File size: 7,761 Bytes
5bd4cc6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f48f564
5bd4cc6
 
 
 
7b039e4
6bcadea
 
5bd4cc6
 
 
7b039e4
5bd4cc6
 
6bcadea
5bd4cc6
 
 
 
 
 
 
 
 
 
 
 
7b039e4
 
5bd4cc6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8491f0c
5bd4cc6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5328f65
5bd4cc6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f8eab3b
5bd4cc6
7b039e4
5bd4cc6
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
---
license: other
license_name: seallms
license_link: https://huggingface.co/SeaLLMs/SeaLLM-13B-Chat/blob/main/LICENSE
language:
- en
- zh
- id
- vi
- th
- ms
tags:
- sea
- multilingual

---

# *SeaLLMs-v3 - Large Language Models for Southeast Asia*

<p align="center">
<a href="https://damo-nlp-sg.github.io/SeaLLMs/" target="_blank" rel="noopener">Website</a>
&nbsp;&nbsp;
<a href="https://huggingface.co/SeaLLMs/SeaLLMs-v3-7B" target="_blank" rel="noopener">Model</a>
&nbsp;&nbsp;
<a href="https://huggingface.co/spaces/SeaLLMs/SeaLLM-Chat" target="_blank" rel="noopener"> 🤗 DEMO</a>
&nbsp;&nbsp;
<a href="https://github.com/DAMO-NLP-SG/SeaLLMs" target="_blank" rel="noopener">Github</a>
&nbsp;&nbsp;
<a href="https://arxiv.org/pdf/2407.19672" target="_blank" rel="noopener">[NEW] Technical Report</a>
</p>


We introduce **SeaLLMs-v3**, the latest series of the SeaLLMs (Large Language Models for Southeast Asian languages) family. It achieves state-of-the-art performance among models with similar sizes, excelling across a diverse array of tasks such as world knowledge, mathematical reasoning, translation, and instruction following. In the meantime, it was specifically enhanced to be more trustworthy, exhibiting reduced hallucination and providing safe responses, particularly in queries closed related to Southeast Asian culture.

## 🔥 Highlights

- State-of-the-art performance compared to open-source models of similar sizes, evaluated across various dimensions such as human exam questions, instruction-following, mathematics, and translation.
- Significantly enhanced instruction-following capability, especially in multi-turn settings.
- Ensures safety in usage with significantly reduced instances of hallucination and sensitivity to local contexts.

## Uses

SeaLLMs is tailored for handling a wide range of languages spoken in the SEA region, including English, Chinese, Indonesian, Vietnamese, Thai, Tagalog, Malay, Burmese, Khmer, Lao, Tamil, and Javanese.

This page introduces the **SeaLLMs-v3-7B** model, which can be fine-tuned for your specific downstream tasks, especially in SEA languages.
Note that this is a base model, if you are looking for a model that can be directly applicable to your downstream applications, you may want to check the chat version model: **[SeaLLMs-v3-7B-Chat](https://huggingface.co/SeaLLMs/SeaLLMs-v3-7B-Chat)**.

## Evaluation

We evaluate SeaLLMs-v3-7B using human exam questions and mathematics.

#### Multilingual World Knowledge - M3Exam

[M3Exam](https://arxiv.org/abs/2306.05179) consists of local exam questions collected from each country. It reflects the model's world knowledge (e.g., with language or social science subjects) and reasoning abilities (e.g., with mathematics or natural science subjects).

| Model                  |        en |        zh |        id |        th |        vi |       avg |   avg_sea |
| :--------------------- | --------: | --------: | --------: | --------: | --------: | --------: | --------: |
| Gemma-7B               |     0.732 |     0.519 |     0.475 |     0.460 |     0.594 |     0.556 |     0.510 |
| Sailor-7B-Chat         |     0.660 |     0.652 |     0.475 |     0.462 |     0.513 |     0.552 |     0.483 |
| SeaLLM-7B-v2.5         |     0.758 |     0.581 |     0.499 |     0.502 |     0.622 |     0.592 |     0.541 |
| Sailor-14B             |     0.748 |     0.840 |     0.536 |     0.528 |     0.621 |     0.655 |     0.562 |
| Sailor-14B-Chat        |     0.749 |     0.843 |     0.553 |     0.566 |     0.637 |     0.670 |     0.585 |
| Qwen2-7B               | **0.815** |     0.874 |     0.530 |     0.479 |     0.628 |     0.665 |     0.546 |
| Qwen2-7B-Instruct      |     0.809 | **0.880** |     0.558 |     0.555 |     0.624 |     0.685 |     0.579 |
| **SeaLLMs-v3-7B**      |     0.809 |     0.863 |     0.545 |     0.530 |     0.628 |     0.675 |     0.568 |
| **SeaLLMs-v3-7B-Chat** |     0.809 |     0.874 | **0.558** | **0.569** | **0.649** | **0.692** | **0.592** |

#### Multilingual World Knowledge - MMLU

[MMLU](https://arxiv.org/abs/2009.03300) questions are translated to SEA languages for evaluation, which primarily tests the cross-lingual alignment of the model as the required knowledge is still mainly Western-focused.

| Model                  |        en |        zh |        id |        th |        vi |       avg |   avg_sea |
| :--------------------- | --------: | --------: | --------: | --------: | --------: | --------: | --------: |
| Gemma-7B               |     0.634 |     0.509 |     0.545 |     0.490 |     0.494 |     0.535 |     0.510 |
| Sailor-7B-Chat         |     0.558 |     0.472 |     0.484 |     0.414 |     0.462 |     0.478 |     0.454 |
| SeaLLM-7B-v2.5         |     0.652 |     0.544 |     0.565 |     0.479 |     0.528 |     0.553 |     0.524 |
| Sailor-14B             |     0.618 |     0.564 |     0.570 |     0.482 |     0.535 |     0.554 |     0.529 |
| Sailor-14B-Chat        |     0.627 |     0.561 |     0.567 |     0.496 |     0.541 |     0.558 |     0.535 |
| Qwen2-7B               |     0.710 |     0.642 |     0.602 |     0.520 |     0.566 |     0.608 |     0.563 |
| Qwen2-7B-Instruct      |     0.708 |     0.635 |     0.599 |     0.524 |     0.568 |     0.607 |     0.564 |
| **SeaLLMs-v3-7B**      |     0.706 | **0.654** |     0.617 |     0.536 | **0.587** |     0.620 |     0.580 |
| **SeaLLMs-v3-7B-Chat** | **0.713** |     0.647 | **0.625** | **0.544** |     0.578 | **0.622** | **0.582** |


#### Multilingual Math - MGSM

We evaluate the multilingual math capability by utilizing the [MGSM](https://arxiv.org/abs/2210.03057) dataset with a **5-shot prompting** approach. MGSM originally contains English, Chinese and Thai testing sets only, we use Google Translate to translate the same English questions into other SEA languages. Note that we adopt the tradition of each country to represent the number, e.g., in Indonesian and Vietnamese, dots are used as thousands separators and commas as decimal separators, the opposite of the English system.

| MGSM              |       en |       id |       ms |       th |       vi |       zh |      avg |
| :---------------- | -------: | -------: | -------: | -------: | -------: | -------: | -------: |
| Gemma-7B          |     64.8 |     41.2 |     43.2 |     38.0 |     34.0 |     39.6 |     43.5 |
| Sailor-7B         |     34.4 |     25.2 |     22.8 |     24.8 |     22.4 |     26.4 |     26.0 |
| Meta-Llama-3-8B   |     56.8 |     36.0 |     33.6 |     34.8 |     33.6 |     43.6 |     39.7 |
| GLM-4-9B          |     78.0 |     53.6 | **57.2** |     46.0 | **56.8** |     69.6 |     60.2 |
| Qwen2-7B          | **79.6** |     58.8 |     56.8 |     54.8 |     54.8 |     69.2 |     62.3 |
| **SeaLLMs-v3-7B** |     78.8 | **59.2** |     56.8 | **56.8** |     54.8 | **72.0** | **63.1** |

## Acknowledgement to Our Linguists

We would like to express our special thanks to our professional and native linguists, Tantong Champaiboon, Nguyen Ngoc Yen Nhi and Tara Devina Putri, who helped build, evaluate, and fact-check our sampled pretraining and SFT dataset as well as evaluating our models across different aspects, especially safety.


## Citation

If you find our project useful, we hope you would kindly star our repo and cite our work as follows: 

```
@article{damonlp2024seallm3,
  author = {Wenxuan Zhang*, Hou Pong Chan*, Yiran Zhao*, Mahani Aljunied*,
            Jianyu Wang*, Chaoqun Liu, Yue Deng, Zhiqiang Hu, Weiwen Xu,
            Yew Ken Chia, Xin Li, Lidong Bing},
  title = {SeaLLMs 3: Open Foundation and Chat Multilingual Large Language Models for Southeast Asian Languages},
  year = {2024},
  url = {https://arxiv.org/abs/2407.19672}
}
```

Corresponding Author: [email protected]