`i.ddlmZmZddlZddlmZddlmZddl m Z m Z m Z m Z mZmZddlmZmZddlmZerd d lmZGd d ed ZGdde d ZGdde d ZGdde ZdgZdS))OptionalUnionN) BatchFeature) ImageInput) ImagesKwargsMultiModalDataProcessingKwargsProcessorMixin TextKwargsUnpack)PreTokenizedInput TextInput)is_vision_available) smart_resizeceZdZUeed<dS)Emu3TextKwargsreturn_for_image_generationN)__name__ __module__ __qualname__bool__annotations__|/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/transformers/models/emu3/processing_emu3.pyrr s!%%%%%%rrF)totalc$eZdZUeed<eed<dS)Emu3ImagesKwargsratio image_areaN)rrrstrrintrrrr r $s" JJJOOOOOrr c:eZdZUeed<eed<dddddddZd S) Emu3ProcessorKwargs text_kwargs images_kwargsF)rreturn_mm_token_type_idsz1:1i)r!r")r'r(N)rrrrrr _defaultsrrrr&r&)sX####,1(-      IIIrr&c eZdZdZddgZdZdZ dfd Z ddee d ee e e e e e e fd eed efd Zdd ZdZde fdZxZS) Emu3Processora Constructs a Emu3 processor which wraps a Emu3 image processor and a GPT2 tokenizer into a single processor. [`Emu3Processor`] offers all the functionalities of [`Emu3ImageProcessor`] and [`GPT2TokenizerFast`]. See the [`~Emu3Processor.__call__`] and [`~Emu3Processor.decode`] for more information. Args: image_processor ([`Emu3ImageProcessor`]): The image processor is a required input. tokenizer ([`Emu3TokenizerFast`]): The tokenizer is a required input. chat_template (`str`, *optional*): A Jinja template which will be used to convert lists of messages in a chat into a tokenizable string. image_processor tokenizer) GPT2TokenizerGPT2TokenizerFastEmu3ImageProcessorNc |j|_|j|_|j|_|j|_|j|_|j|_|j |_ d|_ t |||dS)N) chat_template) image_tokenimage_token_id boi_tokenimage_start_token eoi_tokenimage_end_tokenimage_wrapper_tokenfake_token_around_image eof_token bos_tokendownsample_ratiosuper__init__)selfr-r.r4kwargs __class__s rrAzEmu3Processor.__init__Ms}%0'6!*!4(2'0'D$",", ! )=QQQQQrimagestextrCreturnc 0t|tr|g}n?t|ts*t|dtstdjt fdjji|}|ddd}|ddd }|dd d } |r|td |s||td i} j } j j } |s|j |fi|d} t| j} g}|D]}j|vrwt#| }|\}}|jz}|jz}||d zz}| |d|jd|z| }|j|d }j|}j|vw||fd|D}nX|rV|| j\}}| |d|jfd|D}||ggt1|z| d<|ddd }|ddd}j|fi|ddd i}||dg|rYt5j|d}t5j|d}d ||jk<||d<t?i|| |S)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 `kwargs` 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* cFg|]}|djS)rK)replacer5).0samplerBs r z*Emu3Processor.__call__..s*ccc&FNN?D4DEEcccrc*g|]}j|Sr)r>)rNrO image_promptrBs rrPz*Emu3Processor.__call__..s+QQQ&t~=v=|==QQQr image_sizesreturn_tensorsr)image) modalities input_idsmm_token_type_ids)data tensor_type) isinstancer#list TypeError _merge_kwargsr&r. init_kwargspop ValueErrorr8r=r:r-iterrSr5nextr?r<rMr>appendcalculate_generate_sizelen_check_special_mm_tokensnparray zeros_liker6tolistr)rBrErFaudiovideosrC output_kwargsrr!r"image_featuresimage_start_tokensimage_end_tokensrSprompt_stringsrO image_sizeheightwidthimage_seq_lengthimage_placeholderrTr) text_inputs array_idsrXrRs` @r__call__zEmu3Processor.__call__^s*T dC  a6DDD$'' a 47C0H0H a_`` `**   "&."<   '4M&B&F&FGdfk&l&l#o.227DAA"?377 dKK & h6+=fgg g* pt|noo o $ 68"nDd.BDD+ Jv/A1T1&[[M/OQRSSF $888F&&00%%f----ccccTbcccDD) J 88 DLabbMFE0`&``5`$B^``LQQQQQDQQQD.4e_,=D ,IN= )'}599:JDQQ#0#?#C#CD^`e#f#f $dnT__]=-I__Z^___  %%dKWI%NNN # J[!9::I " k+.F G G BC i4+>> ?/@/G/G/I/IK+ ,!BK!B>!BP^____rc Vi}|g}|D]j\}}t|||jj|jj|jj\}}||jz}||jz}||dzz}||kdgt|z}|||dtdi|S)a Computes the number of placeholder tokens needed for multimodal inputs with the given sizes. Args: image_sizes (`list[list[int]]`, *optional*): The input sizes formatted as (height, width) per each image. Returns: `MultiModalData`: A `MultiModalData` object holding number of tokens per each of the provided input modalities, along with other useful data. Nr)num_image_tokensnum_image_patchesr) rr-spatial_factor min_pixels max_pixelsr?rdrfupdater ) rBrSrC vision_datar|rtrurvr}s r_get_num_multimodal_tokensz(Emu3Processor._get_num_multimodal_tokenss  "! !, : :  ,(7(3(3 !!  4#88!66#)UQY#7  ''(89999!"c+&6&6 6    4D[lmm n n n,, ,,,rctt|d\}}||z}||z dz}tt||z|z }tt||z|z } || fS)N:g?)mapr$splitround) rBr!r"r~rurt current_area target_ratio token_height token_widths rrez%Emu3Processor.calculate_generate_sizesCS!1!122 vv~ "\1c9 5,!6!GHHII % 4~ EFFGG [((rc (|jj|fi|SN)r- postprocess)rBrErCs rrzEmu3Processor.postprocesss /t#/AA&AAArr)NNNN)rrr__doc__ attributestokenizer_classimage_processor_classrArrrrrr\r r&rrzrrer __classcell__)rDs@rr,r,8s: $[1Jrs"#"""""""222222%%%%%%rrrrrrrrrrrrrrrrCCCCCCCC((((((4333333&&&&&Zu&&&&|5     *%    }B}B}B}B}BN}B}B}B@  r