---
library_name: BERT
task: sentiment-analysis
tags:
- KinyaBert
- sentiment analysis
language: rw
datasets:
- https://github.com/afrisenti-semeval/afrisent-semeval-2023/tree/main/SubtaskA
---
# Kinyarwanda twitter sentiment analysis
A sentiment analysis model in Kinyarwanda trained using tweets and trained using a BERT model trained on Kinyarwanda dataset

# Dataset information
The dataset was trained on 3000 annotated tweets, here is the [link to the dataset](https://github.com/afrisenti-semeval/afrisent-semeval-2023/blob/main/SubtaskA/train/kr_train.tsv)

# Model limitations
The model was trained on a small dataset of tweets, which can present limitations in accuracy on other datasets

# Result
Accuracy: 85%