InvestmentResearchAI/llm-ade-fin_data-subset-earnings-10k
收藏Hugging Face2024-04-17 更新2024-04-19 收录
下载链接：
https://hf-mirror.com/datasets/InvestmentResearchAI/llm-ade-fin_data-subset-earnings-10k
下载链接
链接失效反馈官方服务：
资源简介：
This dataset is curated to train (next-token) the LLM-ADE model, specifically designed to imbue it with financial domain expertise. It consists of 75,849 sequences, amounting to approximately 16.8 million tokens, using the Llama tokenizer. We have deliberately unlabled the sequences wrt the company to reflect real world data and train the model to process knowledge from unlabelled data.
提供机构：
InvestmentResearchAI
原始信息汇总
数据集概述
- 目的：用于训练LLM-ADE模型，增强其在金融领域的专业知识。
- 数据规模：包含75,849个序列，总计约16.8百万个标记。
- 标记工具：使用Llama标记器进行标记。
- 数据特点：序列未标注公司信息，以模拟真实世界数据，训练模型处理未标注数据的能力。
