MIRRA Shards Dataset – Semantic Drift Signatures
收藏官方服务：
资源简介：
This dataset represents modular spectral signatures of semantic drift in language models. Each shard includes a PCA-based frequency profile, drift intensity score, dominant spectra, and recommended correction strategies. The data was generated using the MIRRA-VORTEX engine and exported in a HuggingFace-compatible format. MIRRA (Modular Interpretative Resonance Response Architecture) is a research framework for detecting and correcting drift in semantic AI environments. Primary access is via HuggingFace: https://huggingface.co/datasets/<your-username>/mirra_shards
提供机构：
Zenodo创建时间：
2025-04-30
