Update README.md
Browse files
README.md
CHANGED
@@ -82,9 +82,40 @@ Llama-3-Luxia-Ko-8B ๋ชจ๋ธ ํ์ต์ ์ํด 1TB ์์ค์ ํ๊ตญ์ด ์ฝํผ์ค์
|
|
82 |
NVIDIA H100 80GB * 8EA์ ํ์ฉํ์ฌ ๋ชจ๋ธ ์ฌ์ ํ์ต์ ์งํํ์์ต๋๋ค.
|
83 |
|
84 |
#### Training Hyperparameters
|
85 |
-
|
86 |
-
|
87 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
88 |
|
89 |
### Tokenizer
|
90 |
Llama-3-Tokenizer๋ฅผ ํ๊ตญ์ด ํนํํ๊ธฐ ์ํด ํ๊ตญ์ด ํ ํฐ 17,536๊ฐ๋ฅผ ์ถ๊ฐํ๊ณ ํ์ฉํ์์ต๋๋ค.
|
|
|
82 |
NVIDIA H100 80GB * 8EA์ ํ์ฉํ์ฌ ๋ชจ๋ธ ์ฌ์ ํ์ต์ ์งํํ์์ต๋๋ค.
|
83 |
|
84 |
#### Training Hyperparameters
|
85 |
+
<table>
|
86 |
+
<tr>
|
87 |
+
<td><strong>Model</strong>
|
88 |
+
</td>
|
89 |
+
<td><strong>Params</strong>
|
90 |
+
</td>
|
91 |
+
<td><strong>Context length</strong>
|
92 |
+
</td>
|
93 |
+
<td><strong>GQA</strong>
|
94 |
+
</td>
|
95 |
+
<td><strong>Learning rate</strong>
|
96 |
+
</td>
|
97 |
+
<td><strong>Batch</strong>
|
98 |
+
</td>
|
99 |
+
<td><strong>Precision</strong>
|
100 |
+
</td>
|
101 |
+
</tr>
|
102 |
+
<tr>
|
103 |
+
<td>Llama-3-Luxia-Ko
|
104 |
+
</td>
|
105 |
+
<td>8B
|
106 |
+
</td>
|
107 |
+
<td>8k
|
108 |
+
</td>
|
109 |
+
<td>yes
|
110 |
+
</td>
|
111 |
+
<td>1e-5
|
112 |
+
</td>
|
113 |
+
<td>128
|
114 |
+
</td>
|
115 |
+
<td>bf16
|
116 |
+
</td>
|
117 |
+
</tr>
|
118 |
+
</table>
|
119 |
|
120 |
### Tokenizer
|
121 |
Llama-3-Tokenizer๋ฅผ ํ๊ตญ์ด ํนํํ๊ธฐ ์ํด ํ๊ตญ์ด ํ ํฐ 17,536๊ฐ๋ฅผ ์ถ๊ฐํ๊ณ ํ์ฉํ์์ต๋๋ค.
|