S^(h[ H d Z ddlZddlZddlZddlmZ ddlmZmZm Z m
Z
m Z ddl Z
ddlmZmZmZmZmZ ddlmZmZ ddlmZmZmZmZ dd l m Z e rddl Z e j@ e! Z"d
d d Z# G d
de Z$dZ% e d Z&d Z' G d d Z( G d de Z)y)z Tokenization classes for TAPEX. N) lru_cache)DictListOptionalTupleUnion ) ExplicitEnumPaddingStrategy
TensorTypeadd_end_docstringsis_pandas_available)
AddedTokenPreTrainedTokenizer)ENCODE_KWARGS_DOCSTRING
BatchEncoding TextInputTruncationStrategy)loggingz
vocab.jsonz
merges.txt)
vocab_file merges_filec e Zd ZdZdZy)TapexTruncationStrategyz}
Possible values for the `truncation` argument in [`~TapasTokenizer.__call__`]. Useful for tab-completion in an IDE.
drop_rows_to_fitN)__name__
__module__ __qualname____doc__DROP_ROWS_TO_FIT u/var/www/skyplay_api_hub/venv/lib/python3.12/site-packages/transformers/models/deprecated/tapex/tokenization_tapex.pyr r ( s *r! r a
add_special_tokens (`bool`, *optional*, defaults to `True`):
Whether or not to encode the sequences with the special tokens relative to their model.
padding (`bool`, `str` or [`~file_utils.PaddingStrategy`], *optional*, defaults to `False`):
Activates and controls padding. Accepts the following values:
- `True` or `'longest'`: Pad to the longest sequence in the batch (or no padding if only a single
sequence if provided).
- `'max_length'`: Pad to a maximum length specified with the argument `max_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).
truncation (`bool`, `str`, [`TapexTruncationStrategy`] or [`~tokenization_utils_base.TruncationStrategy`],
*optional*, defaults to `False`):
Activates and controls truncation. Accepts the following values:
- `'drop_rows_to_fit'`: Truncate to a maximum length specified with the argument `max_length` or to the
maximum acceptable input length for the model if that argument is not provided. This will truncate
row by row, removing rows from the table.
- `True` or `'longest_first'`: Truncate to a maximum length specified with the argument `max_length` or
to the maximum acceptable input length for the model if that argument is not provided. This will
truncate token by token, removing a token from the longest sequence in the pair if a pair of
sequences (or a batch of pairs) is provided.
- `'only_first'`: Truncate to a maximum length specified with the argument `max_length` or to the
maximum acceptable input length for the model if that argument is not provided. This will only
truncate the first sequence of a pair if a pair of sequences (or a batch of pairs) is provided.
- `'only_second'`: Truncate to a maximum length specified with the argument `max_length` or to the
maximum acceptable input length for the model if that argument is not provided. This will only
truncate the second sequence of a pair if a pair of sequences (or a batch of pairs) is provided.
- `False` or `'do_not_truncate'` (default): No truncation (i.e., can output batch with sequence lengths
greater than the model maximum admissible input size).
max_length (`int`, *optional*):
Controls the maximum length to use by one of the truncation/padding parameters. If left unset or set to
`None`, this will use the predefined model maximum length if a maximum length is required by one of the
truncation/padding parameters. If the model has no specific maximum input length (like XLNet)
truncation/padding to a maximum length will be deactivated.
stride (`int`, *optional*, defaults to 0):
If set to a number along with `max_length`, the overflowing tokens returned when
`return_overflowing_tokens=True` will contain some tokens from the end of the truncated sequence
returned to provide some overlap between truncated and overflowing sequences. The value of this
argument defines the number of overlapping tokens.
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` or [`~file_utils.TensorType`], *optional*):
If set, will return tensors instead of list of python integers. Acceptable values are:
- `'tf'`: Return TensorFlow `tf.constant` objects.
- `'pt'`: Return PyTorch `torch.Tensor` objects.
- `'np'`: Return Numpy `np.ndarray` objects.
c t t t d t d dz t t t d t d dz z t t t d t d dz z } | dd }d }t d
D ]1 }|| vs| j | |j d
|z |dz
}3 |D cg c]
}t | }}t t
| | S c c}w ) a3
Returns list of utf-8 byte and a mapping to unicode strings. We specifically avoids mapping to whitespace/control
characters the bpe code barfs on. The reversible bpe codes work on unicode strings. This means you need a large #
of unicode characters in your vocab if you want to avoid UNKs. When you're at something like a 10B token dataset
you end up needing around 5K for decent coverage. This is a significant percentage of your normal, say, 32K bpe
vocab. To avoid that, we want lookup tables between utf-8 bytes and unicode strings.
!~ ¡ ¬ ® ÿNr )listrangeordappendchrdictzip)bscsnbs r" bytes_to_unicoder7 f s
U3s8SX\
*+d5TCIPQM3R.SSVZ[`adeiajloptluxyly[zV{{
AB A
4[ B; IIaL IIdQh
FA
Q#a& B B
s C4c b t }| d }| dd D ] }|j ||f |} |S )z
Return set of symbol pairs in a word. Word is represented as tuple of symbols (symbols being variable-length
strings).
r r& N)setadd)wordpairs prev_charchars r" get_pairsr? } sF
EEQIQR
9d#$ Lr! c 8 e Zd ZdZdefdZdefdZdedefdZ y )
IndexedRowTableLinearizezQ
FORMAT: col: col1 | col2 | col 3 row 1 : val1 | val2 | val3 row 2 : ...
table_contentc d|v rd|v sJ | j | j |d dz }t |d D ]! \ }}|| j ||dz dz z
}# |j S )zs
Given a table, TableLinearize aims at converting it into a flatten sequence with special symbols.
headerrows r& ) row_index)PROMPT_MESSAGEprocess_header enumerate process_rowstrip)selfrB table_stri row_examples r"
process_tablez&IndexedRowTableLinearize.process_table s =(V}-DYdFYFYYD''
h(?@3F '
f(= > NNA{ ))+Q)G#M MI N r! headersc * ddj | z S )z}
Given a list of headers, TableLinearize aims at converting it into a flatten sequence with special symbols.
zcol : | )join)rM rR s r" rI z'IndexedRowTableLinearize.process_header s %**W---r! rowrG c d}g }|D ]> }t |t r|j t | .|j | @ |dj | z
}dt | z dz |z S )zq
Given a row, TableLinearize aims at converting it into a flatten sequence with special symbols.
rT zrow z : )
isinstanceintr/ strrU )rM rV rG row_strrow_cell_values
cell_values r" rK z$IndexedRowTableLinearize.process_row su 3J*c* &&s:7 &&z2 3
5::o..I&.88r! N)
r r r r r rQ r rI rZ rK r r! r" rA rA s2 !4 !.d . 9t 9 9r! rA c & z e Zd ZdZeZddgZ dF fd Z dGdee d e
ee d
ee fd Z dHdee d e
ee d e d
ee f fd
Z
dGdee d e
ee d
ee fdZdIdZed Zd Zd Zd Zd Zd Zd ZdGdede
e d
ee fdZ ee e dJde d ed f d e
e e ee f d e eee f d e d e e ee f d!e e ee!f d"e
e d#e d$e
e d%e
e ee"f d&e
e d'e
e d(e d)e d*e d+e d,e d
e#f$d- Z$ dKde d ed f d e
e e ee f d e eee f d e d e e ee f d!e e ee!f d"e
e d#e d$e
e d%e
e ee"f d&e
e d'e
e d(e d)e d*e d+e d,e d
e#f$d.Z% ee e dLde d ed f d e
ee d ee d e d e e ee f d!e e ef d"e
e d$e
e d%e
e ee"f d&e
e d'e
e d(e d)e d*e d+e d,e d
e#f"d/ Z&ddde jN e!jP dddddddddddfde d ed f d e
ee d e
ee d e d0e d1e!d"e
e d#e d$e
e d%e
e ee"f d&e
e d'e
e d(e d)e d*e d+e d,e d
e#f$d2Z) ee e ddde jN e!jP ddddddddddfde d ed f d e
e e ee f d e
e eee f d e d0e d1e!d"e
e d#e d$e
e d%e
e d&e
e d'e
e d(e d)e d+e d,e d
e#f"d3 Z* ee dMdd d e
e d e
e d e d e e ee f d!e e ee!e+f d"e
e d%e
e ee"f d
ee fd4 Z, ee e dNdd d e
e d e
e d e d e e ee f d!e e ef d"e
e d$e
e d%e
e ee"f d&e
e d'e
e d)e d*e d+e d,e d
e#f d5 Z-ddde jN e!jP dddddddddddfdd d e
e d e
e d e d0e d1e!d"e
e d#e d$e
e d%e
e ee"f d&e
e d'e
e d(e d)e d*e d+e d,e d
e#f$d6Z. dOd e eee f d e d e e ee f d!e e ee!f d"e
e d#e d$e
e d%e
e ee"f d&e
e d'e
e d(e d)e d*e d+e d,e d
e#f d7Z/ dPd ee d e d e e ee f d!e e ef d"e
e d$e
e d%e
e ee"f d&e
e d'e
e d(e d)e d*e d+e d,e d
e#f d8Z0de jN e!jP dddddddddddfd ee d e d0e d1e!d"e
e d#e d$e
e d%e
e ee"f d&e
e d'e
e d(e d)e d*e d+e d,e d
e#f d9Z1 dQd ed e d e e ee f d!e e ee!e+f d"e
e d%e
e ee"f d
ee fd:Z2 dRd ed e d e e ee f d!e e ef d"e
e d$e
e d%e
e ee"f d&e
e d'e
e d)e d*e d+e d,e d
e#f d;Z3de jN e!jP dddddddddddfd ed e d0e d1e!d"e
e d#e d$e
e d%e
e ee"f d&e
e d'e
e d(e d)e d*e d+e d,e d
e#f d<Z4de ee!e+f dfd=Z5d>e6d?ed efd@Z7dA Z8 dSd>e6d?ed e
e eee f fdBZ9dGd>e6d?efdCZ:d>e6d?ed edDe;fdEZ< xZ=S )TTapexTokenizera
Construct a TAPEX tokenizer. Based on byte-level Byte-Pair-Encoding (BPE).
This tokenizer can be used to flatten one or more table(s) and concatenate them with one or more related sentences
to be used by TAPEX models. The format that the TAPEX tokenizer creates is the following:
sentence col: col1 | col2 | col 3 row 1 : val1 | val2 | val3 row 2 : ...
The tokenizer supports a single table + single query, a single table and multiple queries (in which case the table
will be duplicated for every query), a single query and multiple tables (in which case the query will be duplicated
for every table), and multiple tables and queries. In other words, you can provide a batch of tables + questions to
the tokenizer for instance to prepare them for the model.
Tokenization itself is based on the BPE algorithm. It is identical to the one used by BART, RoBERTa and GPT-2.
This tokenizer inherits from [`PreTrainedTokenizer`] which contains most of the main methods. Users should refer to
this superclass for more information regarding those methods.
Args:
vocab_file (`str`):
Path to the vocabulary file.
merges_file (`str`):
Path to the merges file.
do_lower_case (`bool`, *optional*, defaults to `True`):
Whether or not to lowercase the input when tokenizing.
errors (`str`, *optional*, defaults to `"replace"`):
Paradigm to follow when decoding bytes to UTF-8. See
[bytes.decode](https://docs.python.org/3/library/stdtypes.html#bytes.decode) for more information.
bos_token (`str`, *optional*, defaults to `""`):
The beginning of sequence token that was used during pretraining. Can be used a sequence classifier token.
When building a sequence using special tokens, this is not the token that is used for the beginning of
sequence. The token used is the `cls_token`.
eos_token (`str`, *optional*, defaults to `""`):
The end of sequence token.
When building a sequence using special tokens, this is not the token that is used for the end of sequence.
The token used is the `sep_token`.
sep_token (`str`, *optional*, defaults to `""`):
The separator token, which is used when building a sequence from multiple sequences, e.g. two sequences for
sequence classification or for a text and a question for question answering. It is also used as the last
token of a sequence built with special tokens.
cls_token (`str`, *optional*, defaults to `""`):
The classifier token which is used when doing sequence classification (classification of the whole sequence
instead of per-token classification). It is the first token of the sequence when built with special tokens.
unk_token (`str`, *optional*, defaults to `""`):
The unknown token. A token that is not in the vocabulary cannot be converted to an ID and is set to be this
token instead.
pad_token (`str`, *optional*, defaults to `""`):
The token used for padding, for example when batching sequences of different lengths.
mask_token (`str`, *optional*, defaults to `""`):
The token used for masking values. This is the token used when training this model with masked language
modeling. This is the token which the model will try to predict.
add_prefix_space (`bool`, *optional*, defaults to `False`):
Whether or not to add an initial space to the input. This allows to treat the leading word just as any
other word. (BART tokenizer detect beginning of words by the preceding space).
max_cell_length (`int`, *optional*, defaults to 15):
Maximum number of characters per cell when linearizing a table. If this number is exceeded, truncation
takes place.
input_idsattention_maskTFc J t |t rt |dd n|}t |t rt |dd n|}t |t rt |dd n|}t |t rt |dd n|}t | t rt | dd n| } t |
t rt |
dd n|
}
t | t rt | dd n| } t |d 5 }t j
| | _ d d d | j j D ci c] \ }}||
c}}| _ || _ t | _ | j j D ci c] \ }}||
c}}| _ t |d 5 }|j j d dd }d d d D cg c] }t |j }}t! t# |t% t' | | _ i | _ | | _ || _ t1 j2 d | _ t7 | p d ||||||| |||
| | |
d
| |
| _ t= | _ y # 1 sw Y sxY wc c}}w c c}}w # 1 sw Y xY wc c}w ) NF)lstriprstripTutf-8encoding
r& zJ's|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+)
r r
do_lower_caseerrors bos_token eos_token unk_token sep_token cls_token pad_token
mask_tokenadd_prefix_spacemax_cell_lengthr ) rY r[ r openjsonloadencoderitemsdecoderrl r7 byte_encoder byte_decoderreadsplittupler1 r2 r- len bpe_rankscachert rk recompilepatsuper__init__ru rA table_linearize)rM r r rk rl rm rn rp rq ro rr rs rt ru kwargs vocab_handlekv
merges_handle
bpe_mergesmerge __class__s r" r zTapexTokenizer.__init__ sq " JTT]_bIcJyuEir IST]_bIcJyuEir IST]_bIcJyuEir IST]_bIcJyuEir IST]_bIcJyuEir IST]_bIcJyuEir KUU_adJeZ
4Fku
*w
/ 3< 99\2DL 3)- ););)=>A1> , ..2.?.?.E.E.G HdaQT H
+
0 @M&++-33D9!B?J @8BCueEKKM*C
C c*eC
O.D EF
0 * :: op
!#' ! - +
" / 7 9 K 3 3> I @ @Cs* /I;/
J8
J#J
J ;JJ N token_ids_0 token_ids_1returnc | | j g|z | j gz S | j g}| j g}||z |z |z |z |z S )a
Build model inputs from a sequence or a pair of sequence for sequence classification tasks by concatenating and
adding special tokens. A TAPEX sequence has the following format:
- single sequence: ` X `
- pair of sequences: ` A B `
Args:
token_ids_0 (`List[int]`):
List of IDs to which the special tokens will be added.
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
Returns:
`List[int]`: List of [input IDs](../glossary#input-ids) with the appropriate special tokens.
) cls_token_id sep_token_id)rM r r clsseps r" build_inputs_with_special_tokensz/TapexTokenizer.build_inputs_with_special_tokens> sg " %%&48I8I7JJ J ! ![ 3&,{:S@@r! already_has_special_tokensc |rt | ||d S |dgdgt | z z dgz S dgdgt | z z ddgz dgt | z z dgz S )a
Args:
Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding
special tokens using the tokenizer `prepare_for_model` method.
token_ids_0 (`List[int]`):
List of IDs.
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
already_has_special_tokens (`bool`, *optional*, defaults to `False`):
Whether or not the token list is already formatted with special tokens for the model.
Returns:
`List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
T)r r r r& r )r get_special_tokens_maskr )rM r r r r s r" r z&TapexTokenizer.get_special_tokens_maskU s &72 '[]a 3
31#K 001QC7 7sqcC ,,-A61#K@P:PQUVTWWWr! c | j g}| j g}|t ||z |z dgz S t ||z |z |z |z |z dgz S )a
Args:
Create a mask from the two sequences passed to be used in a sequence-pair classification task. TAPEX does not:
make use of token type ids, therefore a list of zeros is returned.
token_ids_0 (`List[int]`):
List of IDs.
token_ids_1 (`List[int]`, *optional*):
Optional second list of IDs for sequence pairs.
Returns:
`List[int]`: List of zeros.
r )r r r )rM r r r r s r" $create_token_type_ids_from_sequencesz3TapexTokenizer.create_token_type_ids_from_sequencesn sm ! ! s[(3./1#5 53$s*S0;>DEKKr! c |j d| j }|s|r&t | dkD r|d j sd|z }||fS )Nrt r rF )poprt r isspace)rM textis_split_into_wordsr rt s r" prepare_for_tokenizationz'TapexTokenizer.prepare_for_tokenization sO !::&8$:O:OP #3#d)a-PTUVPWP_P_Pa:Df~ r! c , t | j S N)r ry rM s r"
vocab_sizezTapexTokenizer.vocab_size s 4<< r! c B t | j fi | j S r )r1 ry added_tokens_encoderr s r" get_vocabzTapexTokenizer.get_vocab s DLL>D$=$=>>r! c | j v r j | S t | }t | }|s|S t | fd }| j vrn|\ }}g }d}|t | k r |j
|| } |j ||| | }|| |k( r6|t | dz
k r%||dz |k( r|j ||z |dz
}n|j || |dz
}|t | k rt | }|}t | dk( rn t | }dj | }| j |< |S # t $ r |j ||d Y iw xY w)Nc N j j | t d S )Ninf)r getfloat)pairrM s r" z$TapexTokenizer.bpe.. s 1C1CD%PU,1W r! keyr r& rF ) r r r? minr r indexextend
ValueErrorr/ rU )
rM tokenr; r< bigramfirstsecondnew_wordrO js
` r" bpezTapexTokenizer.bpe s DJJ ::e$ $U|$L$WXFT^^+ "ME6HAc$i-
5!,A
OOD1I.A7e#CIM(9d1q5kV>S OOEFN3FA OODG,FA c$i- XHD4yA~!$9 : xx~
5 + " OODH-s 1E E)(E)c g }t j j | D ]a }dj fd|j d D }|j d j
| j d D c |S )zTokenize a string.rX c 3 < K | ] }j | y wr )r| ).0r6 rM s r" z+TapexTokenizer._tokenize.. s! )*!!!$ s rf c 3 K | ] }| y wr r )r bpe_tokens r" r z+TapexTokenizer._tokenize.. s TIi Ts rF )r findallr rU encoder r r )rM r
bpe_tokensr s ` r" _tokenizezTapexTokenizer._tokenize s}
ZZ$/ UEGG .3ll7.C E
T% 9N9Ns9S T T U
r! c | j j || j j | j S )z0Converts a token (str) in an id using the vocab.)ry r ro )rM r s r" _convert_token_to_idz#TapexTokenizer._convert_token_to_id s, || t||'7'7'GHHr! c 8 | j j | S )z=Converts an index (integer) in a token (str) using the vocab.)r{ r )rM r s r" _convert_id_to_tokenz#TapexTokenizer._convert_id_to_token s || &&r! c dj | }t |D cg c] }| j | c} j d| j }|S c c}w )z:Converts a sequence of tokens (string) in a single string.rX rf )rl )rU bytearrayr} decoderl )rM tokensr cs r" convert_tokens_to_stringz'TapexTokenizer.convert_tokens_to_string sP wwv=1$++A.=>EEgVZVaVaEb >s Asave_directoryfilename_prefixc P t j j | st j d| d y t j j ||r|dz ndt d z }t j j ||r|dz ndt d z }t |dd 5 }|j t j | j d
d d
dz d d d d}t |dd 5 }|j d t | j j d D ]M \ }} || k7 rt j d| d | }|j dj | dz |dz
}O d d d ||fS # 1 sw Y xY w# 1 sw Y ||fS xY w)NzVocabulary path (z) should be a directory-rX r r wrf rg r TF)indent sort_keys ensure_asciiri r z#version: 0.2
c | d S )Nr& r )kvs r" r z0TapexTokenizer.save_vocabulary.. s
Y[\]Y^ r! r zSaving vocabulary to zZ: BPE merge indices are not consecutive. Please check that the tokenizer is not corrupted!rF r& )ospathisdirloggererrorrU VOCAB_FILES_NAMESrv writerw dumpsry sortedr rz warning)
rM r r r
merge_filefr writerr token_indexs
r" save_vocabularyz TapexTokenizer.save_vocabulary s ww}}^, LL,^,< ,!" %
(
(4(( #5 %% #5 -&;&;*C+E'= + !
& ST Tr! c d}d}t |t j rd}n5t |t t f r t |d t j rd}|t |t
rd}n9t |t t f r#t
| dk( st |d t
rd}|s t d |s t d t |t t f xs t |t t f }|rB | j di d|d|d|d |d
|d |d |d
| d|
d| d| d|
d|d|d|d||S | j di d|d|d|d |d
|d |d |d
| d|
d| d| d|
d|d|d|d||S )NFTr zdtable input must of type `pd.DataFrame` (single example), `List[pd.DataFrame]` (batch of examples). zRquery input must of type `str` (single example), `List[str]` (batch of examples). r r r r r r r r r r r r r r r r r )
rY pd DataFramer, r r[ r r batch_encode_plus encode_plus)rM r r r r r r r r r r r r r r r r r r valid_table valid_query
is_batcheds r" r z TapexTokenizer.source_call_func= s= , eR\\ * K
e}
-*U1Xr||2T K =Juc 2 K
e}
-5zQ*U1Xs"; " v
qr r e}5YEDRW=9Y
)4)) $6
&
& $6 . '< '< +D ,F (> , #
( $4## $6
&
& $6 . '< '< +D ,F (> , #
r! c | j d|||||d|\ }}}} | j di d|d|d|d|d|d|d|d |d
| d |
d | d
| d|
d|d|d||S )z}
This method is deprecated, `__call__` should be used instead.
r r r r r r r r r padding_strategytruncation_strategyr r r r r r r r r r r )"_get_padding_truncation_strategies_batch_encode_plus)rM r r r r r r r r r r r r r r r r r r r s r" r z TapexTokenizer.batch_encode_plus s : ElDDkDk E
!! 1 E
E
A-z6 't&&
2
.
!4
"
2
*
#8
#8
'@
(B
$:
(
#
r! r r c \ |r t d t |t j r%t |t t
f r|gt
| z }t |t t
f r t |t r|gt
| z }| j ||||||||| | | |
|||
| }t | S )Nzreturn_offset_mapping is not available when using Python tokenizers. To use this feature, change your tokenizer to one deriving from transformers.PreTrainedTokenizerFast.)r r r r r r r r r r r r r r r r )
NotImplementedErrorrY r r r, r r r[ _batch_prepare_for_modelr )rM r r r r r r r r r r r r r r r r r r
batch_outputss r" r z!TapexTokenizer._batch_encode_plus s * "%8
eR\\ *z%$ /O Gc%j(E edE] +
5#0F Gc%j(E 55 1 - 3! 1"7"7&?'A')! 6
& ]++r! c D i }|dgt | z }t ||| D ] \ }}}| j ||||| }| j r|j }| j | }| j
| j | |t j j |j ||dd| |
||dd| }|j D ]" \ }}||vrg ||< || j | $ | j ||j || | }t ||
}|S )z
This method adds special tokens, truncates sequences if overflowing while taking into account the special
tokens and manages a moving window (with user defined stride) for overflowing tokens.
Nr r Fidsr r r r r r r r r r r r prepend_batch_axisr r r r r tensor_type)r r2 prepare_table_queryrk lowertokenizeprepare_for_modelconvert_tokens_to_idsr
DO_NOT_PADvaluerz r/ padr )rM r r r r r r r r r r r r r r r r r _table_query_answerr r outputsr r$ s r" r z'TapexTokenizer._batch_prepare_for_model sa 0
>Vc%j(F'*5%'@ 1 #FFG++=P]g , D !!zz|]]4(F,,..v6#5'2288.44% #'&+&;*C+E + ##( - G$ &mmo
1
Um+)+M#& c"))%0
17 1@ $**! 1"7 !
&m P
r! c B | j |f|||||||d| }
|
d S )a6
Prepare a table, a string and possible answer for the model. This method does not return token type IDs,
attention masks, etc. which are necessary for the model to work correctly. Use this method if you want to build
your processing on your own, otherwise refer to `__call__`.
)r r r r r r r ra )r ) rM r r r r r r r r r encoded_inputss r" r zTapexTokenizer.encodeO sJ $ *))
1!!)
k**r! c | j d|||||d|\ }}}} | j d||||||||| |
| | |
||d|S )Nr
)r r r r r r r r r r r r r r r r )r _encode_plus)rM r r r r r r r r r r r r r r r r r r s r" r zTapexTokenizer.encode_pluso s * ElDDkDk E
!! 1 E
E
A-z6 !t
1 - 3! 1)"7"7'A#9'
!
r! c ( |r t d | j ||||| }| j r|j }| j | }| j | j
| ||j |j ||| |
d| | |
||| S )N return_offset_mapping is not available when using Python tokenizers. To use this feature, change your tokenizer to one deriving from transformers.PreTrainedTokenizerFast. More information on available tokenizers at https://github.com/huggingface/transformers/pull/2674r Tr r r r r r r r r r r r r r r )r r rk r r r! r" r$ )rM r r r r r r r r r r r r r r r r r r r r s r" r- zTapexTokenizer._encode_plus s * "%H
'' 5&6IV` (
:: | j d||||||d|}|d S )a
Prepare the answer string for the model. This method does not return token type IDs, attention masks, etc.
which are necessary for the model to work correctly. Use this method if you want to build your processing on
your own, otherwise refer to `__call__`.
Args:
answer `str`:
Corresponding answer supervision to the queries for training the model
)r r r r r r ra r )r3 ) rM r r r r r r r encoded_outputss r"
target_encodez TapexTokenizer.target_encode sD & 2$11
1!!)
{++r! c | | j d|||||
d|\ }}}} | j d||||||||| |
| | |
d
|S )z
Prepare a answer string for the model.
Args:
answer `str`:
Corresponding answer supervision to the queries for training the model.
r
)
r r r r r r r r r r r r r r )r _target_encode_plus)rM r r r r r r r r r r r r r r r r s r" r3 z!TapexTokenizer.target_encode_plus s 2 ElDDkDk E
!! 1 E
E
A-z6 (t''
1 - 3! 1)"7"7'A#9'
r! c |
r t d |}| j r|j }| j | }| j | j | ||j |j ||||d|
| | | || S )Nr/ Tr0 )r rk r r r! r" r$ )rM r r r r r r r r r r r r r r r r r r s r" r; z"TapexTokenizer._target_encode_plus s & "%H
::= 1024 tokens). z@Please carefully check the corresponding table with the query : .z7You provide nothing to query with respect to the table.rF )
emptyr, columnsiterrowsvaluestruncate_table_cellsr r truncate_table_rowsr rQ r r ) rM r r r r r rO rV rB linear_table separator joint_inputs r" r z"TapexTokenizer.prepare_table_query s {{'+EMM':didrdrdtDuZ`Z[]`T#**EUDu vM
% %mUF C"&=&N&NN(( vR\(] //==mLL L 2 NNgTUZT[[\]^
B; NNT U \Cr
+4 $ 2 24 8
,)6L& *CF
+ + $V,
34<,,.. ,T 2F1I
3 r! c t |t st |t r|S |j dk7 rL| j | }t | | j k\ r"|d | j }| j | }|S y |S )NrX )rY rZ r rL r r ru r )rM r^
try_tokens
retain_tokensretain_cell_values r" rJ z TapexTokenizer.truncate_cellP s j# &*Z*G # z2J:$"6"66 *+AT-A-A B
$($A$A-$P!(( r! c , | j ||| \ }}| j |||| d}t |d D ]O \ }} | j j | |dz }
t | j
|
} | |kD r n || z }|dz
}Q |d |d= y)a
Args:
table_content:
{"header": xxx, "rows": xxx, "id" (Optionally): xxx}
question:
natural language sentence
answer:
if for training, is the supervision; otherwise will be empty
r rE r& N)estimate_delete_ratiodelete_unrelated_rowsrJ r rK r r ) rM rB rH r r delete_ratioremain_token_lenmaximum_keep_rowsindrP value_stringvalue_token_lens r" rD z"TapexTokenizer.truncate_table_rows_ s *.)C)CMS[]g)h& & ""=(FLQ )-*? @ # C //;;KqQL !$-- "= >O !11 / " #
& !"3"4 5r! c d|vsd|vr t d | j |d }| j j |d }| j |d }t | t | z }||z
}d} t |d D ]' \ }
} | | j j
| d d z z
} ) t | j | } | |k rd
|fS d || z z
|fS ) NrD rE z?The table content should contain both 'header' and 'rows' keys.T)r FrX d rF g g ?)r r r rI r rJ rK )
rM rB rH r question_tokens
header_string
header_tokensused_token_lenrW rZ _rP r[ s
r" rT z$TapexTokenizer.estimate_delete_ratio| s = (F-,G ^_ _ --T-J ,,;;M( UNA{ D 0 0 < <[# NQT T TL U dmmL 9: - -(( ( )O;;=MM Mr! rV c g }g }|t | dk( r t }n |D ch c] }|j }}| |j |j t |j d j d } t |d }
t
|d D ] \ } } | D
ch c] }
t |
j }}
t ||z dk( r#t || z dk( r|j | `|j | dz
| dz
| | dz | dz g |D cg c] } | |vs| }} t t | t t |d |z }t j || }t t |
D ]
} | |v s|d | = d |v rAt | dkD r2t j# d
j% t | |d yyyc c}w c c}
w c c} w ) zC
The argument answer is used only during training.
Nr z?!.,rF rE r r& )r idz Delete {:.2f} rows in table {})r r9 r updater rL rJ r[ r/ r r rZ randomchoicesreversedr- r r format)rM rB rH r rV truncated_unrelated_indicesrelated_indices
answer_setans_ex question_set row_max_len_row_idxrV rM lower_row
drop_itemsdrop_row_indicess r" rU z$TapexTokenizer.delete_unrelated_rows s ')# >S[A - J7=>V&,,.>J> hnn . /8>>&177<= -/0 &}V'< = kMHc7:;tT *;I;9z)*a/C L8P4QUV4V+228< &&1 hlHhYZl\dgh\h'ij
k &A'
!HTcDcH'
# '
893s=QWCX?Y\h?h;ij
!>>*ET {!34 4H++!&)(3 4
= S)9%:Q%> NN;BB3GWCXZghlZmn o &? 7 ? <'
s G)- G.+ G35G3) Treplace