This dataset is a combination of all datasets used for expert finetuning in a curriculum order of Math, Science, Technology, and Engineering based on similarities between each domain’s fundamentals, aiding model learning. This dataset contains a mixture of chat-based and continued-pretrain data.
meta-math/MetaMathQA… See the full description on the dataset page:
https://huggingface.co/datasets/omniomni/omni-stem.