Update README.md

README.md

@@ -48,11 +48,18 @@ from datasets import load_dataset
 ds = load_dataset("se2p/code-readability-merged")
 ```
 
+The snippets are **not** split into train and test (and validation) set. Thus, the whole dataset is in the **train** set:
+```python
+ds = ds['train']
+ds_as_list = ds.to_list() # Convert the dataset to whatever format suits you best
+
+```
+
 The dataset is structured as follows:
 
-```python
+```json
 {
-  "code_snippet": ...,  # Java source code snippet.
+  "code_snippet": ...,  # Java source code snippet
   "score": ...          # Readability score
 }
 ```
@@ -86,8 +93,6 @@ The code_snippet (string) is the code snippet that was rated in a study by multi
 Those could answer based on a five point Likert scale, with 1 being very unreadable and 5 being very readable.
 The score (float) is the averaged rating score of all participants between 1.0 (very unreadable) and 5.0 (very readable).
 
-The snippets are **not** split into train and test (and validation) set.
-
 ## Dataset Creation
 
 ### Curation Rationale