atekrugis/intent-classification-60k

一个精心策划的数据集，包含60,000个用户提示，标注为六个不同的意图类别：CODING（编码）、CHAT（聊天）、REASONING（推理）、SIMPLE（简单）、TOOL（工具）和BASIC（基础）。每个提示均使用GLM-5-Turbo模型在确定性设置下进行一致标注，确保分类方法的一致性。该数据集旨在训练意图分类模型、微调语言模型以及评估对话AI系统。数据集来源包括多个HuggingFace数据集，经过去重、过滤和分层抽样处理，最终通过GLM-5-Turbo模型进行确定性标注。

A curated dataset of 60,000 user prompts labeled with intent categories: CODING, CHAT, REASONING, SIMPLE, TOOL, and BASIC. Each prompt has been consistently labeled using GLM-5-Turbo with deterministic settings, ensuring uniform classification methodology. The dataset is designed for training intent classification models, fine-tuning language models, and evaluating conversational AI systems. The dataset is sourced from multiple HuggingFace datasets, processed through deduplication, filtering, and stratified sampling, and labeled deterministically using the GLM-5-Turbo model.