yusufizzetmurat/fed-pulse-embedding-caches

这是一个用于fed-pulse FOMC（联邦公开市场委员会）文本分析项目的数据集，包含基于不同编码器的嵌入缓存。数据来自美联储FOMC发布的文本（美国政府的公共领域内容），以Parquet格式存储，每个文件包含记录ID、文档ID、事件日期、块索引、块预览和嵌入向量等字段。该数据集旨在支持货币政策和文本分析任务，作为训练和推理的预计算嵌入资源。This is a dataset for the fed-pulse FOMC (Federal Open Market Committee) text analytics project, containing embedding caches based on various encoders. The data is sourced from FOMC releases by the Federal Reserve (US-government public domain), stored in Parquet format with fields such as record_id, doc_id, event_date, chunk_index, chunk_preview, and embedding. It is designed to support monetary policy and text analysis tasks, serving as pre-computed embedding resources for training and inference.