o i.@sddlmZmZddlZddlmZddlmZddl m Z m Z m Z m Z mZmZddlmZmZddlmZer?d d lmZGd d d ed dZGddde d dZGddde d dZGddde ZdgZdS))OptionalUnionN) BatchFeature) ImageInput) ImagesKwargsMultiModalDataProcessingKwargsProcessorMixin TextKwargsUnpack)PreTokenizedInput TextInput)is_vision_available) smart_resizec@seZdZUeed<dS)Emu3TextKwargsreturn_for_image_generationN)__name__ __module__ __qualname__bool__annotations__rre/home/ubuntu/veenaModal/venv/lib/python3.10/site-packages/transformers/models/emu3/processing_emu3.pyr s rF)totalc@seZdZUeed<eed<dS)Emu3ImagesKwargsratio image_areaN)rrrstrrintrrrrr$s  rc@s4eZdZUeed<eed<dddddddZd S) Emu3ProcessorKwargs text_kwargs images_kwargsF)rreturn_mm_token_type_idsz1:1i)rr)r"r#N)rrrrrr _defaultsrrrrr!)s  r!c seZdZdZddgZdZdZ dfdd Z    dd ee d ee e e e e e e fd eed efd dZdddZddZd e fddZZS) 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 sT|j|_|j|_|j|_|j|_|j|_|j|_|j |_ d|_ t j |||ddS)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(r-kwargs __class__rrr:MszEmu3Processor.__init__imagestextr<returnc st|tr |g}nt|tst|dtstdjtfdjji|}|ddd}|ddd }|dd d } |rH|d urHt d |sV|d urV|d urVt d i} j } j j } |s|d urӈj |fi|d} t| j} g}|D]J}j|vrt| }|\}}|j}|j}||d }| |d|jd|| }|j|d }j|}j|vs||q~fdd|D}n.|r|| j\}}| |d|jfdd|D}||ggt|| d<|ddd }|ddd}j|fi|ddd i}j||dgd|rGt|d}t|d}d ||jk<||d<ti|| |dS)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#rNrzGYou should not provide `images` when `return_for_image_generation=True`zOYou must provide either text or images when `return_for_image_generation=False`r* csg|] }|djqS)rD)replacer..0sample)r;rr sz*Emu3Processor.__call__..csg|] }j|qSr)r7rF image_promptr;rrrIs image_sizesreturn_tensorsr$image) modalities input_idsmm_token_type_ids)data tensor_type) isinstancerlist TypeError _merge_kwargsr!r( init_kwargspop ValueErrorr1r6r3r'iterrLr.nextr8r5rEr7appendcalculate_generate_sizelen_check_special_mm_tokensnparray zeros_liker/tolistr)r;r?r@audiovideosr< output_kwargsrrrimage_featuresimage_start_tokensimage_end_tokensrLprompt_stringsrH image_sizeheightwidthimage_seq_lengthimage_placeholderrMr$ text_inputs array_idsrQrrJr__call__^sj *       $  zEmu3Processor.__call__c Ksi}|durCg}|D])\}}t|||jj|jj|jj\}}||j}||j}||d}||q dgt|}|||dt di|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_pixelsr8r]r_updater) r;rLr< vision_datartrmrnrorurrr_get_num_multimodal_tokenss$      z(Emu3Processor._get_num_multimodal_tokensc CsXtt|d\}}||}||d}tt|||}tt|||} || fS)N:g?)mapr splitround) r;rrrvrnrm current_area target_ratio token_height token_widthrrrr^s  z%Emu3Processor.calculate_generate_sizecKs|jj|fi|SN)r' postprocess)r;r?r<rrrrszEmu3Processor.postprocessr)NNNN)rrr__doc__ attributestokenizer_classimage_processor_classr:rrrrr rUr r!rrsr{r^r __classcell__rrr=rr&8s. k" r&)typingrrnumpyraimage_processing_utilsr image_utilsrprocessing_utilsrrr r r r tokenization_utils_baser rutilsrimage_processing_emu3rrrr!r&__all__rrrrs      A