USER2-base
USER2 is a new generation of the Universal Sentence Encoder for Russian, designed for sentence representation with long-context support of up to 8,192 tokens.
Params
149 M
Context
8,192
Downloads 30d
740 K
Likes
29
Download history
daily snapshots · 14 days740 K411 K
Oct 11Oct 15Oct 20Oct 24
Can you run it?
Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.
| File | Quant | Size | Est. VRAM | Verdict on RTX 4090 · 24 GB |
|---|---|---|---|---|
| model.safetensors | f32 | 0.6 GB | 1.2 GB | ✅ Runs comfortably |
Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.
Specifications
- Architecture
- ModernBertModel
- Parameters
- 149 M
- Tensor type
- F32
- Context length
- 8,192
- Vocabulary
- 50,368
- Layers / heads
- 22 / 12
- Licence
- apache-2.0
- First seen on the Hub
- 2025-02-25
- Training datasets
- deepvk/ru-HNP, deepvk/ru-WANLI, deepvk/cultura_ru_ed, Shitao/bge-m3-data, CarlBrendt/Summ_Dialog_News, IlyaGusev/gazeta, its5Q/habr_qna, wikimedia/wikipedia, RussianNLP/wikiomnia
- Added to our catalog
- 2026-10-11
Compare with any sentence-similarity model