---
dataset_info:
  features:
  - name: problem
    dtype: string
  - name: answer
    dtype: string
  - name: source
    dtype: string
  - name: llama8b_solve_rate
    dtype: float64
  splits:
  - name: train
    num_bytes: 3890093.3620152753
    num_examples: 15938
  - name: test
    num_bytes: 432259.7154052612
    num_examples: 1771
  download_size: 2767668
  dataset_size: 4322353.077420536
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
license: apache-2.0
task_categories:
- question-answering
language:
- en
tags:
- math
- RL
size_categories:
- 10K<n<100K
---

Default Math dataset used for [https://github.com/RD211/actors](https://github.com/RD211/actors).

The dataset is a subset of [https://huggingface.co/datasets/SynthLabsAI/Big-Math-RL-Verified](https://huggingface.co/datasets/SynthLabsAI/Big-Math-RL-Verified).

Only includes problems with solve rate between 0.05 and 0.6 and from `aops_forum` and `olympiads`.