Reasoning summary dataset is a dataset that summarizes and adds extra metadata to a reasoning chain or a block from the reasoning chain (for models like qwen3.5+, gemma4, GLM-5.2)
This dataset is designed for training and evaluating reasoning summarization models.
The goal of the dataset is to convert long, messy, or highly detailed reasoning traces into short, user-friendly summaries that explain what the model was… See the full description on the dataset page:
https://huggingface.co/datasets/Qyrou/reasoning-summaries-61k.
