Generate: consistently handle special tokens as tensors (#30624)

* tmp commit

* [test_all] mvp

* missing not

* [test_all] final test fixes

* fix musicgen_melody and rag

* [test_all] empty commit

* PR comments

* Update src/transformers/generation/utils.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

src/transformers/generation/beam_search.py

@@ -218,8 +218,8 @@ class BeamSearchScorer(BeamScorer):
         next_scores: torch.FloatTensor,
         next_tokens: torch.LongTensor,
         next_indices: torch.LongTensor,
-        pad_token_id: Optional[int] = None,
-        eos_token_id: Optional[Union[int, List[int]]] = None,
+        pad_token_id: Optional[Union[int, torch.Tensor]] = None,
+        eos_token_id: Optional[Union[int, List[int], torch.Tensor]] = None,
         beam_indices: Optional[torch.LongTensor] = None,
         group_index: Optional[int] = 0,
         decoder_prompt_len: Optional[int] = 0,
@@ -245,8 +245,10 @@ class BeamSearchScorer(BeamScorer):
         next_beam_tokens = torch.zeros((batch_size, self.group_size), dtype=next_tokens.dtype, device=device)
         next_beam_indices = torch.zeros((batch_size, self.group_size), dtype=next_indices.dtype, device=device)
 
+        if eos_token_id is not None and not isinstance(eos_token_id, torch.Tensor):
             if isinstance(eos_token_id, int):
                 eos_token_id = [eos_token_id]
+            eos_token_id = torch.tensor(eos_token_id)
 
         for batch_idx in range(batch_size):
             batch_group_idx = batch_idx * self.num_beam_groups + group_index
@@ -322,15 +324,17 @@ class BeamSearchScorer(BeamScorer):
         final_beam_tokens: torch.LongTensor,
         final_beam_indices: torch.LongTensor,
         max_length: int,
-        pad_token_id: Optional[int] = None,
-        eos_token_id: Optional[Union[int, List[int]]] = None,
+        pad_token_id: Optional[Union[int, torch.Tensor]] = None,
+        eos_token_id: Optional[Union[int, List[int], torch.Tensor]] = None,
         beam_indices: Optional[torch.LongTensor] = None,
         decoder_prompt_len: Optional[int] = 0,
     ) -> Tuple[torch.LongTensor]:
         batch_size = len(self._beam_hyps) // self.num_beam_groups
 
+        if eos_token_id is not None and not isinstance(eos_token_id, torch.Tensor):
             if isinstance(eos_token_id, int):
                 eos_token_id = [eos_token_id]
+            eos_token_id = torch.tensor(eos_token_id)
 
         # finalize all open beam hypotheses and add to generated hypotheses
         for batch_group_idx, beam_hyp in enumerate(self._beam_hyps):
@@ -513,8 +517,8 @@ class ConstrainedBeamSearchScorer(BeamScorer):
         next_tokens: torch.LongTensor,
         next_indices: torch.LongTensor,
         scores_for_all_vocab: torch.FloatTensor,
-        pad_token_id: Optional[int] = None,
-        eos_token_id: Optional[Union[int, List[int]]] = None,
+        pad_token_id: Optional[Union[int, torch.Tensor]] = None,
+        eos_token_id: Optional[Union[int, List[int], torch.Tensor]] = None,
         beam_indices: Optional[torch.LongTensor] = None,
         decoder_prompt_len: Optional[int] = 0,
     ) -> Tuple[torch.Tensor]:
@@ -578,8 +582,10 @@ class ConstrainedBeamSearchScorer(BeamScorer):
         next_beam_tokens = torch.zeros((batch_size, self.group_size), dtype=next_tokens.dtype, device=device)
         next_beam_indices = torch.zeros((batch_size, self.group_size), dtype=next_indices.dtype, device=device)
 
+        if eos_token_id is not None and not isinstance(eos_token_id, torch.Tensor):
             if isinstance(eos_token_id, int):
                 eos_token_id = [eos_token_id]
+            eos_token_id = torch.tensor(eos_token_id)
 
         for batch_idx, beam_hyp in enumerate(self._beam_hyps):
             if self._done[batch_idx]:
@@ -811,15 +817,17 @@ class ConstrainedBeamSearchScorer(BeamScorer):
         final_beam_tokens: torch.LongTensor,
         final_beam_indices: torch.LongTensor,
         max_length: int,
-        pad_token_id: Optional[int] = None,
-        eos_token_id: Optional[Union[int, List[int]]] = None,
+        pad_token_id: Optional[Union[int, torch.Tensor]] = None,
+        eos_token_id: Optional[Union[int, List[int], torch.Tensor]] = None,
         beam_indices: Optional[torch.LongTensor] = None,
         decoder_prompt_len: Optional[int] = 0,
     ) -> Tuple[torch.LongTensor]:
         batch_size = len(self._beam_hyps)
 
+        if eos_token_id is not None and not isinstance(eos_token_id, torch.Tensor):
             if isinstance(eos_token_id, int):
                 eos_token_id = [eos_token_id]
+            eos_token_id = torch.tensor(eos_token_id)
 
         # finalize all open beam hypotheses and add to generated hypotheses
         for batch_idx, beam_hyp in enumerate(self._beam_hyps):

src/transformers/generation/logits_process.py

@@ -108,8 +108,8 @@ class MinLengthLogitsProcessor(LogitsProcessor):
     Args:
         min_length (`int`):
             The minimum length below which the score of `eos_token_id` is set to `-float("Inf")`.
-        eos_token_id (`Union[int, List[int]]`):
-            The id of the *end-of-sequence* token. Optionally, use a list to set multiple *end-of-sequence* tokens.
+        eos_token_id (`Union[int, List[int], torch.Tensor]`):
+            The id(s) of the *end-of-sequence* token.
 
     Examples:
 
@@ -137,14 +137,14 @@ class MinLengthLogitsProcessor(LogitsProcessor):
     ```
     """
 
-    def __init__(self, min_length: int, eos_token_id: Union[int, List[int]]):
+    def __init__(self, min_length: int, eos_token_id: Union[int, List[int], torch.Tensor]):
         if not isinstance(min_length, int) or min_length < 0:
             raise ValueError(f"`min_length` has to be a non-negative integer, but is {min_length}")
 
+        if not isinstance(eos_token_id, torch.Tensor):
             if isinstance(eos_token_id, int):
                 eos_token_id = [eos_token_id]
-        if not all(isinstance(i, int) for i in eos_token_id) or any(i < 0 for i in eos_token_id):
-            logger.warning(f"`eos_token_id` has to be a list of positive integers, but is {eos_token_id}")
+            eos_token_id = torch.tensor(eos_token_id)
 
         self.min_length = min_length
         self.eos_token_id = eos_token_id
@@ -152,8 +152,8 @@ class MinLengthLogitsProcessor(LogitsProcessor):
     @add_start_docstrings(LOGITS_PROCESSOR_INPUTS_DOCSTRING)
     def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor:
         vocab_tensor = torch.arange(scores.shape[-1], device=scores.device)
-        eos_token_id = torch.tensor(self.eos_token_id, device=scores.device)
-        eos_token_mask = torch.isin(vocab_tensor, eos_token_id)
+        self.eos_token_id = self.eos_token_id.to(scores.device)
+        eos_token_mask = torch.isin(vocab_tensor, self.eos_token_id)
         scores_processed = scores.clone()
         if input_ids.shape[-1] < self.min_length:
             scores_processed = torch.where(eos_token_mask, -math.inf, scores)
@@ -171,8 +171,8 @@ class MinNewTokensLengthLogitsProcessor(LogitsProcessor):
             input length.
         min_new_tokens (`int`):
             The minimum *new* tokens length below which the score of `eos_token_id` is set to `-float("Inf")`.
-        eos_token_id (`Union[int, List[int]]`):
-            The id of the *end-of-sequence* token. Optionally, use a list to set multiple *end-of-sequence* tokens.
+        eos_token_id (`Union[int, List[int], torch.Tensor]`):
+            The id(s) of the *end-of-sequence* token.
 
     Examples:
 
@@ -195,7 +195,9 @@ class MinNewTokensLengthLogitsProcessor(LogitsProcessor):
     ```
     """
 
-    def __init__(self, prompt_length_to_skip: int, min_new_tokens: int, eos_token_id: Union[int, List[int]]):
+    def __init__(
+        self, prompt_length_to_skip: int, min_new_tokens: int, eos_token_id: Union[int, List[int], torch.Tensor]
+    ):
         for arg_name, arg_value in [
             ("prompt_length_to_skip", prompt_length_to_skip),
             ("min_new_tokens", min_new_tokens),
@@ -203,10 +205,10 @@ class MinNewTokensLengthLogitsProcessor(LogitsProcessor):
             if not isinstance(arg_value, int) or arg_value < 0:
                 raise ValueError(f"`{arg_name}` has to be a positive integer, but is {arg_value}")
 
+        if not isinstance(eos_token_id, torch.Tensor):
             if isinstance(eos_token_id, int):
                 eos_token_id = [eos_token_id]
-        if not all(isinstance(i, int) for i in eos_token_id) or any(i < 0 for i in eos_token_id):
-            logger.warning(f"`eos_token_id` has to be a list of positive integers, but is {eos_token_id}")
+            eos_token_id = torch.tensor(eos_token_id)
 
         self.prompt_length_to_skip = prompt_length_to_skip
         self.min_new_tokens = min_new_tokens
@@ -217,8 +219,8 @@ class MinNewTokensLengthLogitsProcessor(LogitsProcessor):
         new_tokens_length = input_ids.shape[-1] - self.prompt_length_to_skip
         scores_processed = scores.clone()
         vocab_tensor = torch.arange(scores.shape[-1], device=scores.device)
-        eos_token_id = torch.tensor(self.eos_token_id, device=scores.device)
-        eos_token_mask = torch.isin(vocab_tensor, eos_token_id)
+        self.eos_token_id = self.eos_token_id.to(scores.device)
+        eos_token_mask = torch.isin(vocab_tensor, self.eos_token_id)
         if new_tokens_length < self.min_new_tokens:
             scores_processed = torch.where(eos_token_mask, -math.inf, scores)
 
@@ -1195,8 +1197,8 @@ class NoBadWordsLogitsProcessor(SequenceBiasLogitsProcessor):
     Args:
         bad_words_ids (`List[List[int]]`):
             List of list of token ids that are not allowed to be generated.
-        eos_token_id (`Union[int, List[int]]`):
-            The id of the *end-of-sequence* token. Optionally, use a list to set multiple *end-of-sequence* tokens.
+        eos_token_id (`Union[int, List[int], torch.Tensor]`, *optional*):
+            The id(s) of the *end-of-sequence* token.
 
     Examples:
 
@@ -1233,15 +1235,19 @@ class NoBadWordsLogitsProcessor(SequenceBiasLogitsProcessor):
     ```
     """
 
-    def __init__(self, bad_words_ids: List[List[int]], eos_token_id: Union[int, List[int]]):
+    def __init__(
+        self, bad_words_ids: List[List[int]], eos_token_id: Optional[Union[int, List[int], torch.Tensor]] = None
+    ):
         self.bad_word_ids = bad_words_ids
         self._validate_arguments()
 
         # Filter EOS token from bad_words_ids
-        if eos_token_id is None:
-            eos_token_id = []
+        if eos_token_id is not None:
+            if not isinstance(eos_token_id, torch.Tensor):
                 if isinstance(eos_token_id, int):
                     eos_token_id = [eos_token_id]
+                eos_token_id = torch.tensor(eos_token_id)
+
             bad_words_ids = list(
                 filter(lambda bad_token_seq: all(bad_token_seq != [i] for i in eos_token_id), bad_words_ids)
             )
@@ -1522,9 +1528,8 @@ class ForcedEOSTokenLogitsProcessor(LogitsProcessor):
     Args:
         max_length (`int`):
             The maximum length of the sequence to be generated.
-        eos_token_id (`Union[int, List[int]]`):
-            The id of the token to force as the last generated token when `max_length` is reached. Optionally, use a
-            list to set multiple *end-of-sequence* tokens.
+        eos_token_id (`Union[int, List[int], torch.Tensor]`):
+            The id(s) of the *end-of-sequence* token.
 
     Examples:
 
@@ -1548,15 +1553,22 @@ class ForcedEOSTokenLogitsProcessor(LogitsProcessor):
     ```
     """
 
-    def __init__(self, max_length: int, eos_token_id: Union[int, List[int]]):
+    def __init__(self, max_length: int, eos_token_id: Union[int, List[int], torch.Tensor]):
         self.max_length = max_length
+
+        if not isinstance(eos_token_id, torch.Tensor):
             if isinstance(eos_token_id, int):
                 eos_token_id = [eos_token_id]
+            eos_token_id = torch.tensor(eos_token_id)
         self.eos_token_id = eos_token_id
 
+        if torch.is_floating_point(eos_token_id) or (eos_token_id < 0).any():
+            raise ValueError(f"`eos_token_id` has to be a list of positive integers, but is {eos_token_id}")
+
     @add_start_docstrings(LOGITS_PROCESSOR_INPUTS_DOCSTRING)
     def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor:
         cur_len = input_ids.shape[-1]
+        self.eos_token_id = self.eos_token_id.to(scores.device)
         scores_processed = scores
         if cur_len == self.max_length - 1:
             scores_processed = torch.full_like(scores, -math.inf)
@@ -1595,8 +1607,8 @@ class ExponentialDecayLengthPenalty(LogitsProcessor):
         exponential_decay_length_penalty (`tuple(int, float)`):
             This tuple shall consist of: `(start_index, decay_factor)` where `start_index` indicates where penalty
             starts and `decay_factor` represents the factor of exponential decay
-        eos_token_id (`Union[int, List[int]]`):
-            The id of the *end-of-sequence* token. Optionally, use a list to set multiple *end-of-sequence* tokens.
+        eos_token_id (`Union[int, List[int], torch.Tensor]`):
+            The id(s) of the *end-of-sequence* token.
         input_ids_seq_length (`int`):
             The length of the input sequence.
 
@@ -1656,26 +1668,32 @@ class ExponentialDecayLengthPenalty(LogitsProcessor):
     def __init__(
         self,
         exponential_decay_length_penalty: Tuple[int, float],
-        eos_token_id: Union[int, List[int]],
+        eos_token_id: Union[int, List[int], torch.Tensor],
         input_ids_seq_length: int,
     ):
         self.regulation_start = exponential_decay_length_penalty[0] + input_ids_seq_length
         self.regulation_factor = exponential_decay_length_penalty[1]
+
+        if not isinstance(eos_token_id, torch.Tensor):
             if isinstance(eos_token_id, int):
                 eos_token_id = [eos_token_id]
+            eos_token_id = torch.tensor(eos_token_id)
         self.eos_token_id = eos_token_id
 
+        if torch.is_floating_point(eos_token_id) or (eos_token_id < 0).any():
+            raise ValueError(f"`eos_token_id` has to be a list of positive integers, but is {eos_token_id}")
+
     @add_start_docstrings(LOGITS_PROCESSOR_INPUTS_DOCSTRING)
     def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor:
         cur_len = input_ids.shape[-1]
+        self.eos_token_id = self.eos_token_id.to(scores.device)
         penalties = torch.zeros_like(scores)
         scores_processed = scores
         if cur_len > self.regulation_start:
-            for i in self.eos_token_id:
             penalty_idx = cur_len - self.regulation_start
             # To support negative logits we compute the penalty of the absolute value and add to the original logit
-                penalty = torch.abs(scores[:, i]) * (pow(self.regulation_factor, penalty_idx) - 1)
-                penalties[:, i] = penalty
+            penalty = torch.abs(scores[:, self.eos_token_id]) * (pow(self.regulation_factor, penalty_idx) - 1)
+            penalties[:, self.eos_token_id] = penalty
             scores_processed = scores + penalties
         return scores_processed
 
@@ -1753,7 +1771,7 @@ class SuppressTokensAtBeginLogitsProcessor(LogitsProcessor):
     """
 
     def __init__(self, begin_suppress_tokens, begin_index):
-        self.begin_suppress_tokens = list(begin_suppress_tokens)
+        self.begin_suppress_tokens = torch.tensor(list(begin_suppress_tokens))
         self.begin_index = begin_index
 
     def set_begin_index(self, begin_index):
@@ -1762,8 +1780,8 @@ class SuppressTokensAtBeginLogitsProcessor(LogitsProcessor):
     @add_start_docstrings(LOGITS_PROCESSOR_INPUTS_DOCSTRING)
     def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor:
         vocab_tensor = torch.arange(scores.shape[-1], device=scores.device)
-        begin_suppress_tokens = torch.tensor(self.begin_suppress_tokens, device=scores.device)
-        suppress_token_mask = torch.isin(vocab_tensor, begin_suppress_tokens)
+        self.begin_suppress_tokens = self.begin_suppress_tokens.to(scores.device)
+        suppress_token_mask = torch.isin(vocab_tensor, self.begin_suppress_tokens)
         scores_processed = scores
         if input_ids.shape[-1] == self.begin_index:
             scores_processed = torch.where(suppress_token_mask, -float("inf"), scores)
@@ -1801,13 +1819,13 @@ class SuppressTokensLogitsProcessor(LogitsProcessor):
     """
 
     def __init__(self, suppress_tokens):
-        self.suppress_tokens = list(suppress_tokens)
+        self.suppress_tokens = torch.tensor(list(suppress_tokens))
 
     @add_start_docstrings(LOGITS_PROCESSOR_INPUTS_DOCSTRING)
     def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor:
         vocab_tensor = torch.arange(scores.shape[-1], device=scores.device)
-        suppress_tokens = torch.tensor(self.suppress_tokens, device=scores.device)
-        suppress_token_mask = torch.isin(vocab_tensor, suppress_tokens)
+        self.suppress_tokens = self.suppress_tokens.to(scores.device)
+        suppress_token_mask = torch.isin(vocab_tensor, self.suppress_tokens)
         scores = torch.where(suppress_token_mask, -float("inf"), scores)
         return scores
 
@@ -2268,16 +2286,22 @@ class BarkEosPrioritizerLogitsProcessor(LogitsProcessor):
     </Tip>
 
     Args:
-        eos_token_id (`Union[int, List[int]]`):
-            The id of the *end-of-sequence* token. Optionally, use a list to set multiple *end-of-sequence* tokens.
+        eos_token_id (`Union[int, List[int], torch.Tensor]`):
+            The id(s) of the *end-of-sequence* token.
         min_eos_p (`float`, *optional*):
             Minimum end of speech threshold.
     """
 
-    def __init__(self, eos_token_id: Union[int, List[int]], min_eos_p: float):
+    def __init__(self, eos_token_id: Union[int, List[int], torch.Tensor], min_eos_p: float):
+        if not isinstance(eos_token_id, torch.Tensor):
             if isinstance(eos_token_id, int):
                 eos_token_id = [eos_token_id]
+            eos_token_id = torch.tensor(eos_token_id)
         self.eos_token_id = eos_token_id
+
+        if torch.is_floating_point(eos_token_id) or (eos_token_id < 0).any():
+            raise ValueError(f"`eos_token_id` has to be a list of positive integers, but is {eos_token_id}")
+
         if min_eos_p is not None and min_eos_p <= 0:
             raise ValueError(f"`min_eos_p` has to be a positive float, but is {min_eos_p}")
         self.min_eos_p = min_eos_p
@@ -2285,6 +2309,7 @@ class BarkEosPrioritizerLogitsProcessor(LogitsProcessor):
     @add_start_docstrings(LOGITS_PROCESSOR_INPUTS_DOCSTRING)
     def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor) -> torch.FloatTensor:
         scores_processed = scores
+        self.eos_token_id = self.eos_token_id.to(scores.device)
         if self.min_eos_p:
             probs = torch.nn.functional.softmax(scores.float(), dim=-1)
             # create scores full of -inf except for the eos_token_id

src/transformers/generation/stopping_criteria.py

@@ -470,29 +470,32 @@ class EosTokenCriteria(StoppingCriteria):
     By default, it uses the `model.generation_config.eos_token_id`.
 
     Args:
-        eos_token_id (`Union[int, List[int]]`):
-            The id of the *end-of-sequence* token. Optionally, use a list to set multiple *end-of-sequence* tokens.
+        eos_token_id (`Union[int, List[int], torch.Tensor]`):
+            The id(s) of the *end-of-sequence* token.
     """
 
-    def __init__(self, eos_token_id: Union[int, List[int]]):
+    def __init__(self, eos_token_id: Union[int, List[int], torch.Tensor]):
+        if not isinstance(eos_token_id, torch.Tensor):
             if isinstance(eos_token_id, int):
                 eos_token_id = [eos_token_id]
-        self.eos_token_id = torch.tensor(eos_token_id)
+            eos_token_id = torch.tensor(eos_token_id)
+        self.eos_token_id = eos_token_id
 
     @add_start_docstrings(STOPPING_CRITERIA_INPUTS_DOCSTRING)
     def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> torch.BoolTensor:
+        self.eos_token_id = self.eos_token_id.to(input_ids.device)
         if input_ids.device.type == "mps":
             # https://github.com/pytorch/pytorch/issues/77764#issuecomment-2067838075
             is_done = (
                 input_ids[:, -1]
                 .tile(self.eos_token_id.shape[0], 1)
-                .eq(self.eos_token_id.unsqueeze(1).to(input_ids.device))
+                .eq(self.eos_token_id.unsqueeze(1))
                 .sum(dim=0)
                 .bool()
                 .squeeze()
             )
         else:
-            is_done = torch.isin(input_ids[:, -1], self.eos_token_id.to(input_ids.device))
+            is_done = torch.isin(input_ids[:, -1], self.eos_token_id)
         return is_done
 
 

src/transformers/generation/utils.py

@@ -353,7 +353,7 @@ class GenerationMixin:
     def _prepare_model_inputs(
         self,
         inputs: Optional[torch.Tensor] = None,
-        bos_token_id: Optional[int] = None,
+        bos_token_id: Optional[torch.Tensor] = None,
         model_kwargs: Optional[Dict[str, torch.Tensor]] = None,
     ) -> Tuple[torch.Tensor, Optional[str], Dict[str, torch.Tensor]]:
         """
@@ -417,7 +417,7 @@ class GenerationMixin:
     def _maybe_initialize_input_ids_for_generation(
         self,
         inputs: Optional[torch.Tensor] = None,
-        bos_token_id: Optional[int] = None,
+        bos_token_id: Optional[torch.Tensor] = None,
         model_kwargs: Optional[Dict[str, torch.Tensor]] = None,
     ) -> torch.LongTensor:
         """Initializes input ids for generation, if necessary."""
@@ -449,20 +449,37 @@ class GenerationMixin:
     def _prepare_attention_mask_for_generation(
         self,
         inputs: torch.Tensor,
-        pad_token_id: Optional[int],
-        eos_token_id: Optional[Union[int, List[int]]],
+        pad_token_id: Optional[torch.Tensor],
+        eos_token_id: Optional[torch.Tensor],
     ) -> torch.LongTensor:
+        # No information for attention mask inference -> return default attention mask
+        default_attention_mask = torch.ones(inputs.shape[:2], dtype=torch.long, device=inputs.device)
+        if pad_token_id is None:
+            return default_attention_mask
+
         is_input_ids = len(inputs.shape) == 2 and inputs.dtype in [torch.int, torch.long]
-        is_pad_token_in_inputs = (pad_token_id is not None) and (pad_token_id in inputs)
-        if isinstance(eos_token_id, int):
-            eos_token_id = [eos_token_id]
-        is_pad_token_not_equal_to_eos_token_id = (eos_token_id is None) or (pad_token_id not in eos_token_id)
-
-        # Check if input is input_ids and padded -> only then is attention_mask defined
-        if is_input_ids and is_pad_token_in_inputs and is_pad_token_not_equal_to_eos_token_id:
-            return inputs.ne(pad_token_id).long()
-        else:
-            return torch.ones(inputs.shape[:2], dtype=torch.long, device=inputs.device)
+        if not is_input_ids:
+            return default_attention_mask
+
+        # Otherwise we have may have information -> try to infer the attention mask
+        if inputs.device.type == "mps":
+            # mps does not support torch.isin (https://github.com/pytorch/pytorch/issues/77764)
+            raise ValueError(
+                "Can't infer missing attention mask on `mps` device. Please provide an `attention_mask` or use a different device."
+            )
+
+        is_pad_token_in_inputs = (pad_token_id is not None) and (
+            torch.isin(elements=inputs, test_elements=pad_token_id).any()
+        )
+        is_pad_token_not_equal_to_eos_token_id = (eos_token_id is None) or ~(
+            torch.isin(elements=eos_token_id, test_elements=pad_token_id).any()
+        )
+        can_infer_attention_mask = is_pad_token_in_inputs * is_pad_token_not_equal_to_eos_token_id
+        attention_mask_from_padding = inputs.ne(pad_token_id).long()
+        attention_mask = (
+            attention_mask_from_padding * can_infer_attention_mask + default_attention_mask * ~can_infer_attention_mask
+        )
+        return attention_mask
 
     def _prepare_encoder_decoder_kwargs_for_generation(
         self,
@@ -510,8 +527,7 @@ class GenerationMixin:
         batch_size: int,
         model_input_name: str,
         model_kwargs: Dict[str, torch.Tensor],
-        decoder_start_token_id: Union[int, List[int]] = None,
-        bos_token_id: int = None,
+        decoder_start_token_id: torch.Tensor,
         device: torch.device = None,
     ) -> Tuple[torch.LongTensor, Dict[str, torch.Tensor]]:
         """Prepares `decoder_input_ids` for generation with encoder-decoder models"""
@@ -524,25 +540,24 @@ class GenerationMixin:
         else:
             decoder_input_ids = None
 
-        # 2. Encoder-decoder models expect the `decoder_input_ids` to start with a special token. Let's ensure that.
-        decoder_start_token_id = self._get_decoder_start_token_id(decoder_start_token_id, bos_token_id)
+        # 2. `decoder_start_token_id` must have shape (batch_size, 1)
         if device is None:
             device = self.device
-        if isinstance(decoder_start_token_id, list):
-            if len(decoder_start_token_id) != batch_size:
+        if decoder_start_token_id.ndim == 1:
+            if decoder_start_token_id.shape[0] != batch_size:
                 raise ValueError(
-                    f"`decoder_start_token_id` expcted to have length {batch_size} but got {len(decoder_start_token_id)}"
+                    f"`decoder_start_token_id` expected to have length {batch_size} but got {decoder_start_token_id.shape[0]}"
                 )
-            decoder_input_ids_start = torch.tensor(decoder_start_token_id, dtype=torch.long, device=device)
-            decoder_input_ids_start = decoder_input_ids_start.view(-1, 1)
+            decoder_start_token_id = decoder_start_token_id.view(-1, 1)
         else:
-            decoder_input_ids_start = (
+            decoder_start_token_id = (
                 torch.ones((batch_size, 1), dtype=torch.long, device=device) * decoder_start_token_id
             )
 
+        # 3. Encoder-decoder models expect the `decoder_input_ids` to start with a special token. Let's ensure that.
         # no user input -> use decoder_start_token_id as decoder_input_ids
         if decoder_input_ids is None:
-            decoder_input_ids = decoder_input_ids_start
+            decoder_input_ids = decoder_start_token_id
         # exception: Donut checkpoints have task-specific decoder starts and don't expect a BOS token
         elif self.config.model_type == "vision-encoder-decoder" and "donut" in self.name_or_path.lower():
             pass
@@ -550,14 +565,8 @@ class GenerationMixin:
             pass
         # user input but doesn't start with decoder_start_token_id -> prepend decoder_start_token_id (and adjust
         # decoder_attention_mask if provided)
-        elif (
-            isinstance(decoder_start_token_id, int)
-            and (decoder_input_ids[:, 0] != decoder_start_token_id).all().item()
-        ) or (
-            isinstance(decoder_start_token_id, torch.Tensor)
-            and (decoder_input_ids[:, 0] != decoder_start_token_id[:, 0]).all().item()
-        ):
-            decoder_input_ids = torch.cat([decoder_input_ids_start, decoder_input_ids], dim=-1)
+        elif (decoder_input_ids[:, 0] != decoder_start_token_id[:, 0]).all().item():
+            decoder_input_ids = torch.cat([decoder_start_token_id, decoder_input_ids], dim=-1)
             if "decoder_attention_mask" in model_kwargs:
                 decoder_attention_mask = model_kwargs["decoder_attention_mask"]
                 decoder_attention_mask = torch.cat(
@@ -568,24 +577,6 @@ class GenerationMixin:
 
         return decoder_input_ids, model_kwargs
 
-    def _get_decoder_start_token_id(
-        self, decoder_start_token_id: Union[int, List[int]] = None, bos_token_id: int = None
-    ) -> int:
-        decoder_start_token_id = (
-            decoder_start_token_id
-            if decoder_start_token_id is not None
-            else self.generation_config.decoder_start_token_id
-        )
-        bos_token_id = bos_token_id if bos_token_id is not None else self.generation_config.bos_token_id
-
-        if decoder_start_token_id is not None:
-            return decoder_start_token_id
-        elif bos_token_id is not None:
-            return bos_token_id
-        raise ValueError(
-            "`decoder_start_token_id` or `bos_token_id` has to be defined for encoder-decoder generation."
-        )
-
     @staticmethod
     def _expand_inputs_for_generation(
         expand_size: int = 1,
@@ -729,6 +720,8 @@ class GenerationMixin:
         if generation_config.num_beams > 1:
             if isinstance(generation_config.eos_token_id, list):
                 min_tokens_to_keep = len(generation_config.eos_token_id) + 1
+            elif isinstance(generation_config.eos_token_id, torch.Tensor):
+                min_tokens_to_keep = generation_config.eos_token_id.shape[0] + 1
             else:
                 min_tokens_to_keep = 2
         else:
@@ -1346,6 +1339,61 @@ class GenerationMixin:
             self._static_cache.reset()  # reset the cache for a new generation
         return self._static_cache
 
+    def _prepare_special_tokens(
+        self, generation_config: GenerationConfig, kwargs_has_attention_mask: Optional[bool] = None
+    ):
+        """
+        Prepares the special tokens for generation, overwriting the generation config with their processed versions
+        converted to tensor.
+
+        Note that `generation_config` is changed in place and stops being serializable after this method is called.
+        That is no problem if called within `generate` (`generation_config` is a local copy that doesn't leave the
+        function). However, if called outside `generate`, consider creating a copy of `generation_config` first.
+        """
+
+        # Convert special tokens to tensors (if they exist)
+        def _tensor_or_none(token):
+            if token is None or isinstance(token, torch.Tensor):
+                return token
+            return torch.tensor(token, device=self.device, dtype=torch.long)
+
+        bos_token_id = _tensor_or_none(generation_config.bos_token_id)
+        eos_token_id = _tensor_or_none(generation_config.eos_token_id)
+        pad_token_id = _tensor_or_none(generation_config.pad_token_id)
+        decoder_start_token_id = _tensor_or_none(generation_config.decoder_start_token_id)
+        decoder_start_token_id = decoder_start_token_id if decoder_start_token_id is not None else bos_token_id
+
+        # We can have more than one eos token. Always treat it as a 1D tensor (when it exists).
+        if eos_token_id is not None and eos_token_id.ndim == 0:
+            eos_token_id = eos_token_id.unsqueeze(0)
+
+        # Set pad token if unset (and there are conditions to do so)
+        if pad_token_id is None and eos_token_id is not None:
+            if kwargs_has_attention_mask is not None and not kwargs_has_attention_mask:
+                logger.warning(
+                    "The attention mask and the pad token id were not set. As a consequence, you may observe "
+                    "unexpected behavior. Please pass your input's `attention_mask` to obtain reliable results."
+                )
+            pad_token_id = eos_token_id[0]
+            logger.warning(f"Setting `pad_token_id` to `eos_token_id`:{pad_token_id} for open-end generation.")
+
+        # Sanity checks/warnings
+        if self.config.is_encoder_decoder and decoder_start_token_id is None:
+            raise ValueError(
+                "`decoder_start_token_id` or `bos_token_id` has to be defined for encoder-decoder generation."
+            )
+        if eos_token_id is not None and (torch.is_floating_point(eos_token_id) or (eos_token_id < 0).any()):
+            logger.warning(
+                f"`eos_token_id` should consist of positive integers, but is {eos_token_id}. Your generation will not "
+                "stop until the maximum length is reached. Depending on other flags, it may even crash."
+            )
+
+        # Update generation config with the updated special tokens tensors
+        generation_config.bos_token_id = bos_token_id
+        generation_config.eos_token_id = eos_token_id
+        generation_config.pad_token_id = pad_token_id
+        generation_config.decoder_start_token_id = decoder_start_token_id
+
     @torch.no_grad()
     def generate(
         self,
@@ -1460,50 +1508,24 @@ class GenerationMixin:
         logits_processor = logits_processor if logits_processor is not None else LogitsProcessorList()
         stopping_criteria = stopping_criteria if stopping_criteria is not None else StoppingCriteriaList()
 
-        if generation_config.pad_token_id is None and generation_config.eos_token_id is not None:
-            if model_kwargs.get("attention_mask", None) is None:
-                logger.warning(
-                    "The attention mask and the pad token id were not set. As a consequence, you may observe "
-                    "unexpected behavior. Please pass your input's `attention_mask` to obtain reliable results."
-                )
-            eos_token_id = generation_config.eos_token_id
-            if isinstance(eos_token_id, list):
-                eos_token_id = eos_token_id[0]
-            logger.warning(f"Setting `pad_token_id` to `eos_token_id`:{eos_token_id} for open-end generation.")
-            generation_config.pad_token_id = eos_token_id
+        accepts_attention_mask = "attention_mask" in set(inspect.signature(self.forward).parameters.keys())
+        requires_attention_mask = "encoder_outputs" not in model_kwargs
+        kwargs_has_attention_mask = model_kwargs.get("attention_mask", None) is not None
+        self._prepare_special_tokens(generation_config, kwargs_has_attention_mask)
 
         # 3. Define model inputs
-        # inputs_tensor has to be defined
-        # model_input_name is defined if model-specific keyword input is passed
-        # otherwise model_input_name is None
-        # all model-specific keyword inputs are removed from `model_kwargs`
         inputs_tensor, model_input_name, model_kwargs = self._prepare_model_inputs(
             inputs, generation_config.bos_token_id, model_kwargs
         )
         batch_size = inputs_tensor.shape[0]
 
-        # 4. Define other model kwargs
-        # decoder-only models with inputs_embeds forwarding must use caching (otherwise we can't detect whether we are
-        # generating the first new token or not, and we only want to use the embeddings for the first new token)
-        if not self.config.is_encoder_decoder and model_input_name == "inputs_embeds":
-            model_kwargs["use_cache"] = True
-        else:
-            model_kwargs["use_cache"] = generation_config.use_cache
-
-        accepts_attention_mask = "attention_mask" in set(inspect.signature(self.forward).parameters.keys())
-        requires_attention_mask = "encoder_outputs" not in model_kwargs
-
-        if model_kwargs.get("attention_mask", None) is None and requires_attention_mask and accepts_attention_mask:
-            model_kwargs["attention_mask"] = self._prepare_attention_mask_for_generation(
-                inputs_tensor, generation_config.pad_token_id, generation_config.eos_token_id
-            )
-
-        # decoder-only models should use left-padding for generation
-        if not self.config.is_encoder_decoder:
+        # decoder-only models must use left-padding for batched generation.
+        if not self.config.is_encoder_decoder and not is_torchdynamo_compiling():
             # If `input_ids` was given, check if the last id in any sequence is `pad_token_id`
             # Note: If using, `inputs_embeds` this check does not work, because we want to be more hands-off.
             if (
                 generation_config.pad_token_id is not None
+                and batch_size > 1
                 and len(inputs_tensor.shape) == 2
                 and torch.sum(inputs_tensor[:, -1] == generation_config.pad_token_id) > 0
             ):
@@ -1512,9 +1534,21 @@ class GenerationMixin:
                     "generation results, please set `padding_side='left'` when initializing the tokenizer."
                 )
 
+        # 4. Define other model kwargs
+        # decoder-only models with inputs_embeds forwarding must use caching (otherwise we can't detect whether we are
+        # generating the first new token or not, and we only want to use the embeddings for the first new token)
+        if not self.config.is_encoder_decoder and model_input_name == "inputs_embeds":
+            model_kwargs["use_cache"] = True
+        else:
+            model_kwargs["use_cache"] = generation_config.use_cache
+
+        if not kwargs_has_attention_mask and requires_attention_mask and accepts_attention_mask:
+            model_kwargs["attention_mask"] = self._prepare_attention_mask_for_generation(
+                inputs_tensor, generation_config.pad_token_id, generation_config.eos_token_id
+            )
+
         if self.config.is_encoder_decoder and "encoder_outputs" not in model_kwargs:
-            # if model is encoder decoder encoder_outputs are created
-            # and added to `model_kwargs`
+            # if model is encoder decoder encoder_outputs are created and added to `model_kwargs`
             model_kwargs = self._prepare_encoder_decoder_kwargs_for_generation(
                 inputs_tensor, model_kwargs, model_input_name, generation_config
             )
@@ -1526,7 +1560,6 @@ class GenerationMixin:
                 model_input_name=model_input_name,
                 model_kwargs=model_kwargs,
                 decoder_start_token_id=generation_config.decoder_start_token_id,
-                bos_token_id=generation_config.bos_token_id,
                 device=inputs_tensor.device,
             )
         else:
@@ -2631,9 +2664,6 @@ class GenerationMixin:
         return_dict_in_generate = generation_config.return_dict_in_generate
         sequential = generation_config.low_memory
 
-        if isinstance(eos_token_id, int):
-            eos_token_id = [eos_token_id]
-
         batch_size = len(beam_scorer._beam_hyps)
         num_beams = beam_scorer.num_beams
 
@@ -2757,7 +2787,7 @@ class GenerationMixin:
             next_token_scores = next_token_scores.view(batch_size, num_beams * vocab_size)
 
             # Sample 1 + len(eos_token_id) next tokens for each beam so we have at least 1 non eos token per beam.
-            n_eos_tokens = len(eos_token_id) if eos_token_id else 0
+            n_eos_tokens = eos_token_id.shape[0] if eos_token_id is not None else 0
             next_token_scores, next_tokens = torch.topk(
                 next_token_scores, max(2, 1 + n_eos_tokens) * num_beams, dim=1, largest=True, sorted=True
             )
@@ -2901,9 +2931,6 @@ class GenerationMixin:
         output_logits = generation_config.output_logits
         return_dict_in_generate = generation_config.return_dict_in_generate
 
-        if isinstance(eos_token_id, int):
-            eos_token_id = [eos_token_id]
-
         batch_size = len(beam_scorer._beam_hyps)
         num_beams = beam_scorer.num_beams
 
@@ -3124,9 +3151,6 @@ class GenerationMixin:
         output_logits = generation_config.output_logits
         return_dict_in_generate = generation_config.return_dict_in_generate
 
-        if isinstance(eos_token_id, int):
-            eos_token_id = [eos_token_id]
-
         num_beams = beam_scorer.num_beams
         num_beam_groups = beam_scorer.num_beam_groups
         num_sub_beams = num_beams // num_beam_groups
@@ -3229,7 +3253,7 @@ class GenerationMixin:
                 next_token_scores = next_token_scores.view(batch_size, group_size * vocab_size)
 
                 # Sample 1 + len(eos_token_id) next tokens for each beam so we have at least 1 non eos token per beam.
-                n_eos_tokens = len(eos_token_id) if eos_token_id else 0
+                n_eos_tokens = eos_token_id.shape[0] if eos_token_id is not None else 0
                 next_token_scores, next_tokens = torch.topk(
                     next_token_scores, max(2, 1 + n_eos_tokens) * group_size, dim=1, largest=True, sorted=True
                 )
@@ -3409,9 +3433,6 @@ class GenerationMixin:
         output_logits = generation_config.output_logits
         return_dict_in_generate = generation_config.return_dict_in_generate
 
-        if isinstance(eos_token_id, int):
-            eos_token_id = [eos_token_id]
-
         batch_size = len(constrained_beam_scorer._beam_hyps)
         num_beams = constrained_beam_scorer.num_beams
 
@@ -3501,7 +3522,7 @@ class GenerationMixin:
             next_token_scores = next_token_scores.view(batch_size, num_beams * vocab_size)
 
             # Sample 1 + len(eos_token_id) next tokens for each beam so we have at least 1 non eos token per beam.
-            n_eos_tokens = len(eos_token_id) if eos_token_id else 0
+            n_eos_tokens = eos_token_id.shape[0] if eos_token_id is not None else 0
             next_token_scores, next_tokens = torch.topk(
                 next_token_scores, max(2, 1 + n_eos_tokens) * num_beams, dim=1, largest=True, sorted=True
             )

src/transformers/models/musicgen/modeling_musicgen.py

@@ -18,7 +18,7 @@ import inspect
 import math
 import random
 from dataclasses import dataclass
-from typing import TYPE_CHECKING, Any, Dict, Optional, Tuple, Union
+from typing import TYPE_CHECKING, Any, Dict, List, Optional, Tuple, Union
 
 import torch
 import torch.nn as nn
@@ -2587,6 +2587,24 @@ class MusicgenForConditionalGeneration(PreTrainedModel):
                 break
         return torch.ones((batch_size, 1), dtype=torch.long, device=self.device) * bos_token_id
 
+    def _get_decoder_start_token_id(
+        self, decoder_start_token_id: Union[int, List[int]] = None, bos_token_id: int = None
+    ) -> int:
+        decoder_start_token_id = (
+            decoder_start_token_id
+            if decoder_start_token_id is not None
+            else self.generation_config.decoder_start_token_id
+        )
+        bos_token_id = bos_token_id if bos_token_id is not None else self.generation_config.bos_token_id
+
+        if decoder_start_token_id is not None:
+            return decoder_start_token_id
+        elif bos_token_id is not None:
+            return bos_token_id
+        raise ValueError(
+            "`decoder_start_token_id` or `bos_token_id` has to be defined for encoder-decoder generation."
+        )
+
     @torch.no_grad()
     def generate(
         self,

src/transformers/models/musicgen_melody/modeling_musicgen_melody.py

@@ -18,7 +18,7 @@ import inspect
 import math
 import random
 from dataclasses import dataclass
-from typing import TYPE_CHECKING, Any, Dict, Optional, Tuple, Union
+from typing import TYPE_CHECKING, Any, Dict, List, Optional, Tuple, Union
 
 import torch
 import torch.nn as nn
@@ -2452,6 +2452,25 @@ class MusicgenMelodyForConditionalGeneration(PreTrainedModel):
             param.requires_grad = False
         self.text_encoder._requires_grad = False
 
+    # Copied from transformers.models.musicgen.modeling_musicgen.MusicgenForConditionalGeneration._get_decoder_start_token_id
+    def _get_decoder_start_token_id(
+        self, decoder_start_token_id: Union[int, List[int]] = None, bos_token_id: int = None
+    ) -> int:
+        decoder_start_token_id = (
+            decoder_start_token_id
+            if decoder_start_token_id is not None
+            else self.generation_config.decoder_start_token_id
+        )
+        bos_token_id = bos_token_id if bos_token_id is not None else self.generation_config.bos_token_id
+
+        if decoder_start_token_id is not None:
+            return decoder_start_token_id
+        elif bos_token_id is not None:
+            return bos_token_id
+        raise ValueError(
+            "`decoder_start_token_id` or `bos_token_id` has to be defined for encoder-decoder generation."
+        )
+
     @torch.no_grad()
     def generate(
         self,

src/transformers/models/rag/modeling_rag.py

@@ -1458,6 +1458,9 @@ class RagTokenForGeneration(RagPreTrainedModel):
         generation_config = copy.deepcopy(generation_config)
         model_kwargs = generation_config.update(**kwargs)  # All unused kwargs must be model kwargs
 
+        kwargs_has_attention_mask = model_kwargs.get("attention_mask", None) is not None
+        self._prepare_special_tokens(generation_config, kwargs_has_attention_mask)
+
         # set default parameters
         n_docs = n_docs if n_docs is not None else self.config.n_docs
 

tests/generation/test_utils.py

@@ -14,6 +14,7 @@
 # limitations under the License.
 
 
+import copy
 import inspect
 import tempfile
 import unittest
@@ -168,7 +169,9 @@ class GenerationTesterMixin:
         encoder_outputs["last_hidden_state"] = encoder_outputs.last_hidden_state.repeat_interleave(
             num_interleave, dim=0
         )
-        input_ids = torch.zeros_like(input_ids[:, :1]) + model._get_decoder_start_token_id()
+        generation_config = copy.deepcopy(model.generation_config)
+        model._prepare_special_tokens(generation_config)
+        input_ids = torch.zeros_like(input_ids[:, :1]) + generation_config.decoder_start_token_id
         attention_mask = None
         return encoder_outputs, input_ids, attention_mask
 

tests/models/seamless_m4t/test_modeling_seamless_m4t.py

@@ -414,9 +414,11 @@ class SeamlessM4TModelWithSpeechInputTest(ModelTesterMixin, unittest.TestCase):
         encoder_outputs["last_hidden_state"] = encoder_outputs.last_hidden_state.repeat_interleave(
             num_interleave, dim=0
         )
+        generation_config = copy.deepcopy(model.generation_config)
+        model._prepare_special_tokens(generation_config)
         input_ids = (
             torch.zeros(input_ids.shape[:2], dtype=torch.int64, layout=input_ids.layout, device=input_ids.device)
-            + model._get_decoder_start_token_id()
+            + generation_config.decoder_start_token_id
         )
         attention_mask = None
         return encoder_outputs, input_ids, attention_mask

tests/models/seamless_m4t_v2/test_modeling_seamless_m4t_v2.py

@@ -430,9 +430,11 @@ class SeamlessM4Tv2ModelWithSpeechInputTest(ModelTesterMixin, unittest.TestCase)
         encoder_outputs["last_hidden_state"] = encoder_outputs.last_hidden_state.repeat_interleave(
             num_interleave, dim=0
         )
+        generation_config = copy.deepcopy(model.generation_config)
+        model._prepare_special_tokens(generation_config)
         input_ids = (
             torch.zeros(input_ids.shape[:2], dtype=torch.int64, layout=input_ids.layout, device=input_ids.device)
-            + model._get_decoder_start_token_id()
+            + generation_config.decoder_start_token_id
         )
         attention_mask = None
         return encoder_outputs, input_ids, attention_mask

tests/models/speech_to_text/test_modeling_speech_to_text.py

@@ -645,7 +645,9 @@ class Speech2TextModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTest
             num_interleave, dim=0
         )
         input_ids = input_ids[:, :, 0]
-        input_ids = torch.zeros_like(input_ids[:, :1], dtype=torch.long) + model._get_decoder_start_token_id()
+        generation_config = copy.deepcopy(model.generation_config)
+        model._prepare_special_tokens(generation_config)
+        input_ids = torch.zeros_like(input_ids[:, :1]) + generation_config.decoder_start_token_id
         attention_mask = None
         return encoder_outputs, input_ids, attention_mask
 

tests/models/whisper/test_modeling_whisper.py

@@ -833,10 +833,10 @@ class WhisperModelTest(ModelTesterMixin, GenerationTesterMixin, PipelineTesterMi
         encoder_outputs["last_hidden_state"] = encoder_outputs.last_hidden_state.repeat_interleave(
             num_interleave, dim=0
         )
+        generation_config = copy.deepcopy(model.generation_config)
+        model._prepare_special_tokens(generation_config)
         input_ids = input_ids[:, :, 0]
-        input_ids = torch.zeros_like(input_ids[:, :1], dtype=torch.long) + torch.tensor(
-            [model._get_decoder_start_token_id()], device=input_ids.device
-        )
+        input_ids = torch.zeros_like(input_ids[:, :1], dtype=torch.long) + generation_config.decoder_start_token_id
         attention_mask = None
         return encoder_outputs, input_ids, attention_mask