m2v-embeddinggemma-european
A Model2Vec static embedding model distilled from google/embeddinggemma-300m, pruned to European languages only.
What is this?
- Distilled EmbeddingGemma (308M param encoder, based on Gemma 3) into a static token embedding lookup table
- Pruned all non-European script tokens (CJK, Arabic, Hebrew, Thai, Devanagari, Korean, Japanese, etc.)
Stats
| Before pruning | After pruning | |
|---|---|---|
| Vocabulary | 255,732 tokens | 177,926 tokens |
| Model size | ~127 MB | ~87 MB |
| Embedding dim | 256 | 256 |
30.4% of tokens were removed (non-European scripts).
License
Subject to the Gemma Terms of Use.
- Downloads last month
- 5
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for flipbitsnotburgers/m2v-embeddinggemma-european
Base model
google/embeddinggemma-300m