# BLIP-2

## Overview

BLIP-2 モデルは、[BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://huggingface.co/papers/2301.12597) で提案されました。
Junnan Li, Dongxu Li, Silvio Savarese, Steven Hoi.・サバレーゼ、スティーブン・ホイ。 BLIP-2 は、軽量の 12 層 Transformer をトレーニングすることで、フリーズされた事前トレーニング済み画像エンコーダーと大規模言語モデル (LLM) を活用します。
それらの間にエンコーダーを配置し、さまざまな視覚言語タスクで最先端のパフォーマンスを実現します。最も注目すべき点は、BLIP-2 が 800 億パラメータ モデルである [Flamingo](https://huggingface.co/papers/2204.14198) を 8.7% 改善していることです。
ゼロショット VQAv2 ではトレーニング可能なパラメーターが 54 分の 1 に減少します。

論文の要約は次のとおりです。

*大規模モデルのエンドツーエンドのトレーニングにより、視覚と言語の事前トレーニングのコストはますます法外なものになってきています。この論文では、市販の凍結済み事前トレーニング画像エンコーダと凍結された大規模言語モデルから視覚言語の事前トレーニングをブートストラップする、汎用的で効率的な事前トレーニング戦略である BLIP-2 を提案します。 BLIP-2 は、2 段階で事前トレーニングされた軽量の Querying Transformer でモダリティのギャップを橋渡しします。最初のステージでは、フリーズされた画像エンコーダーから学習する視覚言語表現をブートストラップします。第 2 段階では、凍結された言語モデルから視覚から言語への生成学習をブートストラップします。 BLIP-2 は、既存の方法よりもトレーニング可能なパラメーターが大幅に少ないにもかかわらず、さまざまな視覚言語タスクで最先端のパフォーマンスを実現します。たとえば、私たちのモデルは、トレーニング可能なパラメーターが 54 分の 1 少ないゼロショット VQAv2 で、Flamingo80B を 8.7% 上回っています。また、自然言語の命令に従うことができる、ゼロショット画像からテキストへの生成というモデルの新しい機能も実証します*

<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/blip2_architecture.jpg"
alt="drawing" width="600"/> 

 BLIP-2 アーキテクチャ。 元の論文から抜粋。 

このモデルは、[nielsr](https://huggingface.co/nielsr) によって提供されました。
元のコードは [ここ](https://github.com/salesforce/LAVIS/tree/5ee63d688ba4cebff63acee04adaef2dee9af207) にあります。

## Usage tips

- BLIP-2 は、画像とオプションのテキスト プロンプトを指定して条件付きテキストを生成するために使用できます。推論時には、 `generate` メソッドを使用することをお勧めします。
- [Blip2Processor](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Processor) を使用してモデル用の画像を準備し、予測されたトークン ID をデコードしてテキストに戻すことができます。

## Resources

BLIP-2 の使用を開始するのに役立つ公式 Hugging Face およびコミュニティ (🌎 で示されている) リソースのリスト。

- 画像キャプション、ビジュアル質問応答 (VQA)、およびチャットのような会話のための BLIP-2 のデモ ノートブックは、[こちら](https://github.com/NielsRogge/Transformers-Tutorials/tree/master/BLIP-2) にあります。

ここに含めるリソースの送信に興味がある場合は、お気軽にプル リクエストを開いてください。審査させていただきます。リソースは、既存のリソースを複製するのではなく、何か新しいものを示すことが理想的です。

## Blip2Config[[transformers.Blip2Config]]

#### transformers.Blip2Config[[transformers.Blip2Config]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/configuration_blip_2.py#L107)

This is the configuration class to store the configuration of a Blip 2Model. It is used to instantiate a Blip 2
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 [Salesforce/blip2-opt-2.7b](https://huggingface.co/Salesforce/blip2-opt-2.7b)

Configuration objects inherit from [PreTrainedConfig](/docs/transformers/v5.8.0/ja/main_classes/configuration#transformers.PreTrainedConfig) and can be used to control the model outputs. Read the
documentation from [PreTrainedConfig](/docs/transformers/v5.8.0/ja/main_classes/configuration#transformers.PreTrainedConfig) for more information.

Example:

```python
>>> from transformers import (
...     Blip2VisionConfig,
...     Blip2QFormerConfig,
...     OPTConfig,
...     Blip2Config,
...     Blip2ForConditionalGeneration,
... )

>>> # Initializing a Blip2Config with Salesforce/blip2-opt-2.7b style configuration
>>> configuration = Blip2Config()

>>> # Initializing a Blip2ForConditionalGeneration (with random weights) from the Salesforce/blip2-opt-2.7b style configuration
>>> model = Blip2ForConditionalGeneration(configuration)

>>> # Accessing the model configuration
>>> configuration = model.config

>>> # We can also initialize a Blip2Config from a Blip2VisionConfig, Blip2QFormerConfig and any PreTrainedConfig

>>> # Initializing BLIP-2 vision, BLIP-2 Q-Former and language model configurations
>>> vision_config = Blip2VisionConfig()
>>> qformer_config = Blip2QFormerConfig()
>>> text_config = OPTConfig()

>>> config = Blip2Config(vision_config=vision_config, qformer_config=qformer_config, text_config=text_config)
```

**Parameters:**

vision_config (`Union[dict, ~configuration_utils.PreTrainedConfig]`, *optional*) : The config object or dictionary of the vision backbone.

qformer_config (`dict`, *optional*) : Dictionary of configuration options used to initialize [Blip2QFormerConfig](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2QFormerConfig).

text_config (`Union[dict, ~configuration_utils.PreTrainedConfig]`, *optional*) : The config object or dictionary of the text backbone.

num_query_tokens (`int`, *optional*, defaults to 32) : The number of query tokens passed through the Transformer.

image_text_hidden_size (`int`, *optional*, defaults to 256) : Dimensionality of the hidden state of the image-text fusion layer.

image_token_index (`int`, *optional*) : The image token index used as a placeholder for input images.

initializer_factor (`float`, *optional*, defaults to `1.0`) : A factor for initializing all weight matrices (should be kept to 1, used internally for initialization testing).

initializer_range (`float`, *optional*, defaults to `0.02`) : The standard deviation of the truncated_normal_initializer for initializing all weight matrices.

## Blip2VisionConfig[[transformers.Blip2VisionConfig]]

#### transformers.Blip2VisionConfig[[transformers.Blip2VisionConfig]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/configuration_blip_2.py#L29)

This is the configuration class to store the configuration of a Blip 2Model. It is used to instantiate a Blip 2
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 [Salesforce/blip2-opt-2.7b](https://huggingface.co/Salesforce/blip2-opt-2.7b)

Configuration objects inherit from [PreTrainedConfig](/docs/transformers/v5.8.0/ja/main_classes/configuration#transformers.PreTrainedConfig) and can be used to control the model outputs. Read the
documentation from [PreTrainedConfig](/docs/transformers/v5.8.0/ja/main_classes/configuration#transformers.PreTrainedConfig) for more information.

Example:

```python
>>> from transformers import Blip2VisionConfig, Blip2VisionModel

>>> # Initializing a Blip2VisionConfig with Salesforce/blip2-opt-2.7b style configuration
>>> configuration = Blip2VisionConfig()

>>> # Initializing a Blip2VisionModel (with random weights) from the Salesforce/blip2-opt-2.7b style configuration
>>> model = Blip2VisionModel(configuration)

>>> # Accessing the model configuration
>>> configuration = model.config
```

**Parameters:**

hidden_size (`int`, *optional*, defaults to `1408`) : Dimension of the hidden representations.

intermediate_size (`int`, *optional*, defaults to `6144`) : Dimension of the MLP representations.

num_hidden_layers (`int`, *optional*, defaults to `39`) : Number of hidden layers in the Transformer decoder.

num_attention_heads (`int`, *optional*, defaults to `16`) : Number of attention heads for each attention layer in the Transformer decoder.

image_size (`Union[int, list[int], tuple[int, int]]`, *optional*, defaults to `224`) : The size (resolution) of each image.

patch_size (`Union[int, list[int], tuple[int, int]]`, *optional*, defaults to `14`) : The size (resolution) of each patch.

hidden_act (`str`, *optional*, defaults to `gelu`) : The non-linear activation function (function or string) in the decoder. For example, `"gelu"`, `"relu"`, `"silu"`, etc.

layer_norm_eps (`float`, *optional*, defaults to `1e-06`) : The epsilon used by the layer normalization layers.

attention_dropout (`Union[float, int]`, *optional*, defaults to `0.0`) : The dropout ratio for the attention probabilities.

initializer_range (`float`, *optional*, defaults to `1e-10`) : The standard deviation of the truncated_normal_initializer for initializing all weight matrices.

qkv_bias (`bool`, *optional*, defaults to `True`) : Whether to add a bias to the queries, keys and values.

## Blip2QFormerConfig[[transformers.Blip2QFormerConfig]]

#### transformers.Blip2QFormerConfig[[transformers.Blip2QFormerConfig]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/configuration_blip_2.py#L64)

This is the configuration class to store the configuration of a Blip 2Model. It is used to instantiate a Blip 2
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 [Salesforce/blip2-opt-2.7b](https://huggingface.co/Salesforce/blip2-opt-2.7b)

Configuration objects inherit from [PreTrainedConfig](/docs/transformers/v5.8.0/ja/main_classes/configuration#transformers.PreTrainedConfig) and can be used to control the model outputs. Read the
documentation from [PreTrainedConfig](/docs/transformers/v5.8.0/ja/main_classes/configuration#transformers.PreTrainedConfig) for more information.

Examples:

```python
>>> from transformers import Blip2QFormerConfig, Blip2QFormerModel

>>> # Initializing a BLIP-2 Salesforce/blip2-opt-2.7b style configuration
>>> configuration = Blip2QFormerConfig()

>>> # Initializing a model (with random weights) from the Salesforce/blip2-opt-2.7b style configuration
>>> model = Blip2QFormerModel(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
```

**Parameters:**

vocab_size (`int`, *optional*, defaults to `30522`) : Vocabulary size of the model. Defines the number of different tokens that can be represented by the `input_ids`.

hidden_size (`int`, *optional*, defaults to `768`) : Dimension of the hidden representations.

num_hidden_layers (`int`, *optional*, defaults to `12`) : Number of hidden layers in the Transformer decoder.

num_attention_heads (`int`, *optional*, defaults to `12`) : Number of attention heads for each attention layer in the Transformer decoder.

intermediate_size (`int`, *optional*, defaults to `3072`) : Dimension of the MLP representations.

hidden_act (`str`, *optional*, defaults to `gelu`) : The non-linear activation function (function or string) in the decoder. For example, `"gelu"`, `"relu"`, `"silu"`, etc.

hidden_dropout_prob (`Union[float, int]`, *optional*, defaults to `0.1`) : The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.

attention_probs_dropout_prob (`Union[float, int]`, *optional*, defaults to `0.1`) : The dropout ratio for the attention probabilities.

max_position_embeddings (`int`, *optional*, defaults to `512`) : The maximum sequence length that this model might ever be used with.

initializer_range (`float`, *optional*, defaults to `0.02`) : The standard deviation of the truncated_normal_initializer for initializing all weight matrices.

layer_norm_eps (`float`, *optional*, defaults to `1e-12`) : The epsilon used by the layer normalization layers.

pad_token_id (`int`, *optional*, defaults to `0`) : Token id used for padding in the vocabulary.

cross_attention_frequency (`int`, *optional*, defaults to 2) : The frequency of adding cross-attention to the Transformer layers.

encoder_hidden_size (`int`, *optional*, defaults to `1408`) : Dimension of the hidden representations.

use_qformer_text_input (`bool`, *optional*, defaults to `False`) : Whether to use BERT-style embeddings.

## Blip2Processor[[transformers.Blip2Processor]]

#### transformers.Blip2Processor[[transformers.Blip2Processor]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/processing_blip_2.py#L45)

Constructs a Blip2Processor which wraps a image processor and a tokenizer into a single processor.

[Blip2Processor](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Processor) offers all the functionalities of [BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor) and `GPT2Tokenizer`. See the
[~BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor) and `~GPT2Tokenizer` for more information.

**Parameters:**

image_processor (`BlipImageProcessor`) : The image processor is a required input.

tokenizer (`GPT2Tokenizer`) : The tokenizer is a required input.

num_query_tokens (`int`, *optional*) : Number of tokens used by the Qformer as queries, should be same as in model's config.

## Blip2VisionModel[[transformers.Blip2VisionModel]]

#### transformers.Blip2VisionModel[[transformers.Blip2VisionModel]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L483)

The bare Blip 2 Model outputting raw hidden-states without any specific head on top.

This model inherits from [PreTrainedModel](/docs/transformers/v5.8.0/ja/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.Blip2VisionModel.forwardhttps://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L503[{"name": "pixel_values", "val": ": torch.FloatTensor | None = None"}, {"name": "interpolate_pos_encoding", "val": ": bool = False"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **pixel_values** (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`, *optional*) --
  The tensors corresponding to the input images. Pixel values can be obtained using
  [BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor). See `BlipImageProcessor.__call__()` for details ([Blip2Processor](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Processor) uses
  [BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor) for processing images).
- **interpolate_pos_encoding** (`bool`, *optional*, defaults to `False`) --
  Whether to interpolate the pre-trained position encodings.0[BaseModelOutputWithPooling](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPooling) or `tuple(torch.FloatTensor)`A [BaseModelOutputWithPooling](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPooling) 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 ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) and inputs.
The [Blip2VisionModel](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2VisionModel) 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)`) -- Sequence of hidden-states at the output of the last layer of the model.
- **pooler_output** (`torch.FloatTensor` of shape `(batch_size, hidden_size)`) -- Last layer hidden-state of the first token of the sequence (classification token) after further processing
  through the layers used for the auxiliary pretraining task. E.g. for BERT-family of models, this returns
  the classification token after processing through a linear layer and a tanh activation function. The linear
  layer weights are trained from the next sentence prediction (classification) objective during pretraining.
- **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.

**Parameters:**

config ([Blip2VisionConfig](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2VisionConfig)) : 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/v5.8.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

**Returns:**

`[BaseModelOutputWithPooling](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPooling) or `tuple(torch.FloatTensor)``

A [BaseModelOutputWithPooling](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPooling) 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 ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) and inputs.

## Blip2QFormerModel[[transformers.Blip2QFormerModel]]

#### transformers.Blip2QFormerModel[[transformers.Blip2QFormerModel]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L879)

BLIP-2 Querying Transformer (Q-Former).

This model inherits from [PreTrainedModel](/docs/transformers/v5.8.0/ja/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.Blip2QFormerModel.forwardhttps://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L959[{"name": "query_embeds", "val": ": FloatTensor"}, {"name": "query_length", "val": ": int | None = None"}, {"name": "attention_mask", "val": ": torch.FloatTensor | None = None"}, {"name": "encoder_hidden_states", "val": ": torch.FloatTensor | None = None"}, {"name": "encoder_attention_mask", "val": ": torch.FloatTensor | None = None"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **query_embeds** (`torch.FloatTensor`  of shape `(batch_size, sequence_length, hidden_size)`) --
  Hidden states to be used in the attention computation. If cross-attention,
  will be used for the query (i.e., key and value will use the encoder_hidden_states).
- **query_length** (`int`, *optional*) --
  Length of the query, usually based on the number of query tokens.
  If no value is provided, query_length will be inferred by the query_embeds.
- **attention_mask** (`torch.FloatTensor` 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)
- **encoder_hidden_states** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*) --
  Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention
  if the model is configured as a decoder.
- **encoder_attention_mask** (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*) --
  Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used in
  the cross-attention if the model is configured as a decoder. Mask values selected in `[0, 1]`:

  - 1 for tokens that are **not masked**,
  - 0 for tokens that are **masked**.0[BaseModelOutputWithPoolingAndCrossAttentions](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPoolingAndCrossAttentions) or `tuple(torch.FloatTensor)`A [BaseModelOutputWithPoolingAndCrossAttentions](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPoolingAndCrossAttentions) 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 ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) and inputs.
The [Blip2QFormerModel](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2QFormerModel) 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)`) -- Sequence of hidden-states at the output of the last layer of the model.
- **pooler_output** (`torch.FloatTensor` of shape `(batch_size, hidden_size)`) -- Last layer hidden-state of the first token of the sequence (classification token) after further processing
  through the layers used for the auxiliary pretraining task. E.g. for BERT-family of models, this returns
  the classification token after processing through a linear layer and a tanh activation function. The linear
  layer weights are trained from the next sentence prediction (classification) objective during pretraining.
- **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.
- **cross_attentions** (`tuple(torch.FloatTensor)`, *optional*, returned when `output_attentions=True` and `config.add_cross_attention=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 of the decoder's cross-attention layer, after the attention softmax, used to compute the
  weighted average in the cross-attention heads.
- **past_key_values** (`Cache`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`) -- It is a `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.

**Parameters:**

config ([Blip2QFormerConfig](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2QFormerConfig)) : 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/v5.8.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

**Returns:**

`[BaseModelOutputWithPoolingAndCrossAttentions](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPoolingAndCrossAttentions) or `tuple(torch.FloatTensor)``

A [BaseModelOutputWithPoolingAndCrossAttentions](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPoolingAndCrossAttentions) 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 ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) and inputs.

## Blip2Model[[transformers.Blip2Model]]

#### transformers.Blip2Model[[transformers.Blip2Model]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L1045)

BLIP-2 Model for generating text and image features. The model consists of a vision encoder, Querying Transformer
(Q-Former) and a language model.

This model inherits from [PreTrainedModel](/docs/transformers/v5.8.0/ja/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.Blip2Model.forwardhttps://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L1246[{"name": "pixel_values", "val": ": FloatTensor"}, {"name": "input_ids", "val": ": FloatTensor"}, {"name": "attention_mask", "val": ": torch.LongTensor | None = None"}, {"name": "decoder_input_ids", "val": ": torch.LongTensor | None = None"}, {"name": "decoder_attention_mask", "val": ": torch.LongTensor | None = None"}, {"name": "labels", "val": ": torch.LongTensor | None = None"}, {"name": "interpolate_pos_encoding", "val": ": bool = False"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **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
  [BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor). See `BlipImageProcessor.__call__()` for details ([Blip2Processor](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Processor) uses
  [BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor) for processing images).
- **input_ids** (`torch.FloatTensor` of shape `(batch_size, sequence_length)`) --
  Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.

  Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.8.0/ja/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.8.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and
  [PreTrainedTokenizer.__call__()](/docs/transformers/v5.8.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details.

  [What are input IDs?](../glossary#input-ids)
- **attention_mask** (`torch.LongTensor` 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)
- **decoder_input_ids** (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*) --
  Indices of decoder input sequence tokens in the vocabulary.

  Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.8.0/ja/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.8.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and
  [PreTrainedTokenizer.__call__()](/docs/transformers/v5.8.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details.

  [What are decoder input IDs?](../glossary#decoder-input-ids)
- **decoder_attention_mask** (`torch.BoolTensor` of shape `(batch_size, target_sequence_length)`, *optional*) --
  Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also
  be used by default.

  Only relevant in case an encoder-decoder language model (like T5) is used.
- **labels** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) --
  Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
  config.vocab_size]` or -100 (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]`.
- **interpolate_pos_encoding** (`bool`, *optional*, defaults to `False`) --
  Whether to interpolate the pre-trained position encodings.0`Blip2ForConditionalGenerationModelOutput` or `tuple(torch.FloatTensor)`A `Blip2ForConditionalGenerationModelOutput` 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 ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) and inputs.
The [Blip2Model](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Model) 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`, *optional*, returned when `labels` is provided, `torch.FloatTensor` of shape `(1,)`) -- Language modeling loss from the language model.
- **logits** (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`) -- Prediction scores of the language modeling head of the language model.
- **vision_outputs** (`~modeling_outputs.BaseModelOutputWithPooling`, *optional*, defaults to `None`) -- Outputs of the vision encoder.
- **qformer_outputs** (`~modeling_outputs.BaseModelOutputWithPoolingAndCrossAttentions`, *optional*, defaults to `None`) -- Outputs of the Q-Former (Querying Transformer).
- **language_model_outputs** (`CausalLMOutputWithPast` or `Seq2SeqLMOutput`) -- Outputs of the language model.

Examples:

```python
>>> from PIL import Image
>>> import httpx
>>> from io import BytesIO
>>> from transformers import Blip2Processor, Blip2Model
>>> import torch

>>> device = "cuda" if torch.cuda.is_available() else "cpu"

>>> processor = Blip2Processor.from_pretrained("Salesforce/blip2-opt-2.7b")
>>> model = Blip2Model.from_pretrained("Salesforce/blip2-opt-2.7b", dtype=torch.float16)
>>> model.to(device)
>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> with httpx.stream("GET", url) as response:
...     image = Image.open(BytesIO(response.read()))

>>> prompt = "Question: how many cats are there? Answer:"
>>> inputs = processor(images=image, text=prompt, return_tensors="pt").to(device, torch.float16)

>>> outputs = model(**inputs)
```

**Parameters:**

config ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) : 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/v5.8.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

**Returns:**

``Blip2ForConditionalGenerationModelOutput` or `tuple(torch.FloatTensor)``

A `Blip2ForConditionalGenerationModelOutput` 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 ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) and inputs.
#### get_text_features[[transformers.Blip2Model.get_text_features]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L1088)

- **last_hidden_state** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`) -- Sequence of hidden-states at the output of the last layer of the model.
- **pooler_output** (`torch.FloatTensor` of shape `(batch_size, hidden_size)`) -- Last layer hidden-state of the first token of the sequence (classification token) after further processing
  through the layers used for the auxiliary pretraining task. E.g. for BERT-family of models, this returns
  the classification token after processing through a linear layer and a tanh activation function. The linear
  layer weights are trained from the next sentence prediction (classification) objective during pretraining.
- **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.

Examples:
```python
>>> import torch
>>> from transformers import AutoTokenizer, Blip2Model

>>> model = Blip2Model.from_pretrained("Salesforce/blip2-opt-2.7b")
>>> tokenizer = AutoTokenizer.from_pretrained("Salesforce/blip2-opt-2.7b")

>>> inputs = tokenizer(["a photo of a cat"], padding=True, return_tensors="pt")
>>> with torch.inference_mode():
...     text_features = model.get_text_features(**inputs)
```

**Parameters:**

input_ids (`torch.Tensor` of shape `(batch_size, sequence_length)`) : Indices of input sequence tokens in the vocabulary. Padding will be ignored by default.  Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.8.0/ja/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.8.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.8.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details.  [What are input IDs?](../glossary#input-ids)

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)

decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*) : Indices of decoder input sequence tokens in the vocabulary.  Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.8.0/ja/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.8.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/v5.8.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details.  [What are decoder input IDs?](../glossary#decoder-input-ids)  T5 uses the `pad_token_id` as the starting token for `decoder_input_ids` generation. If `past_key_values` is used, optionally only the last `decoder_input_ids` have to be input (see `past_key_values`).  To know more on how to prepare `decoder_input_ids` for pretraining take a look at [T5 Training](./t5#training).

decoder_attention_mask (`torch.BoolTensor` of shape `(batch_size, target_sequence_length)`, *optional*) : Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also be used by default.

labels (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*) : Labels for computing the masked language modeling loss. Indices should either be in `[0, ..., config.vocab_size]` or -100 (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]`.

**Returns:**

`[BaseModelOutputWithPooling](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPooling) or `tuple(torch.FloatTensor)``

A [BaseModelOutputWithPooling](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPooling) 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 ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) and inputs.
#### get_image_features[[transformers.Blip2Model.get_image_features]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L1150)

- **last_hidden_state** (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`) -- Sequence of hidden-states at the output of the last layer of the model.
- **pooler_output** (`torch.FloatTensor` of shape `(batch_size, hidden_size)`) -- Last layer hidden-state of the first token of the sequence (classification token) after further processing
  through the layers used for the auxiliary pretraining task. E.g. for BERT-family of models, this returns
  the classification token after processing through a linear layer and a tanh activation function. The linear
  layer weights are trained from the next sentence prediction (classification) objective during pretraining.
- **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.

Examples:
```python
>>> import torch
>>> from transformers import AutoProcessor, Blip2Model
>>> from transformers.image_utils import load_image

>>> model = Blip2Model.from_pretrained("Salesforce/blip2-opt-2.7b")

>>> processor = AutoProcessor.from_pretrained("Salesforce/blip2-opt-2.7b")
>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = load_image(url)

>>> inputs = processor(images=image, return_tensors="pt")
>>> with torch.inference_mode():
...     image_outputs = model.get_image_features(**inputs)
```

**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 [BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor). See `BlipImageProcessor.__call__()` for details ([Blip2Processor](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Processor) uses [BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor) for processing images).

interpolate_pos_encoding (`bool`, *optional*, defaults to `False`) : Whether to interpolate the pre-trained position encodings.

**Returns:**

`[BaseModelOutputWithPooling](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPooling) or `tuple(torch.FloatTensor)``

A [BaseModelOutputWithPooling](/docs/transformers/v5.8.0/ja/main_classes/output#transformers.modeling_outputs.BaseModelOutputWithPooling) 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 ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) and inputs.
#### get_qformer_features[[transformers.Blip2Model.get_qformer_features]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L1181)

Examples:

```python
>>> import torch
>>> from transformers import AutoProcessor, Blip2Model
>>> from transformers.image_utils import load_image

>>> processor = Blip2Processor.from_pretrained("Salesforce/blip2-opt-2.7b")
>>> model = Blip2Model.from_pretrained("Salesforce/blip2-opt-2.7b")

>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> image = load_image(url)

>>> inputs = processor(images=image, return_tensors="pt")
>>> with torch.inference_mode():
...     qformer_outputs = model.get_qformer_features(**inputs)
```

**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 [BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor). See `BlipImageProcessor.__call__()` for details ([Blip2Processor](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Processor) uses [BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor) for processing images).

interpolate_pos_encoding (`bool`, *optional*, defaults to `False`) : Whether to interpolate the pre-trained position encodings.

**Returns:**

`qformer_outputs (`torch.FloatTensor`)`

The Q-Former model's last layer hidden states.

## Blip2ForConditionalGeneration[[transformers.Blip2ForConditionalGeneration]]

#### transformers.Blip2ForConditionalGeneration[[transformers.Blip2ForConditionalGeneration]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L1559)

BLIP-2 Model for generating text given an image and an optional text prompt. The model consists of a vision
encoder, Querying Transformer (Q-Former) and a language model.

One can optionally pass `input_ids` to the model, which serve as a text prompt, to make the language model continue
the prompt. Otherwise, the language model starts generating text from the [BOS] (beginning-of-sequence) token.

Note that Flan-T5 checkpoints cannot be cast to float16. They are pre-trained using bfloat16.

This model inherits from [PreTrainedModel](/docs/transformers/v5.8.0/ja/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.Blip2ForConditionalGeneration.forwardhttps://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L1692[{"name": "pixel_values", "val": ": FloatTensor"}, {"name": "input_ids", "val": ": LongTensor"}, {"name": "attention_mask", "val": ": torch.LongTensor | None = None"}, {"name": "decoder_input_ids", "val": ": torch.LongTensor | None = None"}, {"name": "decoder_attention_mask", "val": ": torch.LongTensor | None = None"}, {"name": "inputs_embeds", "val": ": torch.FloatTensor | None = None"}, {"name": "labels", "val": ": torch.LongTensor | None = None"}, {"name": "interpolate_pos_encoding", "val": ": bool = False"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **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
  [BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor). See `BlipImageProcessor.__call__()` for details ([Blip2Processor](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Processor) uses
  [BlipImageProcessor](/docs/transformers/v5.8.0/ja/model_doc/blip#transformers.BlipImageProcessor) for processing images).
- **input_ids** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) --
  Indices of input sequence tokens in the vocabulary of the language model. Input tokens can optionally be
  provided to serve as text prompt, which the language model can continue.

  Indices can be obtained using [Blip2Processor](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Processor). See `Blip2Processor.__call__()` for details.

  [What are input IDs?](../glossary#input-ids)
- **attention_mask** (`torch.LongTensor` 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)
- **decoder_input_ids** (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*) --
  Indices of decoder input sequence tokens in the vocabulary.

  Indices can be obtained using [AutoTokenizer](/docs/transformers/v5.8.0/ja/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/v5.8.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and
  [PreTrainedTokenizer.__call__()](/docs/transformers/v5.8.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.__call__) for details.

  [What are decoder input IDs?](../glossary#decoder-input-ids)
- **decoder_attention_mask** (`torch.BoolTensor` of shape `(batch_size, target_sequence_length)`, *optional*) --
  Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also
  be used by default.

  Only relevant in case an encoder-decoder language model (like T5) is used.
- **inputs_embeds** (`torch.FloatTensor` 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.
- **labels** (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*) --
  Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
  config.vocab_size]` or -100 (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]`.
- **interpolate_pos_encoding** (`bool`, *optional*, defaults to `False`) --
  Whether to interpolate the pre-trained position encodings.0`Blip2ForConditionalGenerationModelOutput` or `tuple(torch.FloatTensor)`A `Blip2ForConditionalGenerationModelOutput` 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 ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) and inputs.
The [Blip2ForConditionalGeneration](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2ForConditionalGeneration) 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`, *optional*, returned when `labels` is provided, `torch.FloatTensor` of shape `(1,)`) -- Language modeling loss from the language model.
- **logits** (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`) -- Prediction scores of the language modeling head of the language model.
- **vision_outputs** (`~modeling_outputs.BaseModelOutputWithPooling`, *optional*, defaults to `None`) -- Outputs of the vision encoder.
- **qformer_outputs** (`~modeling_outputs.BaseModelOutputWithPoolingAndCrossAttentions`, *optional*, defaults to `None`) -- Outputs of the Q-Former (Querying Transformer).
- **language_model_outputs** (`CausalLMOutputWithPast` or `Seq2SeqLMOutput`) -- Outputs of the language model.

Examples:

Prepare processor, model and image input

```python
>>> from PIL import Image
>>> import httpx
>>> from io import BytesIO
>>> from transformers import Blip2Processor, Blip2ForConditionalGeneration
>>> import torch

>>> device = "cuda" if torch.cuda.is_available() else "cpu"

>>> processor = Blip2Processor.from_pretrained("Salesforce/blip2-opt-2.7b")
>>> model = Blip2ForConditionalGeneration.from_pretrained(
...     "Salesforce/blip2-opt-2.7b", load_in_8bit=True, device_map={"": 0}, dtype=torch.float16
... )  # doctest: +IGNORE_RESULT

>>> url = "http://images.cocodataset.org/val2017/000000039769.jpg"
>>> with httpx.stream("GET", url) as response:
...     image = Image.open(BytesIO(response.read()))
```

Image captioning (without providing a text prompt):

```python
>>> inputs = processor(images=image, return_tensors="pt").to(device, torch.float16)

>>> generated_ids = model.generate(**inputs)
>>> generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0].strip()
>>> print(generated_text)
two cats laying on a couch
```

Visual question answering (prompt = question):

```python
>>> prompt = "Question: how many cats are there? Answer:"
>>> inputs = processor(images=image, text=prompt, return_tensors="pt").to(device="cuda", dtype=torch.float16)

>>> generated_ids = model.generate(**inputs)
>>> generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0].strip()
>>> print(generated_text)
two
```

Note that int8 inference is also supported through [bitsandbytes](https://github.com/TimDettmers/bitsandbytes).
This greatly reduces the amount of memory used by the model while maintaining the same performance.

```python
>>> model = Blip2ForConditionalGeneration.from_pretrained(
...     "Salesforce/blip2-opt-2.7b", load_in_8bit=True, device_map={"": 0}, dtype=torch.bfloat16
... )  # doctest: +IGNORE_RESULT

>>> inputs = processor(images=image, text=prompt, return_tensors="pt").to(device="cuda", dtype=torch.bfloat16)

>>> generated_ids = model.generate(**inputs)
>>> generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0].strip()
>>> print(generated_text)
two
```

**Parameters:**

config ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) : 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/v5.8.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

**Returns:**

``Blip2ForConditionalGenerationModelOutput` or `tuple(torch.FloatTensor)``

A `Blip2ForConditionalGenerationModelOutput` 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 ([Blip2Config](/docs/transformers/v5.8.0/ja/model_doc/blip-2#transformers.Blip2Config)) and inputs.
#### generate[[transformers.Blip2ForConditionalGeneration.generate]]

[Source](https://github.com/huggingface/transformers/blob/v5.8.0/src/transformers/models/blip_2/modeling_blip_2.py#L1842)

Overrides `generate` function to be able to use the model as a conditional generator.

**Parameters:**

pixel_values (`torch.FloatTensor` of shape (batch_size, num_channels, height, width)) : Input images to be processed.

input_ids (`torch.LongTensor` of shape (batch_size, sequence_length), *optional*) : The sequence used as a prompt for the generation.

attention_mask (`torch.LongTensor` of shape (batch_size, sequence_length), *optional*) : Mask to avoid performing attention on padding token indices

inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`) : Embedded representation of the inputs. Should be float, not int tokens.

interpolate_pos_encoding (`bool`, *optional*, defaults to `False`) : Whether to interpolate the positional encoding of the image embeddings.

**Returns:**

`captions (list)`

A list of strings of length batch_size * num_captions.