^jCddlZddlZddlZddlmZddlmZmZm Z ddl m Z m Z m Z mZmZer ddlZddlmZmZGdd e Zeed Gd d e Zy)N)GenerationConfig)add_end_docstringsis_torch_availablerequires_backends)ArgumentHandlerDatasetPipelinePipelineExceptionbuild_pipeline_init_args),MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMES0MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMESceZdZdZddZy)%TableQuestionAnsweringArgumentHandlerzB Handles arguments for the TableQuestionAnsweringPipeline Nc t|dddl}| td|t|tr&|j d|j d|g}nt|t rt|dkDrrtd|Dstdd|D|dj d|dj d|}nptd |djd tt|tst|tjr|Std t|d ||d g}|D]C}t|d|jr|d td|j|d|d<E|S)Npandasrz(Keyword argument `table` cannot be None.querytablec3<K|]}t|tywN) isinstancedict.0ds z/var/www/ramen.bs-engineer-server.com/venv/lib/python3.12/site-packages/transformers/pipelines/table_question_answering.py zATableQuestionAnsweringArgumentHandler.__call__..-s>1:a.>sz:Keyword argument `table` should be a list of dict, but is c32K|]}t|ywr)typers rrzATableQuestionAnsweringArgumentHandler.__call__../sUmbcVZ[\V]UmszIf keyword argument `table` is a list of dictionaries, each dictionary should have a `table` and `query` key, but only dictionary has keys z `table` and `query` keys.zZInvalid input. Keyword argument `table` should be either of type `dict` or `list`, but is ))rrzTable cannot be None.)rr ValueErrorrrgetlistlenallkeysr types GeneratorTyper DataFrame)selfrrkwargspdtqa_pipeline_inputstqa_pipeline_inputs r__call__z.TableQuestionAnsweringArgumentHandler.__call__s $) =GH H ]%&599W+=+IeiiX_N`Nl',g#E4(SZ!^>>>$TUmglUmTno8<<(4qg9N9Z*/'$JJOPQ(--/IZZtv$E7)CzRWY^YlYlGm  u+a) .3U#C"D "5 X 092<<H%g.6$%<==.0ll;Mg;V.W"7+  X#")NN)__name__ __module__ __qualname____doc__r0r1rrrs -#r1rT) has_tokenizerceZdZdZdZdZdZdZdZdZ e dZ e ffd Z dZd Zfd Zdd Zdd Zdd ZdZxZS)TableQuestionAnsweringPipelinea Table Question Answering pipeline using a `ModelForTableQuestionAnswering`. This pipeline is only available in PyTorch. Unless the model you're using explicitly sets these generation parameters in its configuration files (`generation_config.json`), the following default values will be used: - max_new_tokens: 256 Example: ```python >>> from transformers import pipeline >>> oracle = pipeline(model="google/tapas-base-finetuned-wtq") >>> table = { ... "Repository": ["Transformers", "Datasets", "Tokenizers"], ... "Stars": ["36542", "4512", "3934"], ... "Contributors": ["651", "77", "34"], ... "Programming language": ["Python", "Python", "Rust, Python and NodeJS"], ... } >>> oracle(query="How many stars does the transformers repository have?", table=table) {'answer': 'AVERAGE > 36542', 'coordinates': [(0, 1)], 'cells': ['36542'], 'aggregator': 'AVERAGE'} ``` Learn more about the basics of using a pipeline in the [pipeline tutorial](../pipeline_tutorial) This tabular question answering pipeline can currently be loaded from [`pipeline`] using the following task identifier: `"table-question-answering"`. The models that this pipeline can use are models that have been fine-tuned on a tabular question answering task. See the up-to-date list of available models on [huggingface.co/models](https://huggingface.co/models?filter=table-question-answering). z table,queryTF)max_new_tokensc t|di|||_tj}|j t |j|t|jjddxr!t|jjdd|_ t|jjdrd|_ yd|_ y)Naggregation_labelsnum_aggregation_labelstapasr6)super__init__ _args_parserrcopyupdatercheck_model_typegetattrmodelconfig aggregatehasattrr )r+ args_parserr,mapping __class__s rrAz'TableQuestionAnsweringPipeline.__init__}s "6"'BGGICD g& !2!24H$O T[ JJ  7U  'tzz'8'8:NOG UY r1c &|jdi|S)Nr6)rG)r+inputss rbatch_inferencez.TableQuestionAnsweringPipeline.batch_inferencestzz#F##r1c g}g}d}|djd}|dj|j}|dj|j}|dj|j}d} t|D]+} |'| dddf} t j | j j} || } t| jdD]{} | dddfj| }| dddfj| dz }| dddfj| dz }|dk\s]|dk\sc|dk(sit|||f| | <}tj| jtjj|j| dddf<|| }|| }|| } |j|jd|jd| jd }|j }|j"r|j%|j&|j%|tj(j+| }|j,|jtj.j|j,jz}t1j2t4}t7|j9jD]\} }| dddfj| }| dddfj| dz }| dddfj| dz }|dk\s`|dk\sf|dk(sl|||fj%||Dcic],}|t j:||j=d kD.}}.tj>tA|d}|j"s|fS|tj>tA|dfScc}w) z Inference used for models that need to process sequences in a sequential fashion, like the SQA models which handle conversational query related to a table. N input_idsrattention_masktoken_type_idsrr)rRrSrT)logitsg?)!shapetodevicerangenp zeros_likecpunumpytolistinttorch from_numpyr longrG unsqueezerVrIappendlogits_aggregation distributions Bernoulliprobsfloat32 collections defaultdictr$ enumeratesqueezearraymeancattuple)r+rO all_logitsall_aggregations prev_answers batch_sizerRrSrTtoken_type_ids_exampleindexprev_labels_example model_labelsi segment_idcol_idrow_idinput_ids_exampleattention_mask_exampleoutputsrVdist_per_token probabilitiescoords_to_probspcolrowkey logits_batchs rsequential_inferencez3TableQuestionAnsweringPipeline.sequential_inferences   K(..q1 ;'**4;;7  0144T[[A 0144T[[A!%:&. iE'&&|11!45NA!71!=!D!D!Fq!IJ3AqD9@@B1EIF3AqD9@@B1EIF{v{zQ*-lFF;K.L*M Q N05/?/? /M/R/RSXS]S]/^/a/abfbmbm/n&q!t, )% 0 %3E%: "%3E%: "jj+55a85??B5??B!G ^^F~~ ''(B(BC   f %"00::&:IN*003I3N3Nu}}3]3`3`$$++4M*55d;O!-"7"7"9"@"@"BC :13AqD9@@B1E ,QT299;A>B,QT299;A>B!8qZ1_#S#J/66q9  :YhhQTC/#*>!?!D!D!F!LLhLh]. i`yyz!2A6 &*nn o<SXYiSjlmIn:oo is91O>cr|j|i|}t||fi|}t|dk(r|dS|S)a Answers queries according to a table. The pipeline accepts several types of inputs which are detailed below: - `pipeline(table, query)` - `pipeline(table, [query])` - `pipeline(table=table, query=query)` - `pipeline(table=table, query=[query])` - `pipeline({"table": table, "query": query})` - `pipeline({"table": table, "query": [query]})` - `pipeline([{"table": table, "query": query}, {"table": table, "query": query}])` The `table` argument should be a dict or a DataFrame built from that dict, containing the whole table: Example: ```python data = { "actors": ["brad pitt", "leonardo di caprio", "george clooney"], "age": ["56", "45", "59"], "number of movies": ["87", "53", "69"], "date of birth": ["7 february 1967", "10 june 1996", "28 november 1967"], } ``` This dictionary can be passed in as such, or can be converted to a pandas DataFrame: Example: ```python import pandas as pd table = pd.DataFrame.from_dict(data) ``` Args: table (`pd.DataFrame` or `Dict`): Pandas DataFrame or dictionary that will be converted to a DataFrame containing all the table values. See above for an example of dictionary. query (`str` or `list[str]`): Query or list of queries that will be sent to the model alongside the table. sequential (`bool`, *optional*, defaults to `False`): Whether to do inference sequentially or as a batch. Batching is faster, but models like SQA require the inference to be done sequentially to extract relations within sequences, given their conversational nature. padding (`bool`, `str` or [`~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` or [`TapasTruncationStrategy`], *optional*, defaults to `False`): Activates and controls truncation. Accepts the following values: - `True` or `'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. - `False` or `'do_not_truncate'` (default): No truncation (i.e., can output batch with sequence lengths greater than the model maximum admissible input size). Return: A dictionary or a list of dictionaries containing results: Each result is a dictionary with the following keys: - **answer** (`str`) -- The answer of the query given the table. If there is an aggregator, the answer will be preceded by `AGGREGATOR >`. - **coordinates** (`list[tuple[int, int]]`) -- Coordinates of the cells of the answers. - **cells** (`list[str]`) -- List of strings made up of the answer cell values. - **aggregator** (`str`) -- If the model has an aggregator, this returns the aggregator. rr)rBr@r0r%)r+argsr,pipeline_inputsresultsrMs rr0z'TableQuestionAnsweringPipeline.__call__sLV,$++T>#/97;7M7M 34)djj))LLLOLG)5RYZ r1cj|d}|d}|d}|jdk(r|jr|dd\}}|jj|||}|\}} t | D cic])\} } | |j j j| +} } } |j j j} t | D cic]\} } | | k7s | | | dz}} } n*|d}|jj||}|d}i} i}g}t |D]\}}|Dcgc]}|j|}}| j|d}|j|d}|d j|z||Dcgc]}|j|c}d }|r||d <|j|tdk(rMtd |j jd |jj!|dDcgc]}d|i}}t|dkDr|S|dScc} } wcc} } wcc}wcc}wcc}w)Nrrrr?rz > rrz, )answer coordinatescells aggregatorzTable question answeringz Empty answerT)skip_special_tokensrr)r rIrconvert_logits_to_predictionsrmrGrHr=no_aggregation_label_indexiatr#joinrer%r name_or_path batch_decode)r+rrOrrrV logits_agg predictionsanswer_coordinates_batchagg_predictionsr{pred aggregatorsno_agg_label_indexaggregators_prefixanswersrxr coordinaterraggregator_prefixrs r postprocessz*TableQuestionAnsweringPipeline.postprocessVsc~.g& * 99 ~~%,Ra[" "nnJJ6SY[ef ( %'"G&/0H&I '"{ALM::.MM(__UB7 $6$:$:5"$E!/$))E2BB#.FQR eii 3R +5F<(v& '6{a'(BDJJD[D[]kll8<8S8ST[qu8S8vwf&)wGwg,*w: :?x&N Sxs$$.H H  H !H&5H+ 8 H0)NNN)TN)F)r2r3r4r5default_input_names_pipeline_calls_generate_load_processor_load_image_processor_load_feature_extractor_load_tokenizerr_default_generation_configrrArPrr0rrrr __classcell__)rMs@rr9r9Msr D(#O!#O!1"$I#J Z$ApFPd5& "(;r1r9)rkr(r^r[ generationrutilsrrrbaser r r r r ramodels.auto.modeling_autorrrr9r6r1rrsk ) ba 2#O2#j,4@Ap;Xp;Bp;r1