# Qwen2-VL[[Qwen2-VL]]

## Overview[[Overview]]

[Qwen2-VL](https://qwenlm.github.io/blog/qwen2-vl/) 모델은 알리바바 리서치의 Qwen팀에서 개발한 [Qwen-VL](https://huggingface.co/papers/2308.12966) 모델의 주요 업데이트 버전입니다.

블로그의 요약은 다음과 같습니다:

*이 블로그는 지난 몇 년간 Qwen-VL에서 중대한 개선을 거쳐 발전된 Qwen2-VL 모델을 소개합니다. 중요 개선 사항은 향상된 이미지 이해, 고급 비디오 이해, 통합 시각 에이전트 기능, 확장된 다언어 지원을 포함하고 있습니다.모델 아키텍처는 Naive Dynamic Resolution 지원을 통해 임의의 이미지 해상도를 처리할 수 있도록 최적화되었으며, 멀티모달 회전 위치 임베딩(M-ROPE)을 활용하여 1D 텍스트와 다차원 시각 데이터를 효과적으로 처리합니다. 이 업데이트된 모델은 시각 관련 작업에서 GPT-4o와 Claude 3.5 Sonnet 같은 선도적인 AI 시스템과 경쟁력 있는 성능을 보여주며, 텍스트 능력에서는 오픈소스 모델 중 상위권에 랭크되어 있습니다. 이러한 발전은 Qwen2-VL을 강력한 멀티모달 처리 및 추론 능력이 필요한 다양한 응용 분야에서 활용할 수 있는 다재다능한 도구로 만들어줍니다.*

<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/qwen2_vl_architecture.jpeg"
alt="drawing" width="600"/>

 Qwen2-VL 구조. 출처: 블로그 게시글 

이 모델은 [simonJJJ](https://huggingface.co/simonJJJ)에 의해 기여되었습니다.

## 사용 예시[[Usage example]]

### 단일 미디어 추론[[Single Media inference]]

이 모델은 이미지와 비디오를 모두 인풋으로 받을 수 있습니다. 다음은 추론을 위한 예제 코드입니다.

```python

import torch
from transformers import Qwen2VLForConditionalGeneration, AutoTokenizer, AutoProcessor

# 사용 가능한 장치에서 모델을 반 정밀도(half-precision)로 로드
model = Qwen2VLForConditionalGeneration.from_pretrained("Qwen/Qwen2-VL-7B-Instruct", device_map="auto")
processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct")

conversation = [
    {
        "role":"user",
        "content":[
            {
                "type":"image",
                "url": "https://qianwen-res.oss-cn-beijing.aliyuncs.com/Qwen-VL/assets/demo.jpeg"
            },
            {
                "type":"text",
                "text":"Describe this image."
            }
        ]
    }
]

inputs = processor.apply_chat_template(
    conversation,
    add_generation_prompt=True,
    tokenize=True,
    return_dict=True,
    return_tensors="pt"
).to(model.device)

# 추론: 아웃풋 생성
output_ids = model.generate(**inputs, max_new_tokens=128)
generated_ids = [output_ids[len(input_ids):] for input_ids, output_ids in zip(inputs.input_ids, output_ids)]
output_text = processor.batch_decode(generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=True)
print(output_text)

# 비디오
conversation = [
    {
        "role": "user",
        "content": [
            {"type": "video", "path": "/path/to/video.mp4"},
            {"type": "text", "text": "What happened in the video?"},
        ],
    }
]

inputs = processor.apply_chat_template(
    conversation,
    fps=1,
    add_generation_prompt=True,
    tokenize=True,
    return_dict=True,
    return_tensors="pt"
).to(model.device)

# 추론: 아웃풋 생성
output_ids = model.generate(**inputs, max_new_tokens=128)
generated_ids = [output_ids[len(input_ids):] for input_ids, output_ids in zip(inputs.input_ids, output_ids)]
output_text = processor.batch_decode(generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=True)
print(output_text)
```

### 배치 혼합 미디어 추론[[Batch Mixed Media Inference]]

이 모델은 이미지, 비디오, 텍스트 등 다양한 유형의 데이터를 혼합하여 배치 입력으로 처리할 수 있습니다. 다음은 예제입니다.

```python

# 첫번째 이미지에 대한 대화
conversation1 = [
    {
        "role": "user",
        "content": [
            {"type": "image", "path": "/path/to/image1.jpg"},
            {"type": "text", "text": "Describe this image."}
        ]
    }
]

# 두 개의 이미지에 대한 대화
conversation2 = [
    {
        "role": "user",
        "content": [
            {"type": "image", "path": "/path/to/image2.jpg"},
            {"type": "image", "path": "/path/to/image3.jpg"},
            {"type": "text", "text": "What is written in the pictures?"}
        ]
    }
]

# 순수 텍스트로만 이루어진 대화
conversation3 = [
    {
        "role": "user",
        "content": "who are you?"
    }
]

# 혼합된 미디어로 이루어진 대화
conversation4 = [
    {
        "role": "user",
        "content": [
            {"type": "image", "path": "/path/to/image3.jpg"},
            {"type": "image", "path": "/path/to/image4.jpg"},
            {"type": "video", "path": "/path/to/video.jpg"},
            {"type": "text", "text": "What are the common elements in these medias?"},
        ],
    }
]

conversations = [conversation1, conversation2, conversation3, conversation4]
# 배치 추론을 위한 준비
ipnuts = processor.apply_chat_template(
    conversations,
    fps=1,
    add_generation_prompt=True,
    tokenize=True,
    return_dict=True,
    return_tensors="pt"
).to(model.device)

# 배치 추론
output_ids = model.generate(**inputs, max_new_tokens=128)
generated_ids = [output_ids[len(input_ids):] for input_ids, output_ids in zip(inputs.input_ids, output_ids)]
output_text = processor.batch_decode(generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=True)
print(output_text)
```

### 사용 팁[[Usage Tips]]

#### 이미지 해상도 트레이드오프[[Image Resolution trade-off]]

이 모델은 다양한 해상도의 입력을 지원합니다. 디폴트로 입력에 대해 네이티브(native) 해상도를 사용하지만, 더 높은 해상도를 적용하면 성능이 향상될 수 있습니다. 다만, 이는 더 많은 연산 비용을 초래합니다. 사용자는 최적의 설정을 위해 최소 및 최대 픽셀 수를 조정할 수 있습니다.

```python
min_pixels = 224*224
max_pixels = 2048*2048
processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct", min_pixels=min_pixels, max_pixels=max_pixels)
```

제한된 GPU RAM의 경우, 다음과 같이 해상도를 줄일 수 있습니다:

```python
min_pixels = 256*28*28
max_pixels = 1024*28*28 
processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct", min_pixels=min_pixels, max_pixels=max_pixels)
```
이렇게 하면 각 이미지가 256~1024개의 토큰으로 인코딩됩니다. 여기서 28은 모델이 14 크기의 패치(patch)와 2의 시간 패치(temporal patch size)를 사용하기 때문에 나온 값입니다 (14 × 2 = 28).

#### 다중 이미지 인풋[[Multiple Image Inputs]]

기본적으로 이미지와 비디오 콘텐츠는 대화에 직접 포함됩니다. 여러 개의 이미지를 처리할 때는 이미지 및 비디오에 라벨을 추가하면 참조하기가 더 쉬워집니다. 사용자는 다음 설정을 통해 이 동작을 제어할 수 있습니다:

```python
conversation = [
    {
        "role": "user",
        "content": [
            {"type": "image"}, 
            {"type": "text", "text": "Hello, how are you?"}
        ]
    },
    {
        "role": "assistant",
        "content": "I'm doing well, thank you for asking. How can I assist you today?"
    },
    {
        "role": "user",
        "content": [
            {"type": "text", "text": "Can you describe these images and video?"}, 
            {"type": "image"}, 
            {"type": "image"}, 
            {"type": "video"}, 
            {"type": "text", "text": "These are from my vacation."}
        ]
    },
    {
        "role": "assistant",
        "content": "I'd be happy to describe the images and video for you. Could you please provide more context about your vacation?"
    },
    {
        "role": "user",
        "content": "It was a trip to the mountains. Can you see the details in the images and video?"
    }
]

# 디폴트:
prompt_without_id = processor.apply_chat_template(conversation, add_generation_prompt=True)
# 예상 아웃풋: '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n<|im_start|>user\n<|vision_start|><|image_pad|><|vision_end|>Hello, how are you?<|im_end|>\n<|im_start|>assistant\nI'm doing well, thank you for asking. How can I assist you today?<|im_end|>\n<|im_start|>user\nCan you describe these images and video?<|vision_start|><|image_pad|><|vision_end|><|vision_start|><|image_pad|><|vision_end|><|vision_start|><|video_pad|><|vision_end|>These are from my vacation.<|im_end|>\n<|im_start|>assistant\nI'd be happy to describe the images and video for you. Could you please provide more context about your vacation?<|im_end|>\n<|im_start|>user\nIt was a trip to the mountains. Can you see the details in the images and video?<|im_end|>\n<|im_start|>assistant\n'

# id 추가
prompt_with_id = processor.apply_chat_template(conversation, add_generation_prompt=True, add_vision_id=True)
# 예상 아웃풋: '<|im_start|>system\nYou are a helpful assistant.<|im_end|>\n<|im_start|>user\nPicture 1: <|vision_start|><|image_pad|><|vision_end|>Hello, how are you?<|im_end|>\n<|im_start|>assistant\nI'm doing well, thank you for asking. How can I assist you today?<|im_end|>\n<|im_start|>user\nCan you describe these images and video?Picture 2: <|vision_start|><|image_pad|><|vision_end|>Picture 3: <|vision_start|><|image_pad|><|vision_end|>Video 1: <|vision_start|><|video_pad|><|vision_end|>These are from my vacation.<|im_end|>\n<|im_start|>assistant\nI'd be happy to describe the images and video for you. Could you please provide more context about your vacation?<|im_end|>\n<|im_start|>user\nIt was a trip to the mountains. Can you see the details in the images and video?<|im_end|>\n<|im_start|>assistant\n'

```

#### 빠른 생성을 위한 Flash-Attention 2[[Flash-Attention 2 to speed up generation]]

첫번째로, Flash Attention 2의 최신 버전을 설치합니다:

```bash
pip install -U flash-attn --no-build-isolation
```

또한, Flash-Attention 2를 지원하는 하드웨어가 필요합니다. 자세한 내용은 공식 문서인 [flash attention repository](https://github.com/Dao-AILab/flash-attention)에서 확인할 수 있습니다. FlashAttention-2는 모델이 `torch.float16` 또는 `torch.bfloat16` 형식으로 로드된 경우에만 사용할 수 있습니다.

Flash Attention-2를 사용하여 모델을 로드하고 실행하려면, 다음과 같이 모델을 로드할 때 `attn_implementation="flash_attention_2"` 옵션을 추가하면 됩니다:

```python
from transformers import Qwen2VLForConditionalGeneration

model = Qwen2VLForConditionalGeneration.from_pretrained(
    "Qwen/Qwen2-VL-7B-Instruct", 
    dtype=torch.bfloat16, 
    attn_implementation="flash_attention_2",
)
```

## Qwen2VLConfig[[transformers.Qwen2VLConfig]]

#### transformers.Qwen2VLConfig[[transformers.Qwen2VLConfig]]

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen2_vl/configuration_qwen2_vl.py#L136)

This is the configuration class to store the configuration of a Qwen2VLModel. It is used to instantiate a Qwen2 Vl
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 [Qwen/Qwen2-VL-7B-Instruct](https://huggingface.co/Qwen/Qwen2-VL-7B-Instruct)

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

Example:

```python
>>> from transformers import Qwen2VLForConditionalGeneration, Qwen2VLConfig

>>> # Initializing a Qwen2VL style configuration
>>> configuration = Qwen2VLConfig()

>>> # Initializing a model from the Qwen2-VL-7B style configuration
>>> model = Qwen2VLForConditionalGeneration(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.

image_token_id (`int`, *optional*, defaults to `151655`) : The image token index used as a placeholder for input images.

video_token_id (`int`, *optional*, defaults to `151656`) : The video token index used as a placeholder for input videos.

vision_start_token_id (`int`, *optional*, defaults to `151652`) : Token ID that marks the start of a visual segment in the multimodal input sequence.

vision_end_token_id (`int`, *optional*, defaults to `151653`) : Token ID that marks the end of a visual segment in the multimodal input sequence.

tie_word_embeddings (`bool`, *optional*, defaults to `False`) : Whether to tie weight embeddings according to model's `tied_weights_keys` mapping.

## Qwen2VLImageProcessor[[transformers.Qwen2VLImageProcessor]]

#### transformers.Qwen2VLImageProcessor[[transformers.Qwen2VLImageProcessor]]

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen2_vl/image_processing_qwen2_vl.py#L92)

Constructs a Qwen2VLImageProcessor image processor.

preprocesstransformers.Qwen2VLImageProcessor.preprocesshttps://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen2_vl/image_processing_qwen2_vl.py#L140[{"name": "images", "val": ": typing.Union[ForwardRef('PIL.Image.Image'), numpy.ndarray, ForwardRef('torch.Tensor'), list['PIL.Image.Image'], list[numpy.ndarray], list['torch.Tensor']]"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.models.qwen2_vl.image_processing_qwen2_vl.Qwen2VLImageProcessorKwargs]"}]- **images** (`Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, list[PIL.Image.Image], list[numpy.ndarray], list[torch.Tensor]]`) --
  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`.
- **min_pixels** (`int`, *kwargs*, *optional*, defaults to `56 * 56`) --
  The min pixels of the image to resize the image.
- **max_pixels** (`int`, *kwargs*, *optional*, defaults to `28 * 28 * 1280`) --
  The max pixels of the image to resize the image.
- **patch_size** (`int`, *kwargs*, *optional*, defaults to 14) --
  The spatial patch size of the vision encoder.
- **temporal_patch_size** (`int`, *kwargs*, *optional*, defaults to 2) --
  The temporal patch size of the vision encoder.
- **merge_size** (`int`, *kwargs*, *optional*, defaults to 2) --
  The merge size of the vision encoder to llm encoder.
- **return_tensors** (`str` or [TensorType](/docs/transformers/main/ko/internal/file_utils#transformers.TensorType), *optional*) --
  Returns stacked tensors if set to `'pt'`, otherwise returns a list of tensors.
- ****kwargs** (`ImagesKwargs`, *optional*) --
  Additional image preprocessing options. Model-specific kwargs 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:**

min_pixels (`int`, *kwargs*, *optional*, defaults to `56 * 56`) : The min pixels of the image to resize the image.

max_pixels (`int`, *kwargs*, *optional*, defaults to `28 * 28 * 1280`) : The max pixels of the image to resize the image.

patch_size (`int`, *kwargs*, *optional*, defaults to 14) : The spatial patch size of the vision encoder.

temporal_patch_size (`int`, *kwargs*, *optional*, defaults to 2) : The temporal patch size of the vision encoder.

merge_size (`int`, *kwargs*, *optional*, defaults to 2) : The merge size of the vision encoder to llm encoder.

- ****kwargs** (`ImagesKwargs`, *optional*) : Additional image preprocessing options. Model-specific kwargs are listed above; see the TypedDict class for the complete list of supported arguments.

**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.

## Qwen2VLImageProcessorFast[[transformers.Qwen2VLImageProcessor]]

#### transformers.Qwen2VLImageProcessor[[transformers.Qwen2VLImageProcessor]]

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen2_vl/image_processing_qwen2_vl.py#L92)

Constructs a Qwen2VLImageProcessor image processor.

preprocesstransformers.Qwen2VLImageProcessor.preprocesshttps://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen2_vl/image_processing_qwen2_vl.py#L140[{"name": "images", "val": ": typing.Union[ForwardRef('PIL.Image.Image'), numpy.ndarray, ForwardRef('torch.Tensor'), list['PIL.Image.Image'], list[numpy.ndarray], list['torch.Tensor']]"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.models.qwen2_vl.image_processing_qwen2_vl.Qwen2VLImageProcessorKwargs]"}]- **images** (`Union[PIL.Image.Image, numpy.ndarray, torch.Tensor, list[PIL.Image.Image], list[numpy.ndarray], list[torch.Tensor]]`) --
  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`.
- **min_pixels** (`int`, *kwargs*, *optional*, defaults to `56 * 56`) --
  The min pixels of the image to resize the image.
- **max_pixels** (`int`, *kwargs*, *optional*, defaults to `28 * 28 * 1280`) --
  The max pixels of the image to resize the image.
- **patch_size** (`int`, *kwargs*, *optional*, defaults to 14) --
  The spatial patch size of the vision encoder.
- **temporal_patch_size** (`int`, *kwargs*, *optional*, defaults to 2) --
  The temporal patch size of the vision encoder.
- **merge_size** (`int`, *kwargs*, *optional*, defaults to 2) --
  The merge size of the vision encoder to llm encoder.
- **return_tensors** (`str` or [TensorType](/docs/transformers/main/ko/internal/file_utils#transformers.TensorType), *optional*) --
  Returns stacked tensors if set to `'pt'`, otherwise returns a list of tensors.
- ****kwargs** (`ImagesKwargs`, *optional*) --
  Additional image preprocessing options. Model-specific kwargs 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:**

min_pixels (`int`, *kwargs*, *optional*, defaults to `56 * 56`) : The min pixels of the image to resize the image.

max_pixels (`int`, *kwargs*, *optional*, defaults to `28 * 28 * 1280`) : The max pixels of the image to resize the image.

patch_size (`int`, *kwargs*, *optional*, defaults to 14) : The spatial patch size of the vision encoder.

temporal_patch_size (`int`, *kwargs*, *optional*, defaults to 2) : The temporal patch size of the vision encoder.

merge_size (`int`, *kwargs*, *optional*, defaults to 2) : The merge size of the vision encoder to llm encoder.

- ****kwargs** (`ImagesKwargs`, *optional*) : Additional image preprocessing options. Model-specific kwargs are listed above; see the TypedDict class for the complete list of supported arguments.

**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.

## Qwen2VLProcessor[[transformers.Qwen2VLProcessor]]

#### transformers.Qwen2VLProcessor[[transformers.Qwen2VLProcessor]]

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen2_vl/processing_qwen2_vl.py#L44)

Constructs a Qwen2VLProcessor which wraps a image processor, a tokenizer, and a video processor into a single processor.

[Qwen2VLProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLProcessor) offers all the functionalities of [Qwen2VLImageProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLImageProcessor), `Qwen2Tokenizer`, and `Qwen2VLVideoProcessor`. See the
[~Qwen2VLImageProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLImageProcessor), `~Qwen2Tokenizer`, and `~Qwen2VLVideoProcessor` for more information.

post_process_image_text_to_texttransformers.Qwen2VLProcessor.post_process_image_text_to_texthttps://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen2_vl/processing_qwen2_vl.py#L169[{"name": "generated_outputs", "val": ""}, {"name": "skip_special_tokens", "val": " = True"}, {"name": "clean_up_tokenization_spaces", "val": " = False"}, {"name": "**kwargs", "val": ""}]- **generated_outputs** (`torch.Tensor` or `np.ndarray`) --
  The output of the model `generate` function. The output is expected to be a tensor of shape `(batch_size, sequence_length)`
  or `(sequence_length,)`.
- **skip_special_tokens** (`bool`, *optional*, defaults to `True`) --
  Whether or not to remove special tokens in the output. Argument passed to the tokenizer's `batch_decode` method.
- **clean_up_tokenization_spaces** (`bool`, *optional*, defaults to `False`) --
  Whether or not to clean up the tokenization spaces. Argument passed to the tokenizer's `batch_decode` method.
- ****kwargs** --
  Additional arguments to be passed to the tokenizer's `batch_decode method`.0`list[str]`The decoded text.

Post-process the output of the model to decode the text.

**Parameters:**

image_processor (`Qwen2VLImageProcessor`) : The image processor is a required input.

tokenizer (`Qwen2Tokenizer`) : The tokenizer is a required input.

video_processor (`Qwen2VLVideoProcessor`) : The video processor is a required input.

chat_template (`str`) : A Jinja template to convert lists of messages in a chat into a tokenizable string.

**Returns:**

``list[str]``

The decoded text.

## Qwen2VLModel[[transformers.Qwen2VLModel]]

#### transformers.Qwen2VLModel[[transformers.Qwen2VLModel]]

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py#L885)

The bare Qwen2 Vl Model outputting raw hidden-states without any specific head on top.

This model inherits from [PreTrainedModel](/docs/transformers/main/ko/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.Qwen2VLModel.forwardhttps://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py#L1203[{"name": "input_ids", "val": ": torch.LongTensor | None = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.LongTensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "inputs_embeds", "val": ": torch.FloatTensor | None = None"}, {"name": "use_cache", "val": ": bool | None = None"}, {"name": "pixel_values", "val": ": torch.Tensor | None = None"}, {"name": "pixel_values_videos", "val": ": torch.FloatTensor | None = None"}, {"name": "image_grid_thw", "val": ": torch.LongTensor | None = None"}, {"name": "video_grid_thw", "val": ": torch.LongTensor | None = None"}, {"name": "rope_deltas", "val": ": torch.LongTensor | None = None"}, {"name": "mm_token_type_ids", "val": ": torch.IntTensor | None = None"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **input_ids** (`torch.LongTensor` 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/main/ko/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/main/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and
  [PreTrainedTokenizer.__call__()](/docs/transformers/main/ko/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)
- **position_ids** (`torch.LongTensor` 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)
- **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/main/ko/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/main/ko/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)`.
- **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.
- **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`).
- **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
  [Qwen2VLImageProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLImageProcessor). See `Qwen2VLImageProcessor.__call__()` for details ([Qwen2VLProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLProcessor) uses
  [Qwen2VLImageProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLImageProcessor) for processing images).
- **pixel_values_videos** (`torch.FloatTensor` of shape `(batch_size, num_frames, num_channels, frame_size, frame_size)`, *optional*) --
  The tensors corresponding to the input video. Pixel values for videos can be obtained using
  `Qwen2VLVideoProcessor`. See `Qwen2VLVideoProcessor.__call__()` for details ([Qwen2VLProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLProcessor) uses
  `Qwen2VLVideoProcessor` for processing videos).
- **image_grid_thw** (`torch.LongTensor` of shape `(num_images, 3)`, *optional*) --
  The temporal, height and width of feature shape of each image in LLM.
- **video_grid_thw** (`torch.LongTensor` of shape `(num_videos, 3)`, *optional*) --
  The temporal, height and width of feature shape of each video in LLM.
- **rope_deltas** (`torch.LongTensor` of shape `(batch_size, )`, *optional*) --
  The rope index difference between sequence length and multimodal rope.
- **mm_token_type_ids** (`torch.IntTensor` of shape `(batch_size, sequence_length)`, *optional*) --
  Indices of input sequence tokens matching each modality. For example text (0), image (1), video (2).
  Multimodal token type ids can be obtained using [AutoProcessor](/docs/transformers/main/ko/model_doc/auto#transformers.AutoProcessor). See [ProcessorMixin.__call__()](/docs/transformers/main/ko/model_doc/tvp#transformers.TvpProcessor.__call__) for details.0`Qwen2VLModelOutputWithPast` or `tuple(torch.FloatTensor)`A `Qwen2VLModelOutputWithPast` 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 ([Qwen2VLConfig](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLConfig)) and inputs.
The [Qwen2VLModel](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLModel) 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/main/ko/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) 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.
- **rope_deltas** (`torch.LongTensor` of shape `(batch_size, )`, *optional*) -- The rope index difference between sequence length and multimodal rope.

**Parameters:**

config ([Qwen2VLConfig](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLConfig)) : 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/main/ko/main_classes/model#transformers.PreTrainedModel.from_pretrained) method to load the model weights.

**Returns:**

``Qwen2VLModelOutputWithPast` or `tuple(torch.FloatTensor)``

A `Qwen2VLModelOutputWithPast` 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 ([Qwen2VLConfig](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLConfig)) and inputs.

## Qwen2VLForConditionalGeneration[[transformers.Qwen2VLForConditionalGeneration]]

#### transformers.Qwen2VLForConditionalGeneration[[transformers.Qwen2VLForConditionalGeneration]]

[Source](https://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py#L1278)

forwardtransformers.Qwen2VLForConditionalGeneration.forwardhttps://github.com/huggingface/transformers/blob/main/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py#L1324[{"name": "input_ids", "val": ": torch.LongTensor | None = None"}, {"name": "attention_mask", "val": ": torch.Tensor | None = None"}, {"name": "position_ids", "val": ": torch.LongTensor | None = None"}, {"name": "past_key_values", "val": ": transformers.cache_utils.Cache | None = None"}, {"name": "inputs_embeds", "val": ": torch.FloatTensor | None = None"}, {"name": "labels", "val": ": torch.LongTensor | None = None"}, {"name": "use_cache", "val": ": bool | None = None"}, {"name": "pixel_values", "val": ": torch.Tensor | None = None"}, {"name": "pixel_values_videos", "val": ": torch.FloatTensor | None = None"}, {"name": "image_grid_thw", "val": ": torch.LongTensor | None = None"}, {"name": "video_grid_thw", "val": ": torch.LongTensor | None = None"}, {"name": "rope_deltas", "val": ": torch.LongTensor | None = None"}, {"name": "mm_token_type_ids", "val": ": torch.IntTensor | None = None"}, {"name": "logits_to_keep", "val": ": int | torch.Tensor = 0"}, {"name": "**kwargs", "val": ": typing_extensions.Unpack[transformers.utils.generic.TransformersKwargs]"}]- **input_ids** (`torch.LongTensor` 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/main/ko/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/main/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and
  [PreTrainedTokenizer.__call__()](/docs/transformers/main/ko/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)
- **position_ids** (`torch.LongTensor` 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)
- **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/main/ko/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/main/ko/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)`.
- **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]`.
- **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`).
- **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
  [Qwen2VLImageProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLImageProcessor). See `Qwen2VLImageProcessor.__call__()` for details ([Qwen2VLProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLProcessor) uses
  [Qwen2VLImageProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLImageProcessor) for processing images).
- **pixel_values_videos** (`torch.FloatTensor` of shape `(batch_size, num_frames, num_channels, frame_size, frame_size)`, *optional*) --
  The tensors corresponding to the input video. Pixel values for videos can be obtained using
  `Qwen2VLVideoProcessor`. See `Qwen2VLVideoProcessor.__call__()` for details ([Qwen2VLProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLProcessor) uses
  `Qwen2VLVideoProcessor` for processing videos).
- **image_grid_thw** (`torch.LongTensor` of shape `(num_images, 3)`, *optional*) --
  The temporal, height and width of feature shape of each image in LLM.
- **video_grid_thw** (`torch.LongTensor` of shape `(num_videos, 3)`, *optional*) --
  The temporal, height and width of feature shape of each video in LLM.
- **rope_deltas** (`torch.LongTensor` of shape `(batch_size, )`, *optional*) --
  The rope index difference between sequence length and multimodal rope.
- **mm_token_type_ids** (`torch.IntTensor` of shape `(batch_size, sequence_length)`, *optional*) --
  Indices of input sequence tokens matching each modality. For example text (0), image (1), video (2).
  Multimodal token type ids can be obtained using [AutoProcessor](/docs/transformers/main/ko/model_doc/auto#transformers.AutoProcessor). See [ProcessorMixin.__call__()](/docs/transformers/main/ko/model_doc/tvp#transformers.TvpProcessor.__call__) for details.

- **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`Qwen2VLCausalLMOutputWithPast` or `tuple(torch.FloatTensor)`A `Qwen2VLCausalLMOutputWithPast` 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 ([Qwen2VLConfig](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLConfig)) and inputs.
The [Qwen2VLForConditionalGeneration](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLForConditionalGeneration) 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/main/ko/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) 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.
- **rope_deltas** (`torch.LongTensor` of shape `(batch_size, )`, *optional*) -- The rope index difference between sequence length and multimodal rope.

Example:

```python
>>> from transformers import AutoProcessor, Qwen2VLForConditionalGeneration

>>> model = Qwen2VLForConditionalGeneration.from_pretrained("Qwen/Qwen2-VL-7B-Instruct")
>>> processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-7B-Instruct")

>>> messages = [
    {
        "role": "user",
        "content": [
            {
                "type": "image",
                "image": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg",
            },
            {"type": "text", "text": "Describe the image."},
        ],
    }
]

>>> inputs = processor.apply_chat_template(
    messages,
    tokenize=True,
    add_generation_prompt=True,
    return_dict=True,
    return_tensors="pt"
)

>>> # Generate
>>> generated_ids = model.generate(**inputs, max_new_tokens=1024)
>>> generated_ids_trimmed = [out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)]
>>> output_text = processor.batch_decode(generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
>>> print(output_text)
```

**Parameters:**

input_ids (`torch.LongTensor` 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/main/ko/model_doc/auto#transformers.AutoTokenizer). See [PreTrainedTokenizer.encode()](/docs/transformers/main/ko/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.encode) and [PreTrainedTokenizer.__call__()](/docs/transformers/main/ko/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)

position_ids (`torch.LongTensor` 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)

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/main/ko/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/main/ko/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)`.

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]`.

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`).

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 [Qwen2VLImageProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLImageProcessor). See `Qwen2VLImageProcessor.__call__()` for details ([Qwen2VLProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLProcessor) uses [Qwen2VLImageProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLImageProcessor) for processing images).

pixel_values_videos (`torch.FloatTensor` of shape `(batch_size, num_frames, num_channels, frame_size, frame_size)`, *optional*) : The tensors corresponding to the input video. Pixel values for videos can be obtained using `Qwen2VLVideoProcessor`. See `Qwen2VLVideoProcessor.__call__()` for details ([Qwen2VLProcessor](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLProcessor) uses `Qwen2VLVideoProcessor` for processing videos).

image_grid_thw (`torch.LongTensor` of shape `(num_images, 3)`, *optional*) : The temporal, height and width of feature shape of each image in LLM.

video_grid_thw (`torch.LongTensor` of shape `(num_videos, 3)`, *optional*) : The temporal, height and width of feature shape of each video in LLM.

rope_deltas (`torch.LongTensor` of shape `(batch_size, )`, *optional*) : The rope index difference between sequence length and multimodal rope.

mm_token_type_ids (`torch.IntTensor` of shape `(batch_size, sequence_length)`, *optional*) : Indices of input sequence tokens matching each modality. For example text (0), image (1), video (2). Multimodal token type ids can be obtained using [AutoProcessor](/docs/transformers/main/ko/model_doc/auto#transformers.AutoProcessor). See [ProcessorMixin.__call__()](/docs/transformers/main/ko/model_doc/tvp#transformers.TvpProcessor.__call__) for details. 

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).

**Returns:**

``Qwen2VLCausalLMOutputWithPast` or `tuple(torch.FloatTensor)``

A `Qwen2VLCausalLMOutputWithPast` 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 ([Qwen2VLConfig](/docs/transformers/main/ko/model_doc/qwen2_vl#transformers.Qwen2VLConfig)) and inputs.