README.md

metadata

dataset_info:
  features:
    - name: question_id
      dtype: string
    - name: category
      dtype: string
    - name: turns
      sequence: string
    - name: ground_truth
      dtype: string
    - name: task
      dtype: string
    - name: livebench_release_date
      dtype: timestamp[s]
    - name: livebench_removal_date
      dtype: string
  splits:
    - name: test
      num_bytes: 305848
      num_examples: 150
  download_size: 149433
  dataset_size: 305848
configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*
arxiv: 2406.19314

Dataset Card for "livebench/data_analysis"

LiveBench is a benchmark for LLMs designed with test set contamination and objective evaluation in mind. It has the following properties:

• 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.
• Each question has verifiable, objective ground-truth answers, allowing hard questions to be scored accurately and automatically, without the use of an LLM judge.
• LiveBench currently contains a set of 18 diverse tasks across 6 categories, and we will release new, harder tasks over time. This is the instruction_following category of livebench.

See more in our paper, leaderboard, and datasheet.