AceReason 11 100K is a dataset designed for fine-tuning and evaluating reasoning capabilities in small- to mid-scale large language models.It includes diverse reasoning tasks spanning mathematics, logic, coding, and scientific reasoning.
The dataset contains random 15000 samples out of 100,000 examples of reasoning prompts and completions.Each example is formatted as JSON objects with fields like:
instruction: The user question or… See the full description on the dataset page:
https://huggingface.co/datasets/darshannere/Random-15k_AceReason-1.1-SFT.
