Note make sure to have from and content the same NEED the from ASSISTANT (uses gpt)

Note ready to use conversations cols with reasoning traces (not specified but probably R1) NEED the from ASSISTANT (uses gpt)

Note non-reasoning

Note SAND-Math (Synthetic Augmented Novel and Difficult Mathematics) is a high-quality, high-difficulty dataset of mathematics problems and solutions.

Note All MetaMathQA data are augmented from the training sets of GSM8K and MATH.

Note a subset of meta-math/MetaMathQA