---
license: apache-2.0
language:
  - lv
library_name: transformers
pipeline_tag: token-classification
base_model: AiLab-IMCS-UL/lv-deberta-base
tags:
  - nlp
  - latvian
  - latvian-nlp
  - sentence-segmentation
  - sentence-boundary-detection
  - sentence-segmenter
  - text-segmentation
  - token-classification
  - deberta-v2
  - ortix
  - baltrix
---

![Ortix](./ortix-card.webp)

# Ortix Seg LV Small

Latvian sentence segmentation model for detecting sentence boundaries in written Latvian text.

This model is part of the Ortix release by Baltrix. It is intended for document cleanup, corpus processing, retrieval preprocessing, and other Latvian text-processing pipelines where sentence boundaries need to be recovered reliably.

The small variant is a 3-layer DeBERTa-v2 token-classification model adapted from `AiLab-IMCS-UL/lv-deberta-base` and optimized for lighter runtime use.

## Use

```bash
pip install ortix-seg
```

```python
from ortix.seg import SentenceSegmenter

segmenter = SentenceSegmenter.from_pretrained("Baltrix-AI/ortix-seg-lv-small")
spans = segmenter.segment("Pirmais teikums. Otrais teikums.")
```

The `ortix-seg` runtime applies the release-time segmentation configuration and Latvian postprocessing rules.

## Links

- Runtime package: https://github.com/Baltrix-AI/ortix-runtime
- Release page: https://baltrix.ai/releases/ortix
- Blog article: https://baltrix.ai/blog/latvian-sentence-segmentation