Update README.md
Browse files
README.md
CHANGED
@@ -11,53 +11,92 @@ tags:
|
|
11 |
- rk3588
|
12 |
- RockChip
|
13 |
---
|
14 |
-
#
|
15 |
-
|
16 |
-
|
17 |
-
- [中文](
|
18 |
-
- [English](
|
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 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
- rk3588
|
12 |
- RockChip
|
13 |
---
|
14 |
+
# 通义千问-1_8B / Qwen-1_8B-Chat_rkLLM
|
15 |
+
|
16 |
+
|
17 |
+
- [中文](#通义千问-1_8B)
|
18 |
+
- [English](#Qwen-1_8B-Chat_rkLLM)
|
19 |
+
## 通义千问-1_8B
|
20 |
+
### 介绍
|
21 |
+
这是从 [Qwen-1_8B](https://huggingface.co/Qwen/Qwen-1_8B/) 转换而来的RKLLM模型,用于Rockchip设备的转换模型。
|
22 |
+
该模型运行在RK3588的NPU上。
|
23 |
+
|
24 |
+
- **模型名称**: Qwen-1_8B
|
25 |
+
- **模型架构**: 同Qwen-1_8B
|
26 |
+
- **发布者**: FydeOS
|
27 |
+
- **日期**: 2024-05-14
|
28 |
+
|
29 |
+
### 模型详情
|
30 |
+
通义千问-1.8B(Qwen-1.8B)是阿里云研发的通义千问大模型系列的18亿参数规模的模型。Qwen-1.8B是基于Transformer的大语言模型, 在超大规模的预训练数据上进行训练得到。预训练数据类型多样,覆盖广泛,包括大量网络文本、专业书籍、代码等。
|
31 |
+
|
32 |
+
### 要求
|
33 |
+
|
34 |
+
|
35 |
+
### 使用指南
|
36 |
+
> 本模型仅支持运行在搭载 rk3588 芯片的相关设备,请确认芯片信息并保证 npu 可用
|
37 |
+
#### OpenFyde 系统
|
38 |
+
> ⚠️:请确认你已经升级到openfyde的最新版本
|
39 |
+
- 点击下载模型文件 `XXX.rkllm`
|
40 |
+
- 新建文件夹 `model/` ,将模型文件放置在此文件夹下
|
41 |
+
- 启动 FydeOS AI ,在设置页面进行相关设置
|
42 |
+
|
43 |
+
#### 非 OpenFyde 系统
|
44 |
+
> ⚠️:请确认你进行了 rkllm 的 npu 相关内核更新
|
45 |
+
- 点击下载模型文件 `XXX.rkllm`
|
46 |
+
- 参考官方文档进行配置:[官方文档](https://github.com/airockchip/rknn-llm)
|
47 |
+
|
48 |
+
### FAQ
|
49 |
+
如遇到问题,敬请查阅issue区,如仍无法解决再提交issue。
|
50 |
+
|
51 |
+
### 限制与注意事项
|
52 |
+
说明模型的局限性和使用中的注意事项。
|
53 |
+
|
54 |
+
### 许可证
|
55 |
+
采用与[Qwen-1_8B](https://huggingface.co/Qwen/Qwen-1_8B/)相同的许可证。
|
56 |
+
|
57 |
+
### 联系方式
|
58 |
+
如需更多信息,请联系:
|
59 |
+
- **电子邮件**: [email protected]
|
60 |
+
- **主页**: [FydeOS AI](https://fydeos.ai/zh/)
|
61 |
+
|
62 |
+
## Qwen-1_8B-Chat_rkLLM
|
63 |
+
|
64 |
+
### Introduction
|
65 |
+
This is a conversion from [Qwen-1_8B](https://huggingface.co/Qwen/Qwen-1_8B/) to the RKLLM format for Rockchip devices. The model runs on the NPU from RK3588.
|
66 |
+
|
67 |
+
- **Model Name**: Qwen-1_8B
|
68 |
+
- **Model Architecture**: Same as Qwen-1_8B
|
69 |
+
- **Publisher**: FydeOS
|
70 |
+
- **Date**: 2024-05-14
|
71 |
+
|
72 |
+
### Model Details
|
73 |
+
The Qwen-1.8B model, part of Alibaba Cloud's larger Qwen model series, has 1.8 billion parameters. It is a large language model based on the Transformer architecture, trained on a vast and varied dataset including extensive internet text, professional literature, and code.
|
74 |
+
|
75 |
+
### Requirements
|
76 |
+
|
77 |
+
### Usage Guide
|
78 |
+
> This model is only supported on devices equipped with the RK3588 chip. Ensure that the chip and NPU are available.
|
79 |
+
#### For OpenFyde Systems
|
80 |
+
> ⚠️: Please make sure you have upgraded to the latest version of OpenFyde.
|
81 |
+
- Click to download the model file `XXX.rkllm`.
|
82 |
+
- Create a new folder `model/` and place the model file in this folder.
|
83 |
+
- Start FydeOS AI and make the necessary settings on the settings page.
|
84 |
+
|
85 |
+
#### For Non-OpenFyde Systems
|
86 |
+
> ⚠️: Ensure you have updated the NPU related kernel for rkllm.
|
87 |
+
- Click to download the model file `XXX.rkllm`.
|
88 |
+
- Refer to the official documentation for setup: [Official Documentation](https://github.com/airockchip/rknn-llm)
|
89 |
+
|
90 |
+
### FAQ
|
91 |
+
Please consult the issue section for any problems, and submit a new issue if the problem remains unsolved.
|
92 |
+
|
93 |
+
### Limitations and Precautions
|
94 |
+
Details on the limitations and precautions to consider when using the model.
|
95 |
+
|
96 |
+
### License
|
97 |
+
Uses the same license as [Qwen-1_8B](https://huggingface.co/Qwen/Qwen-1_8B/).
|
98 |
+
|
99 |
+
### Contact Information
|
100 |
+
For more information, please contact:
|
101 |
+
- **Email**: [email protected]
|
102 |
+
- **Homepage**: [FydeOS AI](https://fydeos.ai/zh/)
|