Generate: remove Marian hack (#25294)

Remove Marian hack

src/transformers/generation/tf_utils.py

@@ -474,27 +474,6 @@ class TFGenerationMixin:
             "A model class needs to define a `prepare_inputs_for_generation` method in order to use `generate`."
         )
 
-    def adjust_logits_during_generation(
-        self, logits, cur_len, max_length, forced_bos_token_id, forced_eos_token_id, **kwargs
-    ):
-        """
-        Implement in subclasses of [`PreTrainedModel`] for custom behavior to adjust the logits in the generate method.
-        """
-        vocab_size = getattr(self.config, "vocab_size", None)
-        if vocab_size is None and self.config.is_encoder_decoder:
-            decoder_config = getattr(self.config, "decoder", None)
-            if decoder_config is not None:
-                vocab_size = getattr(self.config.decoder, "vocab_size", None)
-
-        if cur_len == 1 and forced_bos_token_id is not None:
-            vocab_range = tf.constant(range(vocab_size))
-            return tf.where(vocab_range != forced_bos_token_id, -1e8, logits)
-        elif cur_len == max_length - 1 and forced_eos_token_id is not None:
-            vocab_range = tf.constant(range(vocab_size))
-            return tf.where(vocab_range != forced_eos_token_id, -1e8, logits)
-        else:
-            return logits
-
     def compute_transition_scores(
         self,
         sequences: tf.Tensor,

src/transformers/generation/utils.py

@@ -578,12 +578,6 @@ class GenerationMixin:
         inputs = self._maybe_initialize_input_ids_for_generation(inputs, bos_token_id, model_kwargs)
         return inputs, input_name, model_kwargs
 
-    def adjust_logits_during_generation(self, logits: torch.FloatTensor, **kwargs) -> torch.FloatTensor:
-        """
-        Implement in subclasses of [`PreTrainedModel`] for custom behavior to adjust the logits in the generate method.
-        """
-        return logits
-
     def _maybe_initialize_input_ids_for_generation(
         self,
         inputs: Optional[torch.Tensor] = None,
@@ -3060,9 +3054,6 @@ class GenerationMixin:
                 continue  # don't waste resources running the code we don't need
 
             next_token_logits = outputs.logits[:, -1, :]
-            # hack: adjust tokens for Marian. For Marian we have to make sure that the `pad_token_id`
-            # cannot be generated both before and after the `nn.functional.log_softmax` operation.
-            next_token_logits = self.adjust_logits_during_generation(next_token_logits, cur_len=cur_len)
             next_token_scores = nn.functional.log_softmax(
                 next_token_logits, dim=-1
             )  # (batch_size * num_beams, vocab_size)
@@ -3388,9 +3379,6 @@ class GenerationMixin:
 
             next_token_logits = outputs.logits[:, -1, :]
 
-            # hack: adjust tokens for Marian. For Marian we have to make sure that the `pad_token_id`
-            # cannot be generated both before and after the `nn.functional.log_softmax` operation.
-            next_token_logits = self.adjust_logits_during_generation(next_token_logits, cur_len=cur_len)
             next_token_scores = nn.functional.log_softmax(
                 next_token_logits, dim=-1
             )  # (batch_size * num_beams, vocab_size)
@@ -3751,9 +3739,6 @@ class GenerationMixin:
                 # select outputs of beams of current group only
                 next_token_logits = outputs.logits[batch_group_indices, -1, :]
 
-                # hack: adjust tokens for Marian. For Marian we have to make sure that the `pad_token_id`
-                # cannot be generated both before and after the `nn.functional.log_softmax` operation.
-                next_token_logits = self.adjust_logits_during_generation(next_token_logits, cur_len=cur_len)
                 next_token_scores = nn.functional.log_softmax(
                     next_token_logits, dim=-1
                 )  # (batch_size * group_size, vocab_size)
@@ -4110,9 +4095,6 @@ class GenerationMixin:
                 continue  # don't waste resources running the code we don't need
 
             next_token_logits = outputs.logits[:, -1, :]
-            # hack: adjust tokens for Marian. For Marian we have to make sure that the `pad_token_id`
-            # cannot be generated both before and after the `nn.functional.log_softmax` operation.
-            next_token_logits = self.adjust_logits_during_generation(next_token_logits, cur_len=cur_len)
             next_token_scores = nn.functional.log_softmax(
                 next_token_logits, dim=-1
             )  # (batch_size * num_beams, vocab_size)

src/transformers/models/marian/modeling_marian.py

@@ -1524,10 +1524,6 @@ class MarianMTModel(MarianPreTrainedModel):
     def prepare_decoder_input_ids_from_labels(self, labels: torch.Tensor):
         return shift_tokens_right(labels, self.config.pad_token_id, self.config.decoder_start_token_id)
 
-    def adjust_logits_during_generation(self, logits, cur_len):
-        logits[:, self.config.pad_token_id] = float("-inf")  # never predict pad token.
-        return logits
-
     @staticmethod
     def _reorder_cache(past_key_values, beam_idx):
         reordered_past = ()

src/transformers/models/marian/modeling_tf_marian.py

@@ -1443,18 +1443,3 @@ class TFMarianMTModel(TFMarianPreTrainedModel, TFCausalLanguageModelingLoss):
 
     def prepare_decoder_input_ids_from_labels(self, labels: tf.Tensor):
         return shift_tokens_right(labels, self.config.pad_token_id, self.config.decoder_start_token_id)
-
-    def adjust_logits_during_generation(
-        self, logits, cur_len, max_length, forced_bos_token_id, forced_eos_token_id, **kwargs
-    ):
-        """Never predict pad_token_id. Predict </s> when max_length is reached."""
-        vocab_range = tf.constant(range(self.config.vocab_size))
-        logits = tf.where(vocab_range == self.config.pad_token_id, LARGE_NEGATIVE, logits)
-        if cur_len == 1 and forced_bos_token_id is not None:
-            vocab_range = tf.constant(range(self.config.vocab_size))
-            return tf.where(vocab_range != forced_bos_token_id, LARGE_NEGATIVE, logits)
-        elif cur_len == max_length - 1 and forced_eos_token_id is not None:
-            vocab_range = tf.constant(range(self.config.vocab_size))
-            return tf.where(vocab_range != forced_eos_token_id, LARGE_NEGATIVE, logits)
-        else:
-            return logits