# Sarashina2.2 Vision 3b

URL: https://interfaze.ai/models/sbintuitionssarashina22-vision-3b

[All models](https://interfaze.ai/models)

Sarashina2.2 Vision 3b by sbintuitions, a image-to-text model with multimodal capabilities. Understand and compare multimodal features, benchmarks, and capabilities.

## Comparison

| Feature | Sarashina2.2 Vision 3b | Interfaze |
| --- | --- | --- |
| Input Modalities | image, text | image, text, audio, video, document |
| Native OCR | No | Yes |
| Long Document Processing | No | Yes |
| Language Support | unknown | 162+ |
| Native Speech-to-Text | No | Yes |
| Native Object Detection | No | Yes |
| Guardrail Controls | No | Yes |
| Context Input Size | unknown | 1M |
| Tool Calling | No | Tool calling supported + built in browser, code execution and web search |

### Scaling

| Feature | Sarashina2.2 Vision 3b | Interfaze |
| --- | --- | --- |
| Scaling | Self-hosted/Provider-hosted with quantization | Unlimited |

[Try Interfaze](https://interfaze.ai/dashboard)[Read the Docs](https://interfaze.ai/docs)

View model card on [Hugging Face](https://huggingface.co/sbintuitions/sarashina2.2-vision-3b)

**Sarashina2.2-Vision-3B** is a Japanese Large Vision Language Model trained by [SB Intuitions](https://www.sbintuitions.co.jp).

This model is based on [Sarashina2.2-3B-Instruct](https://huggingface.co/sbintuitions/sarashina2.2-3b-instruct-v0.1) and Image Encoder of [SigLIP](https://huggingface.co/google/siglip-so400m-patch14-384).

## Model Performance

### Japanese Performance

| Model | Params(B) | BussinessSlide VQA\*1 | Heron-Bench\*1 | JDocQA\*1 | JMMMU |
| --- | --- | --- | --- | --- | --- |
| Sarashina2.2-Vision-3B | 3.8 | 3.932 | 3.214 | 3.327 | 0.486 |
| Qwen2.5-VL-3B-Instruct | 3.8 | 3.516 | 2.000 | 3.019 | 0.450 |
| Qwen3-VL-4B-Instruct | 4.4 | 4.105 | 2.330 | 3.596 | 0.493 |
| InternVL3\_5-4B | 4.7 | 3.311 | 1.893 | 2.626 | 0.437 |
| Sarashina2-Vision-14B | 14.4 | 3.110 | 2.184 | \-\*2 | 0.432 |
| Stockmark-2-VL-100B-beta | 96.5 | 3.973 | 2.563 | 3.168 | \-\*2 |

\*1. [gpt-oss-120b](https://huggingface.co/openai/gpt-oss-120b) was used for LLM-as-a-Judge.

\*2. These scores cannot be measured because some input data exceeds the model's `max_position_embeddings`.

### English Performance

| Model | Params(B) | DocVQA | InfoVQA | RealworldQA |
| --- | --- | --- | --- | --- |
| Sarashina2.2-Vision-3B | 3.8 | 0.831 | 0.567 | 0.625 |
| Qwen2.5-VL-3B-Instruct | 3.8 | 0.924 | 0.750 | 0.586 |
| Qwen3-VL-4B-Instruct | 4.4 | 0.948 | 0.798 | 0.712 |
| InternVL3\_5-4B | 4.7 | 0.823 | 0.541 | 0.553 |
| Sarashina2-Vision-14B | 14.4 | 0.729 | 0.490 | 0.519 |

## How to use

### 1\. Install dependencies

```
pip install transformers==4.57.1 torch torchvision pillow protobuf sentencepiece accelerate
```

### 2\. Inference

The following script loads the model and allows inference.

```
import requests
from PIL import Image
from transformers import AutoModelForCausalLM, AutoProcessor, set_seed


model_path = "sbintuitions/sarashina2.2-vision-3b"


processor = AutoProcessor.from_pretrained(model_path, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_path,
    device_map="cuda",
    torch_dtype="auto",
    trust_remote_code=True,
)
set_seed(42)

image_url = "https://huggingface.co/sbintuitions/sarashina2.2-vision-3b/resolve/main/sample.jpg"
message = [
    {
        "role": "user",
        "content": [
            {
                "type": "image",
                "image": image_url,
            },
            {
                "type": "text",
                "text": "これはどこで撮った写真ですか？",
            },
        ],
    }
]
text_prompt = processor.apply_chat_template(message, add_generation_prompt=True)
"""text_prompt: <|user|><|prefix|><|file|><|suffix|>これはどこで撮った写真ですか？</s><|assistant|>"""

image = Image.open(requests.get(image_url, stream=True).raw).convert("RGB")
inputs = processor(
    text=[text_prompt],
    images=[image],
    padding=True,
    return_tensors="pt",
)
inputs = inputs.to(model.device)


output_ids = model.generate(
    **inputs,
    max_new_tokens=512,
    temperature=0.7,
    top_p=0.95,
    repetition_penalty=1.2,
)
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[0])
"""
この写真は、**道後温泉本館（どうごおんせんほんかん）** の入り口を夜景で撮影した写真です。

---
 場所の詳細：
- **名称**：道後温泉本館（Dogo Onsen Honkan）
- **所在地**：〒790-0842 愛媛県松山市道後湯之町1丁目3番5号
- **アクセス**：JR松山駅から市内電車「道後温泉駅」下車すぐ
- **特徴**：日本最古の温泉の一つとして知られる「道後温泉」の中心的な施設。国の重要文化財にも指定されています。

---
 写真の特徴から判断した理由：
- 建物の屋根や装飾が伝統的な和風建築で、「道後温泉」の看板が目立つ。
- 入口の垂れ幕には「道後」「道後」と書かれており、白い鳳凰の模様が描かれている → 道後温泉の象徴的デザイン。
- 夜の照明と石灯籠、提灯風の灯りが日本の温泉地らしい雰囲気を醸し出している。
- 看板に「道後温泉」の文字が明確に表示されている。

---
 補足情報：
道後温泉本館は、夏目漱石の小説『坊っちゃん』の舞台としても有名で、多くの観光客が訪れる人気スポットです。また、2020年にリニューアルされ、現代的な設備も導入されていますが、外観は伝統を残しています。

---
よって、この写真は **愛媛県松山市にある「道後温泉本館」の夜景** です。
"""
```

## Training

**Sarashina2.2-Vision-3B** is created through the following five-stage training process:

### PreTrain

1.  Projector Warmup: To bridge the gap between the text and image embedding spaces within the LLM
2.  Vision Encoder Pretraining: To enhance image comprehension, especially for understanding Japan-specific images and text
3.  Full Model Pretraining: To enhance the model's unified understanding of images and language using interleaved data

### PostTrain

1.  Supervised Fine-Tuning(SFT): To improve the model's ability to follow instructions and respond appropriately to user prompts
2.  Mixed Preference Optimization(MPO): To align the model's outputs with user preferences, ensuring it generates more desirable responses

## Limitations

This model has limited safety training. Therefore, it might generate some meaningless sequences, some inaccurate instances, or biased/objectionable outputs. Before using it, we would like developers to tune models based on human preferences and safety considerations.

## LICENSE

[MIT License](https://huggingface.co/sbintuitions/sarashina2.2-vision-3b/blob/main/LICENSE)

## Want more deterministic results?

[Try Interfaze](https://interfaze.ai/dashboard)[Read the Docs](https://interfaze.ai/docs)