dmis-lab / UltraBERT-Embed-150M

Embeddings151M2.6K50Updated Oct 16, 2026

Downloads

2.6K

Likes

50

Last updated

Oct 16, 2026

License

apache-2.0

UltraBERT-Embed-150M is an English text embedding model built on UltraBERT-150M. It maps text to a 768-dimensional vector, takes inputs up to 32K tokens, follows task instructions on the query side, and embeds code as well as text. On MTEB(eng, v2) it scores 67.24 Mean(Task), …

Tasks & library

  • sentence similarity
  • sentence-transformers

Tags

  • sentence-transformers
  • ultrabert
  • sentence-similarity
  • feature-extraction
  • mteb
  • long-context

Technical specifications

Parameters151M
Librarysentence-transformers
Languagesen
Licenseapache-2.0