outputs["sequences"]
 
-    def _beam_sample(
-        self,
-        input_ids: torch.LongTensor,
-        beam_scorer: BeamScorer,
-        logits_processor: LogitsProcessorList,
-        stopping_criteria: StoppingCriteriaList,
-        logits_warper: LogitsProcessorList,
-        generation_config: GenerationConfig,
-        synced_gpus: bool,
-        **model_kwargs,
-    ) -> Union[GenerateBeamOutput, torch.LongTensor]:
-        r"""
-        Deprecated. Use `._beam_search()` instead, passing the same arguments.
-        """
-
-        logger.warning_once(
-            "Calling `._beam_sample()` directly is deprecated and will be removed in v4.42. Use `._beam_search()` "
-            "instead, passing the same arguments."
-        )
-        return self._beam_search(
-            input_ids=input_ids,
-            beam_scorer=beam_scorer,
-            logits_processor=logits_processor,
-            stopping_criteria=stopping_criteria,
-            logits_warper=logits_warper,
-            generation_config=generation_config,
-            synced_gpus=synced_gpus,
-            **model_kwargs,
-        )
-
     def _group_beam_search(
         self,
         input_ids: torch.LongTensor,

diff --git a/src/transformers/models/llava/configuration_llava.py b/src/transformers/models/llava/configuration_llava.py
@@ -13,8 +13,6 @@
 # limitations under the License.
 """Llava model configuration"""
 
-import warnings
-
 from ...configuration_utils import PretrainedConfig
 from ...utils import logging
 from ..auto import CONFIG_MAPPING
@@ -96,12 +94,6 @@ def __init__(
                 f"Got: {vision_feature_select_strategy}"
             )
 
-        if "vocab_size" in kwargs:
-            warnings.warn(
-                "The `vocab_size` argument is deprecated and will be removed in v4.42, since it can be inferred from the `text_config`. Passing this argument has no effect",
-                FutureWarning,
-            )
-
         self.vision_feature_select_strategy = vision_feature_select_strategy
         self.vision_feature_layer = vision_feature_layer