ZjdSrSSKJr SSKrSSKJr SSKJrJrJr SSK J r SSK J r SS KJr SS KJrJrJrJrJrJr SS KJrJr SS KJr SS KJr SSKJrJ r J!r!J"r" SSK#J$r$ SSK%J&r& SSK'J(r( \"RR"\*5r+"SS\RX5r-"SS\RX5r."SS\RX5r/"SS\RX5r0"SS\RX5r1"SS\RX5r2"SS \RX5r3"S!S"\RX5r4"S#S$\RX5r5SCS%\RXS&\RlS'\RlS(\RlS)\RlS-S*\7S+\74S,jjr8"S-S.\RX5r9"S/S0\RX5r:"S1S2\5r;"S3S4\RX5r<\ "S5S6\55r=\ "S7S8\=55r>\ "S9S:\=55r?\ "S;S<9"S=S>\=55r@\ "S?S<9"S@SA\=55rA/SBQrBg)DzPyTorch MarkupLM model.)CallableN)nn)BCEWithLogitsLossCrossEntropyLossMSELoss)initialization)ACT2FN)GradientCheckpointingLayer)BaseModelOutputBaseModelOutputWithPoolingMaskedLMOutputQuestionAnsweringModelOutputSequenceClassifierOutputTokenClassifierOutput)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack)apply_chunking_to_forward)TransformersKwargsauto_docstringcan_return_tuplelogging)merge_with_config_defaults)capture_outputs)MarkupLMConfigc6^\rSrSrSrU4SjrSSjrSrU=r$)XPathEmbeddings-zConstruct the embeddings from xpath tags and subscripts. We drop tree-id in this version, as its info can be covered by xpath. c >[TU]5 URUl[R"UR UR-UR 5Ul[R"UR5Ul [R"5Ul [R"UR UR-SUR -5Ul [R"SUR -UR 5Ul[R"[!UR5Vs/sH.n[R""UR$UR 5PM0 sn5Ul[R"[!UR5Vs/sH.n[R""UR(UR 5PM0 sn5Ulgs snfs snf)N)super__init__ max_depthrLinearxpath_unit_hidden_size hidden_sizexpath_unitseq2_embeddingsDropouthidden_dropout_probdropoutReLU activationxpath_unitseq2_inner inner2emb ModuleListrange Embeddingmax_xpath_tag_unit_embeddingsxpath_tag_sub_embeddingsmax_xpath_subs_unit_embeddingsxpath_subs_sub_embeddings)selfconfig_ __class__s /root/GenerationalWealth/GenerationalWealth/venv/lib/python3.13/site-packages/transformers/models/markuplm/modeling_markuplm.pyr$XPathEmbeddings.__init__3sd )))+63P3PSWSaSa3acicucu)v&zz&"<"<= '')$&IIf.K.Kdnn.\^_bhbtbt^t$u!1v'9'9#96;M;MN(* t~~. .A VAA6C`C`a. ) %*,t~~. .A VBBFDaDab. * &   s 75G!5G&c /n/n[UR5H_nURURU"USS2SS2U455 URURU"USS2SS2U455 Ma [ R "USS9n[ R "USS9nX4-nURURURURU5555nU$)Ndim) r2r%appendr5r7torchcatr0r,r.r/)r8xpath_tags_seqxpath_subs_seqxpath_tags_embeddingsxpath_subs_embeddingsixpath_embeddingss r<forwardXPathEmbeddings.forwardMs " "t~~&A ! ( ()F)Fq)I.YZ\]_`Y`Ja)b c ! ( ()G)G)J>Z[]^`aZaKb)c d'!& *?R H % *?R H0H>>$,,ttG`G`aqGr7s*tu)r.r,r0r%r7r5r)r/)NN) __name__ __module__ __qualname____firstlineno____doc__r$rK__static_attributes__ __classcell__r;s@r<rr-s  4  rMrcf^\rSrSrSrU4Sjr\S5r\SSj5rS Sjr Sr U=r $) MarkupLMEmbeddings_zGConstruct the embeddings from word, position and token_type embeddings.cX>[TU]5 Xl[R"UR UR URS9Ul[R"URUR 5Ul URUl [U5Ul [R"URUR 5Ul[R "UR UR"S9Ul[R$"UR&5UlUR+S[,R."UR5R1S5SS9 URUl[R"URUR UR2S9Ul g)N) padding_idxeps position_idsrr?F) persistent)r#r$r9rr3 vocab_sizer( pad_token_idword_embeddingsmax_position_embeddingsposition_embeddingsr%rrJtype_vocab_sizetoken_type_embeddings LayerNormlayer_norm_epsr*r+r,register_bufferrCarangeexpandrZr8r9r;s r<r$MarkupLMEmbeddings.__init__bs8  !||F,=,=v?Q?Q_e_r_rs#%<<0N0NPVPbPb#c )) / 7%'\\&2H2H&J\J\%]"f&8&8f>S>STzz&"<"<=   ELL)G)GHOOPWXej  "..#%<<  * *F,>,>DL\L\$  rMcUR5SSnUSn[R"US-X1-S-[RURS9nUR S5R U5$)z We are provided embeddings directly. We cannot infer which are padded so just generate sequential position ids. Args: inputs_embeds: torch.Tensor Returns: torch.Tensor Nr?rdtypedevicer)sizerCrjlongrq unsqueezerk) inputs_embedsrZ input_shapesequence_lengthr]s r<&create_position_ids_from_inputs_embeds9MarkupLMEmbeddings.create_position_ids_from_inputs_embedszsn$((*3B/ %a.|| !O_:Q>ejjYfYmYm %%a(// <>*::;5::V\7N  !![[44uzzd;'4>>*::;5::V\8N)"66|D $ : :> J00P%;>SSVff ^^J/ \\*- rM) rgr9r,r%rZrdrfrbrJ)r)NNNNNN) rNrOrPrQrRr$ staticmethodrxrrKrSrTrUs@r<rWrW_sQQ 0==" 8 8"11rMrWcz^\rSrSrU4SjrS\R S\R S\R 4SjrSrU=r $)MarkupLMSelfOutputc(>[TU]5 [R"URUR5Ul[R "URURS9Ul[R"UR5Ul gNr[) r#r$rr&r(densergrhr*r+r,rls r<r$MarkupLMSelfOutput.__init__s` YYv1163E3EF f&8&8f>S>STzz&"<"<= rM hidden_states input_tensorreturncpURU5nURU5nURX-5nU$Nrr,rgr8rrs r<rKMarkupLMSelfOutput.forward5 =1  ]3 }'CD rMrgrr, rNrOrPrQr$rCTensorrKrSrTrUs@r<rr6> U\\RWR^R^rMrcb^\rSrSrU4SjrS\R S\R 4SjrSrU=r $)MarkupLMIntermediatec>[TU]5 [R"URUR 5Ul[UR[5(a[URUl gURUl gr) r#r$rr&r(intermediate_sizer isinstance hidden_actstrr intermediate_act_fnrls r<r$MarkupLMIntermediate.__init__s` YYv1163K3KL f'' - -'-f.?.?'@D $'-'8'8D $rMrrcJURU5nURU5nU$rrrr8rs r<rKMarkupLMIntermediate.forwards& =1 00? rMrrrUs@r<rrs(9U\\ellrMrcz^\rSrSrU4SjrS\R S\R S\R 4SjrSrU=r $)MarkupLMOutputc(>[TU]5 [R"URUR 5Ul[R"UR URS9Ul[R"UR5Ul gr) r#r$rr&rr(rrgrhr*r+r,rls r<r$MarkupLMOutput.__init__s` YYv779K9KL f&8&8f>S>STzz&"<"<= rMrrrcpURU5nURU5nURX-5nU$rrrs r<rKMarkupLMOutput.forwardrrMrrrUs@r<rrrrMrcb^\rSrSrU4SjrS\R S\R 4SjrSrU=r $)MarkupLMPooleric>[TU]5 [R"URUR5Ul[R "5Ulgr)r#r$rr&r(rTanhr.rls r<r$MarkupLMPooler.__init__s9 YYv1163E3EF '')rMrrc\USS2S4nURU5nURU5nU$)Nr)rr.)r8rfirst_token_tensor pooled_outputs r<rKMarkupLMPooler.forwards6+1a40 #56  6 rM)r.rrrUs@r<rrs($ U\\ellrMrcb^\rSrSrU4SjrS\R S\R 4SjrSrU=r $)MarkupLMPredictionHeadTransformicp>[TU]5 [R"URUR5Ul[ UR[5(a[URUl OURUl [R"URURS9Ul gr) r#r$rr&r(rrrrr transform_act_fnrgrhrls r<r$(MarkupLMPredictionHeadTransform.__init__s~ YYv1163E3EF f'' - -$*6+<+<$=D !$*$5$5D !f&8&8f>S>STrMrrclURU5nURU5nURU5nU$r)rrrgrs r<rK'MarkupLMPredictionHeadTransform.forwards4 =1 --m< }5 rM)rgrrrrUs@r<rrs)UU\\ellrMrc.^\rSrSrU4SjrSrSrU=r$)MarkupLMLMPredictionHeadi#c>[TU]5 [U5Ul[R "UR URSS9Ul[R"[R"UR55Ul g)NT)bias) r#r$r transformrr&r(r`decoder ParameterrCrrrls r<r$!MarkupLMLMPredictionHead.__init__$s[ 8@yy!3!3V5F5FTR LLV->->!?@ rMcJURU5nURU5nU$r)rrrs r<rK MarkupLMLMPredictionHead.forward-s$}5  ]3 rM)rrr)rNrOrPrQr$rKrSrTrUs@r<rr#sArMrcb^\rSrSrU4SjrS\R S\R 4SjrSrU=r $)MarkupLMOnlyMLMHeadi4cB>[TU]5 [U5Ulgr)r#r$r predictionsrls r<r$MarkupLMOnlyMLMHead.__init__5s 3F;rMsequence_outputrc(URU5nU$rr)r8rprediction_scoress r<rKMarkupLMOnlyMLMHead.forward9s ,,_=  rMrrrUs@r<rr4s(<!u||! !!rMrmodulequerykeyvalueattention_maskscalingr,c [R"XRSS55U-nUbX-n[RR US[R S9RUR5n[RRXURS9n[R"X5n U RSS5R5n X4$)Nrr?)rArp)ptrainingr) rCmatmul transposer functionalsoftmaxfloat32torpr,r contiguous) rrrrrrr,kwargs attn_weights attn_outputs r<eager_attention_forwardr?s<<}}Q':;gEL!#4 ==((2U]](SVVW\WbWbcL==((6??([L,,|3K''1-88:K  $$rMc ^\rSrSrU4SjrS S\R S\RS-S\\ S\ \R \R S-44Sjjr S r U=r $) MarkupLMSelfAttentioniVc6>[TU]5 URUR-S:wa7[ US5(d&[ SURSURS35eXlURUl[URUR- 5UlURUR-Ul [R"URUR5Ul [R"URUR5Ul [R"URUR5Ul[R"UR 5UlUR UlURS-Ulg)Nrembedding_sizezThe hidden size (z6) is not a multiple of the number of attention heads ()g)r#r$r(num_attention_headshasattr ValueErrorr9r|attention_head_size all_head_sizerr&rrrr*attention_probs_dropout_probr,attention_dropoutrrls r<r$MarkupLMSelfAttention.__init__WsD     : : :a ?PVXhHiHi#F$6$6#78 445Q8   #)#=#= #&v'9'9F[TU]5 [U5Ul[ U5Ulgr)r#r$rr8routputrls r<r$MarkupLMAttention.__init__s& )&1 (0 rMNrrrrc ZUnUR"U4SU0UD6upURX5nU$Nr)r8r)r8rrrresidualr:s r<rKMarkupLMAttention.forwardsE !99  )    M< rM)rr8r) rNrOrPrQr$rCrr rrrKrSrTrUs@r<rrsV148 || ))D0 +,    rMrc ^\rSrSrU4SjrS S\R S\RS-S\\ S\R 4Sjjr S r S r U=r $) MarkupLMLayeric>[TU]5 URUlSUl[ U5Ul[ U5Ul[U5Ul g)Nr) r#r$chunk_size_feed_forward seq_len_dimr attentionr intermediaterrrls r<r$MarkupLMLayer.__init__sI '-'E'E$*6208$V, rMNrrrrc UR"U4SU0UD6n[URURURU5nU$r)rrfeed_forward_chunkrr)r8rrrs r<rKMarkupLMLayer.forwardsW   )  2  # #T%A%A4CSCSUb rMcJURU5nURX!5nU$r)rr)r8attention_outputintermediate_output layer_outputs r<r  MarkupLMLayer.feed_forward_chunks)"//0@A{{#6I rM)rrrrrr)rNrOrPrQr$rCrr rrrKr rSrTrUs@r<rrs[-48||))D0+,   $rMrc |^\rSrSrU4SjrS S\R S\RS-S\\ S\ 4Sjjr S r U=r $) MarkupLMEncoderic>[TU]5 Xl[R"[ UR 5Vs/sHn[U5PM sn5UlSUl gs snf)NF) r#r$r9rr1r2num_hidden_layersrlayergradient_checkpointing)r8r9rIr;s r<r$MarkupLMEncoder.__init__sR  ]]5IaIaCb#cCbaM&$9Cb#cd &+#$dsA&Nrrrrc NURH nU"UU40UD6nM [US9$)N)last_hidden_state)r+r )r8rrr layer_modules r<rKMarkupLMEncoder.forwards> !JJL(M'+  rM)r9r,r+r)rNrOrPrQr$rCrr rrr rKrSrTrUs@r<r(r(sR,48 || ))D0 +,     rMr(cj^\rSrSr%\\S'Sr\\S.r \ R"5U4Sj5r Sr U=r$)MarkupLMPreTrainedModelir9markuplm)r attentionscr>[TU]U5 [U[5(a![R "UR 5 g[U[5(a\[R"UR[R"URRS5RS55 gg)zInitialize the weightsr?r^N)r# _init_weightsrrinitzeros_rrWcopy_r]rCrjrrk)r8rr;s r<r7%MarkupLMPreTrainedModel._init_weightss~ f% f6 7 7 KK $  2 3 3 JJv**ELL9L9L9R9RSU9V,W,^,^_f,g h4rM)rNrOrPrQr__annotations__base_model_prefixrr_can_record_outputsrCno_gradr7rSrTrUs@r<r3r3s9 "&+  ]]_iirMr3cX^\rSrSrSU4SjjrSrSr\\\ SS\ RS-S\ RS-S\ RS-S \ RS-S \ RS-S \ RS-S \ RS-S \ \S\\-4Sjj555rSrU=r$) MarkupLMModelic>[TU]U5 Xl[U5Ul[ U5UlU(a [U5OSUlUR5 g)z^ add_pooling_layer (bool, *optional*, defaults to `True`): Whether to add a pooling layer N) r#r$r9rWrr(encoderrpooler post_init)r8r9add_pooling_layerr;s r<r$MarkupLMModel.__init__sK   ,V4&v. 0AnV,t  rMc.URR$rrrb)r8s r<get_input_embeddings"MarkupLMModel.get_input_embeddingss...rMc$XRlgrrJ)r8rs r<set_input_embeddings"MarkupLMModel.set_input_embeddingss */'rMNrrErFrrr]rurrc UbUb [S5eUb"URX5 UR5n O"UbUR5SSn O [S5eUb URO URn Uc[R "XS9nUc$[R "U [RU S9nURS5RS5n U RURS 9n S U - S -n URUUUUUUS 9n UR"U U 40UD6n U S nURbURU5OSn[UUS9$)ao xpath_tags_seq (`torch.LongTensor` of shape `(batch_size, sequence_length, config.max_depth)`, *optional*): Tag IDs for each token in the input sequence, padded up to config.max_depth. xpath_subs_seq (`torch.LongTensor` of shape `(batch_size, sequence_length, config.max_depth)`, *optional*): Subscript IDs for each token in the input sequence, padded up to config.max_depth. Examples: ```python >>> from transformers import AutoProcessor, MarkupLMModel >>> processor = AutoProcessor.from_pretrained("microsoft/markuplm-base") >>> model = MarkupLMModel.from_pretrained("microsoft/markuplm-base") >>> html_string = " Page Title " >>> encoding = processor(html_string, return_tensors="pt") >>> outputs = model(**encoding) >>> last_hidden_states = outputs.last_hidden_state >>> list(last_hidden_states.shape) [1, 4, 768] ```NzDYou cannot specify both input_ids and inputs_embeds at the same timer?z5You have to specify either input_ids or inputs_embeds)rqrorr)rpg?g)rrErFr]rrur)r/ pooler_output)r%warn_if_padding_and_no_attention_maskrrrqrCrrrsrtrrprrDrEr )r8rrErFrrr]rurrvrqextended_attention_maskembedding_outputencoder_outputsrrs r<rKMarkupLMModel.forward soJ  ]%>cd d  "  6 6y Q#..*K  &',,.s3KTU U%.%:!! @T@T  !"ZZ CN  !"[[EJJvVN"0":":1"="G"G"J"9"<"<4::"<"N#&)@#@H"L??))%)' + ,,  #   *!,8< 8O O4UY )-'  rM)r9rrDrE)T)NNNNNNN)rNrOrPrQr$rKrNrrrrC LongTensorr rrrr rKrSrTrUs@r<rBrBs  /0 .2262637260426K ##d*K ((4/K ((4/ K ))D0 K ((4/ K &&-K ((4/K +,K  + +K  K rMrBc^\rSrSrU4Sjr\\SS\RS-S\RS-S\RS-S\RS-S\RS-S \RS-S \RS-S \RS-S \RS-S \ \ S\ \R\ -4Sjj55r SrU=r$)MarkupLMForQuestionAnsweringi\c>[TU]U5 URUl[USS9Ul[ R "URUR5UlUR5 gNF)rG) r#r$ num_labelsrBr4rr&r( qa_outputsrFrls r<r$%MarkupLMForQuestionAnswering.__init___sU   ++%fF ))F$6$68I8IJ rMNrrErFrrr]rustart_positions end_positionsrrc UR"U4UUUUUUS.U D6n U Sn URU 5n U RSSS9upURS5R 5nURS5R 5nSnUbU b[ UR 55S:aURS5n[ U R 55S:aU RS5n UR S5nURSU5 U RSU5 [US9nU"X5nU"X5nUU-S- n[UUUU RU RS 9$) a xpath_tags_seq (`torch.LongTensor` of shape `(batch_size, sequence_length, config.max_depth)`, *optional*): Tag IDs for each token in the input sequence, padded up to config.max_depth. xpath_subs_seq (`torch.LongTensor` of shape `(batch_size, sequence_length, config.max_depth)`, *optional*): Subscript IDs for each token in the input sequence, padded up to config.max_depth. Examples: ```python >>> from transformers import AutoProcessor, MarkupLMForQuestionAnswering >>> import torch >>> processor = AutoProcessor.from_pretrained("microsoft/markuplm-base-finetuned-websrc") >>> model = MarkupLMForQuestionAnswering.from_pretrained("microsoft/markuplm-base-finetuned-websrc") >>> html_string = " My name is Niels " >>> question = "What's his name?" >>> encoding = processor(html_string, questions=question, return_tensors="pt") >>> with torch.no_grad(): ... outputs = model(**encoding) >>> answer_start_index = outputs.start_logits.argmax() >>> answer_end_index = outputs.end_logits.argmax() >>> predict_answer_tokens = encoding.input_ids[0, answer_start_index : answer_end_index + 1] >>> processor.decode(predict_answer_tokens).strip() 'Niels' ```rErFrrr]rurrr?r@N) ignore_indexr)loss start_logits end_logitsrr5) r4r]splitsqueezerlenrrclamp_rrrr5)r8rrErFrrr]rur_r`routputsrlogitsrerf total_loss ignored_indexloss_fct start_lossend_losss r<rK$MarkupLMForQuestionAnswering.forwardisxZ--   ))))%'    "!*1#)<<r<#: #++B/::< ''+668   &=+D?'')*Q."1"9"9""==%%'(1, - 5 5b 9 (--a0M  " "1m 4  M 2']CH!,@J :H$x/14J+%!!//))   rM)r4r\r]) NNNNNNNNN)rNrOrPrQr$rrrCrrrrrrKrSrTrUs@r<rYrY\s *..2.2.2.2,0-1/3-1T <<$&T  t+T  t+ T  t+ T  t+ T llT)T ||d*T ,T ||d*T +,T  u|| ; ;T T rMrYzC MarkupLM Model with a `token_classification` head on top. ) custom_introcx^\rSrSrU4Sjr\\SS\RS-S\RS-S\RS-S\RS-S\RS-S \RS-S \RS-S \RS-S \ \ S \ \R\ -4Sjj55r SrU=r$)MarkupLMForTokenClassificationicd>[TU]U5 URUl[USS9UlUR b UR O UR n[R"U5Ul [R"URUR5Ul UR5 gr[)r#r$r\rBr4classifier_dropoutr+rr*r,r&r( classifierrFr8r9rwr;s r<r$'MarkupLMForTokenClassification.__init__s   ++%fF )/)B)B)NF % %TZTnTn zz"45 ))F$6$68I8IJ rMNrrErFrrr]rulabelsrrc 8UR"U4UUUUUUS.U D6n U Sn URU 5n Sn UbF[5nU"U RSURR 5URS55n [ U U U RU RS9$)a xpath_tags_seq (`torch.LongTensor` of shape `(batch_size, sequence_length, config.max_depth)`, *optional*): Tag IDs for each token in the input sequence, padded up to config.max_depth. xpath_subs_seq (`torch.LongTensor` of shape `(batch_size, sequence_length, config.max_depth)`, *optional*): Subscript IDs for each token in the input sequence, padded up to config.max_depth. labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): Labels for computing the token classification loss. Indices should be in `[0, ..., config.num_labels - 1]`. Examples: ```python >>> from transformers import AutoProcessor, AutoModelForTokenClassification >>> import torch >>> processor = AutoProcessor.from_pretrained("microsoft/markuplm-base") >>> processor.parse_html = False >>> model = AutoModelForTokenClassification.from_pretrained("microsoft/markuplm-base", num_labels=7) >>> nodes = ["hello", "world"] >>> xpaths = ["/html/body/div/li[1]/div/span", "/html/body/div/li[1]/div/span"] >>> node_labels = [1, 2] >>> encoding = processor(nodes=nodes, xpaths=xpaths, node_labels=node_labels, return_tensors="pt") >>> with torch.no_grad(): ... outputs = model(**encoding) >>> loss = outputs.loss >>> logits = outputs.logits ```rbrNr?rdrlrr5) r4rxrrr9r\rrr5)r8rrErFrrr]rur{rrkrrrdros r<rK&MarkupLMForTokenClassification.forwardsV--   ))))%'    "!* OOO<  ')H!&&r4;;+A+AB BD %$!//))   rM)rxr,r4r\NNNNNNNN)rNrOrPrQr$rrrCrrrrrrKrSrTrUs@r<rurus *..2.2.2.2,0-1&*D <<$&D  t+D  t+ D  t+ D  t+ D llT)D ||d*D  t#D +,D  u|| ~ -D D rMruz MarkupLM Model transformer with a sequence classification/regression head on top (a linear layer on top of the pooled output) e.g. for GLUE tasks. cx^\rSrSrU4Sjr\\SS\RS-S\RS-S\RS-S\RS-S\RS-S \RS-S \RS-S \RS-S \ \ S \ \R\ -4Sjj55r SrU=r$)!MarkupLMForSequenceClassificationi cr>[TU]U5 URUlXl[ U5UlUR b UR O URn[R"U5Ul [R"URUR5Ul UR5 gr)r#r$r\r9rBr4rwr+rr*r,r&r(rxrFrys r<r$*MarkupLMForSequenceClassification.__init__(s   ++ %f- )/)B)B)NF % %TZTnTn zz"45 ))F$6$68I8IJ rMNrrErFrrr]rur{rrc UR"U4UUUUUUS.U D6n U Sn URU 5n URU 5n Sn UGbURRcUR S:XaSURlOoUR S:aNUR [R:XdUR [R:XaSURlOSURlURRS:XaI[5nUR S:Xa&U"U R5UR55n OU"X5n OURRS:Xa=[5nU"U RSUR 5URS55n O,URRS:Xa[5nU"X5n [U U U R U R"S9$) a  xpath_tags_seq (`torch.LongTensor` of shape `(batch_size, sequence_length, config.max_depth)`, *optional*): Tag IDs for each token in the input sequence, padded up to config.max_depth. xpath_subs_seq (`torch.LongTensor` of shape `(batch_size, sequence_length, config.max_depth)`, *optional*): Subscript IDs for each token in the input sequence, padded up to config.max_depth. labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). Examples: ```python >>> from transformers import AutoProcessor, AutoModelForSequenceClassification >>> import torch >>> processor = AutoProcessor.from_pretrained("microsoft/markuplm-base") >>> model = AutoModelForSequenceClassification.from_pretrained("microsoft/markuplm-base", num_labels=7) >>> html_string = " Page Title " >>> encoding = processor(html_string, return_tensors="pt") >>> with torch.no_grad(): ... outputs = model(**encoding) >>> loss = outputs.loss >>> logits = outputs.logits ```rbrN regressionsingle_label_classificationmulti_label_classificationr?r})r4r,rxr9 problem_typer\rprCrsr|rrhrrrrrr5)r8rrErFrrr]rur{rrkrrlrdros r<rK)MarkupLMForSequenceClassification.forward7sT--   ))))%'       ]3 /  {{''/??a'/;DKK,__q(fllejj.HFLL\a\e\eLe/LDKK,/KDKK,{{''<7"9??a'#FNN$4fnn6FGD#F3D))-JJ+- B @&++b/R))-II,./'!//))   rM)rxr9r,r4r\r)rNrOrPrQr$rrrCrrrrrrKrSrTrUs@r<rr s  *..2.2.2.2,0-1&*T <<$&T  t+T  t+ T  t+ T  t+ T llT)T ||d*T  t#T +,T  u|| 7 7T T rMr)rYrrurBr3)r)CrRcollections.abcrrCrtorch.nnrrrr r8 activationsr modeling_layersr modeling_outputsr r rrrrmodeling_utilsrrprocessing_utilsr pytorch_utilsrutilsrrrr utils.genericrutils.output_capturingrconfiguration_markuplmr get_loggerrNloggerModulerrWrrrrrrrrfloatrrrrr(r3rBrYrur__all__r<rMr<rsJ$ AA&!9G&6RR752   H %/ bii/ dpph   299  RYY  RYY  bii$ ryy "!"))!$% II% <<% % << % LL4' %  %%.3)BII3)n ..D bii 2ioii$f +f f Rb #:b b J V %<V  V r g (?g  g T rM