Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,21 @@
|
|
1 |
-
---
|
2 |
-
license: mit
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: mit
|
3 |
+
datasets:
|
4 |
+
- msarmi9/korean-english-multitarget-ted-talks-task
|
5 |
+
language:
|
6 |
+
- en
|
7 |
+
- ko
|
8 |
+
---
|
9 |
+
|
10 |
+
### 직접 구현한 Transformer 및 RoPE로 영어->한국어 번역 모델 제작
|
11 |
+
|
12 |
+
- 약 13만쌍의 영어-한국어 데이터로 scratch training.
|
13 |
+
```
|
14 |
+
num_epochs = 5
|
15 |
+
batch_size = 64
|
16 |
+
config.intermediate_size = 768*4
|
17 |
+
config.num_attention_heads = 6
|
18 |
+
config.num_hidden_layers = 8
|
19 |
+
```
|
20 |
+
|
21 |
+
<img src="W&B Chart 2025. 2. 2. 오전 1_03_52 (1)" title="Orange is val_loss and Blue is loss"/>
|