Datasets:
Tasks:
Text Generation
Modalities:
Text
Formats:
parquet
Languages:
Chinese
Size:
10M - 100M
License:
Update README.md
Browse files
README.md
CHANGED
@@ -86,7 +86,7 @@ We recorded 100,000 data samples along with their scores, creating the dataset `
|
|
86 |
|
87 |
|
88 |
|
89 |
-
#
|
90 |
After meticulously designed ablation studies, we aimed to contrast the effects between the Chinese-fineweb-edu dataset and traditional Chinese pre-training corpora.
|
91 |
For this purpose, we randomly selected samples from five datasets—CCI2-Data, SkyPile-150B, TeleChat-PTD, IndustryCorpus, and MAP-CC—proportional to the Chinese-fineweb-edu dataset, constructing a comparison dataset named chinese-random-select.
|
92 |
In our experiments, we utilized a model with 2.1 billion parameters, training it for 65k steps on both datasets respectively.
|
@@ -97,6 +97,10 @@ The results distinctly show that the dataset trained on Chinese-fineweb-edu sign
|
|
97 |
<img width="900px" alt="experiment" src="./chinese-fineweb-benchmark.png">
|
98 |
</p>
|
99 |
|
|
|
|
|
|
|
|
|
100 |
|
101 |
**We warmly invite developers and researchers interested in this field to follow and engage with the community, working together to advance the technology. Stay tuned for the open-source release of the dataset!**
|
102 |
|
@@ -191,7 +195,8 @@ Chinese Fineweb Edu 数据集的原始数据来源广泛,涵盖了多个国内
|
|
191 |
</p>
|
192 |
|
193 |
|
194 |
-
通过实验结果可以发现,在训练的靠后阶段,由于进入了第2个epoch
|
|
|
195 |
|
196 |
|
197 |
|
|
|
86 |
|
87 |
|
88 |
|
89 |
+
# Abaltion experiments
|
90 |
After meticulously designed ablation studies, we aimed to contrast the effects between the Chinese-fineweb-edu dataset and traditional Chinese pre-training corpora.
|
91 |
For this purpose, we randomly selected samples from five datasets—CCI2-Data, SkyPile-150B, TeleChat-PTD, IndustryCorpus, and MAP-CC—proportional to the Chinese-fineweb-edu dataset, constructing a comparison dataset named chinese-random-select.
|
92 |
In our experiments, we utilized a model with 2.1 billion parameters, training it for 65k steps on both datasets respectively.
|
|
|
97 |
<img width="900px" alt="experiment" src="./chinese-fineweb-benchmark.png">
|
98 |
</p>
|
99 |
|
100 |
+
The experimental results reveal that in the later stages of training, as it enters the second epoch and the learning rate rapidly decreases, the model trained with the chinese-fineweb-edu data shows a significant increase in accuracy,
|
101 |
+
whereas the model trained with randomly selected data remains at a lower level. This proves that the high-quality data of chinese-fineweb-edu significantly aids in training effectiveness.
|
102 |
+
With the same training duration, it can enhance model capabilities faster and save training resources.
|
103 |
+
This outcome also shares a striking similarity with the data ablation experiments conducted by HuggingFace on fineweb edu.
|
104 |
|
105 |
**We warmly invite developers and researchers interested in this field to follow and engage with the community, working together to advance the technology. Stay tuned for the open-source release of the dataset!**
|
106 |
|
|
|
195 |
</p>
|
196 |
|
197 |
|
198 |
+
通过实验结果可以发现,在训练的靠后阶段,由于进入了第2个epoch,且学习率进入快速下降阶段此时,使用chinese-fineweb-edu训练的模型,准确率有了明显的上升,而使用随机抽取的数据训练,则一直处于较低水平
|
199 |
+
这证明了chinese-fineweb-edu高质量数据对于模型训练效果有显著帮助,在同样训练时间下,能够更快的提升模型能力,节省训练资源,这个结果也和HuggingFace fineweb edu 的数据消融实验有异曲同工之妙。
|
200 |
|
201 |
|
202 |
|