Datasets:

openai

/

gsm8k

like 365

Some of the examples in this dataset have the wrong answer.

I automatically detected some of these with my company's software, and shared them here:
https://huggingface.co/datasets/Cleanlab/bad_data_gsm8k_svamp.csv

Example:

Question: After scoring 14 points, Erin now has three times more points than Sara, who scored 8. How many points did Erin have before?

Answer According to the Dataset: 18