Add dataset skew and update description for file commit chains

README.md

@@ -11,7 +11,7 @@ GitGoodBench Lite is a subset of 900 samples for evaluating the performance of A
 The samples in the dataset are evenly split across the programming languages Python, Java and Kotlin and the sample types merge conflict resolution and file-commit gram.
 This dataset thus contains 150 samples per sample type and programming language.
 
-All data in this dataset are collected from open-source repositories on GitHub with permissive licenses 
+All data in this dataset are collected from 493 unique, open-source GitHub repositories with permissive licenses 
 that have >= 1000 stars, >= 5 branches, >= 10 contributors and are not a fork or archived. We collected the initial list of repositories using [SEART.](https://seart-ghs.si.usi.ch/)
 
 Evaluation is to be performed by exact-match (EM) of diffs for the merge conflict setting and by LLM-as-a-Judge for the file-commit gram setting. [For further details see our paper.]()
@@ -43,6 +43,8 @@ File-commit chain scenarios consist of two commits, the oldest and newest commit
 In total the chain consists of `times_seen_consecutively` commits. The intended use-cases of these scenarios are to evaluate the agent's capacity to create meaningful, cohesive commits or 
 improve the local tree via rebasing. Thus samples of this `sample_type` cover two scenario types.
 
+File-commit chains are at least 3 commits long, all commits only contain files in the specified programming_language and no commit is a merge commit.
+
 A `file_commit_chain` scenario looks as follows:
 ```
 {
@@ -82,6 +84,35 @@ The following table provides per-field details. Columns marked “Yes” under *
 # Dataset statistics
 We provide some statistics on the distribution of “difficulty” within the overall dataset and across different scenario types.
 
+## Dataset Skew
+
+We note that our dataset is skewed towards the top four repositories especially, however skew flattens quickly.
+
+### Distribution Statistics
+- Total number of repositories (count): 493
+- Average (mean) samples per repository: 1.83
+- Standard deviation (std): 2.82
+- Minimum (min): 1
+- 25th percentile (25%): 1
+- Median (50%): 1
+- 75th percentile (75%): 2
+- Maximum (max): 46
+
+### Top-10 Repositories by Sample Count
+
+| Repository                          | Percentage of Total Samples |
+|-------------------------------------|----------------------------:|
+| oss-review-toolkit/ort              | 5.11%                       |
+| stripe/stripe-android               | 3.33%                       |
+| element-hq/element-android          | 2.67%                       |
+| jetbrains/compose-multiplatform     | 1.11%                       |
+| jetbrains/ideavim                   | 0.89%                       |
+| kotlin/dokka                        | 0.89%                       |
+| android/nowinandroid                | 0.78%                       |
+| wikimedia/apps-android-wikipedia    | 0.78%                       |
+| coil-kt/coil                        | 0.78%                       |
+| kotlin/kotlinx.coroutines          | 0.67%                       |
+
 
 ## Overall Difficulty Distribution