---
dataset_info:
  features:
  - name: expected_answer
    dtype: string
  - name: problem
    dtype: string
  - name: generated_solution
    dtype: string
  - name: __index_level_0__
    dtype: int64
  splits:
  - name: train
    num_bytes: 92365102
    num_examples: 7507
  download_size: 42113908
  dataset_size: 92365102
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

This is the filtered version of `unsloth/OpenMathReasoning-mini` [dataset](https://huggingface.co/datasets/unsloth/OpenMathReasoning-mini). Here the dataset removes `expected_answer` that is not a number.