---
license: unlicense
pipeline_tag: sentence-similarity
---
Encoder-model for search query similarity task.

Fast and accurate. 

Sentencepiece tokenizer fitted on 269 million Russian search queries log.

DeBERTaV2 with a short context length to save the memory.

The data set for validation with manual markup consisted of 362 thousand examples.

|![Validation results](https://huggingface.co./fkrasnov2/SBE/blob/main/bvf_recall1k_query_len_eng.svg)|


```python
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained('fkrasnov2/SBE')
tokenizer = AutoTokenizer.from_pretrained('fkrasnov2/SBE')

input_ids = tokenizer.encode("чёрное платье", max_length=model.config.max_position_embeddings, truncation=True, return_tensors='pt')

model.eval()
vector = model(input_ids=input_ids, attention_mask=input_ids>3)[0][0,0]

assert model.config.hidden_size == vector.shape[0]
```