rh1ddlmZmZddlZddlmZddlmZddl m Z m Z m Z m Z mZmZddlmZmZddlmZerd d lmZGd d ed ZGdde d ZGdde d ZGdde ZdgZy))OptionalUnionN) BatchFeature) ImageInput) ImagesKwargsMultiModalDataProcessingKwargsProcessorMixin TextKwargsUnpack)PreTokenizedInput TextInput)is_vision_available) smart_resizeceZdZUeed<y)Emu3TextKwargsreturn_for_image_generationN)__name__ __module__ __qualname__bool__annotations__{/var/www/html/ai-insurance-compliance-backend/venv/lib/python3.12/site-packages/transformers/models/emu3/processing_emu3.pyrr s!%%rrF)totalc"eZdZUeed<eed<y)Emu3ImagesKwargsratio image_areaN)rrrstrrintrrrr r $s JOrr c8eZdZUeed<eed<dddddddZy ) Emu3ProcessorKwargs text_kwargs images_kwargsF)rreturn_mm_token_type_idsz1:1i)r!r")r'r(N)rrrrrr _defaultsrrrr&r&)s/##,1(-   Irr&c eZdZdZddgZdZdZ dfd Z ddee dee e e e e e e fd eed efd Zdd Zd Zde fdZdZdZedZxZS) Emu3Processora Constructs a Emu3 processor which wraps a Emu3 image processor and a GPT2 tokenizer into a single processor. [`Emu3Processor`] offers all the functionalities of [`Emu3ImageProcessor`] and [`GPT2TokenizerFast`]. See the [`~Emu3Processor.__call__`] and [`~Emu3Processor.decode`] for more information. Args: image_processor ([`Emu3ImageProcessor`]): The image processor is a required input. tokenizer ([`Emu3TokenizerFast`]): The tokenizer is a required input. chat_template (`str`, *optional*): A Jinja template which will be used to convert lists of messages in a chat into a tokenizable string. image_processor tokenizer) GPT2TokenizerGPT2TokenizerFastEmu3ImageProcessorc &|j|_|j|_|j|_|j|_|j |_|j|_|j|_ d|_ t|1|||y)N) chat_template) image_tokenimage_token_id boi_tokenimage_start_token eoi_tokenimage_end_tokenimage_wrapper_tokenfake_token_around_image eof_token bos_tokendownsample_ratiosuper__init__)selfr-r.r4kwargs __class__s rrAzEmu3Processor.__init__Ms%00'66!*!4!4(22'0'D'D$",,",, ! )=QrimagestextrCreturnc |t|tr|g}n.t|tst|dts td|jt fd|j ji|}|djdd}|djdd }|djd d } |r | td |s| | td i} |j} |j|j} |s| |j|fi|d} t| j} g}|D]}|j |vrt#| }|\}}||j$z}||j$z}||d zz}| |d||j&d|z| }|j)|j |d }|j*|}|j |vr|j-||Dcgc]}|j)d|j  }}nj|rh|j/|| |j$\}}| |d||j&}|Dcgc]}|j*||}}||ggt1|z| d<|djdd }|djdd}|j |fi|ddd i}|j3||dg|rUt5j6|d}t5j8|d}d |||j:k(<|j=|d<t?i|| |Scc}wcc}w)a Main method to prepare for the model one or several sequences(s) and image(s). This method forwards the `text` and `kwargs` arguments to Emu3TokenizerFast's [`~Emu3TokenizerFast.__call__`] if `text` is not `None` to encode the text. To prepare the image(s), this method forwards the `images` and `kwrags` arguments to CLIPImageProcessor's [`~CLIPImageProcessor.__call__`] if `images` is not `None`. Please refer to the docstring 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. Both channels-first and channels-last formats are supported. 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. Returned when `text` is not `None`. - **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`. rzAInvalid input text. Please provide a string, or a list of stringstokenizer_init_kwargsr'rFr(r!Nr"zGYou should not provide `images` when `return_for_image_generation=True`zOYou must provide either text or images when `return_for_image_generation=False`r*z image_sizesreturn_tensorsr)image) modalities input_idsmm_token_type_ids)data tensor_type) isinstancer#list TypeError _merge_kwargsr&r. init_kwargspop ValueErrorr8r=r:r-iterrKr5nextr?r<replacer>appendcalculate_generate_sizelen_check_special_mm_tokensnparray zeros_liker6tolistr)rBrErFaudiovideosrC output_kwargsrr!r"image_featuresimage_start_tokensimage_end_tokensrKprompt_stringssample image_sizeheightwidthimage_seq_lengthimage_placeholder image_promptrLr) text_inputs array_idsrPs r__call__zEmu3Processor.__call__^sT dC 6DD$' 47C0H_` `***  "&.."<"<  '4M&B&F&FGdfk&l#o.227DA"?377 dK &6+=fg g*t|no o $ 6 67"nn-d.B.B-CD+v/A1T11&[M/vhawtOkOkNlm|@PnPmQRbQc)d%#^^D,<,<>OQRSF $/x8F&&&0%%f- .Ucc&FNN?D4D4DEcDc) 88 DLaLabMFE01&5'$B^B^A_`LLPQ&t~~&vh|n=QDQ.4e_,=D ,IN= )'}599:JDQ#0#?#C#CD^`e#f $dnnT_]=-I_Z^_  %%dKWI%N #[!9:I " k+.F G BC i4+>+>> ?/@/G/G/IK+ ,!BK!B>!BP^__+d Rs ##L4L9c i}|g}|D]\}}t|||jj|jj|jj\}}||j z}||j z}||dzz}|j |dgt|z}|j||dtdi|S)a Computes the number of placeholder tokens needed for multimodal inputs with the given sizes. Args: image_sizes (`list[list[int]]`, *optional*): The input sizes formatted as (height, width) per each image. Returns: `MultiModalData`: A `MultiModalData` object holding number of tokens per each of the provided input modalities, along with other useful data. r)num_image_tokensnum_image_patchesr) rr-spatial_factor min_pixels max_pixelsr?r]r_updater ) rBrKrC vision_datarwrnrorprxs r_get_num_multimodal_tokensz(Emu3Processor._get_num_multimodal_tokenss  "! !, :  ,((77((33((33 !  4#8#88!6!66#)UQY#7  ''(89 :"#c+&6 6    4D[lm n, ,,rctt|jd\}}||z}||z dz}tt||z|z }tt||z|z } || fS)N:g?)mapr$splitround) rBr!r"ryrorn current_area target_ratio token_height token_widths rr^z%Emu3Processor.calculate_generate_sizespCS!12 vv~ "\1c9 5,!6!GHI % 4~ EFG [((rc <|jj|fi|SN)r- postprocess)rBrErCs rrzEmu3Processor.postprocesss /t##//A&AArc:|jj|i|S)z This method forwards all its arguments to Emu3TokenizerFast's [`~PreTrainedTokenizer.batch_decode`]. Please refer to the docstring of this method for more information. )r. batch_decoderBargsrCs rrzEmu3Processor.batch_decodes +t~~**D;F;;rc:|jj|i|S)z This method forwards all its arguments to Emu3TokenizerFast's [`~PreTrainedTokenizer.decode`]. Please refer to the docstring of this method for more information. )r.decoders rrzEmu3Processor.decodes %t~~$$d5f55rc|jj}|jj}ttj ||zSr)r.model_input_namesr-rTdictfromkeys)rBtokenizer_input_namesimage_processor_input_namess rrzEmu3Processor.model_input_namess? $ @ @&*&:&:&L&L#DMM"7:U"UVWWrr)NNNN)rrr__doc__ attributestokenizer_classimage_processor_classrArrrrrrTr r&rrur~r^rrrpropertyr __classcell__)rDs@rr,r,8s $[1Jrsp"#2%rrC(3&Zu&|5 *% QXNQXh  r