mys/simula-batched-qa-exp

该数据集是一个由AfterImage开发的OpenSimula检查点子树，用于生成多样本单问答（QA）数据，结合策略语料库。它采用Simula启发的机制设计理念，专注于通过分类法、策略和元提示生成合成数据。数据集包含manifest.json、taxonomy_bundle.json、sampling_strategy.json和run_config.json等文件，用于定义生成过程的各个方面。

This dataset is an OpenSimula checkpoint subtree developed by AfterImage, designed for generating multi-sample single-QA (Question-Answering) data with a policy corpus. It follows Simula-inspired mechanism design principles, focusing on synthetic data generation through taxonomies, strategies, and meta-prompts. The dataset includes files like manifest.json, taxonomy_bundle.json, sampling_strategy.json, and run_config.json, which define various aspects of the generation process.