lv-mbert-large-conloan-lv-ext

Description

This model is part of a Bachelor's thesis at the University of Latvia: "Contextual approach to Latvian loanword detection: dataset creation and classification experiments".

It is a fine-tuned version of lv-mbert-large-conloan-lv-ext on the extended dataset.

Classes and Labels

Dataset Type: {Baseline (Binary) / Extended (Contrastive)} Labels:

• O: Outside
• LOAN: Borrowing (Materiālie aizguvumi)
• CS: Code-switching (Koda maiņa)
• NE: Named Entities (Nosauktās entitātes)

Performance (k-fold average)

• F1 Score: {0.897}
• Std Dev (σ): {0.006}

Usage

from transformers import pipeline
nlp = pipeline("ner", model="jorenchik/lv-mbert-large-conloan-lv-ext")
nlp("Šodienas mītings bija ļoti produktīvs.")