Skip to content

Model files: EmbeddingGemma 2 (v1)

Pre-release
Pre-release

Choose a tag to compare

@TeamBorealis TeamBorealis released this 25 Oct 06:08
· 599 commits to main since this release

Model files for Borealis semantic search (optional; downloaded only after the user turns it on in the app).

File Bytes SHA-256
embeddinggemma-2-Q8_0.gguf 309,855,456 2188ac1deca4b77dffefd603c2776a9d76d9d74ec01841392982ebb840b09135
embeddinggemma-2-Q6_K-te-Q4_K.gguf 211,161,312 14b38ee0dc8556da0d420f8c920ecea9c0588a13ce440448697b7e0467413913
  • embeddinggemma-2-Q8_0.gguf is redistributed unmodified from ggml-org/embeddinggemma-2-GGUF @ 58faf90.
  • embeddinggemma-2-Q6_K-te-Q4_K.gguf was modified by the Borealis project: re-quantised from embeddinggemma-2-BF16.gguf with llama.cpp llama-quantize (commit 2639a3b, b11642): llama-quantize --token-embedding-type q4_k embeddinggemma-2-BF16.gguf embeddinggemma-2-Q6_K-te-Q4_K.gguf Q6_K 3.

EmbeddingGemma 2 weights: Copyright Google LLC / Google DeepMind, Apache License 2.0 (see embeddinggemma-2-LICENSE.txt and embeddinggemma-2-NOTICE.txt). "Gemma" is a trademark of Google LLC; Borealis is not affiliated with or endorsed by Google.

This is not an app release.