Skip to content
@Toloka

Toloka

Data labeling platform for ML

Pinned Loading

  1. beemo Public

    Benchmark for fine-grained machine-generated text detection. 6.5k texts written by humans, generated by ten open-source instruction-finetuned LLMs and edited by expert annotators.

    6 1

  2. u-math Public

    Official evaluation code for the U-MATH and μ-MATH benchmarks. These datasets are designed to test the mathematical reasoning and meta-evaluation capabilities of LLMs on university-level problems.

    Python 9 2

  3. crowd-kit Public

    Control the quality of your labeled data with the Python tools you already know.

    Python 227 18

Repositories

Showing 10 of 28 repositories

Top languages

Loading…