Dataset Card for "codex_math_qa_alpaca_style"

This dataset consists of code responses generated by codex-davinci-002 for solving math word problems from math_qa . This dataset is equivalent to theblackcat102/codex-math-qa , but has been slightly modified to fit the Alpaca format.

Numbers:

• Prompts : 28050
• Tokens : 6626950 using the EleutherAI/gpt-neox-20b tokenizer (counting instruction+input+output)