Update README.md
Browse files
README.md
CHANGED
|
@@ -62,7 +62,8 @@ Using Magpie, one-round dialogue tasks were synthesized for:
|
|
| 62 |
|
| 63 |
Five-round dialogue data was generated to simulate conversational styles typical of everyday interactions, enhancing the model's performance in realistic communication scenarios.
|
| 64 |
|
| 65 |
-
#### **
|
|
|
|
| 66 |
|
| 67 |
Math problem data, including detailed reasoning steps in the answers, was generated using **deepseek-v2.5**.Increased Training Data Size and Content Diversity
|
| 68 |
|
|
@@ -248,6 +249,8 @@ smoltalk-chinese 数据集的构建过程严格遵循高标准，确保数据的
|
|
| 248 |
<img width="900px" alt="experiment" src="./score.png">
|
| 249 |
</p>
|
| 250 |
|
|
|
|
|
|
|
| 251 |
**我们诚邀对这一领域感兴趣的开发者和研究者关注和联系社区，共同推动技术的进步。敬请期待数据集的开源发布！**
|
| 252 |
|
| 253 |
|
|
|
|
| 62 |
|
| 63 |
Five-round dialogue data was generated to simulate conversational styles typical of everyday interactions, enhancing the model's performance in realistic communication scenarios.
|
| 64 |
|
| 65 |
+
#### **
|
| 66 |
+
4. Math Problems from the Math23K Chinese Version**
|
| 67 |
|
| 68 |
Math problem data, including detailed reasoning steps in the answers, was generated using **deepseek-v2.5**.Increased Training Data Size and Content Diversity
|
| 69 |
|
|
|
|
| 249 |
<img width="900px" alt="experiment" src="./score.png">
|
| 250 |
</p>
|
| 251 |
|
| 252 |
+
训练的模型在 opencsg/csg-wukong-2b-smoltalk-chinese
|
| 253 |
+
|
| 254 |
**我们诚邀对这一领域感兴趣的开发者和研究者关注和联系社区，共同推动技术的进步。敬请期待数据集的开源发布！**
|
| 255 |
|
| 256 |
|