Commit
•
87b915e
1
Parent(s):
f6ed53d
Update README.md
Browse files
README.md
CHANGED
|
@@ -32,4 +32,13 @@ configs:
|
|
| 32 |
data_files:
|
| 33 |
- split: test
|
| 34 |
path: data/test-*
|
|
|
|
| 35 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
data_files:
|
| 33 |
- split: test
|
| 34 |
path: data/test-*
|
| 35 |
+
arxiv: 2406.19314
|
| 36 |
---
|
| 37 |
+
# Dataset Card for "livebench/language"
|
| 38 |
+
LiveBench is a benchmark for LLMs designed with test set contamination and objective evaluation in mind. It has the following properties:
|
| 39 |
+
- LiveBench is designed to limit potential contamination by releasing new questions monthly, as well as having questions based on recently-released datasets, arXiv papers, news articles, and IMDb movie synopses.
|
| 40 |
+
- Each question has verifiable, objective ground-truth answers, allowing hard questions to be scored accurately and automatically, without the use of an LLM judge.
|
| 41 |
+
- LiveBench currently contains a set of 18 diverse tasks across 6 categories, and we will release new, harder tasks over time.
|
| 42 |
+
This is the instruction_following category of livebench.
|
| 43 |
+
|
| 44 |
+
See more in our [paper](https://arxiv.org/abs/2406.19314), [leaderboard](https://livebench.ai/), and [datasheet](https://github.com/LiveBench/LiveBench/blob/main/docs/DATASHEET.md).
|