ritwikraha/

reasoning

This dataset aims to be a compilation of simple preference pairs for reasoning tasks generated synthetically using gemini-2.5-flash It has been generated using this colab notebook.

	Dataset Details




	
	Uses

For RL and reasoning tasks.

Use it as a preference dataset for DPO tuning.

Though a detailed schema isn’t provided in its card, it follows a standard Hugging Face DatasetDict. Once… See the full description on the dataset page: https://huggingface.co/datasets/ritwikraha/reasoning.