qianguo sun
commited on
Commit
·
68ff9a9
1
Parent(s):
8848aa0
更新readme
Browse files
README.md
CHANGED
@@ -1,3 +1,30 @@
|
|
1 |
---
|
2 |
license: gpl-3.0
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: gpl-3.0
|
3 |
+
language:
|
4 |
+
- en
|
5 |
+
- zh
|
6 |
+
|
7 |
+
inference: false
|
8 |
---
|
9 |
+
|
10 |
+
# Ziya-LLaMA-13B-v1
|
11 |
+
|
12 |
+
- Main Page:[Fengshenbang](https://fengshenbang-lm.com/)
|
13 |
+
- Github: [Fengshenbang-LM](https://github.com/IDEA-CCNL/Fengshenbang-LM)
|
14 |
+
|
15 |
+
(LLaMA权重的许可证限制,我们无法直接发布完整的模型权重,用户需要参考[使用说明](https://huggingface.co/IDEA-CCNL/Ziya-LLaMA-13B-v1)进行合并)
|
16 |
+
|
17 |
+
# 姜子牙系列模型
|
18 |
+
|
19 |
+
- [Ziya-LLaMA-13B-v1](https://huggingface.co/IDEA-CCNL/Ziya-LLaMA-13B-v1)
|
20 |
+
- [Ziya-LLaMA-7B-Reward](https://huggingface.co/IDEA-CCNL/Ziya-LLaMA-7B-Reward)
|
21 |
+
- [Ziya-LLaMA-13B-Pretrain-v1](https://huggingface.co/IDEA-CCNL/Ziya-LLaMA-13B-Pretrain-v1)
|
22 |
+
- [Ziya-BLIP2-14B-Visual-v1](https://huggingface.co/IDEA-CCNL/Ziya-BLIP2-14B-Visual-v1)
|
23 |
+
|
24 |
+
## 简介 Brief Introduction
|
25 |
+
我们对Ziya-LLaMA-13B-v1模型进行继续优化,推出开源版本Ziya-LLaMA-13B-v1.1。通过调整微调数据的比例和采用更优的强化学习策略,本版本在问答准确性、数学能力以及安全性等方面得到了明显提升,详细能力分析如图1所示。
|
26 |
+
<img src="https://huggingface.co/IDEA-CCNL/Ziya-LLaMA-13B-v1.1/blob/main/pk.png" width=1000 height=600>
|
27 |
+
## 软件依赖
|
28 |
+
```
|
29 |
+
pip install torch==1.12.1 tokenizers==0.13.3 git+https://github.com/huggingface/transformers
|
30 |
+
```
|