# KOSMOS-2 ## Overview The KOSMOS-2 model was proposed in [Kosmos-2: Grounding Multimodal Large Language Models to the World](https://huggingface.co/papers/2306.14824) by Zhiliang Peng, Wenhui Wang, Li Dong, Yaru Hao, Shaohan Huang, Shuming Ma, Furu Wei. KOSMOS-2 is a Transformer-based causal language model and is trained using the next-word prediction task on a web-scale dataset of grounded image-text pairs [GRIT](https://huggingface.co/datasets/zzliang/GRIT). The spatial coordinates of the bounding boxes in the dataset are converted to a sequence of location tokens, which are appended to their respective entity text spans (for example, `a snowman` followed by ``). The data format is similar to “hyperlinks” that connect the object regions in an image to their text span in the corresponding caption. The abstract from the paper is the following: *We introduce Kosmos-2, a Multimodal Large Language Model (MLLM), enabling new capabilities of perceiving object descriptions (e.g., bounding boxes) and grounding text to the visual world. Specifically, we represent refer expressions as links in Markdown, i.e., ``[text span](bounding boxes)'', where object descriptions are sequences of location tokens. Together with multimodal corpora, we construct large-scale data of grounded image-text pairs (called GrIT) to train the model. In addition to the existing capabilities of MLLMs (e.g., perceiving general modalities, following instructions, and performing in-context learning), Kosmos-2 integrates the grounding capability into downstream applications. We evaluate Kosmos-2 on a wide range of tasks, including (i) multimodal grounding, such as referring expression comprehension, and phrase grounding, (ii) multimodal referring, such as referring expression generation, (iii) perception-language tasks, and (iv) language understanding and generation. This work lays out the foundation for the development of Embodiment AI and sheds light on the big convergence of language, multimodal perception, action, and world modeling, which is a key step toward artificial general intelligence. Code and pretrained models are available at https://aka.ms/kosmos-2.* Overview of tasks that KOSMOS-2 can handle. Taken from the original paper. ## Example ```python >>> from PIL import Image >>> import requests >>> from transformers import AutoProcessor, Kosmos2ForConditionalGeneration >>> model = Kosmos2ForConditionalGeneration.from_pretrained("microsoft/kosmos-2-patch14-224") >>> processor = AutoProcessor.from_pretrained("microsoft/kosmos-2-patch14-224") >>> url = "https://huggingface.co/microsoft/kosmos-2-patch14-224/resolve/main/snowman.jpg" >>> image = Image.open(requests.get(url, stream=True).raw) >>> prompt = " An image of" >>> inputs = processor(text=prompt, images=image, return_tensors="pt") >>> generated_ids = model.generate( ... pixel_values=inputs["pixel_values"], ... input_ids=inputs["input_ids"], ... attention_mask=inputs["attention_mask"], ... image_embeds=None, ... image_embeds_position_mask=inputs["image_embeds_position_mask"], ... use_cache=True, ... max_new_tokens=64, ... ) >>> generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0] >>> processed_text = processor.post_process_generation(generated_text, cleanup_and_extract=False) >>> processed_text ' An image of a snowman warming himself by a fire.' >>> caption, entities = processor.post_process_generation(generated_text) >>> caption 'An image of a snowman warming himself by a fire.' >>> entities [('a snowman', (12, 21), [(0.390625, 0.046875, 0.984375, 0.828125)]), ('a fire', (41, 47), [(0.171875, 0.015625, 0.484375, 0.890625)])] ``` This model was contributed by [Yih-Dar SHIEH](https://huggingface.co/ydshieh). The original code can be found [here](https://github.com/microsoft/unilm/tree/master/kosmos-2). ## Kosmos2Config[[transformers.Kosmos2Config]] #### transformers.Kosmos2Config[[transformers.Kosmos2Config]] [Source](https://github.com/huggingface/transformers/blob/v5.5.3/src/transformers/models/kosmos2/configuration_kosmos2.py#L80) This is the configuration class to store the configuration of a Kosmos2Model. It is used to instantiate a Kosmos2 model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the [microsoft/kosmos-2-patch14-224](https://huggingface.co/microsoft/kosmos-2-patch14-224) Configuration objects inherit from [PreTrainedConfig](/docs/transformers/v5.5.3/en/main_classes/configuration#transformers.PreTrainedConfig) and can be used to control the model outputs. Read the documentation from [PreTrainedConfig](/docs/transformers/v5.5.3/en/main_classes/configuration#transformers.PreTrainedConfig) for more information. Example: ```python >>> from transformers import Kosmos2Config, Kosmos2Model >>> # Initializing a Kosmos-2 kosmos-2-patch14-224 style configuration >>> configuration = Kosmos2Config() >>> # Initializing a model (with random weights) from the kosmos-2-patch14-224 style configuration >>> model = Kosmos2Model(configuration) >>> # Accessing the model configuration >>> configuration = model.config ``` **Parameters:** text_config (`Union[dict, ~configuration_utils.PreTrainedConfig]`, *optional*) : The config object or dictionary of the text backbone. vision_config (`Union[dict, ~configuration_utils.PreTrainedConfig]`, *optional*) : The config object or dictionary of the vision backbone. latent_query_num (`int`, *optional*, defaults to 64) : The number of latent query tokens that represent the image features used in the text decoder component. tie_word_embeddings (`bool`, *optional*, defaults to `True`) : Whether to tie weight embeddings according to model's `tied_weights_keys` mapping. ## Kosmos2ImageProcessor ## Kosmos2Processor[[transformers.Kosmos2Processor]] #### transformers.Kosmos2Processor[[transformers.Kosmos2Processor]] [Source](https://github.com/huggingface/transformers/blob/v5.5.3/src/transformers/models/kosmos2/processing_kosmos2.py#L87) Constructs a Kosmos2Processor which wraps a image processor and a tokenizer into a single processor. [Kosmos2Processor](/docs/transformers/v5.5.3/en/model_doc/kosmos-2#transformers.Kosmos2Processor) offers all the functionalities of `image_processor_class` and `tokenizer_class`. See the `~image_processor_class` and `~tokenizer_class` for more information. __call__transformers.Kosmos2Processor.__call__https://github.com/huggingface/transformers/blob/v5.5.3/src/transformers/models/kosmos2/processing_kosmos2.py#L137[{"name": "images", "val": ": typing.Union[ForwardRef('PIL.Image.Image'), numpy.ndarray, ForwardRef('torch.Tensor'), list['PIL.Image.Image'], list[numpy.ndarray], list['torch.Tensor'], NoneType] = None"}, {"name": "text", "val": ": str | list[str] = None"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.models.kosmos2.processing_kosmos2.Kosmos2ProcessorKwargs]"}]- **images** (`Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, list[PIL.Image.Image], list[numpy.ndarray], list[torch.Tensor]]`, *optional*) -- Image to preprocess. Expects a single or batch of images with pixel values ranging from 0 to 255. If passing in images with pixel values between 0 and 1, set `do_rescale=False`. - **text** (`Union[str, list[str]]`, *optional*) -- The sequence or batch of sequences to be encoded. Each sequence can be a string or a list of strings (pretokenized string). If you pass a pretokenized input, set `is_split_into_words=True` to avoid ambiguity with batched inputs. - **add_eos_token** (`bool`, *kwargs*, defaults to `False`) -- Whether or not to include `EOS` token id in the encoding when `add_special_tokens=True`. - **bboxes** (`Union[list[tuple[int]], *kwargs*, list[tuple[float]], list[list[tuple[int]]], list[list[tuple[float]]]]`, *optional*) -- The bounding bboxes associated to `texts`. - **num_image_tokens** (`int`, *kwargs*, *optional* defaults to 64) -- The number of (consecutive) places that are used to mark the placeholders to store image information. This should be the same as `latent_query_num` in the instance of `Kosmos2Config` you are using. - **first_image_token_id** (`int`, *kwargs*, *optional*) -- The token id that will be used for the first place of the subsequence that is reserved to store image information. If unset, will default to `self.tokenizer.unk_token_id + 1`. - **return_tensors** (`str` or [TensorType](/docs/transformers/v5.5.3/en/internal/file_utils#transformers.TensorType), *optional*) -- If set, will return tensors of a particular framework. Acceptable values are: - `'pt'`: Return PyTorch `torch.Tensor` objects. - `'np'`: Return NumPy `np.ndarray` objects. - ****kwargs** ([ProcessingKwargs](/docs/transformers/v5.5.3/en/main_classes/processors#transformers.ProcessingKwargs), *optional*) -- Additional processing options for each modality (text, images, videos, audio). Model-specific parameters are listed above; see the TypedDict class for the complete list of supported arguments.0`~image_processing_base.BatchFeature`- **data** (`dict`) -- Dictionary of lists/arrays/tensors returned by the __call__ method ('pixel_values', etc.). - **tensor_type** (`Union[None, str, TensorType]`, *optional*) -- You can give a tensor_type here to convert the lists of integers in PyTorch/Numpy Tensors at initialization. **Parameters:** image_processor (`image_processor_class`) : The image processor is a required input. tokenizer (`tokenizer_class`) : The tokenizer is a required input. num_patch_index_tokens (`int`, *optional*, defaults to 1024) : The number of tokens that represent patch indices. **Returns:** ``~image_processing_base.BatchFeature`` - **data** (`dict`) -- Dictionary of lists/arrays/tensors returned by the __call__ method ('pixel_values', etc.). - **tensor_type** (`Union[None, str, TensorType]`, *optional*) -- You can give a tensor_type here to convert the lists of integers in PyTorch/Numpy Tensors at initialization. ## Kosmos2Model[[transformers.Kosmos2Model]] #### transformers.Kosmos2Model[[transformers.Kosmos2Model]] [Source](https://github.com/huggingface/transformers/blob/v5.5.3/src/transformers/models/kosmos2/modeling_kosmos2.py#L1362) KOSMOS-2 Model for generating text and image features. The model consists of a vision encoder and a language model. This model inherits from [PreTrainedModel](/docs/transformers/v5.5.3/en/main_classes/model#transformers.PreTrainedModel). Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.) This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior. forwardtransformers.Kosmos2Model.forwardhttps://github.com/huggingface/transformers/blob/v5.5.3/src/transformers/models/kosmos2/modeling_kosmos2.py#L1414[{"name": "pixel_values", "val": ": torch.Tensor | None = None"}, {"name": "input_ids", "val": ": torch.Tensor | None = None"}, {"name": "image_embeds_position_mask", "val": ": torch.Tensor | None = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "image_embeds", "val": ": torch.Tensor | None = None"}, {"name": "inputs_embeds", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.Tensor | None = None"}, {"name": "use_cache", "val": ": bool | None = None"}, {"name": "interpolate_pos_encoding", "val": ": bool = False"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **pixel_values** (`torch.Tensor` of shape `(batch_size, num_channels, image_size, image_size)`, *optional*) -- The tensors corresponding to the input images. Pixel values can be obtained using `image_processor_class`. See `image_processor_class.__call__` for details (`processor_class` uses `image_processor_class` for processing images). - **input_ids** (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) -- Indices of input sequence tokens in the vocabulary. Padding will be ignored by default. Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.5.3/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.5.3/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.5.3/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details. [What are input IDs?](../glossary#input-ids) - **image_embeds_position_mask** (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) -- Mask to indicate the location in a sequence to insert the image features . Mask values selected in `[0, 1]`: - 1 for places where to put the image features, - 0 for places that are not for image features (i.e. for text tokens). - **attention_mask** (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) -- Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`: - 1 for tokens that are **not masked**, - 0 for tokens that are **masked**. [What are attention masks?](../glossary#attention-mask) - **past_key_values** (`~cache_utils.Cache`, *optional*) -- Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in the `past_key_values` returned by the model at a previous stage of decoding, when `use_cache=True` or `config.use_cache=True`. Only [Cache](/docs/transformers/v5.5.3/en/internal/generation_utils#transformers.Cache) instance is allowed as input, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). If no `past_key_values` are passed, [DynamicCache](/docs/transformers/v5.5.3/en/internal/generation_utils#transformers.DynamicCache) will be initialized by default. The model will output the same cache format that is fed as input. If `past_key_values` are used, the user is expected to input only unprocessed `input_ids` (those that don't have their past key value states given to this model) of shape `(batch_size, unprocessed_length)` instead of all `input_ids` of shape `(batch_size, sequence_length)`. - **image_embeds** (`torch.FloatTensor` of shape `(batch_size, latent_query_num, hidden_size)`, *optional*) -- Sequence of hidden-states at the output of `Kosmos2ImageToTextProjection`. - **inputs_embeds** (`torch.Tensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) -- Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model's internal embedding lookup matrix. - **position_ids** (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) -- Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.n_positions - 1]`. [What are position IDs?](../glossary#position-ids) - **use_cache** (`bool`, *optional*) -- If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). - **interpolate_pos_encoding** (`bool`, *optional*, defaults to `False`) -- Whether to interpolate the pre-trained position encodings.0`Kosmos2ModelOutput` or `tuple(torch.FloatTensor)`A `Kosmos2ModelOutput` or a tuple of `torch.FloatTensor` (if `return_dict=False` is passed or when `config.return_dict=False`) comprising various elements depending on the configuration ([Kosmos2Config](/docs/transformers/v5.5.3/en/model_doc/kosmos-2#transformers.Kosmos2Config)) and inputs. The [Kosmos2Model](/docs/transformers/v5.5.3/en/model_doc/kosmos-2#transformers.Kosmos2Model) forward method, overrides the `__call__` special method. Although the recipe for forward pass needs to be defined within this function, one should call the `Module` instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them. - **last_hidden_state** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*, defaults to `None`) -- Sequence of hidden-states at the output of the last layer of the model. - **past_key_values** (`Cache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`) -- It is a [Cache](/docs/transformers/v5.5.3/en/internal/generation_utils#transformers.Cache) instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). Contains pre-computed hidden-states (key and values in the self-attention blocks and optionally if `config.is_encoder_decoder=True` in the cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding. - **hidden_states** (`tuple[torch.FloatTensor]`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`) -- Tuple of `torch.FloatTensor` (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer plus the optional initial embedding outputs. - **attentions** (`tuple[torch.FloatTensor]`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`) -- Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`. Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads. - **image_embeds** (`torch.FloatTensor` of shape `(batch_size, latent_query_num, hidden_size)`, *optional*) -- Sequence of hidden-states at the output of `Kosmos2ImageToTextProjection`. - **projection_attentions** (`tuple(torch.FloatTensor)`, *optional*) -- Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`. Attentions weights given by `Kosmos2ImageToTextProjection`, after the attention softmax, used to compute the weighted average in the self-attention heads. - **vision_model_output** (`BaseModelOutputWithPooling`, *optional*) -- The output of the `Kosmos2VisionModel`. Examples: ```python >>> from PIL import Image >>> import httpx >>> from io import BytesIO >>> from transformers import AutoProcessor, Kosmos2Model >>> model = Kosmos2Model.from_pretrained("microsoft/kosmos-2-patch14-224") >>> processor = AutoProcessor.from_pretrained("microsoft/kosmos-2-patch14-224") >>> url = "https://huggingface.co/microsoft/kosmos-2-patch14-224/resolve/main/snowman.jpg" >>> with httpx.stream("GET", url) as response: ... image = Image.open(BytesIO(response.read())) >>> text = ( ... " An image of a snowman" ... " warming himself by a fire" ... "" ... ) >>> inputs = processor(text=text, images=image, return_tensors="pt", add_eos_token=True) >>> last_hidden_state = model( ... pixel_values=inputs["pixel_values"], ... input_ids=inputs["input_ids"], ... attention_mask=inputs["attention_mask"], ... image_embeds_position_mask=inputs["image_embeds_position_mask"], ... ).last_hidden_state >>> list(last_hidden_state.shape) [1, 91, 2048] ``` **Parameters:** config ([Kosmos2Config](/docs/transformers/v5.5.3/en/model_doc/kosmos-2#transformers.Kosmos2Config)) : Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the [from_pretrained()](/docs/transformers/v5.5.3/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights. **Returns:** ``Kosmos2ModelOutput` or `tuple(torch.FloatTensor)`` A `Kosmos2ModelOutput` or a tuple of `torch.FloatTensor` (if `return_dict=False` is passed or when `config.return_dict=False`) comprising various elements depending on the configuration ([Kosmos2Config](/docs/transformers/v5.5.3/en/model_doc/kosmos-2#transformers.Kosmos2Config)) and inputs. #### get_image_features[[transformers.Kosmos2Model.get_image_features]] [Source](https://github.com/huggingface/transformers/blob/v5.5.3/src/transformers/models/kosmos2/modeling_kosmos2.py#L1382) - **last_hidden_state** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) -- Sequence of hidden-states at the output of the last layer of the model. - **pooler_output** (`torch.FloatTensor` of shape `(batch_size, hidden_size)`, *optional*) -- Last layer hidden-state after a pooling operation on the spatial dimensions. - **hidden_states** (`tuple[torch.FloatTensor, ...]`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`) -- Tuple of `torch.FloatTensor` (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer plus the optional initial embedding outputs. - **attentions** (`tuple[torch.FloatTensor, ...]`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`) -- Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`. Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads. - **projection_attentions** (`tuple(torch.FloatTensor)`) -- Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`. Attentions weights given by `Kosmos2ImageToTextProjection`, after the attention softmax, used to compute the weighted average in the self-attention heads. **Parameters:** pixel_values (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`) : The tensors corresponding to the input images. Pixel values can be obtained using `image_processor_class`. See `image_processor_class.__call__` for details (`processor_class` uses `image_processor_class` for processing images). interpolate_pos_encoding (`bool`, *optional*, defaults to `False`) : Whether to interpolate the pre-trained position encodings. **Returns:** ``BaseModelOutputWithProjectionAttentions` or `tuple(torch.FloatTensor)`` A `BaseModelOutputWithProjectionAttentions` or a tuple of `torch.FloatTensor` (if `return_dict=False` is passed or when `config.return_dict=False`) comprising various elements depending on the configuration ([Kosmos2Config](/docs/transformers/v5.5.3/en/model_doc/kosmos-2#transformers.Kosmos2Config)) and inputs. ## Kosmos2ForConditionalGeneration[[transformers.Kosmos2ForConditionalGeneration]] #### transformers.Kosmos2ForConditionalGeneration[[transformers.Kosmos2ForConditionalGeneration]] [Source](https://github.com/huggingface/transformers/blob/v5.5.3/src/transformers/models/kosmos2/modeling_kosmos2.py#L1513) KOSMOS-2 Model for generating text and bounding boxes given an image. The model consists of a vision encoder and a language model. This model inherits from [PreTrainedModel](/docs/transformers/v5.5.3/en/main_classes/model#transformers.PreTrainedModel). Check the superclass documentation for the generic methods the library implements for all its model (such as downloading or saving, resizing the input embeddings, pruning heads etc.) This model is also a PyTorch [torch.nn.Module](https://pytorch.org/docs/stable/nn.html#torch.nn.Module) subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior. forwardtransformers.Kosmos2ForConditionalGeneration.forwardhttps://github.com/huggingface/transformers/blob/v5.5.3/src/transformers/models/kosmos2/modeling_kosmos2.py#L1541[{"name": "pixel_values", "val": ": torch.Tensor | None = None"}, {"name": "input_ids", "val": ": torch.Tensor | None = None"}, {"name": "image_embeds_position_mask", "val": ": torch.Tensor | None = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "image_embeds", "val": ": torch.Tensor | None = None"}, {"name": "inputs_embeds", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.Tensor | None = None"}, {"name": "labels", "val": ": torch.LongTensor | None = None"}, {"name": "use_cache", "val": ": bool | None = None"}, {"name": "logits_to_keep", "val": ": int | torch.Tensor = 0"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **pixel_values** (`torch.Tensor` of shape `(batch_size, num_channels, image_size, image_size)`, *optional*) -- The tensors corresponding to the input images. Pixel values can be obtained using `image_processor_class`. See `image_processor_class.__call__` for details (`processor_class` uses `image_processor_class` for processing images). - **input_ids** (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) -- Indices of input sequence tokens in the vocabulary. Padding will be ignored by default. Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.5.3/en/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.5.3/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.5.3/en/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details. [What are input IDs?](../glossary#input-ids) - **image_embeds_position_mask** (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) -- Mask to indicate the location in a sequence to insert the image features . Mask values selected in `[0, 1]`: - 1 for places where to put the image features, - 0 for places that are not for image features (i.e. for text tokens). - **attention_mask** (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) -- Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`: - 1 for tokens that are **not masked**, - 0 for tokens that are **masked**. [What are attention masks?](../glossary#attention-mask) - **past_key_values** (`~cache_utils.Cache`, *optional*) -- Pre-computed hidden-states (key and values in the self-attention blocks and in the cross-attention blocks) that can be used to speed up sequential decoding. This typically consists in the `past_key_values` returned by the model at a previous stage of decoding, when `use_cache=True` or `config.use_cache=True`. Only [Cache](/docs/transformers/v5.5.3/en/internal/generation_utils#transformers.Cache) instance is allowed as input, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). If no `past_key_values` are passed, [DynamicCache](/docs/transformers/v5.5.3/en/internal/generation_utils#transformers.DynamicCache) will be initialized by default. The model will output the same cache format that is fed as input. If `past_key_values` are used, the user is expected to input only unprocessed `input_ids` (those that don't have their past key value states given to this model) of shape `(batch_size, unprocessed_length)` instead of all `input_ids` of shape `(batch_size, sequence_length)`. - **image_embeds** (`torch.FloatTensor` of shape `(batch_size, latent_query_num, hidden_size)`, *optional*) -- Sequence of hidden-states at the output of `Kosmos2ImageToTextProjection`. - **inputs_embeds** (`torch.Tensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) -- Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model's internal embedding lookup matrix. - **position_ids** (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) -- Indices of positions of each input sequence tokens in the position embeddings. Selected in the range `[0, config.n_positions - 1]`. [What are position IDs?](../glossary#position-ids) - **labels** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) -- Labels for computing the left-to-right language modeling loss (next word prediction). Indices should be in `[-100, 0, ..., config.vocab_size]` (see `input_ids` docstring) Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]` - **use_cache** (`bool`, *optional*) -- If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). - **logits_to_keep** (`Union[int, torch.Tensor]`, *optional*, defaults to `0`) -- If an `int`, compute logits for the last `logits_to_keep` tokens. If `0`, calculate logits for all `input_ids` (special case). Only last token logits are needed for generation, and calculating them only for that token can save memory, which becomes pretty significant for long sequences or large vocabulary size. If a `torch.Tensor`, must be 1D corresponding to the indices to keep in the sequence length dimension. This is useful when using packed tensor format (single dimension for batch and sequence length).0`Kosmos2ForConditionalGenerationModelOutput` or `tuple(torch.FloatTensor)`A `Kosmos2ForConditionalGenerationModelOutput` or a tuple of `torch.FloatTensor` (if `return_dict=False` is passed or when `config.return_dict=False`) comprising various elements depending on the configuration ([Kosmos2Config](/docs/transformers/v5.5.3/en/model_doc/kosmos-2#transformers.Kosmos2Config)) and inputs. The [Kosmos2ForConditionalGeneration](/docs/transformers/v5.5.3/en/model_doc/kosmos-2#transformers.Kosmos2ForConditionalGeneration) forward method, overrides the `__call__` special method. Although the recipe for forward pass needs to be defined within this function, one should call the `Module` instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them. - **loss** (`torch.FloatTensor` of shape `(1,)`, *optional*, returned when `labels` is provided) -- Language modeling loss (for next-token prediction). - **logits** (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`) -- Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax). - **past_key_values** (`Cache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`) -- It is a [Cache](/docs/transformers/v5.5.3/en/internal/generation_utils#transformers.Cache) instance. For more details, see our [kv cache guide](https://huggingface.co/docs/transformers/en/kv_cache). Contains pre-computed hidden-states (key and values in the self-attention blocks and optionally if `config.is_encoder_decoder=True` in the cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding. - **hidden_states** (`tuple[torch.FloatTensor]`, *optional*, returned when `output_hidden_states=True` is passed or when `config.output_hidden_states=True`) -- Tuple of `torch.FloatTensor` (one for the output of the embeddings, if the model has an embedding layer, + one for the output of each layer) of shape `(batch_size, sequence_length, hidden_size)`. Hidden-states of the model at the output of each layer plus the optional initial embedding outputs. - **attentions** (`tuple[torch.FloatTensor]`, *optional*, returned when `output_attentions=True` is passed or when `config.output_attentions=True`) -- Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`. Attentions weights after the attention softmax, used to compute the weighted average in the self-attention heads. - **image_embeds** (`torch.FloatTensor` of shape `(batch_size, latent_query_num, hidden_size)`, *optional*) -- Sequence of hidden-states at the output of `Kosmos2ImageToTextProjection`. - **projection_attentions** (`tuple(torch.FloatTensor)`, *optional*) -- Tuple of `torch.FloatTensor` (one for each layer) of shape `(batch_size, num_heads, sequence_length, sequence_length)`. Attentions weights given by `Kosmos2ImageToTextProjection`, after the attention softmax, used to compute the weighted average in the self-attention heads. - **vision_model_output** (`BaseModelOutputWithPooling`, *optional*) -- The output of the `Kosmos2VisionModel`. Examples: ```python >>> from PIL import Image >>> import httpx >>> from io import BytesIO >>> from transformers import AutoProcessor, Kosmos2ForConditionalGeneration >>> model = Kosmos2ForConditionalGeneration.from_pretrained("microsoft/kosmos-2-patch14-224") >>> processor = AutoProcessor.from_pretrained("microsoft/kosmos-2-patch14-224") >>> url = "https://huggingface.co/microsoft/kosmos-2-patch14-224/resolve/main/snowman.jpg" >>> with httpx.stream("GET", url) as response: ... image = Image.open(BytesIO(response.read())) >>> prompt = " An image of" >>> inputs = processor(text=prompt, images=image, return_tensors="pt") >>> generated_ids = model.generate( ... pixel_values=inputs["pixel_values"], ... input_ids=inputs["input_ids"], ... attention_mask=inputs["attention_mask"], ... image_embeds=None, ... image_embeds_position_mask=inputs["image_embeds_position_mask"], ... use_cache=True, ... max_new_tokens=64, ... ) >>> generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0] >>> processed_text = processor.post_process_generation(generated_text, cleanup_and_extract=False) >>> processed_text ' An image of a snowman warming himself by a fire.' >>> caption, entities = processor.post_process_generation(generated_text) >>> caption 'An image of a snowman warming himself by a fire.' >>> entities [('a snowman', (12, 21), [(0.390625, 0.046875, 0.984375, 0.828125)]), ('a fire', (41, 47), [(0.171875, 0.015625, 0.484375, 0.890625)])] ``` **Parameters:** config ([Kosmos2Config](/docs/transformers/v5.5.3/en/model_doc/kosmos-2#transformers.Kosmos2Config)) : Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the [from_pretrained()](/docs/transformers/v5.5.3/en/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights. **Returns:** ``Kosmos2ForConditionalGenerationModelOutput` or `tuple(torch.FloatTensor)`` A `Kosmos2ForConditionalGenerationModelOutput` or a tuple of `torch.FloatTensor` (if `return_dict=False` is passed or when `config.return_dict=False`) comprising various elements depending on the configuration ([Kosmos2Config](/docs/transformers/v5.5.3/en/model_doc/kosmos-2#transformers.Kosmos2Config)) and inputs.