Commit
·
d8f761b
1
Parent(s):
2f8298d
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,19 @@
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: apache-2.0
|
| 3 |
---
|
| 4 |
+
Trained polyglot 5.8B with the QLORA method using the [Chatbot_data_for_Korean](https://github.com/songys/Chatbot_data) dataset.
|
| 5 |
+
|
| 6 |
+
The hyper-parameters used for training are as follows.
|
| 7 |
+
|
| 8 |
+
- batch-size: 16
|
| 9 |
+
- max_steps: 3000
|
| 10 |
+
- Learning rate: 3e-4
|
| 11 |
+
- Lora r: 8
|
| 12 |
+
- Lora target modules: query_key_value
|
| 13 |
+
|
| 14 |
+
Prompt Template:
|
| 15 |
+
```
|
| 16 |
+
### 질문: {문장}
|
| 17 |
+
### 응답: {문장}
|
| 18 |
+
### 유형: {일반 또는 연애}
|
| 19 |
+
```
|