`ikjdZddlZddlmZmZddlZddlmZddlmZm Z m Z ddl m Z ddl mZdd lmZmZmZmZmZdd lmZdd lmZmZmZdd lmZmZd dlmZej e!Z"Gddej#Z$Gddej#Z%Gddej#Z&Gddej#Z'Gddej#Z(Gddej#Z)Gddej#Z*GddeZ+Gdd ej#Z,Gd!d"ej#Z-eGd#d$eZ.eGd%d&e.Z/ed'(Gd)d*e.Z0eGd+d,e.Z1Gd-d.ej#Z2eGd/d0e.Z3gd1Z4dS)2zPyTorch LiLT model.N)OptionalUnion)nn)BCEWithLogitsLossCrossEntropyLossMSELoss)ACT2FN)GradientCheckpointingLayer)BaseModelOutputBaseModelOutputWithPoolingQuestionAnsweringModelOutputSequenceClassifierOutputTokenClassifierOutput)PreTrainedModel)apply_chunking_to_forward find_pruneable_heads_and_indicesprune_linear_layer)auto_docstringlogging) LiltConfigc:eZdZfdZ ddZdZdZxZS)LiltTextEmbeddingscttj|j|j|j|_tj|j|j|_ tj|j |j|_ tj |j|j |_ tj|j|_|dt%j|jddt+|dd|_|j|_tj|j|j|j|_ dS) N padding_idxeps position_ids)rF) persistentposition_embedding_typeabsolute)super__init__r Embedding vocab_size hidden_size pad_token_idword_embeddingsmax_position_embeddingsposition_embeddingstype_vocab_sizetoken_type_embeddings LayerNormlayer_norm_epsDropouthidden_dropout_probdropoutregister_buffertorcharangeexpandgetattrr#rselfconfig __class__s z/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/transformers/models/lilt/modeling_lilt.pyr&zLiltTextEmbeddings.__init__+s: !|F,=v?Q_e_rsss#%<0NPVPb#c#c %'\&2H&J\%]%]"f&8f>STTTz&"<==   EL)GHHOOPWXXej    (/v7PR\']']$".#%<  *F,>DL\$ $ $    Ncd|K|4|||j|j}n||}||}n|dd}|+t j|t j|j j}|| |}| |}||z}|j dkr| |}||z }||}||}||fS)Nr!dtypedevicer$)"create_position_ids_from_input_idsrtorC&create_position_ids_from_inputs_embedssizer6zeroslongr r+r/r#r-r0r4) r; input_idstoken_type_idsr inputs_embeds input_shaper/ embeddingsr-s r>forwardzLiltTextEmbeddings.forwardBs=  $#FFyRVRbccff$    $JJ=YY  #..**KK',,..ss3K  !"[EJtO`OghhhN   00;;M $ : :> J J"%::  ': 5 5"&":":<"H"H  - -J^^J// \\*-- <''r?c||}tj|d||z}||zS)a Args: Replace non-padding symbols with their position numbers. Position numbers begin at padding_idx+1. Padding symbols are ignored. This is modified from fairseq's `utils.make_positions`. x: torch.Tensor x: Returns: torch.Tensor rdim)neintr6cumsumtype_asrI)r;rJrmaskincremental_indicess r>rDz5LiltTextEmbeddings.create_position_ids_from_input_idsfs`||K((,,..$|Da888@@FF$N"''))K77r?c|dd}|d}tj|jdz||jzdztj|j}|d|S)z Args: We are provided embeddings directly. We cannot infer which are padded so just generate sequential position ids.: inputs_embeds: torch.Tensor Returns: torch.Tensor Nr!rrAr)rGr6r7rrIrC unsqueezer8)r;rLrMsequence_lengthr s r>rFz9LiltTextEmbeddings.create_position_ids_from_inputs_embedsss$((**3B3/ %a.|  q /D4D"Dq"HPUPZcpcw   %%a((// <<rr*sy     2 "("("("(H 8 8 8 = = = = = = =r?rc&eZdZfdZddZxZS)LiltLayoutEmbeddingscttj|j|jdz|_tj|j|jdz|_tj|j|jdz|_tj|j|jdz|_ |j |_ tj|j |j|j z|j |_tj|j|j|j z|_tj|j|j z|j|_tj|j|_dS)Nr) in_features out_featuresr)r%r&rr'max_2d_position_embeddingsr)x_position_embeddingsy_position_embeddingsh_position_embeddingsw_position_embeddingsr*rr,channel_shrink_ratiobox_position_embeddingsLinearbox_linear_embeddingsr0r1r2r3r4r:s r>r&zLiltLayoutEmbeddings.__init__sJ &(\&2SU[UgklUl%m%m"%'\&2SU[UgklUl%m%m"%'\&2SU[UgklUl%m%m"%'\&2SU[UgklUl%m%m"!.')|  *  &"= =(( ( ( $ &(Y*9KvOj9j& & & "f&8FrOzLiltLayoutEmbeddings.forwards c'+'A'A$qqq!!!Qw-'P'P $(,(B(B4111a=(Q(Q %(,(B(B4111a=(Q(Q %(,(B(B4111a=(Q(Q % % c c cZ[[ab b c!% : :4111a=4PQPQPQSTSTSTVWPW=;X Y Y $ : :4111a=4PQPQPQSTSTSTVWPW=;X Y Y&+i()))%%   ' ' ' #'+&@&@A\&]&]#"&">">|"L"L&AD[&[#&*nn5P&Q&Q#&*ll3N&O&O#**sBB B*B%%B*NN)r\r]r^r&rOr_r`s@r>rbrbsL>>>>>*++++++++r?rbc6eZdZdfd ZddZ d dZxZS) LiltSelfAttentionNct|j|jzdkr0t |ds t d|jd|jd|j|_t |j|jz |_|j|jz|_tj |j|j|_ tj |j|j|_ tj |j|j|_ tj |j|jz|j|jz|_tj |j|jz|j|jz|_tj |j|jz|j|jz|_tj|j|_|pt+|dd|_|jdks |jd kr6|j|_tjd |jzd z |j|_|j|_||_dS) Nrembedding_sizezThe hidden size (z6) is not a multiple of the number of attention heads ()r#r$ relative_keyrelative_key_queryrqr)r%r&r)num_attention_headshasattr ValueErrorrTattention_head_size all_head_sizerrnquerykeyvaluerl layout_query layout_key layout_valuer2attention_probs_dropout_probr4r9r#r,r'distance_embedding layer_idx)r;r<r#rr=s r>r&zLiltSelfAttention.__init__s2    : :a ? ?PVXhHiHi ?8F$688 4888  $*#= #&v'9F( ' -zC C $  '> 9 9T=Y]q=q=q+1+ID (&(l1v7U3UXY3Y[_[s&t&tD #$*$?!"r?rc|dd|j|j|zfz}|j|}|ddddS)Nr!rrqrr )rGrrviewpermute)r;xr new_x_shapes r>transpose_for_scoresz&LiltSelfAttention.transpose_for_scoressSffhhssmt'?AY]^A^&__ AFK yyAq!$$$r?Fc||||j}||||j}||||j}||} |||} |||} || } tj | | dd} tj || dd}|j dks |j dkr4| d}tj |tj|jdd}tj |tj|jdd}||z }|||jzdz }|| j}|j dkrtjd | |}| |z} n?|j dkr4tjd | |}tjd | |}| |z|z} | t-j|jz }|t-j|j|jzz }||z} ||z}|||z}t3jd |}||}|||z}tj ||}|d d dd}| dd|j|jzfz}|j|}|| |z} t3jd | }||}|||z}tj || }|d d dd}| dd|jfz}|j|}|r||f|fn||ff}|S)N)rr!rrrrA)rBzbhld,lrd->bhlrzbhrd,lrd->bhlrrQrrqr )rrrlrrrrrr6matmul transposer#rGr7rIrCrrr,rErBeinsummathsqrtrrSoftmaxr4r contiguousr)r; hidden_states layout_inputsattention_mask head_maskoutput_attentionslayout_value_layerlayout_key_layerlayout_query_layermixed_query_layer key_layer value_layer query_layerattention_scoreslayout_attention_scores seq_lengthposition_ids_lposition_ids_rdistancepositional_embeddingrelative_position_scoresrelative_position_scores_queryrelative_position_scores_keytmp_attention_scorestmp_layout_attention_scoreslayout_attention_probslayout_context_layernew_context_layer_shapeattention_probs context_layeroutputss r>rOzLiltSelfAttention.forwards"66t7H7H7W7W[_[t6uu44T__]5S5SW[Wp4qq!66t7H7H7W7W[_[t6uu JJ}55--dhh}.E.EFF // =0I0IJJ //0ABB  < Y5H5HR5P5PQQ"',/ACSC]C]^`bdCeCe"f"f  '> 9 9T=Y]q=q=q&++--a0J"\*EJ}OcdddiijlnoppN"\*EJ}OcdddiijkmoppN%6H#'#:#:8dFb;bef;f#g#g #7#:#:AR#:#S#S +~==+0<8H+Wk+l+l(#36N#N  -1EEE16>NP[]q1r1r./4|>,,77  - 9O _kBB %--aAq99DDFF "/"4"4"6"6ss";t?Q>S"S* *,CD ! :m1 2O D D "679  r?r{)rNNF)r\r]r^r&rrOr_r`s@r>r}r}sy############J%%%% \\\\\\\\r?r}cPeZdZfdZdejdejdejfdZxZS)LiltSelfOutputcttj|j|j|_tj|j|j|_tj|j |_ dSNr) r%r&rrnr)denser0r1r2r3r4r:s r>r&zLiltSelfOutput.__init__Gsf Yv163EFF f&8f>STTTz&"<== r?r input_tensorreturnc||}||}|||z}|SNrr4r0r;rrs r>rOzLiltSelfOutput.forwardM@ =11  ]33 }|'CDD r?r\r]r^r&r6TensorrOr_r`s@r>rrFi>>>>> U\RWR^r?rceZdZd fd ZdZ d dejdejdeejdeejd ee d e ejf d Z xZ S) LiltAttentionNc0tt||||_t ||_t |_|j}|j|j z|_t ||_ ||_dS)N)r#r) r%r&r}r;routputset pruned_headsr)rl layout_output)r;r<r#rori_hidden_sizer=s r>r&zLiltAttention.__init__Us %fF]irsss $V,, EE ,#/63NN+F33,r?ct|dkrdSt||jj|jj|j\}}t |jj||j_t |jj||j_t |jj ||j_ t |j j |d|j _ |jjt|z |j_|jj|jjz|j_ |j ||_dS)NrrrQ)lenrr;rrrrrrrrrrunion)r;headsindexs r> prune_headszLiltAttention.prune_headsas u::?? F7 490$)2OQUQb  u -TY_eDD *49=%@@ ,TY_eDD .t{/@%QOOO )- (EE (R %"&)"?$)B_"_  -33E::r?Frrrrrrc||||||}||dd|}||dd|}||ff|ddz} | S)Nrr)r;rr) r;rrrrr self_outputsattention_outputlayout_attention_outputrs r>rOzLiltAttention.forwardssyy         ;;|Aq'9=II"&"4"4\!_Q5G"W"W$&=>@rrTs - - - - - -;;;,7;15,1 ||!!23  E-.  $D>  u| r?rcBeZdZfdZdejdejfdZxZS)LiltIntermediatecttj|j|j|_t|jtrt|j|_ dS|j|_ dSr) r%r&rrnr)intermediate_sizer isinstance hidden_actstrr intermediate_act_fnr:s r>r&zLiltIntermediate.__init__sn Yv163KLL f' - - 9'-f.?'@D $ $ $'-'8D $ $ $r?rrcZ||}||}|Sr)rr)r;rs r>rOzLiltIntermediate.forwards, =11 00?? r?rr`s@r>rrs^99999U\elr?rcPeZdZfdZdejdejdejfdZxZS) LiltOutputcttj|j|j|_tj|j|j|_tj |j |_ dSr) r%r&rrnrr)rr0r1r2r3r4r:s r>r&zLiltOutput.__init__sf Yv79KLL f&8f>STTTz&"<== r?rrrc||}||}|||z}|Srrrs r>rOzLiltOutput.forwardrr?rr`s@r>rrrr?rceZdZd fd Z ddejdejdeejdeejdeed e ejf d Z d Z d Z xZ S) LiltLayerNct|j|_d|_t |||_t ||_t||_ |j }|j }|j |j z|_ |j |j z|_ t ||_ t||_||_ ||_ dS)Nrr)r%r&chunk_size_feed_forward seq_len_dimr attentionr intermediaterrr)rrllayout_intermediater)r;r<rrori_intermediate_sizer=s r>r&zLiltLayer.__init__s '-'E$&vCCC,V44 ((  , & 8#/63NN#)#;v?Z#Z #3F#;#; '//,#8   r?Frrrrrrc||||||}|dd}|dd}|dd} t|j|j|j|} t|j|j|j|} | | ff| z} | S)N)rrr)rrfeed_forward_chunkrrlayout_feed_forward_chunk) r;rrrrrself_attention_outputsrrr layer_outputlayout_layer_outputs r>rOzLiltLayer.forwards"&    / "0" " 2!4Q7"8";A">(,0  #T%A4CSUe  8  *D,H$JZ\s  !"5687Br?c\||}|||}|Sr)rrr;rintermediate_outputrs r>rzLiltLayer.feed_forward_chunks2"//0@AA{{#68HII r?c\||}|||}|Sr)rrrs r>rz#LiltLayer.layout_feed_forward_chunks4"667GHH))*=?OPP r?rr)r\r]r^r&r6rrrrrrOrrr_r`s@r>rrs999999*7;15,1 ||!!23  E-.  $D>  u| > r?rceZdZdfd Z ddejdejdeejdeejd eed eed eed e e eje ffd Z xZ S) LiltEncoderNct|_tjfdt jD|_d|_dS)Nc2g|]}t|S)r)r).0ir<s r> z(LiltEncoder.__init__..s&#l#l#lqIf$B$B$B#l#l#lr?F) r%r&r<r ModuleListrangenum_hidden_layerslayergradient_checkpointing)r;r<rr=s ` r>r&zLiltEncoder.__init__sa  ]#l#l#l#lERXRjLkLk#l#l#lmm &+###r?FTrrrrroutput_hidden_states return_dictrcX|rdnd}|rdnd} t|jD]R\} } |r||fz}||| nd} | |||| |} | dd}| dd}|r | | dfz} S|r||fz}|std||| fDSt||| S)Nrrc3K|]}||V dSrr)r vs r> z&LiltEncoder.forward..s4 =  !===r?)last_hidden_stater attentions) enumeraterrr )r;rrrrrrrall_hidden_statesall_self_attentionsr  layer_modulelayer_head_mask layer_outputss r>rOzLiltEncoder.forwardsC#7@BBD$5?bb4(44 P POA|# I$58H$H!.7.CillO(L! M*!,Q/M)!,Q/M  P&9]1=Mrrs,,,,,,7;15,1/4&*2 2 |2 |2 !!23 2 E-. 2 $D> 2 'tn2 d^2  uU\"O3 42 2 2 2 2 2 2 2 r?rcBeZdZfdZdejdejfdZxZS) LiltPoolercttj|j|j|_tj|_dSr)r%r&rrnr)rTanh activationr:s r>r&zLiltPooler.__init__"sC Yv163EFF '))r?rrcr|dddf}||}||}|SNr)rr$)r;rfirst_token_tensor pooled_outputs r>rOzLiltPooler.forward's@+111a40 #566  66 r?rr`s@r>r!r!!s^$$$$$ U\elr?r!c,eZdZUeed<dZdZgZdZdS)LiltPreTrainedModelr<liltTct|tjrT|jjd|jj|j |jj dSdSt|tj r_|jjd|jj|j +|jj|j  dSdSt|tj r?|jj |jj ddSdS)zInitialize the weightsg)meanstdNg?)rrrnweightdatanormal_r<initializer_rangebiaszero_r'rr0fill_)r;modules r> _init_weightsz!LiltPreTrainedModel._init_weights7s) fbi ( ( * M  & &CT[5R & S S S{&  &&((((('&  - - * M  & &CT[5R & S S S!- "6#56<<>>>>>.-  - - * K  " " $ $ $ M  $ $S ) ) ) ) ) * *r?N) r\r]r^r__annotations__base_model_prefixsupports_gradient_checkpointing_no_split_modulesr7rr?r>r*r*0sC &*#*****r?r*cpeZdZdfd ZdZdZdZe ddee j dee j d ee j d ee j d ee j d ee j d ee j dee dee dee de e e j effdZxZS) LiltModelTc(t|||_t||_t ||_t||_|rt|nd|_ | dS)zv add_pooling_layer (bool, *optional*, defaults to `True`): Whether to add a pooling layer N) r%r&r<rrNrblayout_embeddingsrencoderr!pooler post_init)r;r<add_pooling_layerr=s r>r&zLiltModel.__init__Js     ,V44!5f!=!="6** ,=Gj(((4  r?c|jjSrrNr+)r;s r>get_input_embeddingszLiltModel.get_input_embeddings[s ..r?c||j_dSrrE)r;rs r>set_input_embeddingszLiltModel.set_input_embeddings^s*/'''r?c|D]/\}}|jj|j|0dS)z Prunes heads of the model. heads_to_prune: dict of {layer_num: list of heads to prune in this layer} See base class PreTrainedModel N)itemsr@rrr)r;heads_to_prunerrs r> _prune_headszLiltModel._prune_headsasU +0022 C CLE5 L u % / ; ;E B B B B C Cr?NrJrtrrKr rrLrrrrc l||n |jj}| | n |jj} | | n |jj} ||t d|+||||} n.||dd} nt d| \} } ||jn|j}|$tj | dztj |}|tj | | f|}|gt|j dr1|j jddd| f}|| | }|}n!tj | tj |}||| }|||jj}| |||| \}}||| }||||||| | }|d }|j||nd}| s||f|d dzSt-|||j|jS)a bbox (`torch.LongTensor` of shape `(batch_size, sequence_length, 4)`, *optional*): Bounding boxes of each input sequence tokens. Selected in the range `[0, config.max_2d_position_embeddings-1]`. Each bounding box should be a normalized version in (x0, y0, x1, y1) format, where (x0, y0) corresponds to the position of the upper left corner in the bounding box, and (x1, y1) represents the position of the lower right corner. See [Overview](#Overview) for normalization. Examples: ```python >>> from transformers import AutoTokenizer, AutoModel >>> from datasets import load_dataset >>> tokenizer = AutoTokenizer.from_pretrained("SCUT-DLVCLab/lilt-roberta-en-base") >>> model = AutoModel.from_pretrained("SCUT-DLVCLab/lilt-roberta-en-base") >>> dataset = load_dataset("nielsr/funsd-layoutlmv3", split="train") >>> example = dataset[0] >>> words = example["tokens"] >>> boxes = example["bboxes"] >>> encoding = tokenizer(words, boxes=boxes, return_tensors="pt") >>> outputs = model(**encoding) >>> last_hidden_states = outputs.last_hidden_state ```NzDYou cannot specify both input_ids and inputs_embeds at the same timer!z5You have to specify either input_ids or inputs_embeds)rA)rCrK)rJr rKrL)rtr )rrrrrrr)r pooler_outputrr)r<rruse_return_dictr%warn_if_padding_and_no_attention_maskrGrCr6rHrIonesrrNrKr8get_extended_attention_mask get_head_maskrr?r@rAr rr)r;rJrtrrKr rrLrrrrM batch_sizerrCbuffered_token_type_ids buffered_token_type_ids_expandedextended_attention_maskembedding_outputlayout_embedding_outputencoder_outputssequence_outputr(s r>rOzLiltModel.forwardisP2C1N--TXT_Tq$8$D $+Jj &1%cdd d  "  6 6y. Q Q Q#..**KK  &',,..ss3KKTUU U!, J%.%:!! @T <;{T1FSSSD  !"Z*j)A6RRRN  !t(899 [*./*HKZK*X'3J3Q3QR\^h3i3i0!A!&[ SY!Z!Z!Z150P0PQ_al0m0m&&y$+2OPP )-%)' *9* * &,#'"8"8dQ]"8"^"^,,  #2/!5#'  *!,8< 8O O444UY  J#]3oabb6II I)-')7&1     r?)T) NNNNNNNNNN)r\r]r^r&rFrHrLrrr6rrrrr rOr_r`s@r>r=r=Hs"///000CCC-1'+1515/3,004,0/3&*p p EL)p u|$p !. p !. p u|, p EL)p   -p $D>p 'tnp d^p  uU\"$>> ?p p p ^p p p p p r?r=z LiLT Model transformer with a sequence classification/regression head on top (a linear layer on top of the pooled output) e.g. for GLUE tasks. ) custom_introcxeZdZfdZe ddeejdeejdeej deejdeejdeej d eej d eejd ee d ee d ee de e eje ffdZxZS)LiltForSequenceClassificationct||j|_||_t |d|_t ||_|dSNF)rC) r%r& num_labelsr<r=r+LiltClassificationHead classifierrBr:s r>r&z&LiltForSequenceClassification.__init__sg     + f>>> 088 r?NrJrtrrKr rrLlabelsrrrrc | | n |jj} ||||||||| | |  } | d} || }d}|t||j}|jjf|jdkr d|j_nN|jdkr7|jtj ks|jtj kr d|j_n d|j_|jjdkrWt}|jdkr1|| | }n|||}n|jjdkrGt}||d|j|d}n*|jjdkrt!}|||}| s|f| d dz}||f|zn|St#||| j| j S) a bbox (`torch.LongTensor` of shape `(batch_size, sequence_length, 4)`, *optional*): Bounding boxes of each input sequence tokens. Selected in the range `[0, config.max_2d_position_embeddings-1]`. Each bounding box should be a normalized version in (x0, y0, x1, y1) format, where (x0, y0) corresponds to the position of the upper left corner in the bounding box, and (x1, y1) represents the position of the lower right corner. See [Overview](#Overview) for normalization. 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 AutoTokenizer, AutoModelForSequenceClassification >>> from datasets import load_dataset >>> tokenizer = AutoTokenizer.from_pretrained("SCUT-DLVCLab/lilt-roberta-en-base") >>> model = AutoModelForSequenceClassification.from_pretrained("SCUT-DLVCLab/lilt-roberta-en-base") >>> dataset = load_dataset("nielsr/funsd-layoutlmv3", split="train") >>> example = dataset[0] >>> words = example["tokens"] >>> boxes = example["bboxes"] >>> encoding = tokenizer(words, boxes=boxes, return_tensors="pt") >>> outputs = model(**encoding) >>> predicted_class_idx = outputs.logits.argmax(-1).item() >>> predicted_class = model.config.id2label[predicted_class_idx] ```N rtrrKr rrLrrrrr regressionsingle_label_classificationmulti_label_classificationr!rqlosslogitsrr)r<rPr+rdrErC problem_typerbrBr6rIrTrsqueezerrrrrrr;rJrtrrKr rrLrerrrrr\rmrlloss_fctrs r>rOz%LiltForSequenceClassification.forwards"\&1%r_r_sh     15'+6:59371559-1,0/3&*` ` E,-` u|$` !!23 ` !!12 ` u/0 ` E-.`   12` )*` $D>` 'tn` d^`  uU\"$<< =` ` ` ^` ` ` ` ` r?r_cxeZdZfdZe ddeejdeejdeejdeejdeejdeejd eejd eejd ee d ee d ee de e ej e ffdZxZS)LiltForTokenClassificationcZt||j|_t|d|_|j|jn|j}tj||_ tj |j |j|_ | dSra)r%r&rbr=r+classifier_dropoutr3rr2r4rnr)rdrBr;r<rwr=s r>r&z#LiltForTokenClassification.__init__Ws     +f>>> )/)B)NF % %TZTn z"455 )F$68IJJ r?NrJrtrrKr rrLrerrrrc | | n |jj} ||||||||| | |  } | d} || } || }d}|`||j}t}||d|j |d}| s|f| ddz}||f|zn|St||| j | j S)a bbox (`torch.LongTensor` of shape `(batch_size, sequence_length, 4)`, *optional*): Bounding boxes of each input sequence tokens. Selected in the range `[0, config.max_2d_position_embeddings-1]`. Each bounding box should be a normalized version in (x0, y0, x1, y1) format, where (x0, y0) corresponds to the position of the upper left corner in the bounding box, and (x1, y1) represents the position of the lower right corner. See [Overview](#Overview) for normalization. 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 AutoTokenizer, AutoModelForTokenClassification >>> from datasets import load_dataset >>> tokenizer = AutoTokenizer.from_pretrained("SCUT-DLVCLab/lilt-roberta-en-base") >>> model = AutoModelForTokenClassification.from_pretrained("SCUT-DLVCLab/lilt-roberta-en-base") >>> dataset = load_dataset("nielsr/funsd-layoutlmv3", split="train") >>> example = dataset[0] >>> words = example["tokens"] >>> boxes = example["bboxes"] >>> encoding = tokenizer(words, boxes=boxes, return_tensors="pt") >>> outputs = model(**encoding) >>> predicted_class_indices = outputs.logits.argmax(-1) ```Nrgrr!rqrk) r<rPr+r4rdrErCrrrbrrrrps r>rOz"LiltForTokenClassification.forwardes,V&1%ruruTsi     15+/6:59371559-1,0/3&*N N E,-N u'(N !!23 N !!12 N u/0 N E-.N   12N )*N $D>N 'tnN d^N  uU\"$99 :N N N ^N N N N N r?ruc(eZdZdZfdZdZxZS)rcz-Head for sentence-level classification tasks.c4ttj|j|j|_|j|jn|j}tj||_ tj|j|j |_ dSr) r%r&rrnr)rrwr3r2r4rbout_projrxs r>r&zLiltClassificationHead.__init__s Yv163EFF )/)B)NF % %TZTn z"455  &"4f6GHH r?c |dddddf}||}||}tj|}||}||}|Sr&)r4rr6tanhr|)r;featureskwargsrs r>rOzLiltClassificationHead.forwardsj QQQ111W  LLOO JJqMM JqMM LLOO MM!  r?)r\r]r^__doc__r&rOr_r`s@r>rcrcsR77IIIIIr?rcceZdZfdZe ddeejdeejdeejdeejdeejdeejd eejd eejd eejd ee d ee dee de e ej e ffdZxZS)LiltForQuestionAnsweringct||j|_t|d|_t j|j|j|_| dSra) r%r&rbr=r+rrnr) qa_outputsrBr:s r>r&z!LiltForQuestionAnswering.__init__sj     +f>>> )F$68IJJ r?NrJrtrrKr rrLstart_positions end_positionsrrrrc | | n |jj} ||||||||| | |  } | d}||}|dd\}}|d}|d}d}|| t|dkr|d}t| dkr| d} |d}| d|}| d|} t|}|||}||| }||zdz }| s||f| ddz}||f|zn|St|||| j | j S) a bbox (`torch.LongTensor` of shape `(batch_size, sequence_length, 4)`, *optional*): Bounding boxes of each input sequence tokens. Selected in the range `[0, config.max_2d_position_embeddings-1]`. Each bounding box should be a normalized version in (x0, y0, x1, y1) format, where (x0, y0) corresponds to the position of the upper left corner in the bounding box, and (x1, y1) represents the position of the lower right corner. See [Overview](#Overview) for normalization. Examples: ```python >>> from transformers import AutoTokenizer, AutoModelForQuestionAnswering >>> from datasets import load_dataset >>> tokenizer = AutoTokenizer.from_pretrained("SCUT-DLVCLab/lilt-roberta-en-base") >>> model = AutoModelForQuestionAnswering.from_pretrained("SCUT-DLVCLab/lilt-roberta-en-base") >>> dataset = load_dataset("nielsr/funsd-layoutlmv3", split="train") >>> example = dataset[0] >>> words = example["tokens"] >>> boxes = example["bboxes"] >>> encoding = tokenizer(words, boxes=boxes, return_tensors="pt") >>> 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] >>> predicted_answer = tokenizer.decode(predict_answer_tokens) ```Nrgrrr!rQ) ignore_indexrq)rl start_logits end_logitsrr)r<rPr+rsplitrorrrGclamprrrr)r;rJrtrrKr rrLrrrrrrr\rmrr total_loss ignored_indexrq start_lossend_lossrs r>rOz LiltForQuestionAnswering.forwards^&1%rrs15+/6:593715596:48,0/3&*_ _ E,-_ u'(_ !!23 _ !!12 _ u/0 _ E-._   12_ "%"23_   01_ $D>_ 'tn_ d^_  uU\"$@@ A_ _ _ ^_ _ _ _ _ r?r)rr_rur=r*)5rrtypingrrr6rtorch.nnrrr activationsr modeling_layersr modeling_outputsr r rrrmodeling_utilsr pytorch_utilsrrrutilsrrconfiguration_liltr get_loggerr\loggerModulerrbr}rrrrrrr!r*r=r_rurcr__all__rr?r>rs """""""" AAAAAAAAAA!!!!!!999999.-----llllllllll,,,,,,,,******  H % %V=V=V=V=V=V=V=V=r5+5+5+5+5+295+5+5+pGGGGG GGGV     RY   11111BI111j     ry           99999*999x: : : : : "): : : |        *****/***.Q Q Q Q Q #Q Q Q h n n n n n $7n n  n b_ _ _ _ _ !4_ _ _ FRY,l l l l l 2l l l ^   r?