Datasets:

Modalities:
Text
Formats:
json
ArXiv:
Libraries:
Datasets
Dask
License:
CASIA-LM commited on
Commit
a33e2d7
·
verified ·
1 Parent(s): 3d55011

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +2 -2
README.md CHANGED
@@ -97,7 +97,7 @@ As illustrated in Figure 8, the sample counts and corresponding proportions acro
97
 
98
 
99
 
100
- \paragraph{Quality-Related Domain Distribution}
101
  In order to explore the domain distribution across different quality intervals, we perform an analysis focusing on the quality-related domain distribution. Specifically, we calculate the proportions of various domains within each quality interval. As shown in Figure 5, this table provides a detailed breakdown of domain proportions across different quality intervals. From the results, we observe that the distribution of domain data within each quality interval aligns closely with their overall distribution in the dataset. Based on the proportions in Figure 5, researchers can filter domain-specific data within targeted quality intervals, enabling the extraction of higher-quality domain-specific data subsets.
102
 
103
 
@@ -108,7 +108,7 @@ In order to explore the domain distribution across different quality intervals,
108
  </div>
109
 
110
 
111
- \subsection{Data Toxicity Analysis}
112
 
113
  <div align="center">
114
  <img src="./picture/toxicity_distribution.pdf" width="67%" />
 
97
 
98
 
99
 
100
+ **Quality-Related Domain Distribution**
101
  In order to explore the domain distribution across different quality intervals, we perform an analysis focusing on the quality-related domain distribution. Specifically, we calculate the proportions of various domains within each quality interval. As shown in Figure 5, this table provides a detailed breakdown of domain proportions across different quality intervals. From the results, we observe that the distribution of domain data within each quality interval aligns closely with their overall distribution in the dataset. Based on the proportions in Figure 5, researchers can filter domain-specific data within targeted quality intervals, enabling the extraction of higher-quality domain-specific data subsets.
102
 
103
 
 
108
  </div>
109
 
110
 
111
+ #### Data Toxicity Analysis
112
 
113
  <div align="center">
114
  <img src="./picture/toxicity_distribution.pdf" width="67%" />