electroglyph/jina5_small_distill

distilled embeddings from jinaai/jina-embeddings-v5-text-small this is using data from UltraX-FineWeb-ProX-Doc from openbmb/UltraX-Preview dataset each row is approximately 256-512 tokens, chunks always begin and end on a sentence boundary.