U 9%e@s<dZddlZddlmZmZmZddlZddlZddlmZddl m Z m Z m Z ddl mZddlmZmZmZmZmZdd lmZdd lmZdd lmZmZmZmZmZmZd d l m!Z!erddl"Z"ddl#m$Z$e%e&Z'dZ(dZ)ddgZ*Gdddej+Z,Gdddej+Z-Gdddej+Z.Gdddej+Z/Gdddej+Z0Gdddej+Z1Gdddej+Z2d=d#d$Z3Gd%d&d&ej+Z4Gd'd(d(eZ5d>d)d*Z6Gd+d,d,ej+Z7d-Z8d.Z9Gd/d0d0ej+Z:ed1e8Gd2d3d3e5Z;ed4e8Gd5d6d6e5ZdS)?z PyTorch LayoutLMv2 model.N)OptionalTupleUnion)nn)BCEWithLogitsLossCrossEntropyLossMSELoss)ACT2FN)BaseModelOutputBaseModelOutputWithPoolingQuestionAnsweringModelOutputSequenceClassifierOutputTokenClassifierOutput)PreTrainedModel)apply_chunking_to_forward)add_start_docstrings%add_start_docstrings_to_model_forwardis_detectron2_availableloggingreplace_return_docstringsrequires_backends)LayoutLMv2Config)META_ARCH_REGISTRYz!microsoft/layoutlmv2-base-uncasedrz"microsoft/layoutlmv2-large-uncasedcs(eZdZdZfddZddZZS)LayoutLMv2EmbeddingszGConstruct the embeddings from word, position and token_type embeddings.cstt|tj|j|j|jd|_t|j |j|_ t|j |j |_ t|j |j |_t|j |j|_t|j |j|_t|j|j|_tj|j|jd|_t|j|_|jdt|j ddddS)N) padding_idxeps position_ids)rF persistent)superr__init__r EmbeddingZ vocab_size hidden_sizeZ pad_token_idword_embeddingsZmax_position_embeddingsposition_embeddingsZmax_2d_position_embeddingsZcoordinate_sizex_position_embeddingsy_position_embeddingsZ shape_sizeh_position_embeddingsw_position_embeddingsZtype_vocab_sizetoken_type_embeddings LayerNormlayer_norm_epsDropouthidden_dropout_probdropoutregister_buffertorcharangeexpandselfconfig __class__q/var/www/html/Darija-Ai-API/env/lib/python3.8/site-packages/transformers/models/layoutlmv2/modeling_layoutlmv2.pyr$BszLayoutLMv2Embeddings.__init__c Cs$zt||dddddf}||dddddf}||dddddf}||dddddf}Wn,tk r}ztd|W5d}~XYnX||dddddf|dddddf}||dddddf|dddddf}tj||||||gdd} | S)Nrrr z;The `bbox` coordinate values should be within 0-1000 range.r dim)r)r* IndexErrorr+r,r4cat) r8bboxZleft_position_embeddingsZupper_position_embeddingsZright_position_embeddingsZlower_position_embeddingser+r,spatial_position_embeddingsr<r<r=!_calc_spatial_position_embeddingsTs( 22 z6LayoutLMv2Embeddings._calc_spatial_position_embeddings)__name__ __module__ __qualname____doc__r$rF __classcell__r<r<r:r=r?s rcs6eZdZfddZddZddZd d d ZZS) LayoutLMv2SelfAttentioncst|j|jdkr>t|ds>td|jd|jd|j|_|j|_t|j|j|_|j|j|_ |j |_ |j |_ |jrt j |jd|j dd|_t td d |j |_t td d |j |_n6t |j|j |_t |j|j |_t |j|j |_t |j|_dS) NrZembedding_sizezThe hidden size (z6) is not a multiple of the number of attention heads ()r Fbiasr)r#r$r&num_attention_headshasattr ValueErrorfast_qkvintattention_head_size all_head_sizehas_relative_attention_biashas_spatial_attention_biasrLinear qkv_linear Parameterr4zerosq_biasv_biasquerykeyvaluer0Zattention_probs_dropout_probr2r7r:r<r=r$os& z LayoutLMv2SelfAttention.__init__cCs6|dd|j|jf}|j|}|ddddS)Nr rr>rr )sizerPrUviewpermute)r8xZ new_x_shaper<r<r=transpose_for_scoress z,LayoutLMv2SelfAttention.transpose_for_scorescCs|jr||}tj|ddd\}}}||jkrN||j}||j}qd|dd}||jj|}||jj|}n||}| |}| |}|||fS)Nr r r?rr)r ) rSrZr4chunkZ ndimensionr]r^rcr_r`ra)r8 hidden_statesZqkvqkvZ_szr<r<r= compute_qkvs      z#LayoutLMv2SelfAttention.compute_qkvNFcCs||\}}} ||} ||} || } | t|j} t| | dd} |jr`| |7} |j rn| |7} | | tj t| jj} tjj| dtjd| }||}|dk r||}t|| }|dddd}|dd|jf}|j|}|r||fn|f}|S)Nr )r@dtyperr>rr )rmrfmathsqrtrUr4matmul transposerWrXfloatZ masked_fill_toboolfinforominrZ functionalZsoftmaxZfloat32Ztype_asr2rd contiguousrbrVrc)r8riattention_mask head_maskoutput_attentionsrel_pos rel_2d_posrjrkrlZ query_layerZ key_layerZ value_layerZattention_scoresZattention_probsZ context_layerZnew_context_layer_shapeoutputsr<r<r=forwards0        zLayoutLMv2SelfAttention.forward)NNFNN)rGrHrIr$rfrmrrKr<r<r:r=rLns rLcs&eZdZfddZdddZZS)LayoutLMv2Attentioncs"tt||_t||_dSN)r#r$rLr8LayoutLMv2SelfOutputoutputr7r:r<r=r$s  zLayoutLMv2Attention.__init__NFc Cs<|j||||||d}||d|}|f|dd} | S)Nr}r~rr)r8r) r8rirzr{r|r}r~Z self_outputsattention_outputrr<r<r=rs zLayoutLMv2Attention.forward)NNFNNrGrHrIr$rrKr<r<r:r=rs rcs$eZdZfddZddZZS)rcsBtt|j|j|_tj|j|jd|_t|j |_ dSNr) r#r$rrYr&denser.r/r0r1r2r7r:r<r=r$s zLayoutLMv2SelfOutput.__init__cCs&||}||}|||}|Srrr2r.r8ri input_tensorr<r<r=rs  zLayoutLMv2SelfOutput.forwardrr<r<r:r=rs rcs0eZdZfddZejejdddZZS)LayoutLMv2IntermediatecsBtt|j|j|_t|jt r6t |j|_ n|j|_ dSr) r#r$rrYr&intermediate_sizer isinstanceZ hidden_actstrr intermediate_act_fnr7r:r<r=r$s   zLayoutLMv2Intermediate.__init__)rireturncCs||}||}|Sr)rr)r8rir<r<r=rs  zLayoutLMv2Intermediate.forwardrGrHrIr$r4TensorrrKr<r<r:r=rs rcs4eZdZfddZejejejdddZZS)LayoutLMv2OutputcsBtt|j|j|_tj|j|jd|_t |j |_ dSr) r#r$rrYrr&rr.r/r0r1r2r7r:r<r=r$s zLayoutLMv2Output.__init__)rirrcCs&||}||}|||}|Srrrr<r<r=r s  zLayoutLMv2Output.forwardrr<r<r:r=rs rcs.eZdZfddZd ddZddZZS) LayoutLMv2Layercs:t|j|_d|_t||_t||_t||_ dS)Nr) r#r$chunk_size_feed_forward seq_len_dimr attentionr intermediaterrr7r:r<r=r$s    zLayoutLMv2Layer.__init__NFc CsL|j||||||d}|d}|dd} t|j|j|j|} | f| } | S)N)r|r}r~rr)rrfeed_forward_chunkrr) r8rirzr{r|r}r~Zself_attention_outputsrr layer_outputr<r<r=rs$   zLayoutLMv2Layer.forwardcCs||}|||}|Sr)rr)r8rZintermediate_outputrr<r<r=r9s  z"LayoutLMv2Layer.feed_forward_chunk)NNFNN)rGrHrIr$rrrKr<r<r:r=rs  rT c Csd}|r0|d}||dk|7}t|}nt| t|}|d}||k}|t||t||||tj}t |t ||d}|t |||7}|S)a Adapted from Mesh Tensorflow: https://github.com/tensorflow/mesh/blob/0cb87fe07da627bf0b7e60475d59f95ed6b5be3d/mesh_tensorflow/transformer/transformer_layers.py#L593 Translate relative position to a bucket number for relative attention. The relative position is defined as memory_position - query_position, i.e. the distance in tokens from the attending position to the attended-to position. If bidirectional=False, then positive relative positions are invalid. We use smaller buckets for small absolute relative_position and larger buckets for larger absolute relative_positions. All relative positions >=max_distance map to the same bucket. All relative positions <=-max_distance map to the same bucket. This should allow for more graceful generalization to longer sequences than the model has been trained on. Args: relative_position: an int32 Tensor bidirectional: a boolean - whether the attention is bidirectional num_buckets: an integer max_distance: an integer Returns: a Tensor with the same shape as relative_position, containing int32 values in the range [0, num_buckets) rr>r) longr4absmax zeros_likelogrtrprurxZ full_likewhere) Zrelative_position bidirectional num_buckets max_distanceretnZ max_exactZis_smallZ val_if_larger<r<r=relative_position_bucket?s  &rcs6eZdZfddZddZddZd d d ZZS) LayoutLMv2Encodercst|_tfddtjD|_j|_j |_ |jrlj |_ j |_ tj |j j dd|_|j rj|_j|_tj |jj dd|_tj |jj dd|_d|_dS)Ncsg|] }tqSr<)r).0_r9r<r= osz.LayoutLMv2Encoder.__init__..FrN)r#r$r9rZ ModuleListrangenum_hidden_layerslayerrWrX rel_pos_bins max_rel_posrYrP rel_pos_biasmax_rel_2d_posrel_2d_pos_binsrel_pos_x_biasrel_pos_y_biasgradient_checkpointingr7r:rr=r$ls  zLayoutLMv2Encoder.__init__cCsN|d|d}t||j|jd}|jj|dddd}|}|S)Nrnr rrrr rr>) unsqueezerrrrweighttrdry)r8rZ rel_pos_matr}r<r<r=!_calculate_1d_position_embeddingssz3LayoutLMv2Encoder._calculate_1d_position_embeddingsc Cs|dddddf}|dddddf}|d|d}|d|d}t||j|jd}t||j|jd}|jj|dddd}|jj|dddd}| }| }||}|S)Nrr rnr rrr>) rrrrrrrrdrry) r8rCZposition_coord_xZposition_coord_yZrel_pos_x_2d_matZrel_pos_y_2d_matZ rel_pos_xZ rel_pos_yr~r<r<r=!_calculate_2d_position_embeddingss(z3LayoutLMv2Encoder._calculate_2d_position_embeddingsNFTc  s|rdnd} rdnd} |jr(||nd} |jr<||nd} t|jD]\} }|r`| |f} |dk rp|| nd}|jr|jrfdd}tj j j |||||| | d}n||||| | d}|d}rJ| |df} qJ|r| |f} |st dd|| | fDSt || | d S) Nr<csfdd}|S)Ncs|fSrr<)inputs)moduler|r<r=custom_forwardszPLayoutLMv2Encoder.forward..create_custom_forward..custom_forwardr<)rrr|)rr=create_custom_forwardsz8LayoutLMv2Encoder.forward..create_custom_forwardrrrcss|]}|dk r|VqdSrr<)rrlr<r<r= sz,LayoutLMv2Encoder.forward..)last_hidden_stateri attentions) rWrrXr enumeraterrZtrainingr4utils checkpointtupler )r8rirzr{r|output_hidden_states return_dictrCrZall_hidden_statesZall_self_attentionsr}r~iZ layer_moduleZlayer_head_maskrZ layer_outputsr<rr=rsV         zLayoutLMv2Encoder.forward)NNFFTNN)rGrHrIr$rrrrKr<r<r:r=rks  rc@s.eZdZdZeZeZdZddZ d ddZ dS) LayoutLMv2PreTrainedModelz An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained models. layoutlmv2cCst|tjr:|jjjd|jjd|jdk r|jj nft|tj rz|jjjd|jjd|j dk r|jj|j  n&t|tj r|jj |jj ddS)zInitialize the weightsg)meanZstdN?)rrrYrdataZnormal_r9Zinitializer_rangerOZzero_r%rr.Zfill_)r8rr<r<r= _init_weightss      z'LayoutLMv2PreTrainedModel._init_weightsFcCst|tr||_dSr)rrr)r8rrar<r<r=_set_gradient_checkpointings z5LayoutLMv2PreTrainedModel._set_gradient_checkpointingN)F) rGrHrIrJr config_class(LAYOUTLMV2_PRETRAINED_MODEL_ARCHIVE_LISTZpretrained_model_archive_mapZbase_model_prefixrrr<r<r<r=rs rcCst|tjjjjr"tjj||S|}t|tj j rtjj|j |j dd|d}tj |j|_tj |j|_|j|_|j|_tjdtj|jjd|_|D]\}}||t||q~|S)NT) num_featuresrZaffineZtrack_running_stats process_grouprrodevice)rr4rmodulesZ batchnormZ _BatchNormZ SyncBatchNormZconvert_sync_batchnorm detectron2ZlayersZFrozenBatchNorm2drrr[rrOZ running_meanZ running_vartensorrrZnum_batches_trackedZnamed_childrenZ add_modulemy_convert_sync_batchnorm)rrZ module_outputnamechildr<r<r=r s(rcs,eZdZfddZddZddZZS)LayoutLMv2VisualBackbonec st||_|jjj}t||j}t|j t j j j sDt |j |_ t|jjjt|jjjkslt t|jjj}|jdt|jjj|dddd|jdt|jjj|ddddd|_trDtdd}|j |jj}ttt|d ||jd tt|d||jdf|_ nt!|jdd |_ t|jd kr|j"|j |jj#|j |jj#|jd kst dS) N pixel_meanrFr! pixel_stdp2z0using `AvgPool2d` instead of `AdaptiveAvgPool2d`)rrr>)$r#r$Zget_detectron2_configcfgZMODELZMETA_ARCHITECTURErgetrbackbonerZmodelingZFPNAssertionErrorlenZ PIXEL_MEANZ PIXEL_STDr3r4rrcout_feature_keyZ$are_deterministic_algorithms_enabledloggerwarningZ output_shapeZstriderZ AvgPool2drpceilimage_feature_pool_shapepoolZAdaptiveAvgPool2dappendZchannels)r8r9Z meta_archmodelZ num_channels input_shapeZbackbone_strider:r<r=r$#s@        z!LayoutLMv2VisualBackbone.__init__cCsVt|r|n|j|j|j}||}||j}||jdd dd }|S)Nr>)Z start_dimr) r4Z is_tensorrrrrrrflattenrsry)r8ZimagesZ images_inputfeaturesr<r<r=rFs    z LayoutLMv2VisualBackbone.forwardcstjr"tjr"tjdks*tdtj}tjtj}|dks\tdfddt |Dfddt |D}|}t |j ||d|_ dS) Nr z/Make sure torch.distributed is set up properly.rzGMake sure the number of processes can be divided by the number of nodescs&g|]}tt||dqSrg)listrrr) node_sizer<r=r[szCLayoutLMv2VisualBackbone.synchronize_batch_norm..csg|]}tjj|dqS))Zranks)r4 distributedZ new_groupr)node_global_ranksr<r=r\s)r) r4rZ is_availableZis_initializedZget_rank RuntimeErrorcudaZ device_countZget_world_sizerrr)r8Z self_rankZ world_sizeZsync_bn_groupsZ node_rankr<)rrr=synchronize_batch_normMs$       z/LayoutLMv2VisualBackbone.synchronize_batch_norm)rGrHrIr$rrrKr<r<r:r=r"s #raM This model is a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) sub-class. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior. Parameters: config ([`LayoutLMv2Config`]): Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the [`~PreTrainedModel.from_pretrained`] method to load the model weights. a Args: input_ids (`torch.LongTensor` of shape `{0}`): Indices of input sequence tokens in the vocabulary. Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for details. [What are input IDs?](../glossary#input-ids) bbox (`torch.LongTensor` of shape `({0}, 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. image (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)` or `detectron.structures.ImageList` whose `tensors` is of shape `(batch_size, num_channels, height, width)`): Batch of document images. attention_mask (`torch.FloatTensor` of shape `{0}`, *optional*): Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`: - 1 for tokens that are **not masked**, - 0 for tokens that are **masked**. [What are attention masks?](../glossary#attention-mask) token_type_ids (`torch.LongTensor` of shape `{0}`, *optional*): Segment token indices to indicate first and second portions of the inputs. Indices are selected in `[0, 1]`: - 0 corresponds to a *sentence A* token, - 1 corresponds to a *sentence B* token. [What are token type IDs?](../glossary#token-type-ids) position_ids (`torch.LongTensor` of shape `{0}`, *optional*): Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.max_position_embeddings - 1]`. [What are position IDs?](../glossary#position-ids) head_mask (`torch.FloatTensor` of shape `(num_heads,)` or `(num_layers, num_heads)`, *optional*): Mask to nullify selected heads of the self-attention modules. Mask values selected in `[0, 1]`: - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*): Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert *input_ids* indices into associated vectors than the model's internal embedding lookup matrix. output_attentions (`bool`, *optional*): Whether or not to return the attentions tensors of all attention layers. See `attentions` under returned tensors for more detail. output_hidden_states (`bool`, *optional*): Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors for more detail. return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. cs$eZdZfddZddZZS)LayoutLMv2Poolercs*tt|j|j|_t|_dSr)r#r$rrYr&rZTanh activationr7r:r<r=r$s zLayoutLMv2Pooler.__init__cCs(|dddf}||}||}|S)Nr)rr)r8riZfirst_token_tensor pooled_outputr<r<r=rs  zLayoutLMv2Pooler.forwardrr<r<r:r=rs rzdThe bare LayoutLMv2 Model transformer outputting raw hidden-states without any specific head on top.cseZdZfddZddZddZddd Zd d Zd d ZdddZ e e de eeddeejeejeejeejeejeejeejeejeeeeeeeeefd ddZZS)LayoutLMv2Modelcst|dt|||_|j|_t||_t||_t |j d|j |_ |jrpt t d|j jd|_t j|j |jd|_t |j|_t||_t||_|dS)Nrr rrr)rr#r$r9has_visual_segment_embeddingr embeddingsrvisualrrYrr& visual_projr[r%rvisual_segment_embeddingr.r/visual_LayerNormr0r1visual_dropoutrencoderrpooler post_initr7r:r<r=r$s      zLayoutLMv2Model.__init__cCs|jjSrrr'r8r<r<r=get_input_embeddingssz$LayoutLMv2Model.get_input_embeddingscCs ||j_dSrr)r8rar<r<r=set_input_embeddingssz$LayoutLMv2Model.set_input_embeddingsNc Cs|dk r|}n|dd}|d}|dkrVtj|tj|jd}|d|}|dkrht|}|dkr||j |}|j |}|j |} |j |} ||| | } |j | } |j| } | S)Nr rrr)rbr4r5rrrZ expand_asrrr'r(rFr-r.r2) r8 input_idsrCrtoken_type_ids inputs_embedsr seq_lengthr(rEr-rr<r<r=_calc_text_embeddingss$        z%LayoutLMv2Model._calc_text_embeddingscCs\|||}|j|}|j|}|||}|jrD||j7}||}||}|Sr) rrrr(rFrrr r )r8imagerCrZvisual_embeddingsr(rErr<r<r=_calc_img_embeddingss      z$LayoutLMv2Model._calc_img_embeddingscCs tjtjdd|ddd||jd|jjddd}tjtjdd|jjddd||jd|jjddd}tj|dd|dd|dd|dddd|dd|dd|dd|ddddgdd d| d}||ddd}|S) Nrir)rrofloor)Z rounding_moder r?) r4divr5ror9rstackrepeatrsrcrb)r8rrCr final_shapeZ visual_bbox_xZ visual_bbox_y visual_bboxr<r<r=_calc_visual_bboxsF     z!LayoutLMv2Model._calc_visual_bboxcCsN|dk r|dk rtdn0|dk r*|S|dk rB|ddStddS)NDYou cannot specify both input_ids and inputs_embeds at the same timer 5You have to specify either input_ids or inputs_embeds)rRrb)r8rrr<r<r=_get_input_shape"s z LayoutLMv2Model._get_input_shapez(batch_size, sequence_length) output_typer) rrCrrzrrr{rr|rrrc  Cs| dk r | n|jj} | dk r | n|jj} | dk r4| n|jj} |||} |dk rV|jn|j} t| }|jjd|jjd|d<t |}t|||}|d|d7<t |}| |jj|| |}tj ||gdd}|dkrtj | | d}tj || d}tj ||gdd}|dkr0tj | tj| d}|dkrd| d}|jjddd|f}|| }tjd|dtj| d| dd}tj ||gdd}|dkrtj tt| dgtj| d}|j|||||d}|j|||d }tj ||gdd}|dd }|j|jd }d |t|jj}|dk r|dkrz|ddd d }||jjd d d d }n$|d kr|dd d }|jt|jd }ndg|jj}|j |||||| | | d}|d}|!|}| s ||f|ddSt"|||j#|j$dS)au Return: Examples: ```python >>> from transformers import AutoProcessor, LayoutLMv2Model, set_seed >>> from PIL import Image >>> import torch >>> from datasets import load_dataset >>> set_seed(88) >>> processor = AutoProcessor.from_pretrained("microsoft/layoutlmv2-base-uncased") >>> model = LayoutLMv2Model.from_pretrained("microsoft/layoutlmv2-base-uncased") >>> dataset = load_dataset("hf-internal-testing/fixtures_docvqa") >>> image_path = dataset["test"][0]["file"] >>> image = Image.open(image_path).convert("RGB") >>> encoding = processor(image, return_tensors="pt") >>> outputs = model(**encoding) >>> last_hidden_states = outputs.last_hidden_state >>> last_hidden_states.shape torch.Size([1, 342, 768]) ``` Nrrr?)rr)rrCrrrrrCrr>)rorr )rCrr{r|rr)rZ pooler_outputrir)%r9r|ruse_return_dictr"rrrr4SizerrBZonesr\rrrr6r5rrrrrrurorwrxr@rnext parametersr r r rir)r8rrCrrzrrr{rr|rrrr visual_shaperrZ final_bboxZvisual_attention_maskZfinal_attention_maskrvisual_position_idsZfinal_position_idsZtext_layout_embZ visual_embZ final_embZextended_attention_maskZencoder_outputssequence_outputrr<r<r=r,s.           zLayoutLMv2Model.forward)N)NN) NNNNNNNNNNN)rGrHrIr$rrrrrr"rLAYOUTLMV2_INPUTS_DOCSTRINGformatrr _CONFIG_FOR_DOCrr4 LongTensor FloatTensorrvrrr rrKr<r<r:r=rsD   %   rax LayoutLMv2 Model with a sequence classification head on top (a linear layer on top of the concatenation of the final hidden state of the [CLS] token, average-pooled initial visual embeddings and average-pooled final visual embeddings, e.g. for document image classification tasks such as the [RVL-CDIP](https://www.cs.cmu.edu/~aharley/rvl-cdip/) dataset. cseZdZfddZddZeedee e dd e e j e e j e e je e je e j e e j e e je e je e j e ee ee eeee fd d d ZZS) #LayoutLMv2ForSequenceClassificationcsNt||j|_t||_t|j|_t |j d|j|_ | dS)Nr r#r$ num_labelsrrrr0r1r2rYr& classifierr r7r:r<r=r$s   z,LayoutLMv2ForSequenceClassification.__init__cCs |jjjSrrrr'rr<r<r=rsz8LayoutLMv2ForSequenceClassification.get_input_embeddingsbatch_size, sequence_lengthr#N rrCrrzrrr{rlabelsr|rrrc  Cs4| dk r | n|jj} |dk r.|dk r.tdn@|dk rL||||} n"|dk rf|dd} ntd|dk r||jn|j}t| }|jjd|jjd|d<t |}t| }|d|d7<t |}|j |jj|||}tj d|dtj |d| dd}|j j|||d}|j ||||||||| | | d }|dk rV|} n|dd} | d}|dddd|f|ddd|df}}|dddddf}|jdd }|jdd }tj|||gdd }||}||}d}| dk r|jjdkrb|jdkr(d |j_n:|jdkrZ| jtj ksP| jtjkrZd |j_nd |j_|jjd krt}|jdkr||| }n ||| }nN|jjd krt}||d|j| d}n|jjd krt}||| }| s |f|dd}|dk r|f|S|St|||j|jdS)a5 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). Returns: Example: ```python >>> from transformers import AutoProcessor, LayoutLMv2ForSequenceClassification, set_seed >>> from PIL import Image >>> import torch >>> from datasets import load_dataset >>> set_seed(88) >>> dataset = load_dataset("rvl_cdip", split="train", streaming=True) >>> data = next(iter(dataset)) >>> image = data["image"].convert("RGB") >>> processor = AutoProcessor.from_pretrained("microsoft/layoutlmv2-base-uncased") >>> model = LayoutLMv2ForSequenceClassification.from_pretrained( ... "microsoft/layoutlmv2-base-uncased", num_labels=dataset.info.features["label"].num_classes ... ) >>> encoding = processor(image, return_tensors="pt") >>> sequence_label = torch.tensor([data["label"]]) >>> outputs = model(**encoding, labels=sequence_label) >>> loss, logits = outputs.loss, outputs.logits >>> predicted_idx = logits.argmax(dim=-1).item() >>> predicted_answer = dataset.info.features["label"].names[4] >>> predicted_idx, predicted_answer (4, 'advertisement') ``` Nr r r!rrrr& rrCrrzrrr{rr|rrr?Z regressionZsingle_label_classificationZmulti_label_classificationr>losslogitsrir) r9r'rRZ%warn_if_padding_and_no_attention_maskrbrrrr4r(rrr5rrrrrBr2r6Z problem_typer5rorTrsqueezerrcrrrir)r8rrCrrzrrr{rr:r|rrrrr+rrr,Zinitial_image_embeddingsrrr-Zfinal_image_embeddingsZcls_final_outputZpooled_initial_image_embeddingsZpooled_final_image_embeddingsr>r=loss_fctrr<r<r=rs9       2       (    z+LayoutLMv2ForSequenceClassification.forward) NNNNNNNNNNNN)rGrHrIr$rrr.r/rrr0rr4r1r2rvrrrrKr<r<r:r=r3s>    r3a LayoutLMv2 Model with a token classification head on top (a linear layer on top of the text part of the hidden states) e.g. for sequence labeling (information extraction) tasks such as [FUNSD](https://guillaumejaume.github.io/FUNSD/), [SROIE](https://rrc.cvc.uab.es/?ch=13), [CORD](https://github.com/clovaai/cord) and [Kleister-NDA](https://github.com/applicaai/kleister-nda). cseZdZfddZddZeedee e dd e e j e e j e e je e je e j e e j e e je e je e j e ee ee eeee fd d d ZZS) LayoutLMv2ForTokenClassificationcsJt||j|_t||_t|j|_t |j |j|_ | dSrr4r7r:r<r=r$zs   z)LayoutLMv2ForTokenClassification.__init__cCs |jjjSrr7rr<r<r=rsz5LayoutLMv2ForTokenClassification.get_input_embeddingsr8r#Nr9c  Cs| dk r | n|jj} |j||||||||| | | d } |dk rF|}n|dd}|d}| dddd|f}||}||}d}| dk rt}||d|j| d}| s|f| dd}|dk r|f|S|St ||| j | j dS)a2 labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): Labels for computing the token classification loss. Indices should be in `[0, ..., config.num_labels - 1]`. Returns: Example: ```python >>> from transformers import AutoProcessor, LayoutLMv2ForTokenClassification, set_seed >>> from PIL import Image >>> from datasets import load_dataset >>> set_seed(88) >>> datasets = load_dataset("nielsr/funsd", split="test") >>> labels = datasets.features["ner_tags"].feature.names >>> id2label = {v: k for v, k in enumerate(labels)} >>> processor = AutoProcessor.from_pretrained("microsoft/layoutlmv2-base-uncased", revision="no_ocr") >>> model = LayoutLMv2ForTokenClassification.from_pretrained( ... "microsoft/layoutlmv2-base-uncased", num_labels=len(labels) ... ) >>> data = datasets[0] >>> image = Image.open(data["image_path"]).convert("RGB") >>> words = data["words"] >>> boxes = data["bboxes"] # make sure to normalize your bounding boxes >>> word_labels = data["ner_tags"] >>> encoding = processor( ... image, ... words, ... boxes=boxes, ... word_labels=word_labels, ... padding="max_length", ... truncation=True, ... return_tensors="pt", ... ) >>> outputs = model(**encoding) >>> logits, loss = outputs.logits, outputs.loss >>> predicted_token_class_ids = logits.argmax(-1) >>> predicted_tokens_classes = [id2label[t.item()] for t in predicted_token_class_ids[0]] >>> predicted_tokens_classes[:5] ['B-ANSWER', 'B-HEADER', 'B-HEADER', 'B-HEADER', 'B-HEADER'] ``` Nr;r rrr>r<) r9r'rrbr2r6rrcr5rrir)r8rrCrrzrrr{rr:r|rrrrrr-r>r=r@rr<r<r=rsDB    z(LayoutLMv2ForTokenClassification.forward) NNNNNNNNNNNN)rGrHrIr$rrr.r/rrr0rr4r1r2rvrrrrKr<r<r:r=rAps>    rAa LayoutLMv2 Model with a span classification head on top for extractive question-answering tasks such as [DocVQA](https://rrc.cvc.uab.es/?ch=17) (a linear layer on top of the text part of the hidden-states output to compute `span start logits` and `span end logits`). cseZdZd fdd ZddZeedee e dd e e j e e j e e je e je e j e e j e e je e je e j e e j e ee ee eeee fd d d ZZS)LayoutLMv2ForQuestionAnsweringTcsBt||j|_||_t||_t|j|j|_ | dSr) r#r$r5rrrrrYr& qa_outputsr )r8r9rr:r<r=r$s   z'LayoutLMv2ForQuestionAnswering.__init__cCs |jjjSrr7rr<r<r=rsz3LayoutLMv2ForQuestionAnswering.get_input_embeddingsr8r#N)rrCrrzrrr{rstart_positions end_positionsr|rrrc Cs| dk r | n|jj} |j||||||||| | | d }|dk rF|}n|dd}|d}|dddd|f}||}|jddd\}}|d}|d}d}| dk rF| dk rFt| dkr| d} t| dkr| d} |d}| d|} | d|} t |d}||| }||| }||d}| sx||f|dd}|dk rt|f|S|St ||||j |j d S) am start_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*): Labels for position (index) of the start of the labelled span for computing the token classification loss. Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the sequence are not taken into account for computing the loss. end_positions (`torch.LongTensor` of shape `(batch_size,)`, *optional*): Labels for position (index) of the end of the labelled span for computing the token classification loss. Positions are clamped to the length of the sequence (`sequence_length`). Position outside of the sequence are not taken into account for computing the loss. Returns: Example: In this example below, we give the LayoutLMv2 model an image (of texts) and ask it a question. It will give us a prediction of what it thinks the answer is (the span of the answer within the texts parsed from the image). ```python >>> from transformers import AutoProcessor, LayoutLMv2ForQuestionAnswering, set_seed >>> import torch >>> from PIL import Image >>> from datasets import load_dataset >>> set_seed(88) >>> processor = AutoProcessor.from_pretrained("microsoft/layoutlmv2-base-uncased") >>> model = LayoutLMv2ForQuestionAnswering.from_pretrained("microsoft/layoutlmv2-base-uncased") >>> dataset = load_dataset("hf-internal-testing/fixtures_docvqa") >>> image_path = dataset["test"][0]["file"] >>> image = Image.open(image_path).convert("RGB") >>> question = "When is coffee break?" >>> encoding = processor(image, question, return_tensors="pt") >>> outputs = model(**encoding) >>> predicted_start_idx = outputs.start_logits.argmax(-1).item() >>> predicted_end_idx = outputs.end_logits.argmax(-1).item() >>> predicted_start_idx, predicted_end_idx (154, 287) >>> predicted_answer_tokens = encoding.input_ids.squeeze()[predicted_start_idx : predicted_end_idx + 1] >>> predicted_answer = processor.tokenizer.decode(predicted_answer_tokens) >>> predicted_answer # results are not very good without further fine-tuning 'council mem - bers conducted by trrf treasurer philip g. kuehn to get answers which the public ... ``` ```python >>> target_start_index = torch.tensor([7]) >>> target_end_index = torch.tensor([14]) >>> outputs = model(**encoding, start_positions=target_start_index, end_positions=target_end_index) >>> predicted_answer_span_start = outputs.start_logits.argmax(-1).item() >>> predicted_answer_span_end = outputs.end_logits.argmax(-1).item() >>> predicted_answer_span_start, predicted_answer_span_end (154, 287) ``` Nr;r rrr?)Z ignore_indexr>)r= start_logits end_logitsrir)r9r'rrbrCsplitr?ryrclamprr rir)r8rrCrrzrrr{rrDrEr|rrrrrr-r>rFrGZ total_lossZ ignored_indexr@Z start_lossZend_lossrr<r<r=r s\J           z&LayoutLMv2ForQuestionAnswering.forward)T) NNNNNNNNNNNNN)rGrHrIr$rrr.r/rr r0rr4r1r2rvrrrrKr<r<r:r=rBsB     rB)Trr)N)?rJrptypingrrrr4Ztorch.utils.checkpointrZtorch.nnrrrZ activationsr Zmodeling_outputsr r r rrZmodeling_utilsrZ pytorch_utilsrrrrrrrrZconfiguration_layoutlmv2rrZdetectron2.modelingrZ get_loggerrGrZ_CHECKPOINT_FOR_DOCr0rModulerrLrrrrrrrrrrZLAYOUTLMV2_START_DOCSTRINGr.rrr3rArBr<r<r<r=sv        /]+ , B <} . {