Data Collator
Data collators are objects that will form a batch by using a list of dataset elements as input. These elements are of
the same type as the elements of train_dataset
or eval_dataset
.
To be able to build batches, data collators may apply some processing (like padding). Some of them (like DataCollatorForLanguageModeling) also apply some random data augmentation (like random masking) on the formed batch.
Examples of use can be found in the example scripts or example notebooks.
Default data collator
transformers.default_data_collator
< source >( features: typing.List[InputDataClass] return_tensors = 'pt' )
Very simple data collator that simply collates batches of dictlike objects and performs special handling for potential keys named:
label
: handles a single value (int or float) per objectlabel_ids
: handles a list of values per object
Does not do any additional preprocessing: property names of the input object will be used as corresponding inputs to the model. See glue and ner for example of how it’s useful.
DefaultDataCollator
class transformers.DefaultDataCollator
< source >( return_tensors: str = 'pt' )
Very simple data collator that simply collates batches of dictlike objects and performs special handling for potential keys named:
label
: handles a single value (int or float) per objectlabel_ids
: handles a list of values per object
Does not do any additional preprocessing: property names of the input object will be used as corresponding inputs to the model. See glue and ner for example of how it’s useful.
This is an object (like other data collators) rather than a pure function like default_data_collator. This can be helpful if you need to set a return_tensors value at initialization.
DataCollatorWithPadding
class transformers.DataCollatorWithPadding
< source >( tokenizer: PreTrainedTokenizerBase padding: typing.Union[bool, str, transformers.utils.generic.PaddingStrategy] = True max_length: typing.Optional[int] = None pad_to_multiple_of: typing.Optional[int] = None return_tensors: str = 'pt' )
Parameters
 tokenizer (PreTrainedTokenizer or PreTrainedTokenizerFast) — The tokenizer used for encoding the data.

padding (
bool
,str
or PaddingStrategy, optional, defaults toTrue
) — Select a strategy to pad the returned sequences (according to the model’s padding side and padding index) among:True
or'longest'
(default): Pad to the longest sequence in the batch (or no padding if only a single sequence is provided).'max_length'
: Pad to a maximum length specified with the argumentmax_length
or to the maximum acceptable input length for the model if that argument is not provided.False
or'do_not_pad'
: No padding (i.e., can output a batch with sequences of different lengths).

max_length (
int
, optional) — Maximum length of the returned list and optionally padding length (see above). 
pad_to_multiple_of (
int
, optional) — If set will pad the sequence to a multiple of the provided value.This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).

return_tensors (
str
) — The type of Tensor to return. Allowable values are “np”, “pt” and “tf”.
Data collator that will dynamically pad the inputs received.
DataCollatorForTokenClassification
class transformers.DataCollatorForTokenClassification
< source >( tokenizer: PreTrainedTokenizerBase padding: typing.Union[bool, str, transformers.utils.generic.PaddingStrategy] = True max_length: typing.Optional[int] = None pad_to_multiple_of: typing.Optional[int] = None label_pad_token_id: int = 100 return_tensors: str = 'pt' )
Parameters
 tokenizer (PreTrainedTokenizer or PreTrainedTokenizerFast) — The tokenizer used for encoding the data.

padding (
bool
,str
or PaddingStrategy, optional, defaults toTrue
) — Select a strategy to pad the returned sequences (according to the model’s padding side and padding index) among:True
or'longest'
: Pad to the longest sequence in the batch (or no padding if only a single sequence is provided).'max_length'
: Pad to a maximum length specified with the argumentmax_length
or to the maximum acceptable input length for the model if that argument is not provided.False
or'do_not_pad'
(default): No padding (i.e., can output a batch with sequences of different lengths).

max_length (
int
, optional) — Maximum length of the returned list and optionally padding length (see above). 
pad_to_multiple_of (
int
, optional) — If set will pad the sequence to a multiple of the provided value.This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).

label_pad_token_id (
int
, optional, defaults to 100) — The id to use when padding the labels (100 will be automatically ignore by PyTorch loss functions). 
return_tensors (
str
) — The type of Tensor to return. Allowable values are “np”, “pt” and “tf”.
Data collator that will dynamically pad the inputs received, as well as the labels.
DataCollatorForSeq2Seq
class transformers.DataCollatorForSeq2Seq
< source >( tokenizer: PreTrainedTokenizerBase model: typing.Optional[typing.Any] = None padding: typing.Union[bool, str, transformers.utils.generic.PaddingStrategy] = True max_length: typing.Optional[int] = None pad_to_multiple_of: typing.Optional[int] = None label_pad_token_id: int = 100 return_tensors: str = 'pt' )
Parameters
 tokenizer (PreTrainedTokenizer or PreTrainedTokenizerFast) — The tokenizer used for encoding the data.

model (PreTrainedModel) —
The model that is being trained. If set and has the prepare_decoder_input_ids_from_labels, use it to
prepare the decoder_input_ids
This is useful when using label_smoothing to avoid calculating loss twice.

padding (
bool
,str
or PaddingStrategy, optional, defaults toTrue
) — Select a strategy to pad the returned sequences (according to the model’s padding side and padding index) among:True
or'longest'
: Pad to the longest sequence in the batch (or no padding if only a single sequence is provided).'max_length'
: Pad to a maximum length specified with the argumentmax_length
or to the maximum acceptable input length for the model if that argument is not provided.False
or'do_not_pad'
(default): No padding (i.e., can output a batch with sequences of different lengths).

max_length (
int
, optional) — Maximum length of the returned list and optionally padding length (see above). 
pad_to_multiple_of (
int
, optional) — If set will pad the sequence to a multiple of the provided value.This is especially useful to enable the use of Tensor Cores on NVIDIA hardware with compute capability >= 7.5 (Volta).

label_pad_token_id (
int
, optional, defaults to 100) — The id to use when padding the labels (100 will be automatically ignored by PyTorch loss functions). 
return_tensors (
str
) — The type of Tensor to return. Allowable values are “np”, “pt” and “tf”.
Data collator that will dynamically pad the inputs received, as well as the labels.
DataCollatorForLanguageModeling
class transformers.DataCollatorForLanguageModeling
< source >( tokenizer: PreTrainedTokenizerBase mlm: bool = True mlm_probability: float = 0.15 pad_to_multiple_of: typing.Optional[int] = None tf_experimental_compile: bool = False return_tensors: str = 'pt' )
Parameters
 tokenizer (PreTrainedTokenizer or PreTrainedTokenizerFast) — The tokenizer used for encoding the data.

mlm (
bool
, optional, defaults toTrue
) — Whether or not to use masked language modeling. If set toFalse
, the labels are the same as the inputs with the padding tokens ignored (by setting them to 100). Otherwise, the labels are 100 for nonmasked tokens and the value to predict for the masked token. 
mlm_probability (
float
, optional, defaults to 0.15) — The probability with which to (randomly) mask tokens in the input, whenmlm
is set toTrue
. 
pad_to_multiple_of (
int
, optional) — If set will pad the sequence to a multiple of the provided value. 
return_tensors (
str
) — The type of Tensor to return. Allowable values are “np”, “pt” and “tf”.
Data collator used for language modeling. Inputs are dynamically padded to the maximum length of a batch if they are not all of the same length.
For best performance, this data collator should be used with a dataset having items that are dictionaries or
BatchEncoding, with the "special_tokens_mask"
key, as returned by a PreTrainedTokenizer or a
PreTrainedTokenizerFast with the argument return_special_tokens_mask=True
.
numpy_mask_tokens
< source >( inputs: typing.Any special_tokens_mask: typing.Optional[typing.Any] = None )
Prepare masked tokens inputs/labels for masked language modeling: 80% MASK, 10% random, 10% original.
tf_mask_tokens
< source >( inputs: typing.Any vocab_size mask_token_id special_tokens_mask: typing.Optional[typing.Any] = None )
Prepare masked tokens inputs/labels for masked language modeling: 80% MASK, 10% random, 10% original.
torch_mask_tokens
< source >( inputs: typing.Any special_tokens_mask: typing.Optional[typing.Any] = None )
Prepare masked tokens inputs/labels for masked language modeling: 80% MASK, 10% random, 10% original.
DataCollatorForWholeWordMask
class transformers.DataCollatorForWholeWordMask
< source >( tokenizer: PreTrainedTokenizerBase mlm: bool = True mlm_probability: float = 0.15 pad_to_multiple_of: typing.Optional[int] = None tf_experimental_compile: bool = False return_tensors: str = 'pt' )
Data collator used for language modeling that masks entire words.
 collates batches of tensors, honoring their tokenizer’s pad_token
 preprocesses batches for masked language modeling
This collator relies on details of the implementation of subword tokenization by BertTokenizer, specifically
that subword tokens are prefixed with ##. For tokenizers that do not adhere to this scheme, this collator will
produce an output that is roughly equivalent to .DataCollatorForLanguageModeling
.
Prepare masked tokens inputs/labels for masked language modeling: 80% MASK, 10% random, 10% original. Set ‘mask_labels’ means we use whole word mask (wwm), we directly mask idxs according to it’s ref.
Prepare masked tokens inputs/labels for masked language modeling: 80% MASK, 10% random, 10% original. Set ‘mask_labels’ means we use whole word mask (wwm), we directly mask idxs according to it’s ref.
Prepare masked tokens inputs/labels for masked language modeling: 80% MASK, 10% random, 10% original. Set ‘mask_labels’ means we use whole word mask (wwm), we directly mask idxs according to it’s ref.
DataCollatorForPermutationLanguageModeling
class transformers.DataCollatorForPermutationLanguageModeling
< source >( tokenizer: PreTrainedTokenizerBase plm_probability: float = 0.16666666666666666 max_span_length: int = 5 return_tensors: str = 'pt' )
Data collator used for permutation language modeling.
 collates batches of tensors, honoring their tokenizer’s pad_token
 preprocesses batches for permutation language modeling with procedures specific to XLNet
The masked tokens to be predicted for a particular sequence are determined by the following algorithm:
 Start from the beginning of the sequence by setting
cur_len = 0
(number of tokens processed so far).  Sample a
span_length
from the interval[1, max_span_length]
(length of span of tokens to be masked)  Reserve a context of length
context_length = span_length / plm_probability
to surround span to be masked  Sample a starting point
start_index
from the interval[cur_len, cur_len + context_length  span_length]
and mask tokensstart_index:start_index + span_length
 Set
cur_len = cur_len + context_length
. Ifcur_len < max_len
(i.e. there are tokens remaining in the sequence to be processed), repeat from Step 1.
The masked tokens to be predicted for a particular sequence are determined by the following algorithm:
 Start from the beginning of the sequence by setting
cur_len = 0
(number of tokens processed so far).  Sample a
span_length
from the interval[1, max_span_length]
(length of span of tokens to be masked)  Reserve a context of length
context_length = span_length / plm_probability
to surround span to be masked  Sample a starting point
start_index
from the interval[cur_len, cur_len + context_length  span_length]
and mask tokensstart_index:start_index + span_length
 Set
cur_len = cur_len + context_length
. Ifcur_len < max_len
(i.e. there are tokens remaining in the sequence to be processed), repeat from Step 1.
The masked tokens to be predicted for a particular sequence are determined by the following algorithm:
 Start from the beginning of the sequence by setting
cur_len = 0
(number of tokens processed so far).  Sample a
span_length
from the interval[1, max_span_length]
(length of span of tokens to be masked)  Reserve a context of length
context_length = span_length / plm_probability
to surround span to be masked  Sample a starting point
start_index
from the interval[cur_len, cur_len + context_length  span_length]
and mask tokensstart_index:start_index + span_length
 Set
cur_len = cur_len + context_length
. Ifcur_len < max_len
(i.e. there are tokens remaining in the sequence to be processed), repeat from Step 1.