Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -1208,7 +1208,7 @@ The dataset presents eleven sections (one per regression task), with varying deg
|
|
1208 |
|
1209 |
## Partitions and versions
|
1210 |
|
1211 |
-
The sections are already partitioned into a predefined train-validation-test split for training assessors. Assessors need a particular kind of partitioning (mainly stratified by instance identifier to avoid contamination), so that's why the
|
1212 |
|
1213 |
The **main** branch contains the unaltered datasets, keeping the original values of the task and model characteristics, whereas the **normalised** branch contains the datasets properly normalised.
|
1214 |
|
|
|
1208 |
|
1209 |
## Partitions and versions
|
1210 |
|
1211 |
+
The sections are already partitioned into a predefined train-validation-test split for training assessors. Assessors need a particular kind of partitioning (mainly stratified by instance identifier to avoid contamination), so that's why the subsets are given.
|
1212 |
|
1213 |
The **main** branch contains the unaltered datasets, keeping the original values of the task and model characteristics, whereas the **normalised** branch contains the datasets properly normalised.
|
1214 |
|