Update README.md

README.md

@@ -13,7 +13,7 @@ This dataset supports research aligned with the methodology described in our pap
 ## Key Features
 
 - ✅ **LLVM IR representation** of each function (field: `llvm_ir_function`)
-- ✅ Includes **train**, **validation**, and **test** splits
+- ✅ Includes **train**, **validation**, and **test** splits (see below)
 - ✅ Vulnerability labels (`label`) for supervised learning
 - ✅ Metadata about original source (`dataset`, `file`, `fun_name`)
 - ✅ Structured as **Parquet** files for fast loading and processing
@@ -29,6 +29,14 @@ Each record contains:
 - `label`: Binary label indicating vulnerability (`1` for vulnerable, `0` for non-vulnerable)
 - `split`: Dataset split (`train`, `validation`, `test`)
 
+## Split Information
+
+This dataset is split into **train**, **validation**, and **test** sets, following the exact partitioning strategy used in the experiments described in our paper. The split ensures a fair evaluation of generalization performance by separating functions into disjoint sets with no overlap. This allows researchers to directly reproduce our results or compare against them under consistent conditions.
+
+- `train`: Used to fit model parameters
+- `validation`: Used for model selection and hyperparameter tuning
+- `test`: Used exclusively for final evaluation and benchmarking
+
 ## Format
 
 This dataset is provided in [Apache Parquet](https://parquet.apache.org/) format under the `default` configuration. It follows the [Croissant schema](https://mlcommons.org/croissant/) and includes three predefined splits.
@@ -43,7 +51,6 @@ from datasets import load_dataset
 # Load a specific split
 train_ds = load_dataset("compAgent/CompRealVul_LLVM", split="train")
 print(train_ds[0])
-```
 
 ## Example
 ```json