README.md

---
license: apache-2.0
language:
- en
library_name: transformers
pipeline_tag: image-text-to-text
tags:
- multimodal
- aria
base_model:
- rhymes-ai/Aria-Base-64K
---
<!-- <p align="center">
  <br>Aria</br>
</p>  -->


# Aria-Chat Model Card

<!-- 
- Aria is the **first open multimodal native MoE** model, capable of seamlessly handling various input modalities within a MoE architecture.
- Aria performs **on par with GPT-4o mini and Gemini 1.5 Flash** across a range of multimodal tasks while maintaining strong performance on **text**-only tasks.
- Compared to similar or even larger models, Aria boasts **faster speeds** and **lower costs**. This high efficiency stems from its ability to activate only 3.9B parameters during inference – the **fewest** among models with comparable performance.
 -->
 
## Key features

- **Especially Optimized For Multimodal Chat**: Aria-Chat is especially optimized for open-ended and multi-round dialogs. We hope this version can provide seamless open-source multimodal chat experience.
- **Improved Reliability**: We have improved its reliability for generating long outputs, reducing probabilities for previously-reported bad cases like incomplete responses on Markdown tables, or endless responses on listwise outputs. 
- **Better Multi-Lingual Abilities**: We have optimized its ability on non-English scenarios (Chinese, Spanish, French, Japanese, *etc*), including both multi-lingual OCR and multi-lingual dialogs.

<p align="center">
🔗 <a href="https://rhymes.ai/" target="_blank"> Try Aria!</a> · 📖 <a href="https://www.rhymes.ai/blog-details/aria-first-open-multimodal-native-moe-model" target="_blank">Blog</a> · 📌 <a href="https://arxiv.org/pdf/2410.05993" target="_blank">Paper</a> 
 · ⭐ <a href="https://github.com/rhymes-ai/Aria" target="_blank">GitHub</a> · 🟣 <a href="https://discord.com/invite/u8HxU23myj" target="_blank"> Discord </a>
</p> 


<!-- # Model Info

| Model  | Download  | Parameter | Context Length |
| :---- | :------- | :------------ | :------ |
| Aria | < HF link - TBD> | • Activation: 3.9B (3.5B MoE + 0.4B Visual Encoder) <br> • Total: 25.3B | 64K           | -->

## Benchmark

This checkpoint is not designed for benchmarks, but for real-world open-ended applications. To this end, we evaluated on WildVision-Bench and noticed non-trivial improvements on it:


| Model                     | Score   |
|---------------------------|---------|
| gpt-4o                    | 89.15   |
| **Aria-Chat**     |**81.3** |
| gpt-4-vision-preview      | 79.78   |
| Aria                      | 74.1    |
| Reka-Flash                | 64.65   |
| claude-3-opus-20240229    | 62.03   |
| yi-vl-plus                | 55.05   |
| liuhaotian/llava-v1.6-34b | 51.89   |
| claude-3-sonnet-20240229  | 50.0    |
| claude-3-haiku-20240307   | 37.83   |

## Quick Start
### Installation
```
pip install transformers==4.45.0 accelerate==0.34.1 sentencepiece==0.2.0 torchvision requests torch Pillow
pip install flash-attn --no-build-isolation

# For better inference performance, you can install grouped-gemm, which may take 3-5 minutes to install
pip install grouped_gemm==0.1.6
```

### Inference

Aria has 25.3B total parameters, it can be loaded in one A100 (80GB) GPU with bfloat16 precision.

Here is a code snippet to show you how to use Aria.

```python
import requests
import torch
from PIL import Image
from transformers import AutoModelForCausalLM, AutoProcessor

model_id_or_path = "rhymes-ai/Aria-Chat"

model = AutoModelForCausalLM.from_pretrained(model_id_or_path, device_map="auto", torch_dtype=torch.bfloat16, trust_remote_code=True)

processor = AutoProcessor.from_pretrained(model_id_or_path, trust_remote_code=True)

image_path = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cat.png"

image = Image.open(requests.get(image_path, stream=True).raw)

messages = [
    {
        "role": "user",
        "content": [
            {"text": None, "type": "image"},
            {"text": "what is the image?", "type": "text"},
        ],
    }
]

text = processor.apply_chat_template(messages, add_generation_prompt=True)
inputs = processor(text=text, images=image, return_tensors="pt")
inputs["pixel_values"] = inputs["pixel_values"].to(model.dtype)
inputs = {k: v.to(model.device) for k, v in inputs.items()}

with torch.inference_mode(), torch.cuda.amp.autocast(dtype=torch.bfloat16):
    output = model.generate(
        **inputs,
        max_new_tokens=500,
        stop_strings=["<|im_end|>"],
        tokenizer=processor.tokenizer,
        do_sample=True,
        temperature=0.9,
    )
    output_ids = output[0][inputs["input_ids"].shape[1]:]
    result = processor.decode(output_ids, skip_special_tokens=True)

print(result)
```

### Advanced Inference and Fine-tuning
We provide a [codebase](https://github.com/rhymes-ai/Aria) for more advanced usage of Aria,
including vllm inference, cookbooks, and fine-tuning on custom datasets.


## Citation
If you find our work helpful, please consider citing.
```
@article{aria,
  title={Aria: An Open Multimodal Native Mixture-of-Experts Model}, 
  author={Dongxu Li and Yudong Liu and Haoning Wu and Yue Wang and Zhiqi Shen and Bowen Qu and Xinyao Niu and Guoyin Wang and Bei Chen and Junnan Li},
  year={2024},
  journal={arXiv preprint arXiv:2410.05993},
}
```