Russian sentence encoder for retrieval
According to the model card, USER2 is a Universal Sentence Encoder for Russian built on RuModernBERT and tuned for retrieval and semantic tasks.
Open Source Model Profile · deepvk
USER2-small
USER2-small is a 34.39M-parameter Russian sentence encoder from deepvk. According to the model card, it builds on RuModernBERT with 8,192-token support.
- Publisher
- deepvk
- Task
- sentence-similarity
- Model type
- modernbert
- License
- apache-2.0
- Library
- sentence-transformers
- Publication status
- Approved for indexing
Model overview
USER2-small is published by deepvk as a sentence-similarity embedding model. The captured configuration identifies ModernBertModel and Safetensors metadata reports 34,391,424 parameters. According to the model card, it is a Universal Sentence Encoder for Russian built on RuModernBERT for retrieval and semantic tasks.
Recorded capabilities
8,192 tokens with Matryoshka support
According to the model card, the model supports up to 8,192 tokens and Matryoshka Representation Learning with truncate_dim usage.
34.39M ModernBERT with sentence-transformers
The captured configuration reports ModernBertModel with model type modernbert and 34,391,424 parameters, with sentence-transformers support.
Prefix and RetroMAE training detail
According to the model card, the model expects task prefixes such as classification and uses RetroMAE retrieval-oriented pretraining.
Use cases in the source record
- Russian sentence representation and retrieval workflows with long-context support up to 8,192 tokens.
- Semantic similarity experiments with task prefixes and adjustable Matryoshka embedding dimensions.
Limitations and unknowns
- No evaluation results were extracted from this record.
- Provider state is historical snapshot data, not independently refreshed current availability.
- Training-data and method information comes from the publisher model card and has not been independently verified by Ethen.
Source and provenance
Source: deepvk/USER2-small
Captured: Unknown. Processed: 2026-09-07T19:34:43.629837+00:00.
USER2-small USER2 is a new generation of the U niversal S entence E ncoder for R ussian, designed for sentence representation with long-context support of up to 8,192 tokens. The models are built on top of the RuModernBERT encoders and are fine-tuned for retrieval and semantic tasks. They also support Matryoshka Representation Learning (MRL) — a technique that enables reducing embedding size with minimal loss in representation quality. This is a small model with 34 million parameters. Model Size Context Length Hidden Dim MRL Dims deepvk/USER2-small 34M 8192 384 [32, 64, 128, 256, 384] deepvk/USER2-base 149M 8192 768 [32, 64, 128, 25…
F001F002F003F004F005F006F007F008F010F011F012F013F015F016F018F019F021F023