File size: 2,333 Bytes
60c6d53
c58b35b
60c6d53
eb7b4c3
 
4de21b8
eb7b4c3
 
 
60c6d53
eb7b4c3
 
4de21b8
eb7b4c3
4de21b8
eb7b4c3
 
 
4de21b8
eb7b4c3
 
 
4de21b8
eb7b4c3
 
 
 
 
 
 
 
4de21b8
eb7b4c3
 
 
 
4de21b8
 
eb7b4c3
04cdc7f
 
 
 
c58b35b
 
 
 
4de21b8
eb7b4c3
 
 
 
4de21b8
 
eb7b4c3
 
4de21b8
 
eb7b4c3
 
 
 
 
 
 
4de21b8
eb7b4c3
 
4de21b8
 
eb7b4c3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4de21b8
eb7b4c3
 
 
 
 
 
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
---

license: openrail

datasets:
- LinkSoul/LLaSM-Audio-Instructions
language:
- zh
- en
---


# LLaSM: Large Language and Speech Model

开源,可商用的**中英文双语语音-语言助手 LLaSM 以及中英文语音 SFT 数据集 LLaSM-Audio-Instructions**,第一个支持中英文语音-文本多模态对话的开源可商用对话模型。

<!--
<div align="center">
    <img src="https://huggingface.co/LinkSoul/LLaSM-Cllama2/blob/main/meta/preview.jpg" width="40%">
</div>

--> 
![LLaSM](meta/llasm_preview.jpg)

## 基础演示

![Base Demo](meta/demo.gif)

## 在线试玩

> Talk is cheap, Show you the Demo.
- [Demo 地址 / HuggingFace Spaces](https://huggingface.co/spaces/LinkSoul/LLaSM) 

## 资源下载

- 模型:
  - [LLaSM-Chinese-Llama-2-7B](https://huggingface.co/LinkSoul/LLaSM-Cllama2)
  - [LLaSM-Baichuan-7B](https://huggingface.co/LinkSoul/LLaSM-Baichuan)

- 百度网盘下载:
  - [LLaSM-Chinese-Llama-2-7B](https://pan.baidu.com/s/1PaipNDfqV7f3W1-tl5rwzA?pwd=2549)
  - [LLaSM-Baichuan-7B](https://pan.baidu.com/s/1QZrXA8IJXclN77T4jM7tEw?pwd=y2p7)

- 语言模型:
  - [Chinese-Llama-2-7b](https://github.com/LinkSoul-AI/Chinese-Llama-2-7b)
  - [Baichuan-7B](https://huggingface.co/baichuan-inc/Baichuan-7B)

- 数据集:[LLaSM-Audio-Instructions](https://huggingface.co/datasets/LinkSoul/LLaSM-Audio-Instructions)

## 环境安装
```shell
# clone the repository
git clone https://github.com/LinkSoul-AI/LLaSM
cd LLaSM

# install package
conda create -n llasm python=3.10 -y
conda activate llasm
pip install --upgrade pip
pip install -e .
```

## 快速测试

```shell
export LLASM_DEVICE="cuda:0"
python infer.py \
    --input_audio_file PATH/TO/YOUR/AUDIO \
    --llasm_model PATH/TO/LLaSM/MODEL \
    --llasm_audio_tower PATH/TO/WHISPER/MODEL \
    --llm_type "Chinese_llama2" or "baichuan" \
```

## TODO
- 如何训练
- int4 量化
- docker 部署

## 相关项目
- [Chinese-Llama-2-7B](https://huggingface.co/LinkSoul/Chinese-Llama-2-7b)
- [Whisper](https://ai.meta.com/llama/)
- [baichuan-inc/Baichuan-7B](https://huggingface.co/baichuan-inc/Baichuan-7B)


## 项目协议

[Apache-2.0 license](https://github.com/LinkSoul-AI/LLaSM/blob/main/LICENSE)

## 微信交流群
<!--
<img src="meta/QRcode.jpg" alt="微信交流群" width="300"/>
-->
欢迎加入[微信群](meta/QRcode.jpg)