Self-Learning LLM Models
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/teddy-f-47/self-learning-llm-public
下载链接
链接失效反馈官方服务：
资源简介：
该数据集通过使用不同的预训练模型，包括Mistral-DPO、Mistral-Instruct和Tiny-LLama-Chat，来展示识别潜在用户兴趣（PiUs）方法的有效性。数据集不仅包含了来自不同方法生成的问题，还包含了实验结果的视觉化展示。该数据集涉及了多种规模的模型，其参数量从7B到1.1B不等，主要任务是对潜在用户兴趣的识别以及问题的生成。
This dataset demonstrates the effectiveness of potential user interest (PiUs) identification methods by utilizing various pre-trained models including Mistral-DPO, Mistral-Instruct, and TinyLLaMA-Chat. It includes both questions generated by different approaches and visualizations of experimental results, covering models with parameter scales ranging from 7B to 1.1B. The core tasks of this dataset are potential user interest identification and question generation.
提供机构：
Various providers including Intel for mistral-dpo
