# coding=utf-8 # Copyright 2025 HuggingFace Inc. team. All rights reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. from typing import Optional, Union import numpy as np from ...image_processing_utils import BatchFeature from ...image_utils import ImageInput from ...processing_utils import ImagesKwargs, MultiModalData, ProcessingKwargs, ProcessorMixin, Unpack from ...tokenization_utils_base import PreTokenizedInput, TextInput class Cohere2VisionImagesKwargs(ImagesKwargs, total=False): max_patches: Optional[int] class Cohere2VisionProcessorKwargs(ProcessingKwargs, total=False): images_kwargs: Cohere2VisionImagesKwargs _defaults = { "text_kwargs": { "padding_side": "left", "padding": True, "return_mm_token_type_ids": False, }, } class Cohere2VisionProcessor(ProcessorMixin): r""" Constructs a Cohere2Vision processor which wraps a [`AutoImageProcessor`] and [`PretrainedTokenizerFast`] tokenizer into a single processor that inherits both the image processor and tokenizer functionalities. See the [`~Cohere2VisionProcessor.__call__`] and [`~Cohere2VisionProcessor.decode`] for more information. Args: image_processor ([`AutoImageProcessor`], *optional*): The image processor is a required input. tokenizer ([`PreTrainedTokenizer`, `PreTrainedTokenizerFast`], *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. """ attributes = ["image_processor", "tokenizer"] image_processor_class = "AutoImageProcessor" tokenizer_class = "AutoTokenizer" def __init__( self, image_processor=None, tokenizer=None, chat_template=None, **kwargs, ): super().__init__(image_processor, tokenizer, chat_template=chat_template) self.patch_size = self.image_processor.patch_size self.boi_token = tokenizer.boi_token self.eoi_token = tokenizer.eoi_token self.image_token = tokenizer.image_token self.img_line_break_token = tokenizer.img_line_break_token self.image_token_id = tokenizer.image_token_id self.image_ids = tokenizer.convert_tokens_to_ids( [ self.image_token, self.boi_token, self.eoi_token, self.img_line_break_token, ] ) def __call__( self, images: Optional[ImageInput] = None, text: Optional[Union[TextInput, PreTokenizedInput, list[TextInput], list[PreTokenizedInput]]] = None, **kwargs: Unpack[Cohere2VisionProcessorKwargs], ) -> BatchFeature: """ Main method to prepare for the model one or several sequences(s) and image(s). This method forwards the `text` and `kwargs` arguments to PreTrainedTokenizerFast's [`~PreTrainedTokenizerFast.__call__`] to encode the text. To prepare the vision inputs, this method forwards the `images` and `kwargs` arguments to GotOcr2ImageProcessor's [`~GotOcr2ImageProcessor.__call__`] if `images` is not `None`. 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`. """ if text is None: raise ValueError("You have to specify text.") elif not isinstance(text, (list, tuple)): text = [text] output_kwargs = self._merge_kwargs( Cohere2VisionProcessorKwargs, tokenizer_init_kwargs=self.tokenizer.init_kwargs, **kwargs, ) # Process images image_inputs = {} if images is not None: image_inputs = self.image_processor(images=images, **output_kwargs["images_kwargs"]) batch_num_patches = iter(image_inputs.pop("num_patches")) processed_text = [] for sample in text: while self.image_token in sample: num_patches = next(batch_num_patches) img_patches_per_tile = int(self.patch_size**2) img_string = f"{self.boi_token}" for idx in range(1, num_patches): img_string += "" * img_patches_per_tile + self.img_line_break_token img_string += "" * img_patches_per_tile + self.img_line_break_token img_string += f"{self.eoi_token}" sample = sample.replace(self.image_token, img_string, 1) processed_text.append(sample) text = [sample.replace("", self.image_token) for sample in processed_text] return_tensors = output_kwargs["text_kwargs"].pop("return_tensors", None) return_mm_token_type_ids = output_kwargs["text_kwargs"].pop("return_mm_token_type_ids", False) text_inputs = self.tokenizer(text, **output_kwargs["text_kwargs"], return_tensors=None) if return_mm_token_type_ids: array_ids = np.array(text_inputs["input_ids"]) mm_token_type_ids = np.zeros_like(text_inputs["input_ids"]) mm_token_type_ids[np.isin(array_ids, self.image_ids)] = 1 text_inputs["mm_token_type_ids"] = mm_token_type_ids.tolist() return BatchFeature(data={**text_inputs, **image_inputs}, tensor_type=return_tensors) def _get_num_multimodal_tokens(self, image_sizes=None, **kwargs): """ 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. """ vision_data = {} if image_sizes is not None: images_kwargs = Cohere2VisionProcessorKwargs._defaults.get("images_kwargs", {}) images_kwargs.update(kwargs) num_image_patches = [ self.image_processor.get_number_of_image_patches(*image_size, images_kwargs) for image_size in image_sizes ] token_per_patch = int(self.patch_size**2) num_image_tokens = [ 2 + sum(token_per_patch + 1 for _ in range(num_patches)) for num_patches in num_image_patches ] # Add +2 and +1 for BOI/EOI and image break tokens vision_data.update({"num_image_tokens": num_image_tokens, "num_image_patches": num_image_patches}) return MultiModalData(**vision_data) def batch_decode(self, *args, **kwargs): """ This method forwards all its arguments to PreTrainedTokenizerFast's [`~PreTrainedTokenizer.batch_decode`]. Please refer to the docstring of this method for more information. """ return self.tokenizer.batch_decode(*args, **kwargs) def decode(self, *args, **kwargs): """ This method forwards all its arguments to PreTrainedTokenizerFast's [`~PreTrainedTokenizer.decode`]. Please refer to the docstring of this method for more information. """ return self.tokenizer.decode(*args, **kwargs) @property def model_input_names(self): tokenizer_input_names = self.tokenizer.model_input_names image_processor_input_names = self.image_processor.model_input_names return list(tokenizer_input_names) + list(image_processor_input_names) __all__ = ["Cohere2VisionProcessor"]