TaoTern/TaoChat

TaoChat是一个为Taotern模型家族准备的英文监督微调数据集。它通过TaoData框架使用taochat.yaml配置，从Magpie-Align/Llama-3-Magpie-Air-3M-v0.1数据集处理而来。该版本包含约150万个指令-响应对，以JSONL格式提供，是Taotern模型使用的主要监督微调数据集之一。数据集支持监督微调、指令遵循和聊天模型训练任务，数据实例包括用户指令的input字段、助理响应的output字段以及来源追踪的_source字段。所有数据都在train分割中。数据集是合成的，可能包含风格偏见、任务覆盖不均或幻觉内容，并受上游数据集限制。它适用于语言模型指令调优、聊天助手微调和监督微调数据混合实验。许可证基于上游的llama3标签。TaoChat is an English supervised fine-tuning (SFT) dataset prepared for the Taotern model family. It is processed from Magpie-Align/Llama-3-Magpie-Air-3M-v0.1 with the TaoData framework using the taochat.yaml configuration. This release contains about 1.5 million instruction-response pairs in JSONL format and serves as one of the main SFT datasets used by Taotern models. It supports supervised fine-tuning, instruction following, and chat model training. Each data instance includes input (user instruction), output (assistant response), and _source (provenance tracking). The dataset is synthetic and may contain style biases, uneven task coverage, or hallucinated content, inheriting limitations from the upstream release. It is intended for instruction tuning, chat assistant fine-tuning, and SFT data mixture experiments, with licensing based on the llama3 tag.