---
library_name: transformers
license: mit
language:
- ja
pipeline_tag: image-text-to-text
---

# [EZO model card]

![image/png](https://cdn-uploads.huggingface.co/production/uploads/657e900beaad53ff67ba84db/FrKFjIqieua3tD32CeECS.png)


## [Model Information]
Based on the InternVL2-26B, this model has improved performance, especially in image recognition and Japanese language performance, by employing multiple tuning methods. It excels in Japanese language tasks, it's designed to meet diverse needs globally.

InternVL2-26Bをベースとして、複数のチューニング手法を採用のうえ、画像認識と日本語性能におけるHeronタスクを中心に、性能を向上させたモデルです。
同様の手法でトレーニングを行うことで、世界中の多様なニーズに応えることができる設計となっています。

### [Benchmark Results]

![image/png](https://cdn-uploads.huggingface.co/production/uploads/657e900beaad53ff67ba84db/aAliaz1qk_9We8xNqPfkr.png)

**Terms of Use**:

This model is based on InternVL2-26B and is Mit.
このモデルはInternVL2-26Bをベースにしており、Mitとしています。

- **Developed by:** Axcxept co., ltd.
- **Language(s) (NLP): English, Japanese

## [Usage]

### INSTALL

`
pip install -U transformers==4.37.2 sentencepiece
`


### Model Loading

#### 16-bit (bf16 / fp16)

```python
import torch
from transformers import AutoTokenizer, AutoModel
path = "HODACHI/EZO-InternVL2-26B"
model = AutoModel.from_pretrained(
    path,
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=True,
    trust_remote_code=True).eval().cuda()
```

#### BNB 8-bit Quantization

```python
import torch
from transformers import AutoTokenizer, AutoModel
path = "HODACHI/EZO-InternVL2-26B"
model = AutoModel.from_pretrained(
    path,
    torch_dtype=torch.bfloat16,
    load_in_8bit=True,
    low_cpu_mem_usage=True,
    trust_remote_code=True).eval()
```

#### BNB 4-bit Quantization

> **⚠️ Warning:** Due to significant quantization errors with BNB 4-bit quantization on InternViT-6B, the model may produce nonsensical outputs and fail to understand images. Therefore, please avoid using BNB 4-bit quantization.

#### Multiple GPUs

The reason for writing the code this way is to avoid errors that occur during multi-GPU inference due to tensors not being on the same device. By ensuring that the first and last layers of the large language model (LLM) are on the same device, we prevent such errors.

このようにコードを書く理由は、テンソルが同じデバイス上にないためにマルチGPU推論中に発生するエラーを避けるためです。
ラージ・ランゲージ・モデル（LLM）の最初のレイヤーと最後のレイヤーが同じデバイス上にあるようにすることで、このようなエラーを防ぐことができます。

```python
import math
import torch
from transformers import AutoTokenizer, AutoModel

def split_model(model_name):
    device_map = {}
    world_size = torch.cuda.device_count()
    num_layers = {
        'InternVL2-1B': 24, 'InternVL2-2B': 24, 'InternVL2-4B': 32, 'InternVL2-8B': 32,
        'InternVL2-26B': 48, 'InternVL2-40B': 60, 'InternVL2-Llama3-76B': 80}[model_name]
    # Since the first GPU will be used for ViT, treat it as half a GPU.
    num_layers_per_gpu = math.ceil(num_layers / (world_size - 0.5))
    num_layers_per_gpu = [num_layers_per_gpu] * world_size
    num_layers_per_gpu[0] = math.ceil(num_layers_per_gpu[0] * 0.5)
    layer_cnt = 0
    for i, num_layer in enumerate(num_layers_per_gpu):
        for j in range(num_layer):
            device_map[f'language_model.model.layers.{layer_cnt}'] = i
            layer_cnt += 1
    device_map['vision_model'] = 0
    device_map['mlp1'] = 0
    device_map['language_model.model.tok_embeddings'] = 0
    device_map['language_model.model.embed_tokens'] = 0
    device_map['language_model.output'] = 0
    device_map['language_model.model.norm'] = 0
    device_map['language_model.lm_head'] = 0
    device_map[f'language_model.model.layers.{num_layers - 1}'] = 0

    return device_map

path = "HODACHI/EZO-InternVL2-26B"
device_map = split_model('InternVL2-26B')
model = AutoModel.from_pretrained(
    path,
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=True,
    trust_remote_code=True,
    device_map=device_map).eval()
```

### Inference with Transformers

```python
import numpy as np
import torch
import torchvision.transforms as T
from decord import VideoReader, cpu
from PIL import Image
from torchvision.transforms.functional import InterpolationMode
from transformers import AutoModel, AutoTokenizer

IMAGENET_MEAN = (0.485, 0.456, 0.406)
IMAGENET_STD = (0.229, 0.224, 0.225)

def build_transform(input_size):
    MEAN, STD = IMAGENET_MEAN, IMAGENET_STD
    transform = T.Compose([
        T.Lambda(lambda img: img.convert('RGB') if img.mode != 'RGB' else img),
        T.Resize((input_size, input_size), interpolation=InterpolationMode.BICUBIC),
        T.ToTensor(),
        T.Normalize(mean=MEAN, std=STD)
    ])
    return transform

def find_closest_aspect_ratio(aspect_ratio, target_ratios, width, height, image_size):
    best_ratio_diff = float('inf')
    best_ratio = (1, 1)
    area = width * height
    for ratio in target_ratios:
        target_aspect_ratio = ratio[0] / ratio[1]
        ratio_diff = abs(aspect_ratio - target_aspect_ratio)
        if ratio_diff < best_ratio_diff:
            best_ratio_diff = ratio_diff
            best_ratio = ratio
        elif ratio_diff == best_ratio_diff:
            if area > 0.5 * image_size * image_size * ratio[0] * ratio[1]:
                best_ratio = ratio
    return best_ratio

def dynamic_preprocess(image, min_num=1, max_num=12, image_size=448, use_thumbnail=False):
    orig_width, orig_height = image.size
    aspect_ratio = orig_width / orig_height

    # calculate the existing image aspect ratio
    target_ratios = set(
        (i, j) for n in range(min_num, max_num + 1) for i in range(1, n + 1) for j in range(1, n + 1) if
        i * j <= max_num and i * j >= min_num)
    target_ratios = sorted(target_ratios, key=lambda x: x[0] * x[1])

    # find the closest aspect ratio to the target
    target_aspect_ratio = find_closest_aspect_ratio(
        aspect_ratio, target_ratios, orig_width, orig_height, image_size)

    # calculate the target width and height
    target_width = image_size * target_aspect_ratio[0]
    target_height = image_size * target_aspect_ratio[1]
    blocks = target_aspect_ratio[0] * target_aspect_ratio[1]

    # resize the image
    resized_img = image.resize((target_width, target_height))
    processed_images = []
    for i in range(blocks):
        box = (
            (i % (target_width // image_size)) * image_size,
            (i // (target_width // image_size)) * image_size,
            ((i % (target_width // image_size)) + 1) * image_size,
            ((i // (target_width // image_size)) + 1) * image_size
        )
        # split the image
        split_img = resized_img.crop(box)
        processed_images.append(split_img)
    assert len(processed_images) == blocks
    if use_thumbnail and len(processed_images) != 1:
        thumbnail_img = image.resize((image_size, image_size))
        processed_images.append(thumbnail_img)
    return processed_images

def load_image(image_file, input_size=448, max_num=12):
    image = Image.open(image_file).convert('RGB')
    transform = build_transform(input_size=input_size)
    images = dynamic_preprocess(image, image_size=input_size, use_thumbnail=True, max_num=max_num)
    pixel_values = [transform(image) for image in images]
    pixel_values = torch.stack(pixel_values)
    return pixel_values

# If you have an 80G A100 GPU, you can put the entire model on a single GPU.
# Otherwise, you need to load a model using multiple GPUs, please refer to the `Multiple GPUs` section.
path = 'HODACHI/EZO-InternVL2-26B'
model = AutoModel.from_pretrained(
    path,
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=True,
    trust_remote_code=True).eval().cuda()
tokenizer = AutoTokenizer.from_pretrained(path, trust_remote_code=True, use_fast=False)

# set the max number of tiles in `max_num`
pixel_values = load_image('./examples/image1.jpg', max_num=12).to(torch.bfloat16).cuda()
generation_config = dict(max_new_tokens=1024, do_sample=False)

# pure-text conversation (テキストのみの対話)
question = 'Hello, who are you?'
response, history = model.chat(tokenizer, None, question, generation_config, history=None, return_history=True)
print(f'User: {question}\nAssistant: {response}')

question = 'Can you tell me a story?'
response, history = model.chat(tokenizer, None, question, generation_config, history=history, return_history=True)
print(f'User: {question}\nAssistant: {response}')

# single-image single-round conversation (単一画像、単一ラウンド対話)
question = '<image>\nPlease describe the image shortly.'
response = model.chat(tokenizer, pixel_values, question, generation_config)
print(f'User: {question}\nAssistant: {response}')

# single-image multi-round conversation (単一画像、多ラウンド対話)
question = '<image>\nPlease describe the image in detail.'
response, history = model.chat(tokenizer, pixel_values, question, generation_config, history=None, return_history=True)
print(f'User: {question}\nAssistant: {response}')

question = 'Please write a poem according to the image.'
response, history = model.chat(tokenizer, pixel_values, question, generation_config, history=history, return_history=True)
print(f'User: {question}\nAssistant: {response}')

# multi-image multi-round conversation, combined images (複数画像、複数ラウンド対話、画像のステッチング)
pixel_values1 = load_image('./examples/image1.jpg', max_num=12).to(torch.bfloat16).cuda()
pixel_values2 = load_image('./examples/image2.jpg', max_num=12).to(torch.bfloat16).cuda()
pixel_values = torch.cat((pixel_values1, pixel_values2), dim=0)

question = '<image>\nDescribe the two images in detail.'
response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                               history=None, return_history=True)
print(f'User: {question}\nAssistant: {response}')

question = 'What are the similarities and differences between these two images.'
response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                               history=history, return_history=True)
print(f'User: {question}\nAssistant: {response}')

# multi-image multi-round conversation, separate images (別々の画像による多画像多ラウンド対話)
pixel_values1 = load_image('./examples/image1.jpg', max_num=12).to(torch.bfloat16).cuda()
pixel_values2 = load_image('./examples/image2.jpg', max_num=12).to(torch.bfloat16).cuda()
pixel_values = torch.cat((pixel_values1, pixel_values2), dim=0)
num_patches_list = [pixel_values1.size(0), pixel_values2.size(0)]

question = 'Image-1: <image>\nImage-2: <image>\nDescribe the two images in detail.'
response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                               num_patches_list=num_patches_list,
                               history=None, return_history=True)
print(f'User: {question}\nAssistant: {response}')

question = 'What are the similarities and differences between these two images.'
response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                               num_patches_list=num_patches_list,
                               history=history, return_history=True)
print(f'User: {question}\nAssistant: {response}')

# batch inference, single image per sample (単一画像バッチ処理)
pixel_values1 = load_image('./examples/image1.jpg', max_num=12).to(torch.bfloat16).cuda()
pixel_values2 = load_image('./examples/image2.jpg', max_num=12).to(torch.bfloat16).cuda()
num_patches_list = [pixel_values1.size(0), pixel_values2.size(0)]
pixel_values = torch.cat((pixel_values1, pixel_values2), dim=0)

questions = ['<image>\nDescribe the image in detail.'] * len(num_patches_list)
responses = model.batch_chat(tokenizer, pixel_values,
                             num_patches_list=num_patches_list,
                             questions=questions,
                             generation_config=generation_config)
for question, response in zip(questions, responses):
    print(f'User: {question}\nAssistant: {response}')

# video multi-round conversation (ビデオ・マルチラウンド・ダイアログ)
def get_index(bound, fps, max_frame, first_idx=0, num_segments=32):
    if bound:
        start, end = bound[0], bound[1]
    else:
        start, end = -100000, 100000
    start_idx = max(first_idx, round(start * fps))
    end_idx = min(round(end * fps), max_frame)
    seg_size = float(end_idx - start_idx) / num_segments
    frame_indices = np.array([
        int(start_idx + (seg_size / 2) + np.round(seg_size * idx))
        for idx in range(num_segments)
    ])
    return frame_indices

def load_video(video_path, bound=None, input_size=448, max_num=1, num_segments=32):
    vr = VideoReader(video_path, ctx=cpu(0), num_threads=1)
    max_frame = len(vr) - 1
    fps = float(vr.get_avg_fps())

    pixel_values_list, num_patches_list = [], []
    transform = build_transform(input_size=input_size)
    frame_indices = get_index(bound, fps, max_frame, first_idx=0, num_segments=num_segments)
    for frame_index in frame_indices:
        img = Image.fromarray(vr[frame_index].asnumpy()).convert('RGB')
        img = dynamic_preprocess(img, image_size=input_size, use_thumbnail=True, max_num=max_num)
        pixel_values = [transform(tile) for tile in img]
        pixel_values = torch.stack(pixel_values)
        num_patches_list.append(pixel_values.shape[0])
        pixel_values_list.append(pixel_values)
    pixel_values = torch.cat(pixel_values_list)
    return pixel_values, num_patches_list

video_path = './examples/red-panda.mp4'
pixel_values, num_patches_list = load_video(video_path, num_segments=8, max_num=1)
pixel_values = pixel_values.to(torch.bfloat16).cuda()
video_prefix = ''.join([f'Frame{i+1}: <image>\n' for i in range(len(num_patches_list))])
question = video_prefix + 'What is the red panda doing?'
# Frame1: <image>\nFrame2: <image>\n...\nFrame8: <image>\n{question}
response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                               num_patches_list=num_patches_list, history=None, return_history=True)
print(f'User: {question}\nAssistant: {response}')

question = 'Describe this video in detail. Don\'t repeat.'
response, history = model.chat(tokenizer, pixel_values, question, generation_config,
                               num_patches_list=num_patches_list, history=history, return_history=True)
print(f'User: {question}\nAssistant: {response}')
```

#### Streaming output

Besides this method, you can also use the following code to get streamed output.

```python
from transformers import TextIteratorStreamer
from threading import Thread

# Initialize the streamer
streamer = TextIteratorStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True, timeout=10)
# Define the generation configuration
generation_config = dict(max_new_tokens=1024, do_sample=False, streamer=streamer)
# Start the model chat in a separate thread
thread = Thread(target=model.chat, kwargs=dict(
    tokenizer=tokenizer, pixel_values=pixel_values, question=question,
    history=None, return_history=False, generation_config=generation_config,
))
thread.start()

# Initialize an empty string to store the generated text
generated_text = ''
# Loop through the streamer to get the new text as it is generated
for new_text in streamer:
    if new_text == model.conv_template.sep:
        break
    generated_text += new_text
    print(new_text, end='', flush=True)  # Print each new chunk of generated text on the same line
```

## Finetune

SWIFT from ModelScope community has supported the fine-tuning (Image/Video) of InternVL, please check [this link](https://github.com/modelscope/swift/blob/main/docs/source_en/Multi-Modal/internvl-best-practice.md) for more details.

## Deployment

### LMDeploy

LMDeploy is a toolkit for compressing, deploying, and serving LLM, developed by the MMRazor and MMDeploy teams.

```sh
pip install lmdeploy
```

LMDeploy abstracts the complex inference process of multi-modal Vision-Language Models (VLM) into an easy-to-use pipeline, similar to the Large Language Model (LLM) inference pipeline.

#### A 'Hello, world' example

```python
from lmdeploy import pipeline, TurbomindEngineConfig, ChatTemplateConfig
from lmdeploy.vl import load_image

model = 'HODACHI/EZO-InternVL2-26B'
system_prompt = 'あなたは優秀な、マルチモーダル大規模言語モデルである。'
image = load_image('https://raw.githubusercontent.com/open-mmlab/mmdeploy/main/tests/data/tiger.jpeg')
chat_template_config = ChatTemplateConfig('internvl-internlm2')
chat_template_config.meta_instruction = system_prompt
pipe = pipeline(model, chat_template_config=chat_template_config,
                backend_config=TurbomindEngineConfig(session_len=8192))
response = pipe(('describe this image', image))
print(response.text)
```

If `ImportError` occurs while executing this case, please install the required dependency packages as prompted.

#### Multi-images inference

When dealing with multiple images, you can put them all in one list. Keep in mind that multiple images will lead to a higher number of input tokens, and as a result, the size of the context window typically needs to be increased.

> Warning: Due to the scarcity of multi-image conversation data, the performance on multi-image tasks may be unstable, and it may require multiple attempts to achieve satisfactory results.

```python
from lmdeploy import pipeline, TurbomindEngineConfig, ChatTemplateConfig
from lmdeploy.vl import load_image
from lmdeploy.vl.constants import IMAGE_TOKEN

model = 'HODACHI/EZO-InternVL2-26B'
system_prompt = 'あなたは優秀な、マルチモーダル大規模言語モデルである。'
chat_template_config = ChatTemplateConfig('internvl-internlm2')
chat_template_config.meta_instruction = system_prompt
pipe = pipeline(model, chat_template_config=chat_template_config,
                backend_config=TurbomindEngineConfig(session_len=8192))

image_urls=[
    'https://raw.githubusercontent.com/open-mmlab/mmdeploy/main/demo/resources/human-pose.jpg',
    'https://raw.githubusercontent.com/open-mmlab/mmdeploy/main/demo/resources/det.jpg'
]

images = [load_image(img_url) for img_url in image_urls]
# Numbering images improves multi-image conversations
response = pipe((f'Image-1: {IMAGE_TOKEN}\nImage-2: {IMAGE_TOKEN}\ndescribe these two images', images))
print(response.text)
```

#### Batch prompts inference

Conducting inference with batch prompts is quite straightforward; just place them within a list structure:

```python
from lmdeploy import pipeline, TurbomindEngineConfig, ChatTemplateConfig
from lmdeploy.vl import load_image

model = 'HODACHI/EZO-InternVL2-26B'
system_prompt = 'あなたは優秀な、マルチモーダル大規模言語モデルである。'
chat_template_config = ChatTemplateConfig('internvl-internlm2')
chat_template_config.meta_instruction = system_prompt
pipe = pipeline(model, chat_template_config=chat_template_config,
                backend_config=TurbomindEngineConfig(session_len=8192))

image_urls=[
    "https://raw.githubusercontent.com/open-mmlab/mmdeploy/main/demo/resources/human-pose.jpg",
    "https://raw.githubusercontent.com/open-mmlab/mmdeploy/main/demo/resources/det.jpg"
]
prompts = [('describe this image', load_image(img_url)) for img_url in image_urls]
response = pipe(prompts)
print(response)
```

#### Multi-turn conversation

There are two ways to do the multi-turn conversations with the pipeline. One is to construct messages according to the format of OpenAI and use above introduced method, the other is to use the `pipeline.chat` interface.

```python
from lmdeploy import pipeline, TurbomindEngineConfig, ChatTemplateConfig, GenerationConfig
from lmdeploy.vl import load_image

model = 'HODACHI/EZO-InternVL2-26B'
system_prompt = 'あなたは優秀な、マルチモーダル大規模言語モデルです'
chat_template_config = ChatTemplateConfig('internvl-internlm2')
chat_template_config.meta_instruction = system_prompt
pipe = pipeline(model, chat_template_config=chat_template_config,
                backend_config=TurbomindEngineConfig(session_len=8192))

image = load_image('https://raw.githubusercontent.com/open-mmlab/mmdeploy/main/demo/resources/human-pose.jpg')
gen_config = GenerationConfig(top_k=40, top_p=0.8, temperature=0.8)
sess = pipe.chat(('describe this image', image), gen_config=gen_config)
print(sess.response.text)
sess = pipe.chat('What is the woman doing?', session=sess, gen_config=gen_config)
print(sess.response.text)
```

### [Model Data]
#### Training Dataset]
We extracted high-quality data from Japanese Wikipedia and FineWeb to create instruction data. Our innovative training approach allows for performance improvements across various languages and domains, making the model suitable for global use despite its focus on Japanese data.

日本語のWikiデータおよび、FineWebから良質なデータ、画像分類を抽出し、Instructionデータを作成しました。
このモデルでは日本語に特化させていますが、世界中のどんなユースケースでも利用可能なアプローチです。

https://huggingface.co/datasets/legacy-datasets/wikipedia
https://huggingface.co/datasets/HuggingFaceFW/fineweb
https://huggingface.co/datasets/STIC-LVLM/stic-coco-preference-6k/blob/main/images.zip


#### Data Preprocessing
We used a plain instruction tuning method to train the model on exemplary responses. This approach enhances the model's ability to understand and generate high-quality responses across various languages and contexts.

プレインストラクトチューニング手法を用いて、模範的回答を学習させました。この手法により、モデルは様々な言語やコンテキストにおいて高品質な応答を理解し生成する能力が向上しています。


#### Implementation Information
[Pre-Instruction Training] 

https://huggingface.co/instruction-pretrain/instruction-synthesizer


### [Disclaimer]
このモデルは研究開発のみを目的として提供されるものであり、実験的なプロトタイプとみなされるべきモデルです。
商業的な使用やミッションクリティカルな環境への配備を意図したものではありません。
本モデルの使用は、使用者の責任において行われるものとし、その性能および結果は保証されません。
Axcxept株式会社は、直接的、間接的、特別、偶発的、結果的な損害、または本モデルの使用から生じるいかなる損失に対しても、得られた結果にかかわらず、一切の責任を負いません。
利用者は、本モデルの使用に伴うリスクを十分に理解し、自己の判断で使用するものとします。

### [謝辞]
本ベースモデルを開発してくださったOpenGVLabならびに当該チームの開発者の方々、また自動評価の手法を提供してくださった多数の方々に感謝と尊敬の意を表します。

### [Hardware]
H100 × 8(Running in 8h)

### [We are.]
[![Axcxept logo](https://cdn-uploads.huggingface.co/production/uploads/657e900beaad53ff67ba84db/8OKW86U986ywttvL2RcbG.png)](https://axcxept.com)