dejanseo/

200k-wiki-data-embeddings-gemma

Text embeddings for ~200,000 Wikidata entities, generated from each entity's English label with Google's open google/embeddinggemma-300m model. Vectors are 768-dimensional and unit length (the model normalizes its output), so cosine similarity equals a plain dot product.

	What's inside




	

Column Type Description

entity_id string Wikidata Q-identifier, e.g. Q42

label string English label, e.g. Douglas Adams… See the full description on the dataset page: https://huggingface.co/datasets/dejanseo/200k-wiki-data-embeddings-gemma.