Uzbek NER XLM-RoBERTa

Модель для распознавания сущностей в узбекских текстах.

Classes

  • ORG
  • NAME
  • GEO

Label scheme

BIO: B-ORG, I-ORG, B-NAME, I-NAME, B-GEO, I-GEO, O.

Base model

FacebookAI/xlm-roberta-large

Input processing

Documents are split into overlapping windows:

  • max length: 512
  • stride: 128

Predictions from overlapping windows are aggregated by character offsets.

Downloads last month
13
Safetensors
Model size
0.6B params
Tensor type
F32
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for dashakoryakovskaya/uzbek-ner

Finetuned
(1,105)
this model