princeton-nlp/LLMBar
LLMBar is a challenging meta-evaluation benchmark designed to test the ability of an LLM evaluator in discerning instruction-following outputs. LLMBar consists of 419 instances, where each entry contains an instruction paired with two outputs: one faithfully and correctly follows the instruction and the other deviates from it. There is also a gold preference label indicating which output is objectively better for each instance.
| Author / organization | princeton-nlp |
|---|---|
| License (as declared) | mit |
| Downloads | 162 (as reported by Hugging Face) |
| Likes | 4 |
| Created | 29 October 2023 |
| Last updated at source | 29 October 2023 |
Popularity, last 2 days
162 now (-11 since 2026-10-16). Counts as reported by the source, snapshotted daily by Silytics.
More from princeton-nlp
- princeton-nlp/SWE-bench_VerifiedDataset
- princeton-nlp/SWE-benchDataset
- princeton-nlp/SWE-bench_LiteDataset
- princeton-nlp/CharXivDataset
- princeton-nlp/SWE-bench_MultimodalDataset
- princeton-nlp/prolong-data-64KDataset
Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.