tj8+ddlmZmZddlZddlmZmZmZddl m Z m Z m Z m Z erddlZejeZee dd Gd d e ZdS) )AnyoverloadN)add_end_docstringsis_torch_availablelogging) GenericTensorPipelinePipelineExceptionbuild_pipeline_init_argsT) has_tokenizeraO top_k (`int`, *optional*, defaults to 5): The number of predictions to return. targets (`str` or `list[str]`, *optional*): When passed, the model will limit the scores to the passed targets instead of looking up in the whole vocab. If the provided targets are not in the model vocab, they will be tokenized and the first resulting token will be used (with a warning, and that might be slower). tokenizer_kwargs (`dict`, *optional*): Additional dictionary of keyword arguments passed along to the tokenizer.c eZdZdZdZdZdZ dedej fdZ dedej fdZ defdZ dde eeffd Zd Zdd ZdZddZedededee eeffdZedeededeee eeffdZdeeezdedee eefeee eefzffd ZxZS)FillMaskPipelineFT input_idsreturncNtj||jjkd}|S)NFas_tuple)torchnonzero tokenizer mask_token_id)selfr masked_indexs ^/var/www/html/CA-Chatbot/venv/lib/python3.11/site-packages/transformers/pipelines/fill_mask.pyget_masked_indexz!FillMaskPipeline.get_masked_indexVs'}Y$.2N%NY^___ c||}tj|j}|dkr)t d|jjd|jjddS)Nr fill-maskzNo mask_token (z) found on the input) rnpprodshaper modelbase_model_prefixr mask_token)rrrnumels r_ensure_exactly_one_mask_tokenz/FillMaskPipeline._ensure_exactly_one_mask_tokenZsg,,Y77  *++ 199# ,Q$.";QQQ  9r model_inputsct|tr(|D]#}||dd$dS|dD]}||dS)Nrr) isinstancelistr()rr) model_inputrs rensure_exactly_one_mask_tokenz.FillMaskPipeline.ensure_exactly_one_mask_tokends lD ) ) ?+ Q Q 33K 4LQ4OPPPP Q Q*+6 ? ? 33I>>>> ? ?rNc `|d}|i}|j|fd|i|}|||S)Nptreturn_tensors)rr.)rinputsr1tokenizer_kwargspreprocess_parametersr)s r preprocesszFillMaskPipeline.preprocesslsT  !!N  #! %t~f``^`O_``  **<888rc6|jdi|}|d|d<|S)Nr)r$)rr) model_outputss r_forwardzFillMaskPipeline._forwardxs." 22\22 %1+%> k"rc"||jd|kr |jd}|dd}|d}tj||jjkdd}|d|ddf}|d}| |d|f}||\} } g} | jdd k} tt| | D]\} \}}g}t||D]\}}| }||| }|||| <|tj||jjk}|j|| }|||j|g|d }||| || r| dS| S) NrrlogitsFr)dim.r )skip_special_tokens)scoretoken token_strsequence)r#rrrrsqueezesoftmaxtopk enumerateziptolistnumpycopyr!where pad_token_iddecodeappend)rr8top_k target_idsroutputsrr<probsvalues predictionsresult single_maski_values _predictionsrowvptokensrC propositions r postprocesszFillMaskPipeline.postprocess}s  !j&6q&9E&A&A$Q'E!+.q1 )}Y$.2N%NY^___gghjkk L!!!+,2&&  !#z/*E#jj// l1o* *3C I[I[I]I]4^4^*_*_   &A&CG\22 ( (1"**//11)"1 ,,..A*+|A'4>3N)N O OP >00[0YY()ADNDYDY[\Z]D^D^ltuu  ;'''' MM#      !9  rc t|tr|g} |j}n#t$ri}YnwxYwg}|D]}||}|||dddddd}t |dkrtd|dl|d}td|d |j |d | |tt|}t |dkrtd tj|}|S) NFr T)add_special_tokensreturn_attention_maskreturn_token_type_ids max_length truncationrrzThe specified target token `zd` does not exist in the model vocabulary. We cannot replace it with anything meaningful, ignoring itz:` does not exist in the model vocabulary. Replacing with `z`.z1At least one target must be provided when passed.)r+strr get_vocab Exceptiongetlenloggerwarningconvert_ids_to_tokensrOr,set ValueErrorr!array)rtargetsvocabrQtargetid_rs rget_target_idszFillMaskPipeline.get_target_idss gs # # iG N,,..EE   EEE   # #F))F##C{ NN',*/*/ # + y>>Q&&NNUvUUUl U6UU'+~'K'KC'P'PUUU   c " " " "#j//** z??a  PQQ QXj)) s4 AAci}|||d<i}|||}||d<|||d<|jjtd|jjd|i|fS)Nr3rQrPr z-The tokenizer does not define a `mask_token`.)rvrrr r$r%)rrPrrr3preprocess_paramspostprocess_paramsrQs r_sanitize_parametersz%FillMaskPipeline._sanitize_parameterss  '4D 0 1  ,,W55J/9 | ,  */ w ' > ' /#TZ9;j !"&888rr2kwargsc dSNr7rr2r{s r__call__zFillMaskPipeline.__call__sLOCrc dSr}r7r~s rrzFillMaskPipeline.__call__sX[X[rc tj|fi|}t|trt |dkr|dS|S)a Fill the masked token in the text(s) given as inputs. Args: inputs (`str` or `list[str]`): One or several texts (or one list of prompts) with masked tokens. targets (`str` or `list[str]`, *optional*): When passed, the model will limit the scores to the passed targets instead of looking up in the whole vocab. If the provided targets are not in the model vocab, they will be tokenized and the first resulting token will be used (with a warning, and that might be slower). top_k (`int`, *optional*): When passed, overrides the number of predictions to return. Return: A list or a list of list of `dict`: Each result comes as list of dictionaries with the following keys: - **sequence** (`str`) -- The corresponding input with the mask token prediction. - **score** (`float`) -- The corresponding probability. - **token** (`int`) -- The predicted token id (to replace the masked one). - **token_str** (`str`) -- The predicted token (to replace the masked one). r r)superrr+r,rk)rr2r{rR __class__s rrzFillMaskPipeline.__call__sT,#%''"644V44 fd # # F q(8(81: r)NN)r:N)NNN)__name__ __module__ __qualname___load_processor_load_image_processor_load_feature_extractor_load_tokenizerr r!ndarrayrr(r.dictrgr5r9r`rvrzrrr,r __classcell__)rs@rrrsO!#O2h-BJ "*?-????=A   c= !     ''''R'''R9999*OsOcOd4S>6JOOOXO [tCy[C[Dd3PS8nAUrs  CCCCCCCCCCVVVVVVVVVVVVLLL  H % %4000Y  gggggxgg  gggr