Add dataset card with description, source links, and citation
Browse files
README.md
CHANGED
|
@@ -34,3 +34,21 @@ configs:
|
|
| 34 |
- split: train
|
| 35 |
path: data/train-*
|
| 36 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
- split: train
|
| 35 |
path: data/train-*
|
| 36 |
---
|
| 37 |
+
|
| 38 |
+
# VERDICTS
|
| 39 |
+
|
| 40 |
+
Human labels of model-response correctness for evaluating LLM-as-a-judge. Each row is a model's response to a question from [BFF-Bench](https://huggingface.co/datasets/kensho/BFFBench) (`bffbench`) or [(C)MT-Bench](https://huggingface.co/datasets/kensho/CMTBench) (`mtbmr`), with a human `label`. Pairwise comparisons: [`kensho/VERDICTSPairwise`](https://huggingface.co/datasets/kensho/VERDICTSPairwise).
|
| 41 |
+
|
| 42 |
+
## Citation
|
| 43 |
+
|
| 44 |
+
```bibtex
|
| 45 |
+
@misc{krumdick2026freelabelslimitationsllmasajudge,
|
| 46 |
+
title={No Free Labels: Limitations of LLM-as-a-Judge Without Human Grounding},
|
| 47 |
+
author={Michael Krumdick and Charles Lovering and Varshini Reddy and Seth Ebner and Chris Tanner},
|
| 48 |
+
year={2026},
|
| 49 |
+
eprint={2503.05061},
|
| 50 |
+
archivePrefix={arXiv},
|
| 51 |
+
primaryClass={cs.CL},
|
| 52 |
+
url={https://arxiv.org/abs/2503.05061},
|
| 53 |
+
}
|
| 54 |
+
```
|