rhMNddlmZmZddlmZddlmZmZddlm Z m Z m Z m Z ddl mZmZddlmZerddlZGd d e d ZGd de ZdgZy))OptionalUnion) BatchFeature) ImageInputis_valid_image)MultiModalDataProcessingKwargsProcessorMixinUnpack)PreTokenizedInput TextInput)is_torch_availableNc&eZdZddidddddidZy ) ColQwen2ProcessorKwargspaddinglongestchannels_firstT) data_formatdo_convert_rgbreturn_tensorspt) text_kwargs images_kwargs common_kwargsN)__name__ __module__ __qualname__ _defaults/var/www/html/ai-insurance-compliance-backend/venv/lib/python3.12/site-packages/transformers/models/colqwen2/processing_colqwen2.pyrr#s, y ," +D1 Ir!rF)totalc peZdZdZddgZdZdZ ddeedeeffd Z dd e d e e e ee ee fd eed efd Zd dZdZdZedZed efdZ d d e d eed efdZd e e ee fd eed efdZ d!de dedfde dedfdededde defd df dZxZS)"ColQwen2Processora Constructs a ColQwen2 processor which wraps a Qwen2VLProcessor and special methods to process images and queries, as well as to compute the late-interaction retrieval score. [`ColQwen2Processor`] offers all the functionalities of [`Qwen2VLProcessor`]. See the [`~Qwen2VLProcessor.__call__`] for more information. Args: image_processor ([`Qwen2VLImageProcessor`], *optional*): The image processor is a required input. tokenizer ([`Qwen2TokenizerFast`], *optional*): The tokenizer is a required input. chat_template (`str`, *optional*): A Jinja template which will be used to convert lists of messages in a chat into a tokenizable string. visual_prompt_prefix (`str`, *optional*): A string that gets tokenized and prepended to the image tokens. query_prefix (`str`, *optional*): A prefix to be used for the query. image_processor tokenizerAutoImageProcessor)Qwen2TokenizerQwen2TokenizerFastvisual_prompt_prefix query_prefixc t||||t|dsdn |j|_t|dsdn |j|_|d}||_|d}||_y)N) chat_template image_tokenz <|image_pad|> video_tokenz <|video_pad|>zf<|im_start|>user <|vision_start|><|image_pad|><|vision_end|>Describe the image.<|im_end|><|endoftext|>zQuery: )super__init__hasattrr/r0r+r,)selfr&r'r.r+r,kwargs __class__s r"r2zColQwen2Processor.__init__Hsw )=Q29)]2S?YbYnYn29)]2S?YbYnYn  '$M $8!  $L(r!imagestextr5returnc v|jtfd|jji|}|dj dd}|du}| | t d| | t d|7t |r|g}n^t|trt |drn?t|tr$t|dtrt |dds t d|jgt|z} |jdd |i|d } | d } | |jjd z} d} tt| D]}|j| |vrQ| |j|jd | | j!| zzd| |<| dz } |j| |vrQ| |jd |j| |<|j| fddi|d}t#i|| }|d dddf|d ddd fz}tt%j&|d|j)}t$j*j,j.j1|d|d<|r.|dj3|ddk(d}|j5d|i|S|t|t6r|g}n.t|trt|dt6s t d||j8dz}g}|D]%}|j:|z|z}|j=|'|j|fddi|d}|Sy)a Main method to prepare for the model either (1) one or several texts, either (2) one or several image(s). This method is a custom wrapper around the Qwen2VLProcessor's [`~Qwen2VLProcessor.__call__`] method adapted for the ColQwen2 model. It cannot process both text and images at the same time. When preparing the the text(s), this method forwards the `text` and `kwargs` arguments to Qwen2TokenizerFast's [`~Qwen2TokenizerFast.__call__`]. When preparing the the image(s), this method forwards the `images` and `kwargs` arguments to Qwen2VLImageProcessor's [`~Qwen2VLImageProcessor.__call__`]. Please refer to the doctsring of the above two methods for more information. Args: images (`PIL.Image.Image`, `np.ndarray`, `torch.Tensor`, `list[PIL.Image.Image]`, `list[np.ndarray]`, `list[torch.Tensor]`): The image or batch of images to be prepared. Each image can be a PIL image, NumPy array or PyTorch tensor. In case of a NumPy array/PyTorch tensor, each image should be of shape (C, H, W), where C is a number of channels, H and W are image height and width. text (`str`, `list[str]`, `list[list[str]]`): The sequence or batch of sequences to be encoded. Each sequence can be a string or a list of strings (pretokenized string). If the sequences are provided as list of strings (pretokenized), you must set `is_split_into_words=True` (to lift the ambiguity with a batch of sequences). return_tensors (`str` or [`~utils.TensorType`], *optional*): If set, will return tensors of a particular framework. Acceptable values are: - `'tf'`: Return TensorFlow `tf.constant` objects. - `'pt'`: Return PyTorch `torch.Tensor` objects. - `'np'`: Return NumPy `np.ndarray` objects. - `'jax'`: Return JAX `jnp.ndarray` objects. Returns: [`BatchFeature`]: A [`BatchFeature`] with the following fields: - **input_ids** -- List of token ids to be fed to a model. - **attention_mask** -- List of indices specifying which tokens should be attended to by the model (when `return_attention_mask=True` or if *"attention_mask"* is in `self.model_input_names` and if `text` is not `None`). - **pixel_values** -- Pixel values to be fed to a model. Returned when `images` is not `None`. tokenizer_init_kwargsrsuffixNz&Either text or images must be providedz5Only one of text or images can be processed at a timerzAimages must be an image, list of images or list of list of imagesr7rimage_grid_thwz<|placeholder|>return_token_type_idsF)data pixel_valuesT) batch_first input_idstoken_type_idsilabelsz*Text must be a string or a list of strings r ) _merge_kwargsrr' init_kwargspop ValueErrorr isinstancelistr+lenr& merge_sizeranger/replaceprodrtorchsplittolistnnutilsrnn pad_sequence masked_fillupdatestrquery_augmentation_tokenr,append)r4r7r8audiovideosr5 output_kwargsr<r@ texts_doc image_inputsr= merge_lengthindexi text_inputs return_dataoffsetsrBrF texts_queryqueryaugmented_query batch_querys r"__call__zColQwen2Processor.__call__]sZ+** # "&.."<"<  }-11(DA &d 2 >A s9~.]A**il:'0|';'; ,,.?>RWCXC]C]C_coCo.prs( !   **il: $-Q<#7#78I4K[K[#\IaL ])$..&+ .K ',K{,Kl,KLK""23AqD9KHX[_`>`bfg""Hf#56   $$v t,DGS1I !MNN~66;%'K 4"&"3"3e";f"D""?3 4)$..&+ .K  +r!c i}|tjjdi}|j||jddxs|jj }|Dcgc] }|jj g||"}}|Dcgc] }||dzz } }|j| |dtdi|Scc}wcc}w)a Computes the number of placeholder tokens needed for multimodal inputs with the given sizes. Args: image_sizes (`list[list[int]]`, *optional*): The input sizes formatted as (height, width) per each image. Returns: `MultiModalData`: A `MultiModalData` object holding number of tokens per each of the provided input modalities, along with other useful data. NrrOr>)num_image_tokensnum_image_patchesr )rrgetr[r&rOget_number_of_image_patchesr ) r4 image_sizesr5 vision_datarrO image_sizerq num_patchesrps r"_get_num_multimodal_tokensz,ColQwen2Processor._get_num_multimodal_tokenss  "3==AA/SUVM   (&**<>a$BVBVBaBaJ#.!A$$@@\*\m\! !Sdd; A !=d d   4D[lm n, ,,! es $%B?Cc:|jj|i|S)z This method forwards all its arguments to GemmaTokenizerFast's [`~PreTrainedTokenizer.batch_decode`]. Please refer to the docstring of this method for more information. )r' batch_decoder4argsr5s r"rzzColQwen2Processor.batch_decodes +t~~**D;F;;r!c:|jj|i|S)z This method forwards all its arguments to GemmaTokenizerFast's [`~PreTrainedTokenizer.decode`]. Please refer to the docstring of this method for more information. )r'decoder{s r"r~zColQwen2Processor.decodes %t~~$$d5f55r!c|jj}|jj}ttj ||zSN)r'model_input_namesr&rMdictfromkeys)r4tokenizer_input_namesimage_processor_input_namess r"rz#ColQwen2Processor.model_input_names s? $ @ @&*&:&:&L&L#DMM"7:U"UVWWr!c.|jjS)z Return the query augmentation token. Query augmentation buffers are used as reasoning buffers during inference. )r' pad_token)r4s r"r]z*ColQwen2Processor.query_augmentation_tokens~~'''r!c *|jdd|i|S)a Prepare for the model one or several image(s). This method is a wrapper around the `__call__` method of the ColQwen2Processor's [`ColQwen2Processor.__call__`]. This method forwards the `images` and `kwargs` arguments to the image processor. Args: images (`PIL.Image.Image`, `np.ndarray`, `torch.Tensor`, `list[PIL.Image.Image]`, `list[np.ndarray]`, `list[torch.Tensor]`): The image or batch of images to be prepared. Each image can be a PIL image, NumPy array or PyTorch tensor. In case of a NumPy array/PyTorch tensor, each image should be of shape (C, H, W), where C is a number of channels, H and W are image height and width. return_tensors (`str` or [`~utils.TensorType`], *optional*): If set, will return tensors of a particular framework. Acceptable values are: - `'tf'`: Return TensorFlow `tf.constant` objects. - `'pt'`: Return PyTorch `torch.Tensor` objects. - `'np'`: Return NumPy `np.ndarray` objects. - `'jax'`: Return JAX `jnp.ndarray` objects. Returns: [`BatchFeature`]: A [`BatchFeature`] with the following fields: - **input_ids** -- List of token ids to be fed to a model. - **attention_mask** -- List of indices specifying which tokens should be attended to by the model (when `return_attention_mask=True` or if *"attention_mask"* is in `self.model_input_names` and if `text` is not `None`). - **pixel_values** -- Pixel values to be fed to a model. Returned when `images` is not `None`. r7r rn)r4r7r5s r"process_imagesz ColQwen2Processor.process_imagessBt}}5F5f55r!c *|jdd|i|S)a Prepare for the model one or several texts. This method is a wrapper around the `__call__` method of the ColQwen2Processor's [`ColQwen2Processor.__call__`]. This method forwards the `text` and `kwargs` arguments to the tokenizer. Args: text (`str`, `list[str]`, `list[list[str]]`): The sequence or batch of sequences to be encoded. Each sequence can be a string or a list of strings (pretokenized string). If the sequences are provided as list of strings (pretokenized), you must set `is_split_into_words=True` (to lift the ambiguity with a batch of sequences). return_tensors (`str` or [`~utils.TensorType`], *optional*): If set, will return tensors of a particular framework. Acceptable values are: - `'tf'`: Return TensorFlow `tf.constant` objects. - `'pt'`: Return PyTorch `torch.Tensor` objects. - `'np'`: Return NumPy `np.ndarray` objects. - `'jax'`: Return JAX `jnp.ndarray` objects. Returns: [`BatchFeature`]: A [`BatchFeature`] with the following fields: - **input_ids** -- List of token ids to be fed to a model. - **attention_mask** -- List of indices specifying which tokens should be attended to by the model (when `return_attention_mask=True` or if *"attention_mask"* is in `self.model_input_names` and if `text` is not `None`). r8r r)r4r8r5s r"process_queriesz!ColQwen2Processor.process_queries<s@t}}1$1&11r!query_embeddingsz torch.Tensorpassage_embeddings batch_size output_dtypez torch.dtype output_devicez torch.devicec t|dk(r tdt|dk(r td|dj|djk7r td|dj|djk7r td||dj}g}t dt||D]%}g}t j jjj||||zdd} t dt||D]} t j jjj|| | |zdd} |jt jd| | jd djd |jt j|d j|j|(t j|d S) a[ Compute the late-interaction/MaxSim score (ColBERT-like) for the given multi-vector query embeddings (`qs`) and passage embeddings (`ps`). For ColQwen2, a passage is the image of a document page. Because the embedding tensors are multi-vector and can thus have different shapes, they should be fed as: (1) a list of tensors, where the i-th tensor is of shape (sequence_length_i, embedding_dim) (2) a single tensor of shape (n_passages, max_sequence_length, embedding_dim) -> usually obtained by padding the list of tensors. Args: query_embeddings (`Union[torch.Tensor, list[torch.Tensor]`): Query embeddings. passage_embeddings (`Union[torch.Tensor, list[torch.Tensor]`): Passage embeddings. batch_size (`int`, *optional*, defaults to 128): Batch size for computing scores. output_dtype (`torch.dtype`, *optional*, defaults to `torch.float32`): The dtype of the output tensor. If `None`, the dtype of the input embeddings is used. output_device (`torch.device` or `str`, *optional*, defaults to "cpu"): The device of the output tensor. Returns: `torch.Tensor`: A tensor of shape `(n_queries, n_passages)` containing the scores. The score tensor is saved on the "cpu" device. rzNo queries providedzNo passages providedz/Queries and passages must be on the same devicez-Queries and passages must have the same dtypeT)rC padding_valuez bnd,csd->bcnsr)dimr>r?)rNrKdevicedtyperPrSrVrWrXrYr^einsummaxsumcatto) r4rrrrrscoresrf batch_scores batch_queriesjbatch_passagess r"score_retrievalz!ColQwen2Processor.score_retrieval^s@  A %23 3 ! "a '34 4 A  % %);A)>)E)E ENO O A  $ $(:1(=(C(C CLM M  +A.44L%'q#./< ]A/1L!HHNN..;; Q^4$VW<M1c"45zB !&!3!3!@!@&q1z>:\]"A"##LL-PTTYZT[\]^bbghbi   MM%))La8;;LILL][ \ ]yyQ''r!)NNNNN)NNNNr)Ncpu)rrr__doc__ attributesimage_processor_classtokenizer_classrr\r2rrrr rMr rrrnrxrzr~propertyrr]rrintr __classcell__)r6s@r"r%r%0s$$[1J0>O.2&* ) 'sm ) sm )."^b CCI0$y/4HYCZZ[C 01 C CJ-4<6XX (#(("!6!601!6  !6F 2ItI./ 201 2  2L0449 >(^0D DE>(".$~2F"FG>( >( }- >( ^S01 >( >(r!r%)typingrrfeature_extraction_utilsr image_utilsrrprocessing_utilsr r r r tokenization_utils_baser rrWrrSrr%__all__r r!r"rsO,#45XXC' .e l(l(^  r!