Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,43 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
+
datasets:
|
| 4 |
+
- Azure99/blossom-chat-v2
|
| 5 |
+
- Azure99/blossom-math-v3
|
| 6 |
+
- Azure99/blossom-wizard-v2
|
| 7 |
+
- Azure99/blossom-orca-v2
|
| 8 |
+
language:
|
| 9 |
+
- zh
|
| 10 |
+
- en
|
| 11 |
+
pipeline_tag: text-generation
|
| 12 |
---
|
| 13 |
+
# **BLOSSOM-v4-qwen1_5-7b**
|
| 14 |
+
|
| 15 |
+
### 介绍
|
| 16 |
+
|
| 17 |
+
Blossom是一个对话式语言模型，基于Qwen1.5-7B预训练模型，在Blossom Orca/Wizard/Chat/Math混合数据集上进行指令精调得来。Blossom拥有强大的通用能力及上下文理解能力，此外，训练使用的高质量中英文数据集也进行了开源。
|
| 18 |
+
|
| 19 |
+
训练分为两阶段，第一阶段使用100K Wizard、100K Orca、20K Math单轮指令数据集，训练1个epoch；第二阶段使用50K Blossom chat多轮对话数据集、以及上一阶段中随机采样2%的数据，训练3个epoch。
|
| 20 |
+
|
| 21 |
+
### 推理
|
| 22 |
+
|
| 23 |
+
推理采用对话续写的形式。
|
| 24 |
+
|
| 25 |
+
单轮对话
|
| 26 |
+
|
| 27 |
+
```
|
| 28 |
+
A chat between a human and an artificial intelligence bot. The bot gives helpful, detailed, and polite answers to the human's questions.
|
| 29 |
+
|Human|: 你好
|
| 30 |
+
|Bot|:
|
| 31 |
+
```
|
| 32 |
+
|
| 33 |
+
多轮对话
|
| 34 |
+
|
| 35 |
+
```
|
| 36 |
+
A chat between a human and an artificial intelligence bot. The bot gives helpful, detailed, and polite answers to the human's questions.
|
| 37 |
+
|Human|: 你好
|
| 38 |
+
|Bot|: 你好，有什么我能帮助你的？<|endoftext|>
|
| 39 |
+
|Human|: 介绍下中国的首都吧
|
| 40 |
+
|Bot|:
|
| 41 |
+
```
|
| 42 |
+
|
| 43 |
+
注意：在历史对话的Bot输出结尾，拼接一个<|endoftext|>
|